content
stringlengths 7
2.61M
|
|---|
Transcriptome-proteome integration of archival human renal cell carcinoma biopsies enables identification of molecular mechanisms. Renal cell cancer is among the most common forms of cancer in humans, with around 35,000 deaths attributed to kidney carcinoma in the European Union in 2012 alone. Clear cell renal cell carcinoma (ccRCC) represents the most common form of kidney cancer and the most lethal of all genitourinary cancers. Here, we apply omics technologies to archival core biopsies to investigate the biology underlying ccRCC. Knowledge of these underlying processes should be useful for the discovery and/or confirmation of novel therapeutic approaches and ccRCC biomarker development. From partial or full nephrectomies of 11 patients, paired core biopsies of ccRCC-affected tissue and adjacent ("peritumorous") nontumor tissue were both sampled and subjected to proteomics analyses. We combined proteomics results with our published mRNA sequencing data from the same patients and with published miRNA sequencing data from an overlapping patient cohort from our institution. Statistical analysis and pathway analysis were performed with JMP Genomics and Ingenuity Pathway Analysis (IPA), respectively. Proteomics analysis confirmed the involvement of metabolism and oxidative stress-related pathways in ccRCC, whereas the most affected pathways in the mRNA sequencing data were related to the immune system. Unlike proteomics or mRNA sequencing alone, a combinatorial cross-omics pathway analysis approach captured a broad spectrum of biological processes underlying ccRCC, such as mitochondrial damage, repression of apoptosis, and immune system pathways. Sirtuins, immunoproteasome genes, and CD74 are proposed as potential targets for the treatment of ccRCC. |
<reponame>randiapr/my-hci
package com.apraware.hci.web;
import com.apraware.hci.domain.ModuleApp;
import com.apraware.hci.domain.UserApp;
import com.apraware.hci.domain.UserModules;
import com.apraware.hci.request.UserModulesReq;
import com.apraware.hci.response.UserModuleResponse;
import com.apraware.hci.service.RequestValidationService;
import com.apraware.hci.service.UserModulesService;
import io.swagger.annotations.Api;
import io.swagger.annotations.ApiOperation;
import io.swagger.annotations.ApiParam;
import javax.validation.Valid;
import org.springframework.beans.factory.annotation.Autowired;
import org.springframework.http.HttpStatus;
import org.springframework.http.ResponseEntity;
import org.springframework.validation.BindingResult;
import org.springframework.web.bind.annotation.DeleteMapping;
import org.springframework.web.bind.annotation.GetMapping;
import org.springframework.web.bind.annotation.PathVariable;
import org.springframework.web.bind.annotation.PostMapping;
import org.springframework.web.bind.annotation.PutMapping;
import org.springframework.web.bind.annotation.RequestBody;
import org.springframework.web.bind.annotation.RequestMapping;
import org.springframework.web.bind.annotation.RestController;
/**
*
* @author randi
*/
@RestController
@RequestMapping("/api/v1/userModules")
@Api(value = "User Modules App", tags = "User Modules App")
public class UserModulesController {
@Autowired
private UserModulesService service;
@Autowired
private RequestValidationService reqService;
@ApiOperation(value = "Add new User Module App", response = UserModules.class)
@PostMapping
ResponseEntity<?> create(@RequestBody UserModulesReq userModuleReq, BindingResult result) {
var errorReq = reqService.ValidationService(result);
if (errorReq != null) {
return errorReq;
}
return new ResponseEntity<>(service.create(userModuleReq), HttpStatus.CREATED);
}
@ApiOperation(value = "Update User Module App", response = UserModules.class)
@PutMapping("{id}")
ResponseEntity<?> update(
@ApiParam(value = "User Module App ID", required = true)
@PathVariable(value = "id") Long id, @Valid @RequestBody UserModulesReq req) {
var moduleApp = new ModuleApp();
moduleApp.setId(req.getModuleId());
var userApp = new UserApp();
userApp.setId(req.getUserId());
var userModules = new UserModules();
userModules.setId(id);
userModules.setModuleApp(moduleApp);
userModules.setUserApp(userApp);
return ResponseEntity.ok(service.update(userModules));
}
@ApiOperation(value = "Get All User Module App", response = UserModules.class, responseContainer = "List")
@GetMapping
ResponseEntity<?> getAll() {
return ResponseEntity.ok(service.getAll());
}
@ApiOperation(value = "Get User Module App by User ID", response = UserModuleResponse.class, responseContainer = "List")
@GetMapping("{id}")
ResponseEntity<?> getByUserId(
@ApiParam(value = "User Module App ID", required = true)
@PathVariable(value = "id") Long id) {
return new ResponseEntity<>(service.findByUserId(id), HttpStatus.OK);
}
@ApiOperation(value = "Delete Module App by Module Id")
@DeleteMapping
ResponseEntity<?> delete(
@ApiParam(value = "User Module App ID", required = true)
@PathVariable(value = "id") Long id) {
service.delete(id);
return ResponseEntity.ok("Succeed");
}
}
|
<reponame>usarskyy/ngx-input-number
import { Directive, ElementRef, HostListener, Input } from '@angular/core';
import { InputNumberService } from './input-number.service';
@Directive({ selector: '[inputNumber]' })
export class InputNumberDirective {
@Input() acceptDecimalPlaces = false;
@Input() decimalPlaces = 2;
private element: HTMLInputElement;
constructor(
private inputNumberService: InputNumberService,
private elementRef: ElementRef<HTMLInputElement>) {
this.element = elementRef.nativeElement;
}
@HostListener('keydown', ['$event', '$event.target']) onInput(event: KeyboardEvent, target: HTMLInputElement) {
const key: number = this.inputNumberService.getKeyCode(event);
if ([46, 8, 9, 27, 13].indexOf(key) !== -1 // Allow: Delete, Backspace, Tab, Escape, Enter
|| (key === 65 && event.ctrlKey === true) // Allow: Ctrl+A
|| (key === 67 && event.ctrlKey === true) // Allow: Ctrl+C
|| (key === 86 && event.ctrlKey === true) // Allow: Ctrl+V
|| (key === 88 && event.ctrlKey === true) // Allow: Ctrl+X
|| (key === 90 && event.ctrlKey === true) // Allow: Ctrl+Z
|| (key === 65 && event.metaKey === true) // Cmd+A (Mac)
|| (key === 67 && event.metaKey === true) // Cmd+C (Mac)
|| (key === 86 && event.metaKey === true) // Cmd+V (Mac)
|| (key === 88 && event.metaKey === true) // Cmd+X (Mac)
|| (key === 90 && event.metaKey === true) // Cmd+Z (Mac)
|| (key >= 35 && key <= 39) // Home, End, Left, Right
) {
return;
}
if (this.acceptDecimalPlaces) {
const indexOfDot = this.element.value.indexOf('.');
if (this.inputNumberService.isDecimalIndicator(event)) {
// Allow just one dot
if (indexOfDot < 0) {
// Dont allow dots if will trasnform in a invalid value
if (this.element.value.substr(this.element.selectionStart).length <= this.decimalPlaces) {
return;
} else {
event.preventDefault();
}
}
// Is not a dot/decimal indicator
} else if (indexOfDot > -1) {
// If is trying to insert the value before the dot
if (this.element.selectionStart <= indexOfDot) {
return;
} else {
// If is inserting the value after the dot
const valueAfterDot: string = target.value.substring(indexOfDot + 1);
const quantityCharSelected: number = this.element.selectionEnd - this.element.selectionStart;
const hasSelection: boolean = quantityCharSelected > 0;
if (!hasSelection) {
// Check if already has the maximum of decimal places
if (valueAfterDot.length >= this.decimalPlaces) {
event.preventDefault();
}
}
}
}
}
// Ensure that it is a number and stop the keypress
if ((event.shiftKey || (key < 48 || key > 57))
&& (key < 96 || key > 105)
) {
event.preventDefault();
}
}
@HostListener('paste', ['$event', '$event.target']) onPaste(event: ClipboardEvent, target: HTMLInputElement) {
event.preventDefault();
event.stopPropagation();
const pastedEntry: string = event.clipboardData.getData('text');
const cleanedValue = this.cleanEntry(target, pastedEntry);
this.element.focus();
const inserted = document.execCommand('insertText', false, cleanedValue);
// If something goes wrong on insert text, like firefox: https://bugzilla.mozilla.org/show_bug.cgi?id=1220696
if (!inserted) {
// TODO: Fix another issue with firefox to put the text in the selection start
if (this.element.value !== cleanedValue) {
this.element.value = cleanedValue;
this.element.dispatchEvent(new Event('input'));
this.element.addEventListener('focusout', () => {
this.element.dispatchEvent(new Event('change'));
}, { once: true });
}
}
}
@HostListener('drop', ['$event', '$event.target']) onDrop(event: any, target: HTMLInputElement) {
event.preventDefault();
event.stopPropagation();
const dropedEntry = event.dataTransfer.getData('text');
const cleanedValue = this.cleanEntry(target, dropedEntry);
this.element.focus();
const inserted = document.execCommand('insertText', false, cleanedValue);
// If something goes wrong on insert text, like firefox: https://bugzilla.mozilla.org/show_bug.cgi?id=1220696
if (!inserted) {
if (this.element.value !== cleanedValue) {
this.element.value = cleanedValue;
this.element.dispatchEvent(new Event('input'));
this.element.addEventListener('focusout', () => {
this.element.dispatchEvent(new Event('change'));
}, { once: true });
}
}
}
private cleanEntry(target: HTMLInputElement, pastedEntry: string) {
const indexOfDot = target.value.indexOf('.');
const hasDecimalAlready = indexOfDot > -1;
const quantityCharSelected: number = target.selectionEnd - target.selectionStart;
const isOverwritingEntireValue: boolean = quantityCharSelected === target.value.length;
const keepDecimals = (isOverwritingEntireValue || !hasDecimalAlready) && this.acceptDecimalPlaces;
const isAddingDecimalsNumbers = hasDecimalAlready && (target.selectionStart > indexOfDot);
const hasSelection: boolean = quantityCharSelected > 0;
// Cleaning the data before insert in the field
let cleanedValue = this.inputNumberService
.removeNonNumbers(pastedEntry, keepDecimals, this.decimalPlaces);
if (this.acceptDecimalPlaces && isAddingDecimalsNumbers) {
const quantityOfNewDecimalsAllowed = hasSelection
? quantityCharSelected
: this.decimalPlaces - target.value.substring(indexOfDot + 1).length;
cleanedValue = cleanedValue.substring(0, quantityOfNewDecimalsAllowed);
}
return cleanedValue;
}
}
|
<filename>http/controllers/log_controller.go
package controllers
import (
"github.com/gin-gonic/gin"
"github.com/zhsyourai/URCF-engine/http/controllers/shard"
"github.com/zhsyourai/URCF-engine/http/gin-jwt"
"github.com/zhsyourai/URCF-engine/services/log"
"net/http"
"strconv"
)
func NewLogController(middleware *gin_jwt.JwtMiddleware) *LogController {
return &LogController{
service: log.GetInstance(),
middleware: middleware,
}
}
// LogController is our /log controller.
type LogController struct {
service log.Service
middleware *gin_jwt.JwtMiddleware
}
func (c *LogController) Handler(root *gin.RouterGroup) {
root.GET("/list", c.ListLogHandler)
root.DELETE("/*id", c.CleanLogHandler)
}
func (c *LogController) ListLogHandler(ctx *gin.Context) {
var paging shard.Paging
if ctx.BindQuery(&paging) != nil {
return
}
total, logs, err := c.service.ListAll(paging.Page, paging.Size, paging.Sort, paging.Order)
if err != nil {
ctx.AbortWithError(http.StatusInternalServerError, err)
return
}
ctx.JSON(http.StatusOK, &shard.LogsWithCount{
TotalCount: total,
Items: logs,
})
}
func (c *LogController) CleanLogHandler(ctx *gin.Context) {
idStr := ctx.Param("id")
id, err := strconv.ParseInt(idStr, 10, 0)
if err != nil {
ctx.AbortWithError(http.StatusInternalServerError, err)
return
}
c.service.Clean(id)
ctx.Status(http.StatusOK)
}
|
Sinan: Data-Driven, QoS-Aware Cluster Management for Microservices Cloud applications are increasingly shifting from large monolithic services, to large numbers of loosely-coupled, specialized microservices. Despite their advantages in terms of facilitating development, deployment, modularity, and isolation, microservices complicate resource management, as dependencies between them introduce backpressure effects and cascading QoS violations. We present Sinan, a data-driven cluster manager for interactive cloud microservices that is online and QoS-aware. Sinan leverages a set of scalable and validated machine learning models to determine the performance impact of dependencies between microservices, and allocate appropriate resources per tier in a way that preserves the end-to-end tail latency target. We evaluate Sinan both on dedicated local clusters and large-scale deployments on Google Compute Engine (GCE) across representative end-to-end applications built with microservices, such as social networks and hotel reservation sites. We show that Sinan always meets QoS, while also maintaining cluster utilization high, in contrast to prior work which leads to unpredictable performance or sacrifices resource efficiency. Furthermore, the techniques in Sinan are explainable, meaning that cloud operators can yield insights from the ML models on how to better deploy and design their applications to reduce unpredictable performance. Despite several advantages, such as modular and flexible development and rapid iteration, microservices also introduce new system challenges, especially in resource management, since the complex topologies of microservice dependencies exacerbate queueing effects, and introduce cascading Quality of Service (QoS) violations that are difficult to identify and correct in a timely manner. Current cluster managers are designed for monolithic applications or applications consisting of a few pipelined tiers, and are not expressive enough to capture the complexity of microservices. Given that an increasing number of production cloud services, such as EBay, Netflix, Twitter, and Amazon, are now designed as microservices, addressing their resource management challenges is a pressing need. We take a data-driven approach to tackle the complexity microservices introduce to resource management. Similar machine learning (ML)-driven approaches have been effective at solving resource management problems for large-scale systems in previous work. Unfortunately, these systems are not directly applicable to microservices, as they were designed for monolithic services, and hence do not account for the impact of dependencies between microservices on end-to-end performance. We present Sinan, a scalable and QoS-aware resource manager for interactive cloud microservices. Instead of tasking the user or cloud operator with inferring the impact of dependencies between microservices, Sinan leverages a set of validated ML models to automatically determine the impact of per-tier resource allocations on end-to-end performance, and assign appropriate resources to each tier. Sinan first uses an efficient space exploration algorithm to examine the space of possible resource allocations, especially focusing on corner cases that introduce QoS violations. This yields a training dataset used to train two models: a Convolutional Neural Network (CNN) model for detailed short-term performance prediction, and a Boosted Trees model that evaluates the long-term performance evolution. The combination of the two models allows Sinan to both examine the near-future outcome of a resource allocation, and to account for the system's inertia in building up queues with higher accuracy than a single model examining both time windows. Sinan operates online, adjusting per-tier resources dynamically according to the service's runtime status and end-to-end QoS target. Finally, Sinan is implemented as a centralized resource manager with global visibility into the cluster and application state, and with per-node resource agents that track per-tier performance and resource utilization. We evaluate Sinan using two end-to-end applications from DeathStarBench, built with interactive microservices: a social network and a hotel reservation site. We compare Sinan against both traditionally-employed empirical approaches, such as autoscaling, and previous research on multi-tier service scheduling based on queueing analysis, such as Power-Chief. We demonstrate that Sinan outperforms previous work both in terms of performance and resource efficiency, successfully meeting QoS for both applications under diverse load patterns. On the simpler hotel reservation application, Sinan saves 25.9% on average, and up to 46.0% of the amount of resources used by other QoS-meeting methods. On the more complex social network service, where abstracting application complexity is more essential, Sinan saves 59.0% of resources on average, and up to 68.1%, essentially accommodating twice the amount of requests per second, without the need for more resources. We also validate Sinan's scalability through largescale experiments on approximately 100 container instances on Google Compute Engine (GCE), and demonstrate that the models deployed on the local cluster can be reused on GCE with only minor adjustments instead of retraining. Finally, we demonstrate the explainability benefits of Sinan's models, delving into the insights they can provide for the design of large-scale systems. Specifically, we use an example of Redis's log synchronization, which Sinan helped identify as the source of unpredictable performance out of tens of dependent microservices to show that the system can offer practical and insightful solutions for clusters whose scale make previous empirical approaches impractical. Problem Statement Sinan aims to manage resources for complex, interactive microservices with tail latency QoS constraints in a scalable and resource-efficient manner. Graphs of dependent microservices typically include tens to hundreds of tiers, each with different resource requirements, scaled out and replicated for performance and reliability. Section 2.2 describes some motivating examples of such services with diverse functionality used in this work; other similar examples can be found in. Most cluster managers focus on CPU and memory management. Microservices are by design mostly stateless, hence their performance is defined by their CPU allocation. Given this, Sinan primarily focuses on allocating CPU resources to each tier, both at sub-core and multicore granularity, leveraging Linux cgroups through the Docker API. We also provision each tier with the maximum profiled memory usage to eliminate out of memory errors. Motivating Applications We use two end-to-end interactive applications from DeathStar-Bench : a hotel reservation service, and a social network. Hotel Reservation The service is an online hotel reservation site, whose architecture is shown in Fig. 1. Functionality: The service supports searching for hotels using geolocation, placing reservations, and getting recommendations. It is implemented in Go, and tiers communicate over gRPC. Data backends are implemented in memcached for in-memory caching, and MongoDB, for persistent storage. The database is populated with 80 hotels and 500 active users. Social Network The end-to-end service implements a broadcast-style social network with uni-directional follow relationships, shown in Fig. 2. Inter-microservice messages use Apache Thrift RPCs. Functionality: Users can create posts embedded with text, media, links, and tags to other users, which are then broadcasted to all their followers. The texts and images uploaded by users, specifically, go through image-filter (a CNN classifier) and text-filter services (an SVM classifier), and contents violating the service's ethics guidelines are rejected. Users can also read posts on their timelines. We use the Reed98 social friendship network to populate the user database. User activity follows the behavior of Twitter users reported in, and the distribution of post text length emulates Twitter's text length distribution.. Client requests first reach Nginx, which works as frontend http servers. Then, depending on the type of user request, a number of logic, mid-tiers will be invoked to create a post, read a user's timeline and to follow/unfollow users. At the rightmost of the figure, the requests reach the back-end databases, implemented both with in-memory caching tiers (memcached and Redis), and persistent databases (MongoDB). Management Challenges & the Need for ML Resource management in microservices faces four challenges. Dependencies among tiers Resource management in microservices is additionally complicated by the fact that dependent microservices are not perfect pipelines, and hence can introduce backpressure effects that are hard to detect and prevent. These dependencies can be further exacerbated by the specific RPC and data store API implementation. Therefore, the resource scheduler should have a global view of the microservice graph and be able to anticipate the impact of dependencies on end-to-end performance. 2. System complexity Given that application behaviors change frequently, resource management decisions need to happen online. This means that the resource manager must traverse a space that includes all possible resource allocations per microservice in a practical manner. Prior empirical approaches use resource utilization, or latency measurements to drive allocation decisions. Queueing approaches similarly characterize the system state using queue lengths. Unfortunately these approaches cannot be directly employed in complex microservices with tens of dependent tiers. First, microservice dependencies mean that resource usage across tiers is codependent, so examining fluctuations in individual tiers can attribute poor performance to the wrong tier. Similarly, although queue lengths are accurate indicators of a microservice's system state, obtaining exact queue lengths is hard. First, queues exist across the system stack from the NIC and OS, to the network stack and application. Accurately tracking queue lengths requires application changes and heavy instrumentation, which can negatively impact performance and/or is not possible in public clouds. Second, the application may include third-party software whose source code cannot be instrumented. Alternatively, expecting the user to express each tier's resource sensitivity is problematic, as users already face difficulties correctly reserving resources for simple, monolithic workloads, leading to well-documented underutilization, and the impact of microservice dependencies is especially hard to assess, even for expert developers. 3. Delayed queueing effect Consider a queueing system with processing throughput T o under a latency QoS target, like the one in Fig. 3. T o is a non-decreasing function of the amount of allocated resources R. For input load T i, T o should equal or slightly surpass T i for the system to stably meet QoS, while using the minimum amount of resources R needed. Even when R is reduced, such that T o < T i, QoS will not be immediately violated, since queue accumulation takes time. The converse is also true; by the time QoS is violated, the built-up queue takes a long time to drain, even if resources are upscaled immediately upon detecting the violation (red line). Multi-tier microservices are complex queueing systems with queues both across and within microservices. This delayed queueing effect highlights the need for ML to evaluate the long-term impact of resource allocations, and to proactively prevent the resource manager from reducing resources too aggressively, to avoid latency spikes with long recovery periods. To mitigate a QoS violation, the manager must increase resources proactively (blue line), otherwise the violation becomes unavoidable, even if more resources are allocated a posteriori. 4. Boundaries of resource allocation space Data collection or profiling are essential to the performance of any model. Given the large resource allocation space in microservices, it is essential for any resource manager to quickly identify the boundaries of that space that allow the service to meet its QoS, with the minimum resource amount, so that neither performance nor resource efficiency are sacrificed. Prior work often uses random exploration of the resource space or uses prior system state as the training dataset. Unfortunately, while these approaches work for simpler applications, in microservices they are prone to covariant shift. Random collection blindly explores the entire space, even though many of the explored points may never occur during the system's normal operation, and may not contain any points close to the resource boundary of the service. On the contrary, data from operation logs are biased towards regions that occur frequently in practice but similarly may not include points close to the boundary, as cloud systems often overprovision resources to ensure that QoS is met. To reduce exploration overheads it is essential for a cluster manager to efficiently examine the necessary and sufficient number of points in the resource space that allow it to just meet QoS with the minimum resources. Proposed Approach These challenges suggest that empirical resource management, such as autoscaling or queueing analysis-based approaches for multi-stage applications, such as PowerChief, are prone to unpredictable performance and/or resource inefficiencies. To tackle these challenges, we take a data-driven approach that abstracts away the complexity of microservices from the user, and leverages ML to identify the impact of dependencies on end-to-end performance, and make allocation decisions. We also design an efficient space exploration algorithm that explores the resource allocation space, especially boundary regions that may introduce QoS violations, for different application scenarios. Specifically, Sinan's ML models predict the end-to-end latency and the probability of a QoS violation for a resource configuration, given the system's state and history. The system uses these predictions to maximize resource efficiency, while meeting QoS. At a high level, the workflow of Sinan is as follows: the data collection agent collects training data, using a carefullydesigned algorithm which addresses Challenge 4 (efficiently exploring the resource space). With the collected data, Sinan trains two ML models: a convolution neural network (CNN) model and a boosted trees (BT) model. The CNN handles Challenges 1 and 2 (dependencies between tiers and navigating the system complexity), by predicting the end-to-end tail latency in the near future. The BT model addresses Challenge 3 (delayed queueing effect), by evaluating the probability for a QoS violation further into the future, to account for the system's inertia in building up queues. At runtime, Sinan infers the instantaneous tail latency and the probability for an upcoming QoS violation, and adjusts resources accordingly to satisfy the QoS constraint. If the application or underlying system change at any point in time, Sinan retrains the corresponding models to account for the impact of these changes on end-to-end performance. Machine Learning Models The objective of Sinan's ML models is to accurately predict the performance of the application given a certain resource allocation. The scheduler can then query the model with possible resource allocations for each microservice, and select the one that meets QoS with the least necessary resources. A straightforward way to achieve this is designing an ML model that predicts the immediate end-to-end tail latency as a function of resource allocations and utilization, since QoS is defined in terms of latency, and comparing the predicted latency to the measured latency during deployment is straightforward. The caveat of this approach is the delayed queueing effect described in Sec. 2.3, whereby the impact of an allocation decision would only show up in performance later. As a resolution, we experimented with training a neural network (NN) to predict latency distributions over a future time window: for example, the latency for each second over the next five seconds. However, we found that the prediction accuracy rapidly decreased the further into the future the NN tried to predict, as predictions were based only on the collected current and past metrics (resource utilization and latency), which were accurate enough for immediate-future predictions, but were insufficient to capture how dependencies between microservices would cause performance to evolve later on. Considering the difficulty of predicting latency further into the future, we set an alternative goal: predict the latency of the immediate future, such that imminent QoS violations are identified quickly, but only predict the probability of experiencing a QoS violation later on, instead of the exact latency of each decision interval. This binary classification is a much more contained problem than detailed latency prediction, and still conveys enough information to the resource manager on performance events, e.g., QoS violations, that may require immediate action in the present. An intuitive method for this are multi-task learning NNs that predict the latency of the next interval, and the QoS violation probability in the next few intervals. However, the multi-task NN considerably overpredicts tail latency and QoS violation probability, as shown in Fig. 4. Note that the gap between prediction and ground truth does not indicate a constant difference, which could be easily learned by NNs with strong overfitting capabilities. We attribute the overestimation to interference caused by the semantic gap between the QoS violation probability, a value between 0 and 1, and the latency, a value that is not strictly bounded. To address this, we designed a two-stage model: first, a CNN that predicts the end-to-end latency of the next timestep with high accuracy, and, second, a Boosted Trees (BT) model that estimates the probability for QoS violations further into the future, using the latent variable extracted by CNN. BT is generally less prone to overfitting than CNNs, since it has much fewer tunable hyperparameters than NNs; mainly the number of trees and tree depth. By using two separate models, Sinan is able to optimize each model for the respective objective, and avoid the overprediction issue of using a joint, expensive model for both tasks. We refer to the CNN model as the short-term latency predictor, and the BT model as the long-term violation predictor. Latency Predictor As discussed in Section 2.3, the CNN needs to account for both the dependencies across microservices, and the timeseries pattern of resource usage and application performance. Thus, both the application topology and the timeseries information are encoded in the input of the CNN. The input of the CNN includes the following three parts: 1. an "image" (3D tensor) consisting of per-tier resource utilization within a past time window. The y-axis of the "image" corresponds to different microservices, with consecutive tiers in adjacent rows, the x-axis corresponds to the timeseries, with one timestep per column, and the zaxis (channels) corresponds to resource metrics of different tiers, including CPU usage, memory usage (resident set size and cache memory size) and network usage (number of received and sent packets), which are all retrieved from Docker's cgroup interface. Per-request tracing is not required. 2. a matrix of the end-to-end latency distribution within the past time window, and 3. the examined resource configuration for the next timestep, which is also encoded as a matrix. In each convolutional (Conv) layer of the CNN, a convolutional kernel (k k window) processes information of k adjacent tiers within a time window containing k timestamps. The first few Conv layers in the CNN can thus infer the dependencies of their adjacent tiers over a short time window, and later layers observe the entire graph, and learn interactions across all tiers within the entire time window of interest. The latent representations derived by the convolution layers are then post-processed together with the latency and resource configuration information, through concatenation and fullyconnected (FC) layers to derive the latency predictions. In the remainder of this section, we first discuss the details of the network architecture, and then introduce a custom loss function that improves the prediction accuracy by focusing on the most important latency range. As shown in Fig. 5, the latency predictor takes as input the resource usage history (X RH ), the latency history (X LH ), and the resource allocation under consideration for the next timestep (X RC ), and predicts the end-to-end tail latencies (y L ) (95 th to 99 th percentiles) of the next timestep. X RH is a 3D tensor whose x-axis is the N tiers in the microservices graph, the y-axis is T timestamps (T > 1 accounts for the non-Markovian nature of microservice graph), and channels are F resource usage information related to CPU and memory. The set of necessary and sufficient resource metrics is narrowed down via feature selection. X RC and X LH are 2D matrices. For X RC, the x-axis is the N tiers and the y-axis the CPU limit. For X RH, the x-axis is T timestamps, and the y-axis are vectors of different latency percentiles (95 th to 99 th ). The three inputs are individually processed with Conv and FC layers, and then concatenated to form the latent representation L f, from which the predicted tail latencies L f are derived with another FC layer. The CNN minimizes the difference between predicted and actual latency, using the squared loss function below: where f W () represents the forward function of the CNN, is the ground truth, and n is the number of training samples. Given the spiking behavior of interactive microservices that leads to very high latency, the squared loss in Eq. 1 tends to overfit for training samples with large end-to-end latency, leading to latency overestimation in deployment. Since the latency predictor aims to find the best resource allocation within a tail latency QoS target, the loss should be biased towards training samples whose end-to-end latencies are ≤ QoS. Therefore, we use a scaling function to scale both the predicted and actual end-to-end latency before applying the squared loss function. The scaling function ( ()) is: where the latency range is (0,t), and the hyper-parameter can be tuned for different decay effects. Fig. 6 shows the scaling function with t = 100 and = 0.005, 0.01, 0.02. It is worth mentioning that scaling end-to-end latencies only mitigates overfitting of the predicted latency for the next decision interval, and does not improve predictions further into the future, as described above. We implement all CNN models using MxNet, and trained them with Stochastic Gradient Descent (SGD). Violation Predictor The violation predictor addresses the binary classification task of predicting whether a given allocation will cause a QoS violation further in the future, to filter out undesirable actions. Ensemble methods are good candidates as they are less prone to overfitting. We use Boosted Trees, which realizes an accurate non-linear model by combining a series of simple regression trees. It models the target as the sum of trees, each of which maps features to a score. The final prediction is determined by accumulating scores across all trees. To further reduce the computational cost and memory footprint of Boosted Trees, we reuse the compact latent variable L f extracted from the CNN as its input. Moreover, since the latent variable L f is significantly smaller than X RC, X RH, and X LH in dimensionality, using L f as the input also makes the model more resistant to overfitting. Boosted Trees also takes resource allocations as input. During inference, we simply use the same resource configuration for the next k timesteps to predict whether it will cause a QoS violation k steps in the future. As shown in Fig. 5, each tree leaf represents either a violation or a non-violation with a continuous score. For a given example, we sum the scores for all chosen violation (s V ) and non-violation leaves (s V ) from each tree. The output of BT is the predicted probability of QoS violation (p V ), which can be calculated as p V = e s V e s V +e s NV. For the violation predictor we leverage XGBoost, a gradient tree boosting framework that improves scalability using sparsity-aware approximate split finding. We first train the CNN and then BT using the extracted latent variable from the CNN. The CNN parameters (number of layers, channels per layer, weight decay etc.) and XGBoost (max tree depth) are selected based on the validation accuracy. System Design We first introduce Sinan's overall architecture, and then discuss the data collection process, which is crucial to the effectiveness of the ML models, and Sinan's online scheduler. System Architecture Sinan consists of three components: a centralized scheduler, distributed operators deployed on each server/VM, and a prediction service that hosts the ML models. Figure 7 shows an overview of Sinan's architecture. Sinan makes decisions periodically. In each 1s decision interval (consistent with the granularity at which QoS is defined), the centralized scheduler queries the distributed operators to obtain the CPU, memory, and network utilization of each tier in the previous interval. Resource usage is obtained from Docker's monitoring infrastructure, and only involves a few file reads, incurring negligible overheads. Aside from pertier information, the scheduler also queries the API gateway to get user load statistics from the workload generator. The scheduler sends this data to the hybrid ML model, which is responsible for evaluating the impact of different resource allocations. Resource usage across replicas of the same tier are averaged before being used as inputs to the models. Based on the model's output, Sinan chooses an allocation vector that meets QoS using the least necessary resources, and communicates its decision to the per-node agents for enforcement. Sinan focuses on compute resources, which are most impactful to microservice performance. Sinan explores sub-core allocations in addition to allocating multiple cores per microservice to avoid resource inefficiencies for non-resource demanding tiers, and enable denser colocation. Resource Allocation Space Exploration Representative training data is key to the accuracy of any ML model. Ideally, test data encountered during online deployment should follow the same distribution as the training dataset, so that covariate shift is avoided. Specifically for our problem, the training dataset needs to cover a sufficient spectrum of application behaviors that are likely to occur during online deployment. Because Sinan tries to meet QoS without sacrificing resource efficiency, it must efficiently explore the boundary of the resource allocation space, where points using the minimum amount of resources under QoS reside. We design the data collection algorithm as a multi-armed bandit process, where each tier is an independent arm, with the goal of maximizing the knowledge of the relationship between resources and end-to-end QoS. The data collection algorithm approximates the running state of the application with a tuple (rps, lat cur, lat diff ), where rps is the input requests per second, lat cur is the current tail latency, and lat diff is the tail latency difference from the previous interval, to capture the rate of consuming or accumulating queues. Every tier is considered as an arm that can be played independently, by adjusting its allocated resources. For each tier, we approximate the mapping between its resources and the end-to-end QoS as a Bernoulli distribution, with probability p of meeting the end-to-end QoS, and we define our information gain from assigning certain amount of resources to a tier, as the expected reduction of confidence interval of p for the corresponding Bernoulli distribution. At each step for every tier, we select the operation that maximizes the information gain, as shown in Eq. 3, where op s T is an action selected for tier T at running state s, n are the samples collected for the resulting resource assignment after applying op on tier T at state s, p is the previously-estimated probability of meeting QoS, and p + and p − are the newly-estimated probabilities of meeting QoS, when the new sample meets or violates QoS respectively. Each operation's score is multiplied by a predefined coefficient C op to encourage meeting QoS and reducing overprovisioning. By choosing operations that maximize Equation. 3, the data collection algorithm is incentivized to explore the boundary points that meet QoS with the minimum resource amount, since exploring allocations that definitely meet or violate QoS (with p = 1 or p = 0) has at most 0 information gain. Instead, the algorithm prioritizes exploring resource allocations whose impact on QoS is nondeterministic, like those with p = 0.5. It is also worth noting that the state encoding and information gain definition are simplified approximations of the actual system, with the sole purpose of containing the exploration process in the region of interest. Eventually, we rely on ML to extract the state representation that incorporates inter-tier dependencies in the microservice graph. To prune the action space, Sinan enforces a few rules on both data collection and online scheduling. First, the scheduler is only allowed to select out of a predefined set of operations. Specifically in our setting, the operations include reducing or increasing the CPU allocation by 0.2 up to 1.0 CPU, and increasing or reducing the total CPU allocation of a service by 10% or 30%. These ratios are selected according to the AWS step scaling tutorial ; as long as the granularity of CPU allocations does not change, other resource ratios also work without retraining the model. Second, an upper limit on CPU utilization is enforced on each tier, to avoid overly aggressive resource downsizing that can lead to long queues and dropped requests. Third, when end-to-end tail latency exceeds the expected value, Sinan disables resource reclamations so that the system can recover as soon as possible. A subtle difference from online deployment is that the data collection algorithm explores resource allocations in the tail latency region, where is a small value compared to QoS. The extra allows the data collection process to explore allocations that cause slight QoS violations without the pressure of reverting to states that meet QoS immediately, such that the ML models are aware of boundary cases, and avoid them in deployment. In our setting is 20% of QoS empirically, to adequately explore the allocation space, without causing the tail latency distribution to deviate too much from values that would be seen in deployment. Collecting data exclusively when the system operates nominally, or randomly exploring the allocation space does not fulfill these requirements. Fig. 8 shows the latency distribution in the training dataset, and how the training and validation error of the model changes with respect to the latency range observed in the training dataset, for the Social Network application. In the second figure, the x-axis is the latency of samples in the training dataset, the left y-axis is the root mean squared error RMSE of the CNN, and the right y-axis represents the classification error rate of XGBoost. Each point's y-axis value is the model's training and validation error when trained only with data whose latency is smaller than the corresponding x-value. If the training dataset does not include any samples that violate QoS (500ms), both the CNN and XGBoost experience serious overfitting, greatly mispredicting latencies and QoS violations. Fig. 9 shows data collected using data collection mechanisms that do not curate the dataset's distribution. Specifically, we show the prediction accuracy when the training dataset is collected when autoscaling is in place (a common resource management scheme in most clouds), and when resource allocations are explored randomly. As expected, when using autoscaling, the model does not see enough cases that violate QoS, and hence seriously underestimates latency and causes large spikes in tail latency, forcing the scheduler to use all available resources to prevent further violations. On the other hand, when the model is trained using random profiling, it constantly overestimates latency and prohibits any resource reduction, highlighting the importance of jointly designing the data collection algorithms and the ML models. Incremental and Transfer Learning: Incremental retraining can be applied to accommodate changes to the deployment strategy or microservice updates. In cases where the topology of the microservice graph is not impacted, such as hardware updates and change of public cloud provider, transfer learning techniques such as fine tune can be used to train the ML models in the background with newly collected data. If the topology is changed, the CNN needs to be modified to account for removed and newly-added tiers. Additional resources: Sinan can be extended to other system resources. Several resources, such as network bandwidth and memory capacity act like thresholds, below which performance degrades dramatically, e.g., network bandwidth, or the application experiences out of memory errors, and can be managed with much simpler models, like setting fixed thresholds for memory usage, or scaling proportionally with respect Scale Down Reduce CPU limit of 1 tier Scale Down Batch Reduce CPU limit of k least utilized tiers, Keep current resource allocation Scale Up Increase CPU limit of 1 tier Scale Up All Increase CPU limit of all tiers Scale Up Victim Increase CPU limit of recent victim tiers, that are scaled down in previous t cycles to user load for network bandwidth. Online Scheduler During deployment, the scheduler evaluates resource allocations using the ML models, and selects appropriate allocations that meet the end-to-end QoS without overprovisioning. Evaluating all potential resource allocations online would be prohibitively expensive, especially for complex microservice topologies. Instead, the scheduler evaluates a subset of allocations following the set of heuristics shown in Table 2. For scaling down operations, the scheduler evaluates reducing CPU allocations of single tiers, and batches of tiers, e.g., scaling down the k tiers with lowest cpu utilization, 1 < k ≤ N, N being the number of tiers in the microservice graph. When scaling up is needed, the scheduler examines the impact of scaling up single tiers, all tiers, or the set of tiers that were scaled down in the past t decision intervals, 1 < t < T with T chosen empirically. Finally, the scheduler also evaluates the impact of maintaining the current resource assignment. The scheduler first excludes operations whose predicted tail latency is higher than QoS − RMSE valid. Then it uses the predicted violation probability to filter out risky operations, with two user-defined thresholds, p d and p u (p d < p u ). These thresholds are similar to those used in autoscaling, where the lower threshold triggers scaling down and the higher threshold scaling up; the region between the two thresholds denotes stable operation, where the current resource assignment is kept. Specifically, when the violation probability of holding the current assignment is smaller than p u, the operation is considered acceptable. Similarly, if there exists a scale down operation with violation probability lower than p d, the scale down operation is also considered acceptable. When the violation probability of the hold operation is larger than p u, only scaling up operations with violation probabilities smaller than p u are acceptable; if no such actions exist, all tiers are scaled up to their max amount. We set p u such that the validation study's false negatives are no greater than 1% to eliminate QoS violations, and p d to a value smaller than p u that favors stable resource allocations, so that resources do not fluctuate too frequently unless there are significant fluctuations in utilization and/or user demand. Among all acceptable operations, the scheduler selects the one requiring the least resources. The scheduler also has a safety mechanism for cases where the ML model's predicted latency or QoS violation probability deviate significantly from the ground truth. If a mispredicted QoS violation occurs, Sinan immediately upscales the resources of all tiers. Additionally, given a trust threshold for the model, whenever the number of latency prediction errors or missed QoS violations exceeds the thresholds, the scheduler reduces its trust in the model, and becomes more conservative when reclaiming resources. In practice, Sinan never had to lower its trust to the ML model. Evaluation We first evaluate Sinan's accuracy, and training and inference time, and compare it to other ML approaches. Second, we deploy Sinan on our local cluster, and compare it against autoscaling, a widely-deployed empirical technique to manage resources in production clouds, and PowerChief, a resource manager for multi-stage applications that uses queueing analysis. Third, we show the incremental retraining overheads of Sinan. Fourth, we evaluate Sinan's scalability on a largescale Google Compute Engine (GCE) cluster. Finally, we discuss how interpretable ML can improve the management of cloud systems. Methodology Benchmarks: We use the Hotel Reservation and Social Network benchmarks described in Section 2.2. QoS targets are set with respect to 99% end-to-end latency, 200ms for Hotel Reservation, and 500ms for Social Network. Deployment: Services are deployed with Docker Swarm, with one microservices per container for deployment ease. Locust is used as the workload generator for all experiments. Local cluster: The cluster has four 80-core servers, with 256GB of RAM each. We collected 31302 and 58499 samples for Hotel Reservation and Social Network respectively, using our data collection process, and split them into training and validation sets with a 9:1 ratio, after random shuffling. The data collection agent runs for 16 hours and 8.7 hours for Social Network and Hotel Reservation respectively, and collecting more training samples do not further improve accuracy. GCE cluster: We use 93 container instances on Google Compute Engine (GCE) to run Social Network, with several replicas per microservice tier. 5900 extra training samples are collected on GCE for the transfer learning. Table 3 compares the short-term ML model in Sinan (CNN) against a multilayer perceptron (MLP), and a long short-term memory (LSTM) network, which is traditionally geared towards timeseries predictions. We rearrange the system history X RH to be a 2D tensor with shape T (F * N), and a 1D vector with shape T * F * N for the LSTM and MLP models, respectively. To configure each network's parameters, we increase the number of fully-connected, LSTM, and convolutional layers, as well as the number of channels in each layer for the MLP, LSTM, and Sinan (CNN), until accuracy levels off. Sinan's CNN achieves the lowest RMSE, with the smallest model size. Although the CNN is slightly slower than the LSTM, its inference latency is within 1% of the decision interval (1s), which does not delay online decisions. Table 4 shows a similar validation study for the Boosted Trees model. Specifically, we quantify the accuracy of anticipating a QoS violation over the next 5 intervals (5s), and the number of trees needed for each application. For both applications, the validation accuracy is higher than 94%, demonstrating BT's effectiveness in predicting the performance evolution in the near future. Sinan always runs on a single NVidia Titan XP GPU with average utilization below 2%. Performance and Resource Efficiency We now evaluate Sinan's ability to reduce resource consumption while meeting QoS on the local cluster. We compare Sinan against autoscaling and PowerChief. We experimented with two autoscaling policies: AutoScaleOpt is configured according to, which increases resources by 10% and 30% when utilization is within [60%, 70%) and respectively, and reduces resources by 10% and 30% when utilization is within [30%, 40%) and [0%, 30%). AutoScaleCons is more conservative and optimizes for QoS, using thresholds tuned for the examined applications. It increases resources by 10% and 30% when utilization is within [30%, 50%) and , and reduces resources by 10% when utilization is within [0%, 10%). PowerChief is implemented as in, and estimates the queue length and queueing time ahead of each tier using network traces obtained through Docker. For each service, we run 9 experiments with an increasing number of emulated users sending requests under a Poisson distribution with 1 RPS mean arrival rate. Figure 10 shows the mean and max CPU allocation, and the probability of meeting QoS across all studied mechanisms, where CPU allocation is the aggregate number of CPUs assigned to all tiers averaged over time, the max CPU allocation is the max of the aggregate CPU allocation over time, and the probability of meeting QoS is the fraction of execution time when end-to-end QoS is met. For Hotel Reservation, only Sinan and AutoScaleCons meet QoS at all times, with Sinan additionally reducing CPU usage by 25.9% on average, and up to 46.0%. AutoScaleOpt only meets QoS at low loads, when the number of users is no greater than 1900. At 2200 users, AutoScaleOpt starts to violate QoS by 0.7%, and the probability of meeting QoS drops to 90.3% at 2800 users, and less than 80% beyond 3000 users. Similarly, PowerChief meets QoS for fewer than 2500 users, however the probability of meeting QoS drops to 50.8% at 2800 users, and never exceeds 40% beyond 3000 users. AutoScaleOpt uses 53% the amount of resources Sinan requires on average, at the price of performance unpredictability, and PowerChief By reducing both the average and max CPU allocation, Sinan can yield more resources to colocated tasks, improving the machine's effective utilization. There are three reasons why PowerChief cannot reduce resources similarly and leads to QoS violations. First, as discussed in Sec. 2.3, the complex topology of microservices means that the tier with the longest igress queue, which PowerChief signals as the source of performance issues, is not necessarily the culprit but a symptom. Second, in interactive applications, queueing takes place across the system stack, including the NIC, OS kernel, network processing, and application, making precise queueing time estimations challenging, especially when tracing uses sampling. Finally, the stricter latency targets of microservices, compared to traditional cloud services, indicate that small fluctuations in queueing time can result in major QoS violations due to imperfect pipelining across tiers causing backpressure to amplify across the system. Fig. 11 shows the detailed results for Social Network, for 300 concurrent users under a diurnal load. The three columns each show requests per second (RPS), predicted latency vs. real latency and predicted QoS violation probability, and the realtime CPU allocation. As shown, Sinan's tail latency prediction closely follows the ground truth, and is able to react rapidly to fluctuations in the input load. Incremental Retraining We show the incremental retraining overheads of Sinan's ML models in three different deployment scenarios with the Social Network applications: switching to new server platforms (from the local cluster to a GCE cluster), changing the number of replicas (scale out factor) for all microservices except the backend databases (to avoid data migration overheads), and modifying the application design by introducing encryption in post messages uploaded by users (posts are encrypted with AES before being stored in the databases). Instead of retraining the ML models from scratch, we use the previouslytrained models on the local cluster, and fine-tune them using a small amount of newly-collected data, with the initial learning rate being 1 10 −5, 1 100 of the original value, in order to preserve the learnt weights in the original model and constrain the new solution derived by the SGD algorithm to be in a nearby region of the original one. The results are shown in Fig. 12, in which the y-axis is the RMSE and the x-axis is the number of newly-collected training samples (unit being 1000). Post involve the majority of microservices, and hence are more resource intensive, while others, like ReadUserTimeline involve a much smaller number of tiers, and are easier to allocate resources for. We vary the ratio of Compose-Post:ReadHomeTimeline:ReadUserTimeline requests; the ratios of the W 0, W 1, W 2 and W 3 workloads are 5:80:15, 10:80:10, 1:90:9, and 5:70:25, where W 0 has the same ratio as the training set. The ratios are representative of different social media engagement scenarios. The average CPU allocation and tail latency distribution are shown in Fig. 13 and Fig. 14 Explainable ML For users to trust ML, it is important to interpret its output with respect to the system it manages, instead of treating ML as a black box. We are specifically interested in understanding what makes some features in the model more important than others. The benefits are threefold: 1) debugging the models; 2) identifying and fixing performance issues; 3) filtering out spurious features to reduce model size and speed up inference. 5.6.1. Interpretability methods For the CNN model, we adopt the widely-used ML interpretability approach LIME. LIME interprets NNs by identifying their key input features which contribute most to predictions. Given an input X, LIME perturbs X to obtain a set of artificial samples which are close to X in the feature space. Then, LIME classifies the perturbed samples with the NN, and uses the labeled data to fit a linear regression model. Given that linear regression is easy to interpret, LIME uses it to identify important features based on the regression parameters. Since we are mainly interested in understanding the culprit of the QoS violations, we choose samples X from the timesteps where QoS violations occur. We perturb the features of a given tier or resource by multiplying that feature with different constants. For example, to study the importance of MongoDB, we multiply its utilization history with two constants 0.5 and 0.7, and generate multiple perturbed samples. Then, we construct a dataset with all perturbed and original data to train the linear regression model. Last, we rank the importance of each feature by summing the value of their associated weights. 5.6.2. Interpreting the CNN We used LIME to correct performance issues in Social Network, where tail latency experienced periods of spikes and instability despite the low load, as shown by the red line in Fig. 15. Manual debugging is cumbersome, as it requires delving into each tier, and potentially combinations of tiers to identify the root cause. Instead, we leverage explainable ML to filter the search space. First, we identify the top-5 most important tiers; the results are shown in the w/ Sync part of Table 5. We find that the most important tier for the model's prediction is social-graph Redis, instead of tiers with heavy CPU utilization, like nginx. We then examine the importance of each resource metric for Redis, and find that the most meaningful resources are cache and resident working set size, which correspond to data from disk cached in memory, and non-cached memory, including stacks and heaps. Using these hints, we check the memory configuration and statistics of Redis, and identify that it is set to record logs in persistent storage every minute. For each operation, Redis forks a new process and copies all written memory to disk; during this it stops serving requests. Disabling the log persistence eliminated most of the latency spikes, as shown by the black line in Fig. 15. We further analyze feature importance in the model trained with data from the modified Social Network, and find that the importance of social-graph Redis is significantly reduced, as shown in the Sinan identified Redis as the source of unpredictable performance, and additionally determined the resources that were being saturated, pointing to the issue being in Redis's logging functionality. Disabling logging significanly improved performance, which is also reflected in that tier's importance, as far as meeting QoS is concerned, being reduced. w/o Sync part of Table 5, in agreement with our observation that the service's tail latency is no longer sensitive to that tier. Related Work We now review related work on microservices, cloud management, and the use of machine learning in cloud systems. Microservices: The emergence of microservices has prompted recent work to study their characteristics and system implications. DeathstarBench and uSuite are two representative microservice benchmark suites. Death-StarBench includes several end-to-end applications built with microservices, and explores the system implications of microservices in terms of server design, network and OS overheads, cluster management, programming frameworks, and tail at scale effects. uSuite also introduces a number of multitier applications built with microservices and studies their performance and resource characteristics. Urgaonkar et al. introduced analytical modeling to multi-tier applications, which accurately captured the impact of aspects like concurrency limits and caching policies. The takeaway of all these studies is that, despite their benefits, microservices change several assumptions current cloud infrastructures are designed with, introducing new system challenges both in hardware and software. In terms of resource management, Wechat manages microservices with overload control, by matching the throughput of the upstream and downstream services; PowerChief dynamically power boosts bottleneck services in multi-phase applications, and Suresh et al. leverage overload control and adopt deadline-based scheduling to improve tail latency in multi-tier workloads. Finally, Sriraman et al. present an autotuning framework for microservice concurrency, and show the impact of threading decisions on application performance and responsiveness. Cloud resource management: The prevalence of cloud computing has motivated many cluster management designs. Quasar, Mesos, Torque, and Omega all target resource allocation in large, multi-tenant clusters. Quasar is a QoS-aware cluster manager that leverages machine learning to identify the resource preferences of new, unknown applications, and allocate resources in a way that meets their performance requirements without sacrificing resource efficiency. Mesos is a two-level scheduler that makes resource offers to different tenants, while Omega uses a shared-state approach to scale to larger clusters. More recently, PARTIES leveraged the intuition that resources are fungible to co-locate multiple interactive services on a server, using resource partitioning. Autoscaling is the industry standard for elastically scaling allocations based on utilization. While all these systems improve the performance and/or resource efficiency of the cloud infrastructures they manage, they are designed for monolithic applications, or services with a few tiers, and cannot be directly applied to microservices. ML in cloud systems: There has been growing interest in leveraging ML to tackle system problems, especially resource management. Quasar leverages collaborative filtering to identify appropriate resource allocations for unknown jobs. Autopilot uses an ensemble of models to infer efficient CPU and memory job configurations. Resource central characterizes VM instance behavior and trains a set of ML models offline, which accurately predict CPU utilization, deployment size, lifetime, etc. using random forests and boosting trees. Finally, Seer presented a performance debugging system for microservices, which leverages deep learning to identify patterns of common performance issues and to help locate and resolve them. Conclusion We have presented Sinan, a scalable and QoS-aware resource manager for interactive microservices. Sinan highlights the challenges of managing complex microservices, and leverages a set of validated ML models to infer the impact allocations have on end-to-end tail latency. Sinan operates online and adjusts its decisions to account for application changes. We have evaluated Sinan both on local clusters and public clouds GCE) across different microservices, and showed that it meets QoS without sacrificing resource efficiency. Sinan highlights the importance of automated, data-driven approaches that manage the cloud's complexity in a practical way. |
def msg_type_to_cpp(type):
(base_type, is_array, array_len) = roslib.msgs.parse_type(type)
cpp_type = None
if (roslib.msgs.is_builtin(base_type)):
cpp_type = MSG_TYPE_TO_CPP[base_type]
elif (len(base_type.split('/')) == 1):
if (roslib.msgs.is_header_type(base_type)):
cpp_type = ' ::std_msgs::Header_<ContainerAllocator> '
else:
cpp_type = '%s_<ContainerAllocator> '%(base_type)
else:
pkg = base_type.split('/')[0]
msg = base_type.split('/')[1]
cpp_type = ' ::%s::%s_<ContainerAllocator> '%(pkg, msg)
if (is_array):
if (array_len is None):
return 'std::vector<%s, typename ContainerAllocator::template rebind<%s>::other > '%(cpp_type, cpp_type)
else:
return 'boost::array<%s, %s> '%(cpp_type, array_len)
else:
return cpp_type |
// Legendaries returns how many legendaries are left on the loot table.
func (t *TietoolsLootTable) Legendaries() int {
legos := 0
drops := t.Drops()
t.RLock()
defer t.RUnlock()
for _, d := range drops {
if d.Item == nil {
continue
}
tietool, ok := d.Item.(Tietool)
if !ok {
continue
}
if tietool.Quality == Legendary && d.Weight > 0 {
legos++
}
}
return legos
} |
This invention relates to encoders for converting translational or rotational mechanical movement into electronic position signals. More particularly, this invention relates to analog-to-digital encoders for converting the physical position of a liquid level measuring instrument into digitally encoded electronic position signals.
U.S. patent application Ser. No. 595,185 filed July 11, 1975 for "ANALOG-TO-DIGITAL ENCODER UNIT EMPLOYING A PRE-ENCODED FILM", to be issued as U.S. Pat. No. 3,975,603 on Aug. 17, 1976, the disclosure of which is hereby incorporated by reference, discloses and claims an optical encoder unit for generating digital electrical signals defining the absolute level of a liquid in a tank from the translational position of a perforated tape in a conventional liquid level gauge, which is an improvement over commercially available tank gauge systems. The invention of the referenced patent employs a pre-encoded film strip having a plurality of laterally spaced longitudinally extending tracks with transparent and opaque regions encoded in a predetermined digital code format, which film strip is received on a pair of freely rotatable spools and is guided past an optical detection station by a pair of spaced stationary guide members. A rotatable sprocket having an input shaft adapted to be driven by a shaft coupled to a sprocket in a conventional perforated tape liquid level gauge drives the film strip past the detection station in response to movement of the tape in either direction. A spring-biased pivotally mounted tension arm having a guide in surface contact with the film strip maintains tension therein as the film strip is moved from reel to reel. The spools are mechanically interlinked with a drive belt to provide film take up in either direction.
The optical detection station includes a light source assembly for generating a plurality of laterally spaced light beams and a detector assembly having a corresponding plurality of photosensitive devices each associated to a different one of the light beams and shielded from the remaining beams. The film strip is positioned so that each track occupies the light path between a different one of the beams and photosensitive devices in order to define a plurality of information channels. The electrical position signals output from the photosensitive devices are coupled to a local or remote readout unit provided with electronic circuitry for converting the signal to operator readable digital display values.
While the above-described referenced invention affords several advantages over conventional tank gauge systems, such as non-volatile storage of the liquid level information, local or remote readout, easy installation and removal, and absolute encoding of the liquid level information over a wide range, the specific embodiment disclosed in the above-referenced application is designed for mounting in a particular attitude which is not always convenient to employ. Further, while relatively compact in size, this embodiment is not suitable for use in extremely limited space. |
Next month marks the 100th anniversary of the February Revolution of 1917, the first of two revolutions rocking the Russian Empire that fateful year. The events of February had immense consequences for Eurasia and the world. Historian Sergei Zasorin recalls the Revolution's impact on Poland and its drive for independence.
The February 1917 Revolution, which began on March 8 (February 23 according to the Julian calendar used by the Russian Empire), lasted about a week, and ended in the abdication of Czar Nicholas II, the end of Czarism, the collapse of the Russian Empire and the establishment of the Provisional Government, which attempted to concentrate all legislative and executive authority in its hands.
In a special historical analysis for Sputnik Poland , Dr. Sergei Zasorin, historian and associate professor at the Moscow State Pedagogical University, recalled that among the dramatic consequences of the revolutionary tumult is a topic that still continues to be debated in Poland today, 100 years on. It revolves around the question "of the connection between two phenomena: the overthrow of the monarchy in Russia, and Poland's regaining of its independence," he writes.
In Poland, Zasorin recalled, "some historians and politicians say in a linear manner that Polish sovereignty was artificially provoked by the revolutionary crisis, first in February and then in October 1917. Their opponents are convinced that Poland was already objectively and firmly marching toward independence from Russia since at least the end of the nineteenth century; therefore, they say, the upheaval facing Russia at the start of the 20th century had no effect on this process."
"What is the truth? Perhaps, as always, somewhere in the middle," the historian noted. "In our understanding of this historical problem, I would like to caution against simplistic and extreme conclusions."
Following the third and final partition of Poland among Prussia, the Austrian Empire and the Russian Empire in 1795, the country ceased to exist as an independent state until 1918. Throughout the 19th century, Zasorin noted, Polish national liberation movements fought for independence, "most vividly during the Polish uprisings of 1863-1864 and 1830-1831. Poles also played a decisive role in the dramatic events of the first Russian Revolution of 1905-1907."
But it was during the revolutionary spring of 1917 that the Polish question gained a new strength, the historian added. "After February, the national separatist movement started to be revitalized, and a Polish national elite gained political weight. In the conditions of the liberalization of the political regime, an expansion of rights and freedoms for Russian citizens, the Polish elite saw an intensification of the desire for political independence."
It's worth keeping in mind, Zasorin recalled, that the end of Czarism was complemented by the tremendous uncertainty in the political situation in Russia, thanks to the phenomenon of dual power – with two competing centers of power set up in Petrograd after the February Revolution – the Provisional Government and the Petrograd Soviet.
"The Polish question," the historian noted, "became the subject of a kind of political competition between the two all-Russian centers of power. In fact, both of them, seeking to demonstrate their democratic character, flirted with the most significant national minorities, hoping either for their support in the struggle for political power, or looking for approval from Russia's Entente allies, using the issue of Polish sovereignty as ammunition."
© Sputnik / The provisional government's first session in Mariinsky palace, Petrograd
At first, the Provisional Government favored the preservation of the unity of the Russian state, hoping that the formal provision of equal civil rights and freedoms would slake national minorities' thirst for national recognition. The liberal parties which dominated the government, including the Cadets and Octoberists, each considered it possible to provide Poland with autonomy, but not independence.
However, on March 14, the Petrograd Soviet, then dominated by moderate socialists, including the Mensheviks and left Socialist-Revolutionaries (SRs), published its 'Appeal to the Polish People', in which they spoke openly about Polish independence.
The resolution adopted by the Petrograd Soviet, read as follows: "The Petrograd Soviet of Workers' and Soldiers' Deputies declares that Russian democracy stands for the recognition of national-political self-determination of peoples, and proclaims that Poland has the right to complete independence in national and international affairs. We send our fraternal greetings to the Polish people and wish it success in the forthcoming struggle for the establishment of a democratic, republican order in independent Poland."
© Sputnik / The first Congress of Workers' and Soldiers' Deputies Council at the Tavrichesky Palace in Petrograd, 1917.
This appeal, Zasorin noted, led to an immediate radicalization of the position of the Provisional Government as well. On March 16, 1917, they issued their own Proclamation "to the Poles," and also promised independence – including all three sections of the country divided by the Third Partition.
© Photo : pixabay Happy New 1917! History Buffs Bring the Past to Life in Latest Twitter Craze
"Admittedly, the issue of the exact territorial composition of the future Polish state remained vague," the historian stressed. "The Provisional Government merely noted that it counted on the creation of Poland and a future 'free military alliance,' and postponed the final determination of Poland's and Russia's borders until the meeting of the Constituent Assembly," which was slated to convene in January 1918 following elections.
"This official statement would play a fatal role in the subsequent evolution of the state, and set in motion the process of the disintegration of the empire," Zasorin noted. "In the summer of 197, Finland declared independence, rumblings started in Ukraine about self-determination, and the pace of disintegration only quickened after that. All this led to an internal split within the Provisional Government. When the centrifugal effect turned from the Polish to the Ukrainian question, the Cadets and Octoberists were frightened, leading to their insoluble conflict with the SRs and the collapse of the coalition in July 1917."
"It is significant," the historian added, "that the last Russian Czar himself already tried to play the 'Polish card'. In December 1916, Supreme Commander Nicholas II appealed to the Army and Navy with Order 870, which among the aims of continuing the war first mentioned 'the creation of a free Poland.' Interestingly, this would be the first and last time that the Czar or other royal dignitaries would say anything about the matter. But the words in the Order are also a matter of historical fact, from which it is possible, if desired, to create an argument about a fundamental change in the Czarist position on the Polish question shortly before the revolution, and at a critical moment in the First World War."
© Sputnik / Copy of an illustration depicting Russian Emperor Nicholas II's abdication.
In truth, of course, "the late pre-revolutionary and post-revolutionary Russian authorities abandoned their prerogative for control over Poland not for ideological reasons, but primarily because Polish territories had been occupied by the Central Powers in 1915, and Russia had physically lost this right." In fact, Zasorin noted, "in the competition for Polish sympathies, Russia left Germany and Austro-Hungary far behind."
The altered Russian position, the historian wrote, helped lead to the decision by future Polish Second Republic leader Josef Pilsudski, then commandant of the Polish Legions under the Austro-Hungarian Empire, to disband his units in the summer of 1917.
© Sputnik / Alexey Filippov Necessary to Work Toward Reconciliation When Recalling 1917 Russian Revolution - Putin
Furthermore, just as the February Revolution and the so-called democratization of the Russian Army led to its demoralization and eventual disintegration, so too did it have an impact on Polish soldiers and officers serving under Russia, heightening their sense of nationalism and patriotism. Through 1917, the Polish Rifle Division, for example, saw the spread of the use of the Polish language, as well as Polish national symbols. In March 1917, units of the division stationed in the Kiev garrison took part in a parade using the Polish anthem and flag.
The Polish Lancers, furthermore, gained the right not to swear an oath of allegiance to Russia, using their own text instead, which spoke about the independence and unification of Poland. Zasorin recalled that "at an April 1917 Congress of Soldiers in Kiev, [the Poles] adopted a declaration proclaiming that the continuation of the war was aimed at the restoration of Polish independence on all [occupied] lands. On this basis, it was planned to create a Polish army that would fight on the side of the Entente."
© Sputnik / Workers hold a meeting at a Polish enterprise.
These plans suited the interests of the Provisional Government perfectly well, the historian added. "In March 1917, Minister of War Alexander Guchkov developed a plan on the formation of a Polish Army on Russian territory, providing for the provision of the Polish Division with artillery and engineering units, the creation of a second division, and their eventual unification into a corps. To this end, the Russian General Staff established a military commission for the formation of Polish units. On April 30, the commander of the Polish Infantry Division, Major-General Tadeusz Bylewski, became head this unit."
Ultimately, Zasorin noted, "by the spring and summer of 1917, Poland's drive for independence was greatly stimulated by the democratization processes occurring in Russia, and by the political chaos and inefficiency of the new Russian authorities, the collapse of the army and constant setbacks at the front."
In late 1917, February gave way to October, leading to the collapse of the disorganized and exhausted Provisional Government and the seizure of state power by the Bolsheviks and their allies, who soon proclaimed the world's first socialist state.
As for Poland, the country gradually reemerged as a sovereign republic between 1917 and 1918, and the Second Polish Republic was formally established on November 11, 1918, 123 years after the Third Partition, at the close of the First World War. |
Clinical role of non-contrast magnetic resonance angiography for evaluation of renal artery stenosis. BACKGROUND The association between a gadolinium-based contrast material and nephrogenic systemic fibrosis has been discussed. The purpose of our study was to evaluate whether non-contrast enhanced magnetic resonance angiography (MRA) might provide sufficient information of renal artery stenosis. METHODS AND RESULTS The non-contrast MRA of 26 patients with hypertension was retrospectively reviewed in the present study. The significant renal artery stenosis was visually evaluated by comparing non-contrast MRA with computed tomography or conventional angiographic finding. Difference of the intensities between the proximal and distal aorta was quantitatively evaluated. The sensitivity, specificity, positive predictive value and negative predictive value of non-contrast MRA in the evaluation of the renal artery stenosis was 78%, 91%, 64% and 96%, respectively. The distal abdominal aorta showed less signal intensity than the proximal aorta by 16.9+/-12.2%. CONCLUSIONS Non-contrast MRA is a non-invasive and effective method that allows evaluation of the renal artery stenosis. |
1. Field of the Invention
The present invention relates to an apparatus for managing consumables of an image forming apparatus which are widely used in printers, such as multi-function printers (MFP). More particularly, the present invention relates to an apparatus using a memory for managing consumables of an image forming apparatus.
2. Description of the Related Art
Generally, a facsimile machine, a printer, a copier, and an MFP which combines the functions of a printer, scanner, copier or facsimile machine include an image forming apparatus.
FIG. 1 shows a schematic block diagram of a conventional image forming apparatus and FIG. 2 shows a schematic view of an engine device of the apparatus shown in FIG. 1.
Referring to FIG. 1, the conventional image forming apparatus includes a control unit 10 controlling an overall operation of the apparatus, an interface unit 20 for connecting a computer and the apparatus to receive printing data from the computer, a storage unit 30 for storing a variety of control programs necessary for driving the apparatus and data generated by executing the control programs, an engine device 50 for carrying out a print process, an engine control unit 40 for driving the engine device 50 according to a control of the control unit 10, a high voltage power supply 80 for applying a predetermined voltage to each roller of the engine device 50 according to a control of the control unit 10, and a sensor unit 90 for detecting a printing error like a paper jam or shortage of paper. The engine device 50 shown in FIG. 2 is provided with a charge roller 51, an organic photoconduction cartridge (OPC) 52, a laser scan unit (LSU) 53, a development roller 54, a feed roller 55, a transfer roller 56 and a fusing unit 57, for printing image data on the paper.
The development roller 54 and the transfer roller 56 as shown in FIG. 2 are respectively connected with color development cartridges YC, MC, CC, BC for color prints and an image transfer belt (ITB) as shown in FIG. 3.
Printing procedures of the engine device 50 having the structure of FIG. 2 are typically as follows. The high voltage power supply unit 80 applies voltages to each roller 51, 54, 55, 56, 57 of the engine device 50 according to a control of the control unit 10. Herein, the high voltage power supply (HVPS) 80 generally applies a charge voltage of −1.4 kV to the charge roller 51, a transfer voltage of +2.0 kV to the transfer roller 56, a development voltage of 300V to the development roller 54, a feed voltage of 500V to the feed roller 55. Therefore, the charge roller 51 charged by the high voltage and engaged with the OPC 52 rotates to uniformly charge the photoconductor formed on an outer circumference of the OPC 52. Herein, the LSU 53 receives a control signal from the control unit 10, the control signal allowing an image light corresponding to an image data inputted from control unit 10 to be scanned onto the OPC 52. Accordingly, a laser diode (LD) provided in the LSU 53 is turned on and projects the image light corresponding to the received control signal to the OPC 52. As a result, an electrostatic latent image to be printed, charged by the image light of the LSU 53, is formed on the surface of the OPC 52. Meanwhile, a potential difference is generated between the feed roller 55 to which the high voltage is applied and the development roller 54 to which the voltage lower than that of the feed roller 55 is applied. Accordingly, a toner charged negatively moves to the development roller 54 from the feed roller 55 and is coated on the electrostatic latent image formed on the surface of the OPC 52 to form a toner image. The transfer roller 56 to which the high voltage from HVPS 80 is applied transfers the toner image formed on the electrostatic latent image of the OPC 52 to a paper. The toner image transferred on the paper is fused on the paper by high temperature and high pressure of the fusing unit 57, thereby ending a printing process.
As described above, a plurality of consumables such as development cartridges YC, MC, CC, BC, the ITB unit, the OPC unit and so on such as shown in FIG. 3, are needed for the image forming apparatus to print an image, particularly a color image. In FIG. 3, YC indicates the development cartridge for yellow color, MC indicates the development cartridge for magenta color, CC indicates the development cartridge for cyan color, and BC indicates the development cartridge for black color.
As shown in FIG. 3, key resistances Ry, Rm, Rc, Rb, Ri, Ro provided in the conventional consumables allow the control unit 10 of the image forming apparatus to detect the amounts of colors in the consumables for refilling or replacing the consumables.
The key resistances provided in the respective consumables are respectively connected to pull-up resistances R1 provided in a main board 1 of the image forming apparatus as shown in the FIG. 3, and are also connected to respective ports of the control unit 10. The other terminals of the respective key resistances are grounded, and the other terminals of the respective pull-up resistances are supplied with a power supply voltage Vcc. As shown in FIG. 4, the key resistances vary according to the time of use or consumption of contents (such as ink or toner) of the consumables. The voltage applied to control unit 10 also varies according to the variation of the key resistances. The control unit 10 functions to perceive the consumables, and/or to prevent refill of consumables, and/or to detect the consumables' life span, and so on, through the variation of key resistance.
The conventional apparatus for managing consumables of an image forming apparatus with key resistances provided in the consumables has several disadvantages.
First, the control unit of the image forming apparatus needs as many key resistance sensing ports as the number of consumables to individually detect the key resistances provided in the consumables. For example, as shown in FIG. 3, the control unit 10 of the image forming apparatus needs at least six ports for detecting the individual key resistances of four development cartridges, the ITB, and the OPC. That is, the control unit of the conventional image forming apparatus needs more ports for detecting the key resistance as the number of consumables increases.
Second, a control program of the control unit 10 to individually control the respective ports corresponding to the key resistance of the consumables is very complicated.
Third, signal connection lines and an electrical wire harness to respectively connect the key resistances of the consumables to the ports of the control unit have a complicated structure. Therefore the price of the apparatus increases.
Fourth, it is hard to substantially detect the refill of contents (such as ink or toner of the development cartridge) of the consumables with the key resistances. Therefore, the refill of contents is not practically prevented.
Fifth, if there are many OEM venders manufacturing the consumables, it is hard to identify the OEM venders. |
Achalasia in Anderson-Fabry's Disease induced at term and was uncomplicated, resulting in the vaginal delivery of a healthy girl weighing 3360 g. The previous day, a maternal temperature of 37.5°C had been recorded. Over the next seven days the mother felt weak and complained of a sore throat and sweating. There was a fever with twice daily peaks to 39°C. Cultures from the genital tract, urine, stool and blood grew no pathogens. A course of amoxycillin was without effect. On the fifth postpartum day examination revealed tender hepatosplenomegaly, petechiae on the hard palate and palpable cervical lymph nodes. The facial palsy was unchanged and there were no other neurological signs. The plasma bilirubin was normal, but the alkaline phosphatase was greater than 1373 IU/I and the aspartate aminotransferase was greater than 523 IU/1. The white cell count was 21.2 x 109/1, of which 30% were atypical lymphocytes. The Monospot test for glandular fever was negative and EB virus antibody was not detected. The CMV complement fixation titre was, less than 1: 4 but four weeks later had risen to greater than 1 :128. CMV was isolated from a high vaginal swab five weeks later but not during the illness. The fever subsided 12 days postnatally and the patient slowly recovered normal health. Four weeks after delivery, the liver and spleen were no longer palpable and liver function tests and blood picture were normal. Facial weakness completely resolved. The baby always remained well. At no time was virus isolated from swabs taken from the baby, but CMV IgM was detected in the baby's blood at six weeks and again at nine weeks of age. |
#!/usr/bin/env python3
import sys
import os
from . import __version__, long_version
def _get_usage(prog_name):
usage = '''{version}
Usage: {prog_name} <command> <arguments>
Commands:
[ Creating homology table ]
pretable Create homology pre-table.
This command aligns genomic regions back to the genome
to find homologous regions.
table Convert homology pre-table into homology table.
This command combines overlapping homologous regions
into longer duplications.
[ Analyzing BAM/CRAM files ]
depth Calculate read depth and variance in given genomic windows.
cn Find aggregate and paralog-specific copy number
for given unique and duplicated regions.
cn-using Same as "cn", but use input model parameters.
pool Pool reads from various copies of a given homologous region.
[ Querying homology table ]
view View and filter homology table.
msa Visualize multiple sequence alignment of
homologous regions.
psvs Output PSVs (paralogous-sequence variants)
between homologous regions.
[ General help ]
help Show this help message.
version Show version.
cite Show citation information.
'''
usage = usage.format(version=long_version(), prog_name=prog_name)
return usage
def _get_valid_commands():
return 'pretable table depth cn pool view msa psvs help version'.split()
def _throw_error(prog_name, command):
sys.stderr.write('Error: unknown command "{}"\n\n'.format(command))
sys.stderr.write('Usage: {} <command>\n'.format(prog_name))
sys.stderr.write(' Valid commands: {}.\n'.format(', '.join(_get_valid_commands())))
exit(1)
def _print_citations():
print('Publication in progress, please check later!')
def _process_exceptions(fn, *args, **kwargs):
try:
fn(*args, **kwargs)
except BrokenPipeError:
pass
# Add other exceptions here, if needed.
def main():
prog_name = os.path.basename(sys.argv[0])
if len(sys.argv) < 2:
print(_get_usage(prog_name))
return
command = sys.argv[1]
inner_args = sys.argv[2:]
inner_prog = '{} {}'.format(prog_name, command)
if command == '-h' or command == '--help' or command == 'help':
print(_get_usage(prog_name))
return
if command == '-V' or command == '--version' or command == 'version':
print(long_version())
return
if command == 'cite' or command == 'citation':
_print_citations()
return
if command == 'pretable':
from . import pretable
return _process_exceptions(pretable.main, inner_prog, inner_args)
if command == 'table':
from . import combine_table
return _process_exceptions(combine_table.main, inner_prog, inner_args)
if command == 'depth':
from . import depth
return _process_exceptions(depth.main, inner_prog, inner_args)
if command == 'cn' or command == 'cn-using':
is_new = command == 'cn'
from . import detect_cn
return _process_exceptions(detect_cn.main, inner_prog, inner_args, is_new)
if command == 'pool':
from . import pool_reads
return _process_exceptions(pool_reads.main, inner_prog, inner_args)
if command == 'view':
from . import view
return _process_exceptions(view.main, inner_prog, inner_args)
if command == 'msa':
from . import msa
return _process_exceptions(msa.main, inner_prog, inner_args)
if command == 'psvs':
from . import psvs
return _process_exceptions(psvs.main, inner_prog, inner_args)
if command == 'call':
from . import call_variants
return _process_exceptions(call_variants.main, inner_prog, inner_args)
_throw_error(prog_name, command)
if __name__ == '__main__':
main()
|
<reponame>Abhishek-kumar09/Orekit
/* Copyright 2002-2020 CS GROUP
* Licensed to CS GROUP (CS) under one or more
* contributor license agreements. See the NOTICE file distributed with
* this work for additional information regarding copyright ownership.
* CS licenses this file to You under the Apache License, Version 2.0
* (the "License"); you may not use this file except in compliance with
* the License. You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package org.orekit.gnss.attitude;
import org.hipparchus.RealFieldElement;
import org.hipparchus.util.FastMath;
import org.hipparchus.util.MathUtils;
import org.orekit.frames.Frame;
import org.orekit.time.AbsoluteDate;
import org.orekit.time.FieldAbsoluteDate;
import org.orekit.utils.ExtendedPVCoordinatesProvider;
import org.orekit.utils.TimeStampedAngularCoordinates;
import org.orekit.utils.TimeStampedFieldAngularCoordinates;
/**
* Attitude providers for Beidou Medium Earth Orbit navigation satellites.
* @author <NAME> Java translation
* @since 9.2
*/
public class BeidouMeo extends AbstractGNSSAttitudeProvider {
/** Limit for the Yaw Steering to Orbit Normal switch. */
private static final double BETA_YS_ON = FastMath.toRadians(4.1);
/** Limit for the Orbit Normal to Yaw Steering switch. */
private static final double BETA_ON_YS = FastMath.toRadians(3.9);
/** Simple constructor.
* @param validityStart start of validity for this provider
* @param validityEnd end of validity for this provider
* @param sun provider for Sun position
* @param inertialFrame inertial frame where velocity are computed
*/
public BeidouMeo(final AbsoluteDate validityStart, final AbsoluteDate validityEnd,
final ExtendedPVCoordinatesProvider sun, final Frame inertialFrame) {
super(validityStart, validityEnd, sun, inertialFrame);
}
/** {@inheritDoc} */
@Override
protected TimeStampedAngularCoordinates correctedYaw(final GNSSAttitudeContext context) {
// variation of the β angle over one orbital period (approximately)
final double beta = context.beta(context.getDate());
final double approxPeriod = 2 * FastMath.PI / context.getMuRate();
final double betaVariation = beta - context.beta(context.getDate().shiftedBy(-approxPeriod));
final double delta = context.getOrbitAngleSinceMidnight();
if (FastMath.abs(beta) <= BETA_YS_ON - FastMath.abs(betaVariation)) {
// the β angle is lower than threshold for a complete orbital period
// we are for sure in the Orbit Normal (ON) mode
return context.orbitNormalYaw();
} else if (FastMath.abs(beta) > BETA_ON_YS + FastMath.abs(betaVariation)) {
// the β angle is higher than threshold for a complete orbital period,
// we are for sure in the Yaw Steering mode
return context.nominalYaw(context.getDate());
} else {
// we are in the grey zone, somewhere near a mode switch
final boolean absBetaDecreasing = beta * betaVariation <= 0.0;
if (absBetaDecreasing) {
// we are going towards the β = 0 limit
if (FastMath.abs(beta) >= BETA_YS_ON) {
// we have not yet reached the far limit, we are still in Yaw Steering
return context.nominalYaw(context.getDate());
}
} else {
// we are going away from the β = 0 limit
if (FastMath.abs(beta) <= BETA_ON_YS) {
// we have not yet reached the close limit, we are still in Orbit Normal
return context.orbitNormalYaw();
}
}
// there is a mode switch near the current orbit, it occurs when orbit angle is 90°
// we check what was the β angle at the previous quadrature to see if the switch
// already occurred
final double angleSinceQuadrature =
MathUtils.normalizeAngle(delta - 0.5 * FastMath.PI, FastMath.PI);
final double timeSinceQuadrature = angleSinceQuadrature / context.getMuRate();
final AbsoluteDate quadratureDate = context.getDate().shiftedBy(-timeSinceQuadrature);
final double betaQuadrature = context.beta(quadratureDate);
if (absBetaDecreasing) {
// we are going towards the β = 0 limit
if (FastMath.abs(betaQuadrature) <= BETA_YS_ON) {
// we have switched to Orbit Normal mode since last quadrature
return context.orbitNormalYaw();
}
} else {
// we are going away from the β = 0 limit
if (FastMath.abs(betaQuadrature) <= BETA_ON_YS) {
// β was below switch at last quadrature, we are still in the Orbit Normal mode
return context.orbitNormalYaw();
}
}
return context.nominalYaw(context.getDate());
}
}
/** {@inheritDoc} */
@Override
protected <T extends RealFieldElement<T>> TimeStampedFieldAngularCoordinates<T> correctedYaw(final GNSSFieldAttitudeContext<T> context) {
// variation of the β angle over one orbital period (approximately)
final double beta = context.beta(context.getDate()).getReal();
final double approxPeriod = 2 * FastMath.PI / context.getMuRate().getReal();
final double betaVariation = beta - context.beta(context.getDate().shiftedBy(-approxPeriod)).getReal();
final double delta = context.getOrbitAngleSinceMidnight().getReal();
if (FastMath.abs(beta) <= BETA_YS_ON - FastMath.abs(betaVariation)) {
// the β angle is lower than threshold for a complete orbital period
// we are for sure in the Orbit Normal (ON) mode
return context.orbitNormalYaw();
} else if (FastMath.abs(beta) > BETA_ON_YS + FastMath.abs(betaVariation)) {
// the β angle is higher than threshold for a complete orbital period,
// we are for sure in the Yaw Steering mode
return context.nominalYaw(context.getDate());
} else {
// we are in the grey zone, somewhere near a mode switch
final boolean absBetaDecreasing = beta * betaVariation <= 0.0;
if (absBetaDecreasing) {
// we are going towards the β = 0 limit
if (FastMath.abs(beta) >= BETA_YS_ON) {
// we have not yet reached the far limit, we are still in Yaw Steering
return context.nominalYaw(context.getDate());
}
} else {
// we are going away from the β = 0 limit
if (FastMath.abs(beta) <= BETA_ON_YS) {
// we have not yet reached the close limit, we are still in Orbit Normal
return context.orbitNormalYaw();
}
}
// there is a mode switch near the current orbit, it occurs when orbit angle is 90°
// we check what was the β angle at the previous quadrature to see if the switch
// already occurred
final double angleSinceQuadrature =
MathUtils.normalizeAngle(delta - 0.5 * FastMath.PI, FastMath.PI);
final double timeSinceQuadrature = angleSinceQuadrature / context.getMuRate().getReal();
final FieldAbsoluteDate<T> quadratureDate = context.getDate().shiftedBy(-timeSinceQuadrature);
final double betaQuadrature = context.beta(quadratureDate).getReal();
if (absBetaDecreasing) {
// we are going towards the β = 0 limit
if (FastMath.abs(betaQuadrature) <= BETA_YS_ON) {
// we have switched to Orbit Normal mode since last quadrature
return context.orbitNormalYaw();
}
} else {
// we are going away from the β = 0 limit
if (FastMath.abs(betaQuadrature) <= BETA_ON_YS) {
// β was below switch at last quadrature, we are still in the Orbit Normal mode
return context.orbitNormalYaw();
}
}
return context.nominalYaw(context.getDate());
}
}
}
|
KARACHI (Reuters) - Pakistani grave digger Shahid Baloch is taking no chances.
Like many people in the port city of Karachi, he was caught out by the severity of last summer’s heat wave which killed more than 1,300 people, and has hired a digger to excavate three elongated trenches big enough for 300 bodies.
“Thanks to God, we are better prepared this year,” said Baloch, 28, who works with three brothers at the vast Karachi cemetery run by the charitable organisation Edhi Foundation.
When the heat wave struck in the summer of 2015, hospitals, morgues and graveyards in the city of 20 million people were overwhelmed, and drug addicts, day labourers and the elderly were the biggest victims of the searing heat.
Temperatures hit 44 degrees Celsius (111 Fahrenheit), their highest since 1981 and above normal summer levels of around 37C (99F).
Intervention by the army and charity groups staved off an even worse disaster, locals said, but the crisis exposed the shortcomings of Pakistani emergency services in coping with environmental disasters that scientists say will become more common in the future.
Pakistan’s meteorological office is not predicting a repeat of last year’s extreme conditions, but, like Baloch in the cemetery, officials are preparing for the worst just in case.
“It will not get out of control the way it happen last year,” said Karachi Commissioner Asif Hyder Shah, adding that nearly 60 hospitals now have spare capacity for 1,850 heat wave patients.
Last summer patients slept on ward floors and long queues formed outside Karachi’s main state hospitals at the peak of the heat wave.
Shah said nearly 200 first response centres have been set up across the city, offering basic heat-stroke treatment to swiftly stabilise patients. There are also 700 makeshift relief centres, dishing out drinking water and rehydration salts.
“This will save lives. It’s a comfort,” said street vendor Muhammad Mahmood, 32, after downing a cup of water at one centre. Next to him, children in school uniforms queued to quench their thirst.
Edhi Foundation, at the heart of efforts to limit the suffering caused by the heat wave last year, said it was expanding its huge fleet of ambulances, anchoring extra shelves in its morgue freezer and buying ice machines to keep patients and corpses cool.
Last summer, the Edhi morgue ran out of freezer space after about 650 bodies were brought in the space of a few days. Ambulances left decaying corpses outside in sweltering heat.
Similar macabre scenes plagued Karachi’s cemeteries, where grave diggers refused to work in the baking sun and charged up to five times normal rates for burial plots.
Efforts to prepare for extreme heat have been limited by decades of under-investment in Pakistan’s crumbling electricity grid and water infrastructure, leaving the sprawling city vulnerable in times of crisis.
The problem last year was compounded by power cuts which left people unable to cool themselves with fans and air conditioners, particularly affecting those unable to afford generators.
Some Pakistani politicians pinned some blame on the provincial government and K-Electric, the company that supplies electricity to Karachi, for the high death toll. K-Electric did not respond to requests for comment.
Some Karachi residents said much would depend on whether any future heat wave struck during the Muslim fasting month of Ramadan, when under Pakistani law it is illegal to eat and drink in public places.
Abdul Qayyum Soomro, religious affairs adviser to the chief minister of Sindh province, said officials will meet clerics to discuss whether a fatwa, or religious edict, should be issued allowing people to break the fast for health reasons.
Commissioner Shah said the subject was “extremely sensitive” among a devout population.
“If things get really bad, I may abandon the fast since God says life is most precious,” said a fruit vendor selling mangoes and bananas from a push cart.
Last year, his five-year-old son fell ill from the heat but was only treated at the third hospital they visited. The first two, including Karachi’s biggest, were full. |
/// JSR - Jump to Subroutine
/// Push the address (minus one) of the return point onto the stack and set pc to the target addr
pub fn jsr(&mut self, addr: u16) -> bool {
let pc = self.pc - 1;
self.push_stack((pc >> 8) as u8);
self.push_stack(pc as u8);
self.pc = addr;
false
} |
<reponame>sarahRosannaBusch/petTracker<filename>gpsLibrary-master/library/include/time.h
/****************************************************************************
* Title : Time library
* Filename : time.h
* Author : RL
* Origin Date : 08/25/2015
* Notes : None
*****************************************************************************/
/**************************CHANGE LIST **************************************
*
* Date Software Version Initials Description
* 08/25/15 1.01 RL Platform independent version
*
*****************************************************************************/
/** @file time.h
* @brief Time library for MIkroC
*
* @date 25 Aug 2015
* @author <NAME>
* @copyright GNU Public License
*
* @version .1 - Initial testing and verification
*
* @note Test configuration:
* MCU: STM32F107VC
* Dev.Board: EasyMx Pro v7
* Oscillator: 72 Mhz internal
* Ext. Modules: GPS Click
* SW: ARM 4.5.2
*
*/
#ifndef _TIME_H
#define _TIME_H
/******************************************************************************
* Includes
*******************************************************************************/
#include <stdint.h>
/******************************************************************************
* Preprocessor Constants
*******************************************************************************/
/**
* Some constants
*/
#define Time_secInMn 60 // seconds per minute
#define Time_secInH (Time_secInMn * 60) // seconds per hour
#define Time_secIn24h (Time_secInH * 24) // seconds per day
/******************************************************************************
* Configuration Constants
*******************************************************************************/
/******************************************************************************
* Macros
*******************************************************************************/
/******************************************************************************
* Typedefs
*******************************************************************************/
typedef struct
{
unsigned char ss ; // seconds
unsigned char mn ; // minutes
unsigned char hh ; // hours
unsigned char md ; // day in month, from 1 to 31
unsigned char wd ; // day in week, monday=0, tuesday=1, .... sunday=6
unsigned char mo ; // month number, from 1 to 12 (and not from 0 to 11 as with unix C time !)
unsigned int yy ; // year Y2K compliant, from 1892 to 2038
} TimeStruct;
/******************************************************************************
* Variables
*******************************************************************************/
extern long Time_jd1970 ; // 01/01/1970 julian day number
/******************************************************************************
* Function Prototypes
*******************************************************************************/
#ifdef __cplusplus
extern "C" {
#endif
/*
* public functions
*/
long Time_dateToEpoch(TimeStruct *ts) ;
long Time_dateDiff(TimeStruct *t1, TimeStruct *t2);
void Time_epochToDate(long e, TimeStruct *ts) ;
#ifdef __cplusplus
} // extern "C"
#endif
#endif /*TIME_H_*/
/*** End of File **************************************************************/
|
"""
Реализовать два небольших скрипта:
а) итератор, генерирующий целые числа, начиная с указанного,
б) итератор, повторяющий элементы некоторого списка, определенного заранее.
Подсказка: использовать функцию count() и cycle() модуля itertools. Обратите внимание, что создаваемый цикл
не должен быть бесконечным. Необходимо предусмотреть условие его завершения.
Например, в первом задании выводим целые числа, начиная с 3, а при достижении числа 10 завершаем цикл.
Во втором также необходимо предусмотреть условие, при котором повторение элементов списка будет прекращено.
"""
from itertools import count
def get_list_count(start, stop):
res_list = []
for el in count(start):
if el > stop:
break
else:
res_list.append(el)
return res_list
def main():
my_list = get_list_count(3, 10)
print(my_list)
print("Программа завершена")
if __name__ == "__main__":
main()
|
<gh_stars>1-10
/* istanbul ignore file */
/* tslint:disable */
/* eslint-disable */
export const $s3_Owner = {
properties: {
DisplayName: {
type: 'string',
},
ID: {
type: 'string',
},
},
}; |
<filename>src/components/DrawerMenu.tsx
import React from 'react'
import {
Divider,
Drawer,
IconButton,
List,
ListItem,
ListItemIcon,
ListItemText,
makeStyles
} from '@material-ui/core'
import { FaChevronLeft, FaHamburger } from 'react-icons/fa'
import { useHistory, useLocation } from 'react-router-dom'
import { useNavigator } from '../utils/NavigatorContext'
export default () => {
const {
menuDrawerWidth,
drawer,
toggleMenu,
menu,
maintainIcons,
menuDrawerIcon,
menuDrawerHeader
} = useNavigator()
const classes = useClasses({ drawerWidth: menuDrawerWidth, maintainIcons })
const history = useHistory()
const { pathname } = useLocation()
return (
<Drawer
className={`${classes.drawer} ${drawer ? classes.drawerOpen : classes.drawerClose}`}
variant={maintainIcons ? 'permanent' : 'persistent'}
anchor='left'
open={drawer}
classes={{ paper: `${drawer ? classes.drawerOpen : classes.drawerClose}` }}
>
<div className={classes.drawerHeader}>
{drawer && <div className={classes.drawerHeaderContainer}>{menuDrawerHeader}</div>}
<IconButton onClick={() => toggleMenu()}>
{drawer ? <FaChevronLeft /> : menuDrawerIcon || <FaHamburger />}
</IconButton>
</div>
<Divider />
<List>
{menu
.filter((e) => !e.hidden)
.map(({ route, title, icon, onClick }, i) => {
if (!route && !title) return <Divider key={i} />
return (
<ListItem
button
key={title}
selected={pathname === route}
onClick={() => {
if (onClick && !route) onClick(history)
if (route) history.push(route)
if (!maintainIcons) toggleMenu(false)
}}
>
{icon && <ListItemIcon>{icon}</ListItemIcon>}
<ListItemText primary={title} />
</ListItem>
)
})}
</List>
</Drawer>
)
}
const useClasses = makeStyles((theme) => ({
drawer: ({ drawerWidth }: any) => ({
flexShrink: 0,
[theme.breakpoints.up('sm')]: {
width: drawerWidth
},
width: '100%',
whiteSpace: 'nowrap'
}),
// drawerPaper: ({ drawerWidth }: any) => ({
// [theme.breakpoints.up('sm')]: {
// width: drawerWidth
// },
// width: '100%'
// }),
drawerHeader: {
display: 'flex',
alignItems: 'center',
// padding: theme.spacing(0, 1),
...theme.mixins.toolbar
// justifyContent: 'flex-end'
},
drawerHeaderContainer: {
flex: 1
},
drawerOpen: ({ drawerWidth }: any) => ({
width: drawerWidth,
transition: theme.transitions.create('width', {
easing: theme.transitions.easing.sharp,
duration: theme.transitions.duration.enteringScreen
})
}),
drawerClose: {
transition: theme.transitions.create('width', {
easing: theme.transitions.easing.sharp,
duration: theme.transitions.duration.leavingScreen
}),
overflowX: 'hidden',
width: theme.spacing(5) + 1,
[theme.breakpoints.up('sm')]: {
width: theme.spacing(7) + 1
}
}
}))
|
Health workforce issues and the Global Fund to fight AIDS, Tuberculosis and Malaria: an analytical review Recent studies have shown evidence of a direct and positive causal link between the number of health workers and health outcomes. Several studies have identified an adequate health workforce as one of the key ingredients to achieving improved health outcomes. Global health initiatives are faced with human resources issues as a major, system-wide constraint. This article explores how the Global Fund addresses the challenges of a health workforce bottleneck to the successful implementation of priority disease programmes. Possibilities for investment in human resources in the Global Fund's policy documents and guidelines are reviewed. This is followed by an in-depth study of 35 Global Fund proposals from five African countries: Ethiopia, Ghana, Kenya, Malawi and Tanzania. The discussion presents specific human resources interventions that can be found in proposals. Finally, the comments on human resources interventions in the Global Fund's Technical Review Panel and the budget allocation for human resources for health were examined. Policy documents and guidelines of the Global Fund foster taking account of human resources constraints in recipient countries and interventions to address them. However, the review of actual proposals clearly shows that countries do not often take advantage of their opportunities and focus mainly on short-term, in-service training in their human resources components. The comments of the Technical Review Panel on proposed health system-strengthening interventions reveal a struggle between the Global Fund's goal to fight the three targeted diseases, on the one hand, and the need to strengthen health systems as a prerequisite for success, on the other. In realizing the opportunities the Global Fund provides for human resources interventions, countries should go beyond short-term objectives and link their activities to a long-term development of their human resources for health. Possibilities for investment in human resources in the Global Fund's policy documents and guidelines are reviewed. This is followed by an in-depth study of 35 Global Fund proposals from five African countries: Ethiopia, Ghana, Kenya, Malawi and Tanzania. The discussion presents specific human resources interventions that can be found in proposals. Finally, the comments on human resources interventions in the Global Fund's Technical Review Panel and the budget allocation for human resources for health were examined. Policy documents and guidelines of the Global Fund foster taking account of human resources constraints in recipient countries and interventions to address them. However, the review of actual proposals clearly shows that countries do not often take advantage of their opportunities and focus mainly on short-term, in-service training in their human resources components. The comments of the Technical Review Panel on proposed health system-strengthening interventions reveal a struggle between the Global Fund's goal to fight the three targeted diseases, on the one hand, and the need to strengthen health systems as a prerequisite for success, on the other. In realizing the opportunities the Global Fund provides for human resources interventions, countries should go beyond short-term objectives and link their activities to a long-term development of their human resources for health. Background In the midst of accelerating advances in medicine and health technologies and a growing number of effective and affordable interventions, several low-income countries have experienced a decline in their health outcomes. Rates for child mortality are increasing and life expectancy is decreasing. There is a consensus emerging that one of the key ingredients to achieving improved health outcomes is stronger health systems, including an adequate health workforce. Recent studies also show evidence of a direct and positive causal link between numbers of health workers and health outcomes. The recently launched World health report 2006 suggests a minimum worker density threshold of 2.3 workers (doctors, nurses and midwives) per 1000 population necessary to achieve the United Nations Millennium Development Goals (MDGs) for health. WHO estimates a shortage of 2.4 million doctors, nurses and midwives worldwide, reaching a total of 4.3 million if all the workers required to manage and support their activities are included. The cost of training enough people to meet the shortfall by 2015 is on the order of USD 92 billion, and thereafter a minimum of USD 39 billion per year is required to pay their salaries. The health workforce gap is one of the major bottlenecks to the success of global health initiatives. For example, two studies found that health workforce constraints were the key issue in successful implementation of the Global Alliance for Vaccines & Immunization (GAVI) programme. Findings of these studies have implications for the Global Fund, as the Global Fund's approach has many similarities with GAVI. In light of the GAVI experience, one author suggests that: "staff shortages, poorly motivated staff, and the lack of resources for carrying out routine supervisory visits will inevitably be an obstacle to strengthening... services for the three diseases". This review explores how the Global Fund addresses these challenges. It is outlined along the following key questions: Do the Global Fund policy and its guidelines offer any opportunity for investments in human resources for health? To what extent do countries make use of these opportunities? This review first explores possibilities for investment in human resources in the Global Fund's policy documents and guidelines. Next, five African countries were selected to study how the Global Fund proposals addressed human resources constraints. Thirty-five proposals from Ethiopia, Ghana, Kenya, Malawi and Tanzania that were submitted and approved for funding in the first five Rounds were reviewed in detail. Finally, the comments on human resources interventions in the Global Fund's Technical Review Panel and the budget allocation for human resources for health were examined. Purpose and processes of the Global Fund Since 2001, the Global Fund to Fight AIDS, Tuberculosis and Malaria has played an important part in the world's commitment to improve health. The Global Fund was created fundamentally as a financing agency. Its purpose is: "to attract, manage and disburse additional resources through a new public-private partnership that will make a sustainable and significant contribution to the reduction of infections, illness and death, thereby mitigating the impact caused by HIV/AIDS, tuberculosis and malaria in countries in need, and contributing to poverty reduction as part of the Millennium Development Goals". The scope of the Global Fund comprises a substantial increase in coverage of proven and effective interventions for both the prevention and treatment of the three diseases. Activities supported by the Fund may also include efforts to strengthen health systems and human resources capacity. One principle of the Global Fund is to support the integration of proposals into existing national and international disease programmes and implementation strategies. National ownership based on the equal partnership of local private and public stakeholders is another principle. The Fund seeks the active participation of local representatives from civil society and of those directly affected by the three diseases. Grants are disbursed to and implemented by a local principal recipient chosen in a countryled process. The flow of funds is closely linked to performance and progress, measured according to clearly defined monitoring and evaluation procedures. Countries interested in receiving support from the Fund establish a local public-private partnership, the Country Coordinating Mechanism (CCM). The CCM prepares the proposals based on local needs and gaps in national programmes. To ensure an independent transparent process for the approval of proposals, the Global Fund relies on a Technical Review Panel (TRP) appointed by the board. International health and development experts assess the grant proposals and refer their recommendations to the board. Research on the Global Fund's impact on health systems Even though the Global Fund attracts the interest of external researchers, few studies take a health systems perspective into account. An effort to study the Global Fund's long-term effects on health care systems is undertaken by the "Systemwide Effects of the Fund (SWEF) Research Network". SWEF seeks to understand how monies disbursed by the Global Fund and other significant funding agencies, such as the Multi-Country HIV/AIDS Programme (MAP) and the United States President's Emergency Plan for AIDS Relief (PEPFAR), affect the broader health care system in the recipient countries. SWEF has begun indepth case studies in eight countries to study effects upon the policy environment, the public/private mix, human resources and pharmaceuticals. The first SWEF observations allude to the fact that skills, motivation and distribution of health workers are likely to be affected by Global Fund projects. Interim findings from Benin, Ethiopia and Malawi published in 2005 found that so far, human resources constraints have mainly been observed at the programme management level, which is probably due to the fact that implementation is still at a very early stage. At the programme management level it has also been noted that the employment of staff for the Global Fund on short-term contracts with salaries substantially higher than regular government employees has in some cases led to the exodus of technical staff from the ministry of health (MoH). At the health service delivery level, similar incidences have been observed with regard to health workers moving into higher-paid disease-specific positions. This could potentially weaken community-based services that are not related to one of the three target diseases. One of the preliminary conclusions of the SWEF ongoing research is that health workforce planning must be strengthened in the context of scaling up activities The Global Fund's guidelines on human resources Health systems requirements Since Round 1, proposals have been required to take into account the "institutional and absorptive capacity" of the country. Since then, the Global Fund has entered a kind of tightrope walk by focusing on its clearly defined goal to fight the three targeted diseases, but at the same time recognizing that adequate capacity of the health system is a prerequisite for any successful intervention. As the guidelines have developed over the six Rounds, the necessity of functioning health systems and sufficient provision of human resources have been increasingly emphasized in the context of the required absorptive capacity of the recipient country. Over the past Rounds, possibilities to support the health system through Global Fund resources are extended, but always require a link to the targeted diseases. According to guidelines for Round 1, a proposal "may include interventions to improve national capacity associated with the delivery and monitoring of programmes but should not have capacity building as its main focus." Since Round 4 the linkage of the proposed interventions to health systems strengthening has partly lost its optional character: "proposals should include the broader cross-cutting aspects of systems development that benefit the fight against AIDS, tuberculosis and/or malaria, and should describe how the proposal will have positive system-wide effects." Examples include human capacity development and infrastructure development. Round 4 explicitly permitted an "integrated" proposal addressing a comprehensive response to the three diseases that focuses on system-wide effects. The guidelines for Round 5 also provide the opportunity for a "health system strengthening" proposal. In Round 6, however, the guidelines no longer allow separate health system proposals; activities for the strengthening of health systems now must be integrated within the disease-specific proposals. Integration with national plans and the potential for sustainability Since Round 1, the guidelines have strongly encouraged links to existing national efforts to develop sustainable health systems and broader poverty reduction strategies. In order to demonstrate the potential for sustainability, countries are asked to describe links between the component and broader development policies and programmes such as Poverty Reduction Strategies or Sector-Wide Approaches. Round 4 adds requirements concerning the integration of the proposal with broader efforts to reach the Millennium Development Goals (MDGs) and other international initiatives. Concerning the integration of proposals into public expenditure frameworks, guidelines gradually require more specific information. Round 5 guidelines ask to demonstrate "the ability to service recurrent expenditures" and in Round 6 countries need to describe "any relevant constraints e.g. budget or public sector spending ceilings". With regard to linking the proposals to national health planning, the guidelines for proposals do ask for diseasespecific national plans, whereas integration into an over-all health development plan is not obligatory. However, the recent guidelines ask that proposals report if the current systems will be able to achieve and sustain any planned scale-up of intervention and what constraints exist. If health system constraints exist and adequate means to fully address these constraints are presently not available, applicants are "encouraged to include funding in respect of such activities" to strengthen the health system. Integration with national plans is an important aspect of the potential for sustainability that characterizes a successful proposal, according to criteria laid down for the Technical Review Panel of the Global Fund. High-level political involvement and commitment with respect to the allocation of national resources is another aspect of sustainability. In general, resources sought from the Global Fund should build on or scale up existing efforts and fill existing gaps in national budgets and funding from international donors. Complementarity and supplementation of the planned interventions must be demonstrated. It is a Global Fund policy that its disbursements should not replace existing national and international resources. Restrictions and special requirements for human resources for health Since Round 2, the budget section of the proposal form has included a section on human resources that states: "In cases where human resources (HR) is an important share of the budget, explain to what extent HR spending will strengthen health systems capacity at the patient/target population level, and how these salaries will be sustained after the proposal period is over." This very specific request for sustainability of salaries (which cannot be found for any other activities financed by the Global Fund) is closely linked to macroeconomic policies. It requires an intersectoral approach ensuring government commitment. In addition, proposals are asked to demonstrate a direct link between spending on the health workforce and effects on the patient/target population. Studies have shown this link at the macro level demonstrating a correlation between health outcome indicators (i.e. child mortality) and the density of the health workforce. However, it is quite a different challenge to prove such a direct connection for specific interventions to invest in human resources for health that are described in Global Fund proposals. One of the main obstacles is certainly the long time lag, which is typical for many investments for the health workforce, such as training of health professionals. In the past, the Global Fund guidelines did not clearly define how the link between the health systems-strengthening activities and the disease-oriented goals should be made. It is only in Round 6 that the guidelines aim to provide a common framework for the required linkages between disease-specific and health systems interventions. In summary, though, it is noteworthy that the Global Fund guidelines and proposal forms are in principle flexible and open to funding health workforce interventions. No explicit restrictions are evident for both short-term and long-term investments in human resources for health. Monitoring and evaluation Since Round 4 the Global Fund has also provided a Monitoring and Evaluation Toolkit that was designed to support countries in measuring and reporting programme progress. The initial health workforce indicators suggested were the number of people trained and the number of health personnel with adequate supervision and motivation. The revised second edition, from January 2006, offers a wider range of indicators, such as annual output of trained health workers, number of health facilities fully staffed according to national standards or even patient satisfaction. Still, there is a heavy focus on indicators related to training. Suggested indicators are far from comprehensive, and other indicators to follow up the impact of health workforce interventions may be used. This allows for countryspecific indicators. Involvement of health workforce stakeholders In more than half of the reviewed proposals, some members of the CCM can be identified as stakeholders for human resources issues. This includes mainly representatives of academic institutions, professional associations or the ministry of education. For the majority of the members, however, and particularly with regard to the ministry of health, it is not possible to determine whether a person is an expert in one of the diseases or in health systems or in another category altogether: the composition of CCMs cannot be traced. Nevertheless, it cannot be denied that the composition of the CCMs and the choice of experts to write the proposals influences the content and perspective of the proposed projects. To look further into this aspect, the list of documents attached to the proposals was examined. Here a noteworthy number of documents related to national disease strategies was included in most cases, but rarely any strategic paper on health systems development or human resource development. The report of the Technical Review Panel on Round 5 proposals states: "The TRP is concerned that CCM composition has been built up based on the three diseases, so that many CCMs may lack the expertise to develop strong (health systems) proposals." Guidelines on the CCMs have been changed several times, but the participation of cross-cutting experts has not yet entered the list of requirements for the CCM. Interestingly, for the Technical Review Panel of the Global Fund, a balanced composition of experts with either a diseasespecific or health systems background is carefully monitored and reported. General health workforce concerns and the national context The majority of proposals do take human resources issues into consideration. Out of the 35 reviewed proposals, 25 proposals across all five reviewed countries explicitly recognize human resources constraints. Nine out of 10 that did not tackle the issue are Round 1 and 2 proposals. This suggests a strong and increasing awareness of human resources as essential for the successful implementation of the programmes. Country examples may illustrate the sense of urgency: Tanzania: "The human resource shortage in the health sector is a major concern in Tanzania." (Round 4 Proposal HIV/AIDS) Malawi: "Presently, human resources constraints limit the capacity to implement available malaria control measures"...."Past experience indicates that development efforts failed because of lack of attention to human and institutional capabilities" (Round 2 Proposal Malaria); and even stronger formulations in Round 5: "The health system's civil service suffers from one of the worse staffing shortages in Africa creating a near breakdown in capacity to deliver basic level of health care especially in rural areas. Unless the crisis is resolved, Malawi will not have a sustainable public health system or capacity to expand treatment programs such as ART." (Round 5 Proposal HIV/AIDS and HSS (health system strengthening)) Ethiopia: "The health system is affected by serious shortage of qualified human resources coupled with high turnover of trained staff." (Round 4 Proposal HIV/AIDS) Ghana: "Retention of service providers, their motivation and moral boosting will be crucial for the reduction in the disease burden of the three diseases." (Round 1 Proposal HIV/AIDS) Kenya: "Human resource capacity strengthening is key to all the efforts and will subsequently improve not only the delivery of TB services but also the overall functioning of the general health system in supported areas." (Round 4 Proposal Tuberculosis) In accordance with the Global Fund's guidelines, nearly all proposals describe their links with national disease programmes and how the proposed interventions will fill identified gaps. However, broader development plans for the strengthening of health care systems are rarely included. Only seven out of 35 proposals mention a human resources development plan or a similar strategy. Looking at the development over all five Rounds, the importance of a human resources development plan is emphasized more clearly in the later Rounds. In Ghana for example, the Round 5 proposal for tuberculosis requests the recruitment of three experts to work in the health-workforce development department of the MoH to develop a TB-specific human resources plan. In some cases, over the short timespan of the 5 rounds, progress within a country can be observed. Kenya's Round 2 tuberculosis proposal states that the National Leprosy and Tuberculosis Programme is participating in the TBCTA (Tuberculosis Coalition for Technical Assistance) training initiative that is exploring ways to develop multiyear health workforce development plans. The Round 4 HIV/AIDS proposal expresses a specific intention to develop a human resources plan. In Malawi the development becomes even more visible. Whereas the Round 1 HIV/AIDS proposal mentions that health delivery systems must be developed through incorporating human resources development, the Round 5 proposal on HIV/AIDS and health system strengthening (HSS) can draw on a fully developed and costed Emergency Human Resource Plan that is already partly funded by other donors; consequently the country asks the Global Fund to fill funding gaps. Specific human resources interventions in the proposals Training Training is one of the main topics in proposals. More than 90 % of the 35 proposals reviewed propose activities in this respect. The main focus of the training is in-service training for all types of health care workers. There is training of shopkeepers in good drug dispensary practices; training for prison services health staff; epidemiology courses for staff in public health institutions; data management and reporting skills for the monitoring and evaluation of the programme; and international training in public health for senior programme staff. The Round 2 HIV/TB proposal for mainland Tanzania, for example, states: "Twenty types of interrelated training activities for more than 10 000 participants from community groups/ sites to national facility levels will be conducted." Noticeable are the numerous training programmes for community health workers, lay counsellors, persons living with HIV/AIDS and other volunteers or peer educators who may support health care professionals on the front line. Ethiopia's Round 2 malaria proposal, for example, is to train more than 4000 community health workers and nearly 35 000 mother coordinators at the community level. A large proportion of non-health workers are also included in training for peer education among teachers and teenagers, people in the workplace, mobile population, sex workers and men having sex with men. Most proposals provide an elaborate training strategy, often with a snowballing system in which the first generation is trained with the assistance of local academic institutions or developing partners. However, none of the reviewed proposals link their plans for capacity development to a coordinated country training plan. In fact, such plans seem to be lacking. Furthermore, only about 20 % of the suggested training programmes include the assessment of staff training needs, an evaluation of on-the-jobimpact of new skills or any other kind of follow-up of training activities. A good practice example is provided in the Round 5 HIV/AIDS proposal from Ghana. The provision of follow-up after training includes "using creative approaches such as skills application plans, transfer of learning from classroom to real work life, and the provision of job aids (fact sheets, checklists) and other learning/work guides". Support of pre-service training institutions Few interventions that support pre-service training institutions can be found in the proposals. Updating curricula is a frequent activity but seems to be mainly an isolated exercise for the special Global Fund training programmes that are often limited to in-service training. Chances to revise curricula for all national training institutions are rarely taken; clearly this is also a matter of competence and involvement of stakeholders. One good practice example can be found in Kenya, where the Round 2 proposal on tuberculosis states that the need to incorporate TB control in the teaching curriculum of doctors and nurses is recognized and the National Leprosy and Tuberculosis Programme "has made this recommendation to the institutions." In Round 4 the inclusion of TB/HIV in curricula for middle-and high-level health institutions is a specific activity of the proposal with "number of schools using revised curricula" as an indicator for success. Only one of the reviewed countries suggests a strategy that addresses the very root of the lack of human resources in supporting training institutions. Already in the Round 1 proposal on HIV/AIDS Malawi planned an expanded increase of medical students at its college of medicine and the establishment of a faculty of pharmacy. In the Round 5 HSS component, a strong rationale for the inclusion of pre-service training in the proposal is provided. It is argued that the general lack of labour supply and the critical need of HR managers plus insufficient career opportunities are the main reasons for low staff morale and migration among health workers and that this can be effectively addressed only through strengthening of training institutions. The four main training institutions play a pivotal role in this approach and have already provided "detailed plans for expansion and made output commitments to the MoH". Recruitment The analysis of recruitments planned in the proposals shows a common pattern. In 80 % of the proposals some recruitment is planned for the programme management level. New posts are created for administrators, accountants, procurement and logistics experts or similar positions. Staff are often hired only for the project time and may include external consultants, or secondments from the ministry of health. Few of these positions have a longterm perspective, although they might be supported by other donors after the proposal period is over, as suggested in one of the proposals. On the other hand, greatly increased recruitment is planned for volunteer community health workers and volunteer staff for peer education, counselling and homebased care. In some cases, a small salary for the duration of the project is planned for these volunteers; in other cases non-monetary incentives or no remuneration is proposed. One proposal, for example, defines community health workers (CHW) as being recruited among community volunteers, retired school teachers, traditional birth attendants, etc. These CHW will not be given salaries but "funds will be made available to enable them to carry out their assigned activities" i.e. "enablers" such as free drugs, food packages, travel and transport allowances will be given. Many proposals do not provide detailed information in this matter and further research at the country level is necessary to analyse the use of volunteers and peer educators through Global Fund activities. Apart from an increased pool of volunteers, a common strategy for the programme implementation level is also to obtain the necessary human resources from other partners, such as extended NGOs and faith-based organizations, that are expected to support their own staffing costs with the help of additional donor funding. This review clearly shows that the number of core health workforce categories is rarely increased through Global Fund resources; countries were forced to look for more creative solutions to address human resources constraints. Ghana's Round 5 HIV/AIDS proposal puts it like this: "Scaling up national program entails increasing the number of workers but also extending the type of participants contributing to outcomes." Only in the fourth round can the first example be found in which additional human resources are hired, such as for district TB clinics. The reluctance to recruit human resources for health is caused mainly by restrictions due to public sector freezes, which are mentioned several times in the proposals. An exceptional case of major resources (50% of the requested funds) planned to be used for recruitment and salaries of core health workers such as nurses and physicians is the Round 5 HSS proposal of Malawi. The country justifies its proposal, stating that for Malawi, the "nature of the health workforce crisis is such that the situation warrants measures that might not otherwise be considered as sustainable". Motivation and retention strategies About 60 % of the reviewed proposals consider motivation and retention strategies an important issue. In spite of some proposals that recognize the issue but do not provide any strategies to address it, an overview of the proposals gives a wide range of activities and some creative approaches. Monetary incentives in addition to salaries are a controversial topic and are supported by some proposals and not mentioned by others. Some remuneration strategies for volunteers who will not be paid a salary include travel and transport allowances, provision of motorbikes, and skills-building and training opportunities. Performance based incentives and improved feedback and supervision systems are also suggested activities. The introduction of worker support meetings to help deal with burnout and stress is proposed in support of health care workers dealing with HIV/AIDS patients. One proposal plans to develop a national code of practice to combat stigma. Many health facilities ensure post-exposure prophylaxis and treatment for their staff. In nearly all cases, only one or two single activities are suggested to address motivation issues, but no embedding strategy on motivation and performance is provided and the effectiveness of these activities is neither monitored nor evaluated. As a good practice example the Kenyan Round 4 tuberculosis proposal plans the implementation of a package of incentives to retain staff in hard-to-reach areas. This package includes a limited-stay policy and improved communication and training opportunities. Another comprehensive approach can be found in the Ghana Round 5 HIV/AIDS proposal, where motivation for lab personnel is addressed through a combination of training and capacity development of collaborative research. Again though, no measurement strategy is suggested to evaluate the effects on motivation and performance of health workers or retention rates. Skill mix and regulation A third of the reviewed proposals refer to skill mix as a regulation issue. Most of them tackle the creation of new cadres of volunteer and front-line workers such as lay youth counsellors and care supporters for home-based care. In one case, the formal government approval of a new category of dedicated counsellors as part of the civil service is sought as a Global Fund activity. Some proposals include the adaptation of national standard guidelines and protocols to WHO guidelines, such as the development of guidelines for referral systems in their activities. One outstanding example plans the accreditation of all health services able to provide antiretroviral therapy, according to national guidelines with the support of Global Fund resources. It is obvious, though, that skill mix and regulation of human resources for health is strongly affected by Global Fund activities, especially through the creation of numerous new types of front-line health workers. Long-term effects on health service delivery as well the role of the professional associations in the country coordinating mechanism in this process pose interesting questions for future research. Sustainability According to the Global Fund guidelines, proposals must tackle the issue of health workforce sustainability in cases where the health workforce accounts for an important share of the budget. Several proposals make a statement with regard to sustainability of salaries, but -not surprisingly -cannot provide settled agreements for the time after the proposal period is over. Vague statements may be found, such as "staffing costs will need to be absorbed into the MOHP or other service delivery organization budgets" (Malawi Round 1 HIV/AIDS) or "government should be able to absorb the additional staff" (Kenya Round 4 tuberculosis). The Round 5 HSS proposal by Malawi takes an innovative approach, stating that these human resources investments will "yield significant return with increment upgrading health systems productivity, accessibility and equity". Human resources in the Global Fund's review process of the proposals The Global Fund's Technical Review Panel conducts its review based on the terms of reference laid down by the Global Fund. General characteristics of successful proposals, such as soundness of approach and feasibility, are supplemented by the respective guidelines in force for that specific Round. No specific criteria can be found on health systems or human resources issues, apart from the advice that the potential for the sustainability of the approach should be outlined. Sustainability in this context refers to "the capacity to absorb increased resources such as through innovative approaches to overcoming human resource capacity constraints". This passage was introduced into the guidelines in Round 4 and can also be found in Round 5 and 6 guidelines. Nonetheless, the inability of proposal writers to provide more detailed information on how to ensure sustainability of salaries for human resources did not receive any negative feedback from the TRP. In general the consideration of human resources issues and the integration of human resources components in the proposals is welcomed by the Technical Review Panel of the Global Fund. The TRP counts it as a strength, for example, that "major risks, especially weaknesses in the health system, including human resources" are named in a proposal and specific measures are taken to address them. An inadequate consideration of human resources constraints has repeatedly led to critical comments of the TRP and doubts about the successful achievement of the project goals. A lack of discussion on how ambitious scaling-up targets can be achieved, given the countries' present human resources constraints, is one of the reasons that has led to a rejection of proposals. In proposals that do address human resources issues, key weaknesses are a lack of a comprehensive situation analysis for the health workforce and a lack of overall health workforce development plans. Interventions for the strengthening of the health workforce mainly address single issues without providing an embedding strategy for long-term success. Some activities, for example, aim at a better working environment and improved health worker performance, but no comprehensive motivation and retention strategies are provided along these lines. Numerous training activities are suggested, but they are not part of a coordinated country training plan. In addition, it often remains unclear how the proposed interventions for the health workforce will have positive effects on the patient/target population. Countries struggled to propose clear approaches for the monitoring and evaluation of the success of their health workforce interventions. This is also due to the poor standard of current human resources information systems (HRIS), although this issue has not been specifically addressed by the TRP. Table 1 illustrates an overview of Round 4 and 5 TRP comments on human resources interventions: Human resources in integrated and health system proposals In Round 4, the guidelines introduce a specific option to address cross-cutting health systems issues, including human resources constraints, in an integrated proposal. This option, however, was not widely used by countries. There were only six integrated proposals out of 171 proposals in Round 4 and they were also less successful than the average in the Technical Review Panel. None of them was recommended for funding, although four may be resubmitted in subsequent rounds. One of the integrated proposals from Uganda aims specifically at strengthening human resource development. Expanding the capacity of training institutions, in-service training, an incentive system for hardship areas and strengthened human resources management at all levels were the main features of this outstanding proposal. The TRP classified this proposal as not recommended for funding, reasoning: "it appeared questionable to us that the GFATM is asked to fund a relatively mid-long-term multi-sectoral programme, rather than focusing on gaps that urgently need to be addressed to implement the HIV, malaria and TB programmes that are ongoing... with immediate impact on the delivery of services." Here, the Global Fund's goal for a direct impact in fighting the three targeted diseases clearly outweighed the necessity for long-term investments. For Round 5, the Global Fund decided to invite specific health system strengthening (HSS) components and countries responded more readily: 30 out of the 202 components reviewed in Round 5 were HSS proposals. Again, their success rate was well below the average: only three proposals (10%) are recommended for funding, compared to a 31 % overall success rate. With regard to budgets, this translates into the approval of USD 43 million for HSS proposals out of a total budget of USD 726 million in Round 5. A review of the content of the 30 HSS components shows that human resources is by far the most commonly tackled issue (20 HSS proposals), followed by interventions for the development of health information systems. Specific improvement of the human resources information system cannot be found as a stand-alone intervention in any of the HSS proposals, but is usually part of the overall strengthening of the health information system. An overview of proposed interventions for strengthening health systems can be found in Figure 1 and Table 2. Comments of the Technical Review Panel in Round 5 show quite positive reactions to human resource interventions planned in the HSS proposal. Two of the three successful proposals have a major human resources component. The HSS component from Rwanda assigns 80% of requested funds to human resources and training, the HSS component from Malawi even more than 90% of the budget. Commenting on the HSS proposal from Malawi the TRP reflects that although the human resources constraints were recognized as crucial in Round 1, "at that time the Global Fund was not keen to fund health system components", indicating that the approach of the Panel has undergone some change since the early Rounds. In their report on Round 5 proposals, the TRP found that the invitation of separate HSS proposals was given "insufficient consideration". Applicants were unsure about the precise scope of HSS proposals and "were not given any specific guidance on what an effective linkage between HSS and a disease component should or could look like". This resulted in a high failure rate of Round 5 HSS proposals due to objectives that were too vague and ambitious and a contrived and superficial linkage between HSS activities and the specific diseases. The Technical Review Panel summarizes that the poor quality of these proposals "reflects a confusion in the GFTM as to the precise mandate of the Fund in relation to HSS proposals." The Panel stated a need to refine the Fund's mandate in relation to HSS proposals and raised the question whether to retain HSS proposals as a separate category, especially as the proposal forms had been found "largely unsuitable for the submission of HSS proposals". Consequently, the option of separate HSS proposals was abolished for Round 6; health system interventions must now be completely integrated into the specific disease component section. Budget allocations for human resources The Global Fund's reports on the distribution of grants show that 55% of the resources committed so far have been assigned to sub-Saharan Africa; 58% of the total funding is allocated to HIV/AIDS. Looking at the budget proportions per expenditure targets reveals a decrease in investment in human resources for health and training in Round 4. A recent study of all malaria components over all four Rounds also reveals a decline in allocation to human resources and training towards Round 4. Then again, in Round 5 there is a slight increase in the budget allocation towards human resources for health and training ( Figure 2). Here it must be taken into consideration that the total amount approved for grants in Round 3 was lower than in the other rounds (Table 3). This evolution of the budget allocation for human resources, however, is not readily accessible for interpretation, especially as the definitions of the budget categories "human resources" and "training" have been changed over time. On the one hand, human resources costs are sometimes hidden within other components: training on the application of a new drug might be included in the category "Drugs", for example. On the other hand, the health workforce budget might include resources for workshops and meetings, for example, without having a specified proportion for human resources within these lump sums. An initial study of the malaria proposals in 14 African countries up to and including Round 3 that was conducted by the Roll Back Malaria Department at WHO Geneva found that of the 25% that were declared as the human resources and training proportion of total funds, only around 10% could still be defined as human resources and training costs after a more detailed analysis. In addition it is noteworthy that in the 35 African proposals reviewed, the proportion of funds assigned to the budget categories "human resources" and "training" summed up to only 16% of funds over five Rounds (human resources 6.6%, training 9.2%). This is considerably lower than the average of 22% for the distribution of the total sum of global funds as shown in Figure 2. This finding is surprising, because all five of the reviewed African countries are frequently identified with issues related to human resources constraints. Distribution of grants per Round by expenditure targets in % (provides summary of amounts for all Rounds to date) Figure 2 Distribution of grants per Round by expenditure targets in % (provides summary of amounts for all Rounds to date). Source:. Conclusion The Global Fund clearly provides the opportunity to invest in health systems strengthening and human resources. Policy documents acknowledge that health system capacities and human resources in particular are a necessary prerequisite for the success of interventions that aim to fight the three targeted disease. Changes over time in the Global Fund's guidelines and proposal forms reflect ever more attention to the strengthening of health systems. However, there has been a struggle to accommodate health systems strengthening with the objectives of the Global Fund and its administrative guidelines. The change made in Round 6 guidelines to abandon the possibility for separate health systems proposals can be interpreted as a consequence of this dilemma. In general, most countries do not sufficiently use the possibilities that the Global Fund provides for health system strengthening and human resources interventions, even though a great majority of proposals recognize human resources constraints as a key to success of future interventions. Most proposals include some activities to address human resources constraints. The most frequent activity is training, focused mainly on short-term, in-service training. Support for pre-service training and training institutions is rare and hardly any long-term strategies are proposed to address the lack of adequately trained personnel. Recruitment plans are also frequently included in the Global Fund proposals. In most cases, however, this is limited to a small number of staff at the programme management level rather than addressing the shortages at the service delivery level. This absence may also be due to the Global Fund's specific requirement to demonstrate the sustainability of salaries financed by the Global Fund after the proposal period is over. In the long run this will make it inevitable to address macroeconomic issues, particularly the restrictions imposed on the public sector by the hiring freeze. The comments of the Technical Review Panel on proposed health system-strengthening interventions reveal a struggle between the Global Fund's goal to fight the three targeted diseases, on the one hand, and the need to strengthen health systems as a prerequisite for success, on the other. Unfortunately, in the past the TRP has tended to favour the Global Fund's disease-related objectives over cross-cutting issues. In realizing the opportunities the Global Fund provides for human resources interventions, countries should go beyond short-term objectives and link their activities to a long-term development of their human resources for health. The existence of a human resources development plan, a clear identification of currents gaps and a linkage between human resources and coverage for the targeted disease have been favourite ingredients in proposals requesting funds for human resources. If human resource development plans are not presently available, the framework of the Global Fund permits support to technical assistance to lay those foundations. One step that can be taken at the country level to take advantage of the opportunities the Global Fund provides for strengthening the health workforce is to ensure the involvement of human resources stakeholders and experts in both the CCM and the proposal writing groups. As a preliminary observation on the development of proposals from Round 1 to 5 it can be said that the activities of the Global Fund have revealed and partly exacerbated some long-standing health system weaknesses, especially with regard to human resources. In order to successfully tap into the Global Fund's resources, countries are forced to provide needs-assessment and clear development plans for their human resources for health. Proposals of the five countries reviewed here clearly show some positive developments towards more comprehensive approaches to address human resources constraints. However countries need more encouragement and support to back up their human resources for health through the framework of the Global Fund. |
<reponame>malliina/scrimage
package com.sksamuel.scrimage.composite;
import thirdparty.romainguy.BlendingMode;
public class GreenComposite extends BlenderComposite {
public GreenComposite(double alpha) {
super(BlendingMode.GREEN, alpha);
}
}
|
Changing OrdersPrimary and Secondary Membrane Transporters Revised Cells are surrounded by membranes, which protect them from a hostile environment and allow the maintenance of an internal compartment with a defined composition. In eukaryotes, one or more membranes surround subcompartments, where biochemical processes can take place under compartmentalized conditions. Countless compounds and signals have to be transmitted across these membranes without compromising their important barrier function. Analysis of the growing number of sequenced genomes has revealed that approximately 30 to 40% of encoded proteins contain at least one transmembrane domain, and approximately 10 to 20% of these proteins are thought to be involved in transport processes. Thus, membrane transporters form the largest functional group of proteins, containing more members than, for example, functional groups encoding proteins involved in regulation, transcription, translation, or energy metabolism (for a database and classification of transport proteins see: http://www.biology.ucsd.edu/~msaier/ transport/). Based on their driving force, three classes of active transporters have been defined (Figure 1). Primary transporters use electrical, light, or chemical energy. Well-known examples are the protontranslocating subunits of the respiratory chain, where electron transfer is coupled to the electrogenic transport of protons or sodium ions, or the photosynthetic reaction centers, where absorption of photons is coupled to translocation of protons. Although these systems are extremely important in biology, most primary transport systems use chemical energy, especially ATP, to transport highly diverse substrates. The ATP-binding cassette (ABC) proteins comprise the largest superfamily of primary transporters. Members of this family drive the import as well as export of a wide spectrum of substrates, covering sugars, amino acids, drugs, lipids, peptides and even large proteins. In contrast, secondary transporters use chemiosmotic energy, such as sodium, proton, or charge gradients created by primaryactive transporters, to power substrate transport. Since the driving force (proton or sodium gradient) is normally directed inwards, antiport mechanisms are generally used to export substrates, while symport mechanisms operate for import. The class of secondary transporters contains many different families, but approximately 25% belong to the major facilitator superfamily (MFS). The third class of active transporters comprises the phosphoenolpyruvate/carbohydrate phosphotransferase systems (PTS), which phosphorylate the carbohydrate substrate during transport, thereby maintaining the concentration gradient. The PTS consist of two general phosphocarrier proteins, Enzyme I and HPr, and a transporter, Enzyme II. The transporter has three domains (IIA, IIB, and IIC), which can be separate proteins or be linked in any combination (Figure 1). This class of transporters contains relatively few members and is only found in bac teria. After many years of research and in an impressive tour de force, the crystal structures of several members of the two largest superfamilies of primary (ABC) and secondary transporters (MFS) were solved; this marked a new era in membrane biology. The structures showed striking similarities. These similarities were further demonstrated for the homodimeric ABC transporter LmrA of Lactococcus lactis, which could be converted from an ATP-dependent transporter to a proton symporter after deletion of the ATP-hydrolyzing domain. This review will focus on what we have learned from similarities and differences in the structures and mechanisms of the two largest families of primary (ABC) and secondary (MFS) transporters. Dr. C. van der Does, Prof. Dr. R. Tamp Institute of Biochemistry, Biozentrum Frankfurt Johann Wolfgang Goethe University Marie-Curie-Strasse 9, 60439 Frankfurt am Main (Germany) Fax: (+49)69-798-29495 E-mail : tampe@em.uni-frankfurt.de Figure 1. Examples of the structural organization of primary (the E. coli maltose transporter MalFGK, blue), secondary (the E. coli lactose transporter LacY, pink), and PTS transporters (the E. coli glucose transporter PtsG, green). The E. coli maltose transporter consists of a maltose-binding protein (MBP), two transmembrane domains (MalF and MalG), and the nucleotide-binding domain (MalK). The E. coli glucose transporter system comprises the general phosphocarrier proteins, HPr and Enzyme I, the cytosolic IIA domain, and the fused IIB and IIC domains. |
He didn’t. I’ve never even heard of Misrata before, but for your whole life it was there on a map for you to find and ponder and finally go to. All of us in the profession—the war profession, for lack of a better name—know about that town. It’s there waiting for all of us. But you went to yours, and it claimed you. You went in by boat because the city was besieged by forces loyal to Muammar Qaddafi (another name you probably never gave much thought to during your life) and you must have known this was a bad one. Boat trips are usually such nice affairs, but not this one. How strange to be out on the water off a beautiful coastline with the salt smell and the wind in your face—except this time you’re headed toward a place of violence and killing and destruction. You must have known that the unthinkable had to be considered. You must have known you might not ever get back on that boat alive.
I’m picturing you and your three wounded colleagues in the back of a pickup truck. There are young men with bandannas on their heads and guns in their hands and everyone is screaming and the driver is jamming his overloaded vehicle through the destroyed streets of that city, trying to get you all to the clinic in time.
Tim, man, what can I say? For the first few hours the stories were confused enough that I could imagine maybe none of them were true, but they finally settled into one brief, brutal narrative: while covering rebel forces in the city of Misrata, Libya, you got hit by a piece of shrapnel and bled to death on the way to the clinic. You couldn’t have known this, but your fellow photographer Chris Hondros would die later that evening.
You and I were always talking about risk because she was the beautiful woman we were both in love with, right? The one who made us feel the most special, the most alive? We were always trying to have one more dance with her without paying the price. All those quiet, huddled conversations we had in Afghanistan: where to walk on the patrols, what to do if the outpost gets overrun, what kind of body armor to wear. You were so smart about it, too—so smart about it that I would actually tease you about being scared. Of course you were scared—you were terrified. We both were. We were terrified and we were in love, and in the end, you were the one she chose.
I’m in the truck with you. I’m imagining those last minutes. You’re on your back watching the tops of the buildings jolt by and the blue Mediterranean sky beyond them. I almost drowned once, and when I finally got back to the beach I was all alone and I just lay there watching the clouds go by. I’d never really thought about clouds before, but there they were, all for me, just glorious. Maybe you saw those clouds, too, but you weren’t out of it yet, and you probably knew it. I know what you were thinking: What a silly way to die. What a silly, selfish, ridiculous mistake to have made.
Don’t think that, brother. You had a very specific vision for your work and for your life, and that vision included your death. It didn’t have to, but that’s how it turned out. I’m so sorry, Tim. The conversation we could have had about this crazy stunt of yours! Christ, I would have yelled at you, but you know that. Getting mad was how we kept each other safe, how we kept the other from doing something stupid.
Your vision, though. Let’s talk about that. It’s what you wanted to communicate to the world about this story—about every story. Maybe Misrata wasn’t worth dying for—surely that thought must have crossed your mind in those last moments—but what about all the Misratas of the world? What about Liberia and Darfur and Sri Lanka and all those terrible, ugly stories that you brought such humanity to? That you helped bring the world’s attention to?
After the war in Liberia you rented a house in the capital and lived there for years. Years. Who does that? No one I know except you, my dear friend. That’s part of Misrata, too. That’s also part of what you died for: the decision to live a life that was thrown open to all the beauty and misery and ugliness and joy in the world. Before this last trip you told me that you wanted to make a film about the relationship between young men and violence. You had this idea that young men in combat act in ways that emulate images they’ve seen—movies, photographs—of other men in other wars, other battles. You had this idea of a feedback loop between the world of images and the world of men that continually reinforced and altered itself as one war inevitably replaced another in the long tragic grind of human affairs.
That was a fine idea, Tim—one of your very best. It was an idea that our world very much needs to understand. I don’t know if it was worth dying for—what is?—but it was certainly an idea worth devoting one’s life to. Which is what you did. What a vision you had, my friend. What a goddamned terrible, beautiful vision of things.
Sebastian Junger is a contributing editor of Vanity Fair.
A version of this article will appear in the June 2011 issue. |
<gh_stars>0
///
/// @file WebCore.h
///
/// @brief The main header for the Awesomium C++ API. This header includes most
/// of the common API functions you will need.
///
/// @author
///
/// This file is a part of Awesomium, a Web UI bridge for native apps.
///
/// Website: <http://www.awesomium.com>
///
/// Copyright (C) 2014 Awesomium Technologies LLC. All rights reserved.
/// Awesomium is a trademark of Awesomium Technologies LLC.
///
#ifndef AWESOMIUM_WEB_CORE_H_
#define AWESOMIUM_WEB_CORE_H_
#pragma once
#include <Awesomium/Platform.h>
#include <Awesomium/WebConfig.h>
#include <Awesomium/WebPreferences.h>
#include <Awesomium/WebSession.h>
#include <Awesomium/WebView.h>
#include <Awesomium/Surface.h>
#include <Awesomium/ResourceInterceptor.h>
namespace Awesomium {
///
/// The severity level for a log message. See WebCore::Log
///
enum LogSeverity {
kLogSeverity_Info = 0, ///< Info message
kLogSeverity_Warning, ///< Warning message
kLogSeverity_Error, ///< Error message
kLogSeverity_ErrorReport, ///< Error report message
kLogSeverity_Fatal ///< Fatal error message, terminates application
};
///
/// @brief The core of Awesomium. You should initialize it before doing
/// anything else.
///
/// This singleton class manages the lifetime of all WebViews (see WebView)
/// and maintains useful services like the network stack, inter-process
/// messaging, and Surface creation.
///
class OSM_EXPORT WebCore {
public:
///
/// Creates the WebCore with a certain configuration. You can access the
/// singleton via instance() later. You can only do this once per-process.
///
/// @param config Global configuration settings.
///
/// @return Returns a pointer to the singleton instance.
///
static WebCore* Initialize(const WebConfig& config);
///
/// Destroys the WebCore singleton and cleans up any active WebViews.
///
static void Shutdown();
///
/// Get a pointer to the WebCore singleton.
///
/// @note Will return 0 if the WebCore has not been initialized.
///
static WebCore* instance();
///
/// Create a WebSession which will be used to store all user-data (such as
/// cookies, cache, certificates, local databases, etc).
///
/// @param path The directory path to store the data (will create the path
/// if it doesn't exist, or load it if it already exists).
/// Specify an empty string to use an in-memory store.
///
/// @return Returns a new WebSession instance that you can use with
/// any number of WebViews. You are responsible for calling
/// WebSession::Release when you are done using the session.
///
virtual WebSession* CreateWebSession(const WebString& path,
const WebPreferences& prefs) = 0;
///
/// Creates a new WebView.
///
/// @param width The initial width, in pixels.
/// @param height The initial height, in pixels.
///
/// @param session The session to use for this WebView. Pass 0 to use
/// a default, global, in-memory session.
///
/// @param type The type of WebView to create. See WebViewType for more
/// information.
///
/// @return Returns a pointer to a new WebView instance. You should call
/// WebView::Destroy when you are done with the instance.
///
virtual WebView* CreateWebView(int width,
int height,
WebSession* session = 0,
WebViewType type = kWebViewType_Offscreen) = 0;
///
/// Set the SurfaceFactory to be used to create Surfaces for all offscreen
/// WebViews from this point forward. If you call this, you are responsible
/// for destroying the passed instance after you Shutdown the WebCore.
///
/// @param factory The factory to be used to create all Surfaces. You are
/// responsible for destroying this instance after you
/// call Shutdown.
///
/// @note: If you never call this, a default BitmapSurfaceFactory will be
/// used and all Surfaces will be of type 'BitmapSurface'.
///
virtual void set_surface_factory(SurfaceFactory* factory) = 0;
///
/// Get the current SurfaceFactory instance.
///
virtual SurfaceFactory* surface_factory() const = 0;
///
/// Set the ResourceInterceptor instance.
///
/// @param interceptor The instance that will be used to intercept all
/// resources. You are responsible for destroying this
/// instance after you call Shutdown.
///
virtual void set_resource_interceptor(ResourceInterceptor* interceptor) = 0;
///
/// Get the current ResourceInterceptor instance.
///
virtual ResourceInterceptor* resource_interceptor() const = 0;
///
/// Updates the WebCore, you must call this periodically within your
/// application's main update loop. This method allows the WebCore to
/// conduct operations such as updating the Surface of each WebView,
/// destroying any WebViews that are queued for destruction, and dispatching
/// any queued WebViewListener events.
///
virtual void Update() = 0;
///
/// Log a message to the log (print to the console and log file).
///
virtual void Log(const WebString& message,
LogSeverity severity,
const WebString& file,
int line) = 0;
///
/// Get the version for this build of Awesomium.
///
virtual const char* version_string() const = 0;
///
/// Get the number of bytes actually used by our allocator (TCMalloc).
///
/// @note: Only implemented on Windows.
///
static unsigned int used_memory();
///
/// Get the number of bytes cached by our allocator (TCMalloc).
///
/// @note: Only implemented on Windows.
///
static unsigned int allocated_memory();
///
/// Force TCMalloc to release as much unused memory as possible. Reducing
/// the size of the memory pool may cause a small performance hit.
///
/// @note: Only implemented on Windows.
///
static void release_memory();
protected:
virtual ~WebCore() {}
private:
static WebCore* instance_;
};
} // namespace Awesomium
#endif // AWESOMIUM_WEB_CORE_H_
|
How Religion Can Remain Relevant in 21st Century
Many, including myself, hunger to find meaning in the world that is not purely mechanistic, and is more satisfying than anything traditional religion can offer.
If religion is to ever be taken seriously by people who are capable of critical thinking, it is going to have to drastically realign its world view to become more compatible with reality as a rational person in the 21st Century experiences it. Here are some suggestions.
Quit being literal
My father, who is as good of a Christian and an open-minded humanist as you’ll ever meet, was recently attempting to explain the significance of mass to my eight year old daughter. What prompted the conversation was that my daughter asked, “Why do they eat crackers and drink wine in church?”
My father explained that it was to commemorate the last supper. “What was the last supper?” my daughter asked.
My father explained that it was the last meal that Jesus and his friends had together before he was killed by the Romans. “He was killed?” my daughter asked.
“Yes,” my father continued, “He died on the cross for our sins and was then put in a tomb. A little while after being put in a tomb, Jesus came back to life and went up to heaven.”
My daughter looked at my father quizzically and deep in thought. Finally she spoke. “Do you really believe that?” she asked incredulously.
Quit anthropomorphizing God
You know how the opening chapters of The Bible are full of terrifyingly dull begetting? Here’s some begetting of my own. Horace begot Zeus. Zeus begot Jupiter. Jupiter begot God. And none of them are that different from each other. Michelangelo’s God on the ceiling of the Sistine Chapel, all muscled up in a display of potent masculinity, looks a hell of a lot like his Greek and Roman predecessors.
The real sin of anthropomorphizing God is that God becomes a psychological crutch upon which we can project our shortcomings and thereby make them divine. If I believe that women are inferior to men, then so does God. If I harbor prejudices against other races and creeds, then God does, too. If I am a football player and believe God is firmly on my side, then he is against the team that my team just beat.
I’m reminded of the old comedy skit where a sports reporter is interviewing a quarterback after his team had crushed the Kansas City Chiefs. The quarterback thanks God for the victory. The reporter asks, “So, why do you think God hates the Kansas City Chiefs so much?”
If there is a God, a prime mover, the fount of all that was, is and will be–the alpha and omega–we need to give up assuming we can know who God is. It’s too big, too complex, and we aren’t smart enough to come to any pat conclusions on that front. It would be like a dust mite assuming it knows my shape and motives by the view of things in a wrinkle of skin, between my toes.
Oh, hell, did I just anthropomorphize God with that analogy?
Quit hating on science
That has to stop. Science has its method, and that method has served us exceedingly well, and will continue to do so, so long as we have ethical scientists.
All that occupies this world does so between the very small and the gigantic. At the edges of both, science falls apart. Take the journey to the edges, which is paved all the way by science, and there you will find more mystery than a soul can manage.
Science’s expedition is to answer how. Religion is the why. So long as we have mouths with which to ask “why,” mystery and religion are alive.
“Science without religion is lame, religion without science is blind.” ̴Albert Einstein
Go inside
Or, to state it in the negative, quit looking out to the world for salvation. Quit waiting for the second coming, Armageddon, the apocalypse, a band of angels or an anthropomorphic God to lift you up and skyward, or the winning lottery ticket you have prayed hard for. Sit quietly and look inside. If there is anything divine in this world, that’s where it will be found, as close to you as you are to yourself.
Religion needs to get more mystical and contemplative, and employ technologies and methods to help achieve inner exploration.
Jesus repeatedly used the phrase “kingdom of heaven.” Take “kingdom of heaven” and replace it with “inner divine wisdom” and see what you get.
“Truly I tell you, unless you change and become like children, you will never enter [inner divine wisdom].”
Get cozy with uncertainty
And that’s what it is really all about. There is no shame in admitting that we can’t know it all–it’s not possible. Paradoxically, I find the more I know, the more I don’t, but the more I want to. I don’t think God would have it any other way.
Oops. Am I anthropomorphizing again? |
NEW YORK — First-time documentary filmmakers Tina Brown and Dyana Winkler lugged their cameras to Central Park in New York one day to capture the last few people still passionate about roller skating. Rinks across the country were gone. The activity seemed dead.
They may have come for a funeral but they found something else entirely. Two young African-American skaters approached them and asked them what they were doing. “They said, ‘Skating’s not dead. It just went underground,”‘ Winkler recalled.
Winkler and Brown decided to go find it. Five years and 500 hours of footage later, they’ve emerged with the HBO film “United Skates,” a fascinating look at the rich African-American subculture of roller skating, which is under threat.
“We hope that our viewers will learn something they didn’t know about, fall in love with something they didn’t know about, and maybe be compelled to care enough to protect it,” Winkler said.
The documentary explores how roller rinks were the sites of some of the earliest fights of the civil-rights era and how they later became the launching pads for hip-hop artists.
It shows how unofficial segregation lives on, with so-called “adult nights” that feature metal detectors and masses of police, something not used when whites come to skate. It also shows how rinks are being closed as communities chase more revenue by rezoning for retail use.
“There’s a bigger story to tell and we can use the joyous beauty of roller skating as the sugar to spoon-feed some of these bigger issues. That’s when we started to peel back the layers,” Winkler said.
That day in Central Park changed the trajectory — and the lives — of the filmmakers. The young skaters they met invited the women to come and see what had happened to skating. And so they got on a night bus to Richmond, Virginia.
The duo — one Australian, one American — approached a roller rink at midnight. It was far from funereal: There was a line down the block, music was pumping, skaters were dressed to kill and everyone seemed to know each other.
“We stepped into this world,” said Winkler.
They soon learned that each city had different skate dance styles — Baltimore has “Snapping,” Atlanta has the “Jacknife” and in Texas you do the “Slow Walk” — and how such a tight fellowship among skaters is forged that they will fly across the country to get together.
Embraced by the community, Winkler and Brown never paid for a hotel room or car rental or a meal while crisscrossing the country interviewing some 100 skaters. The skaters themselves opened their homes and drove them around.
The documentary features interviews with hip-hop legends like Salt-N-Pepa, Coolio and Vin Rock of Naughty by Nature. John Legend is an executive producer and the film received the Documentary Audience Award at the Tribeca Film Festival.
The cameras also follow Reggie Brown, a roller-skating ambassador and community advocate. In a phone interview, he explained that roller skating teaches patience, athleticism, purpose, positive reinforcement, determination — and getting up after a fall.
“United Skates” is a documentary made partially by the subjects themselves. Winkler and Brown, who began the project as beginner skaters, enlisted skaters to shoot scenes and used their rink skills to help capture footage.
The cameras capture one suburban Chicago family-owned rink’s gut-wrenching decision to shut its doors — among thousands that have done so in the past decade — and the filmmakers are not shy about hoping their film can stem the tide of closures.
“Obviously if we could save one rink, if we could have one rink reopen because of this film, that’s a huge step forward for this community and we hope that will have a ripple effect,” said Brown. |
President Barack Obama shared a fist bump with a Texas restaurant employee who urged him to do more to promote gay rights in the United States.
Daniel Rugg Webb, a comedian, artist and musician, as well as an employee at Franklin Barbecue in Austin, seized the opportunity to make his voice heard when he found out Potus was visiting the food joint on Thursday, The Austin Chronicle reported.
When the President approached the till, Webb, 32, cast his hand down onto the counter and said: “Equal rights for gay people!”
We’ll tell you what’s true. You can form your own view. From 15p €0.18 $0.18 $0.27 a day, more exclusives, analysis and extras.
Obama responded: “Are you gay?!
To which Webb replied: “Only when I have sex.”
Laughing, the President told Webb to “bump me” and the pair knocked their fists together.
Shape Created with Sketch. Highly Improbable, But Definitely Real, Famous Friendships Show all 12 left Created with Sketch. right Created with Sketch. Shape Created with Sketch. Highly Improbable, But Definitely Real, Famous Friendships 1/12 Barack Obama loves to let his hair down occasionally with a B ball game and a quick blast of hip-hop. Which is probably why he gets on famously with... AFP Getty 2/12 ... Jay-Z, who even admitted he gets texts from the president occasionally. The true mark of friendship. That and a Cuban holiday. Although Obama denied any involvement in the couple's approval to vacation in the embargoed country. Rex Rex Features 3/12 It seems unlikely that many famous faces would admit to being friends with North Korean dictator Kim Jong Un, thanks to his appalling human rights track record. But one man loves him above all others... Reuters REUTERS/Kyodo 4/12 ... Dennis Rodman. "He's my friend and I love him," the former NBA star told a group of journalists at Beijing airport. He was on his way to N.Korea for the dictator's birthday B ball game. Getty Getty 5/12 Wikileaks founder Julian Assange has many a celebrity admirer. But there is one UK-based musician who he lunches with at the Ecuadorian Embassy more than most... AP AP 6/12 ... Rapper MIA. In fact, Assange recently introduced the star via a web link to her audience during a recent gig in New York. Getty Getty Images 7/12 Elton John's a popular man, with one of the most enviable address books in the entertainment industry. But you may be surprised to know... Getty Getty Images 8/12 ... About his highly unlikely but definitely real friendship with Chris Brown. “I met Elton about a year ago, and he's been a good friend to me," he told Page Six in March 2013. Reuters Reuters 9/12 Equally surprising? Former presidential candidate Senator John McCain's random kinship with... AFP AFP 10/12 ... Snooki. Yes, as in Jersey Shore star Snooki. “How's my friend Snooki?” he asked an MTV reporter in 2012... Before he added: "Congratulations again to her on the addition to the family and I still agree with her that we should never, never tax tanning beds. That was one of the worst things about Obamacare that could possibly have happened." Tactical friendship perhaps? Getty Getty 11/12 Bette Midler is a Hollywood legend. So many were surprised to learn that she had forged an unlikely friendship with... Getty GETTY IMAGES 12/12 ... Erm, rapper 50 Cent? “He has really made my life worth living. 50 has been with me through thick and thin," she told MyParkMag. The pair met during a restoration project in Jamaica Queens. Press 1/12 Barack Obama loves to let his hair down occasionally with a B ball game and a quick blast of hip-hop. Which is probably why he gets on famously with... AFP Getty 2/12 ... Jay-Z, who even admitted he gets texts from the president occasionally. The true mark of friendship. That and a Cuban holiday. Although Obama denied any involvement in the couple's approval to vacation in the embargoed country. Rex Rex Features 3/12 It seems unlikely that many famous faces would admit to being friends with North Korean dictator Kim Jong Un, thanks to his appalling human rights track record. But one man loves him above all others... Reuters REUTERS/Kyodo 4/12 ... Dennis Rodman. "He's my friend and I love him," the former NBA star told a group of journalists at Beijing airport. He was on his way to N.Korea for the dictator's birthday B ball game. Getty Getty 5/12 Wikileaks founder Julian Assange has many a celebrity admirer. But there is one UK-based musician who he lunches with at the Ecuadorian Embassy more than most... AP AP 6/12 ... Rapper MIA. In fact, Assange recently introduced the star via a web link to her audience during a recent gig in New York. Getty Getty Images 7/12 Elton John's a popular man, with one of the most enviable address books in the entertainment industry. But you may be surprised to know... Getty Getty Images 8/12 ... About his highly unlikely but definitely real friendship with Chris Brown. “I met Elton about a year ago, and he's been a good friend to me," he told Page Six in March 2013. Reuters Reuters 9/12 Equally surprising? Former presidential candidate Senator John McCain's random kinship with... AFP AFP 10/12 ... Snooki. Yes, as in Jersey Shore star Snooki. “How's my friend Snooki?” he asked an MTV reporter in 2012... Before he added: "Congratulations again to her on the addition to the family and I still agree with her that we should never, never tax tanning beds. That was one of the worst things about Obamacare that could possibly have happened." Tactical friendship perhaps? Getty Getty 11/12 Bette Midler is a Hollywood legend. So many were surprised to learn that she had forged an unlikely friendship with... Getty GETTY IMAGES 12/12 ... Erm, rapper 50 Cent? “He has really made my life worth living. 50 has been with me through thick and thin," she told MyParkMag. The pair met during a restoration project in Jamaica Queens. Press
The exchange was captured in a photograph, which has since been widely shared on the internet. Obama is reported to have spent $300 (£175) in the restaurant.
Webb told the newspaper: “That's my favourite part because it was cool to get a joke in. In all the photos [all over the internet], I look like a dead fish, but it was cool. I do stand-up, so it was nice to have some interaction based on, hopefully, something funny.
“If Rick Perry (the governor of Texas) would've walked in, I would have lost my job. I would've taken that old queen to town,” Webb said.
But Webb added that, although he recognises the level of social progressiveness in the US, he hopes that the President will do more to address the anti-gay views of many of his political contemporaries, including those of Perry.
Video: 2012: Obama makes statement on gay marriage
Webb told Buzzfeed: “It would be interesting if he could call some people out for it. People can use a lot of things - religion, freedom of speech - to be anti-gay, but I need people to understand you can call people out for civil rights things.”
He added: “Even Obama laughing at something as, hopefully, acceptable as sexuality can show the difference.”
We’ll tell you what’s true. You can form your own view.
At The Independent, no one tells us what to write. That’s why, in an era of political lies and Brexit bias, more readers are turning to an independent source. Subscribe from just 15p a day for extra exclusives, events and ebooks – all with no ads.
Subscribe now |
Villa with spectacular view of the sea with swimming pool and rooftop garden is in full Monte Estoril and garage for 3 cars level 0 - garage 50sqm, storage room, suite and entrance hall. 1st Floor - 3 suites 2nd Floor - living room, dining room, kitchen and is the social roof Terrace with swimming pool and garden with 110sqm a few minutes from The beach of Estoril and the train station.
Villa with spectacular view of the sea with swimming pool and rooftop garden is in full M, Lisbon, Portugal is a 330ft2 Lisbon luxury House listed for sale 1,850,000 EUR. This high end Lisbon House is comprised of 4 bedrooms and 4 baths. Find more luxury properties in Lisbon or search for luxury properties for sale in Lisbon. |
/**
* Hide all the view parts of {@linkplain AbsLoadingLayout}.
*/
public final void hideAllViews() {
if (View.VISIBLE == this.mHeaderText.getVisibility()) {
this.mHeaderText.setVisibility(View.INVISIBLE);
}
if (View.VISIBLE == this.mHeaderProgress.getVisibility()) {
this.mHeaderProgress.setVisibility(View.INVISIBLE);
}
if (View.VISIBLE == this.mHeaderImage.getVisibility()) {
this.mHeaderImage.setVisibility(View.INVISIBLE);
}
if (View.VISIBLE == this.mSubHeaderText.getVisibility()) {
this.mSubHeaderText.setVisibility(View.INVISIBLE);
}
} |
Characteristics of proposed 3 and 4 telescope configurations for Darwin and TPF-I The Darwin and TPF-I missions are Infrared free flying interferometer missions based on nulling interferometry. Their main objective is to detect and characterize other Earth-like planets, analyze the composition of their atmospheres and their capability to sustain life, as we know it. Darwin and TPF-I are currently in study phase. A number of mission architectures of 3 and 4 free flying telescopes are evaluated on the basis of the interferometer's response, ability to distinguish multiple planet signatures and starlight rejection capabilities. The characteristics of the new configurations are compared also to the former, more complex Bowtie baseline architectures as well as evaluated on base of their science capability. |
Updated: Monday, Nov. 14, 5:17 p.m.
Individuals involved in a racially charged video in blackface are no longer students at ACU, said Dr. Phil Schubert, president of the university, in an email to students, faculty and parents.
creates a working, learning, or living environment that a reasonable person would find intimidating, hostile, or offensive.
Threats or insinuations that a person’s status or other condition of employment or academic status may be adversely affected because of one’s legally protected characteristic.
Unwelcome verbal or written expressions, derogatory comments, epithets, degrading jokes, or innuendos regarding one’s legally protected characteristic.
Posting objects, pictures, videotapes, audio recordings or literature that may embarrass or offend an individual because of one’s legally protected characteristic. Such material, if used in an educational setting, should be clearly and significantly related to educational purposes.
Bullying, defined as repeated and/or severe aggressive behavior likely to intimidate or intentionally hurt, control or diminish another person, physically or mentally because of one’s legally protected characteristic.
Anyone who violates this policy will be subject to appropriate disciplinary action. Disciplinary measures available to remedy Harassment or retaliation include, but are not limited to, the following: verbal warning/reprimand; written warning/reprimand in employee or student files; requirement of verbal and/or written apology to victim; mandatory education and training on harassment by means of reading assignments, videos, classes or other presentations; referral for psychological assessment or treatment; alternate placement, suspension, probation, termination, or expulsion; or other action university deems appropriate under the circumstance. Additionally, interim remedial measures may become permanent.
Schubert concluded his email, saying harassment will not be tolerated at ACU.
Monday, Nov. 14, 3:39 p.m.
Students have taken to social media to speak out against about a racially charged Snapchat video posted to Twitter this morning.
A student with the Twitter username @JNacole503 posted the photo at 9:38 a.m.
The post was retweeted 67 times and other students replied or copied the video to new Tweets. Some tagged the university Twitter @acuedu and the Optimist Twitter @acuoptimist.
The university Twitter responded saying Title IX has been notified to handle the situation.
Please remove the link to this video, as it serves only one purpose: to further spread sensationalism and hatred. The students involved are no longer ACU students (per email from President Schubert). |
Q:
Utilitarian ethics: are consequences for yourself relevant?
In the movie Into the Wild, the main character (Christopher McCandless) 'moves' to Alaska without most modern comforts. He wants to provide for himself by hunting and collecting eatable plants, but ultimately eats a bad plant (because there is no game and he can't get back) and dies.
For high school I have to write something about his choice for running away from multiple ethical perspectives, among which utilitarianism. This brought me to the following question:
From a purely utilitarian viewpoint, would Chris' death make his choice of living in Alaska less ethical?
Utilitarianism ethics aim for the most possible positive consequences, but do consequences like your own death count? Rationally thinking, this is a 'bad' consequence, thus making it less ethical, but somehow this feels illogical.
By the way, this isn't a question directly from an assignment (in fact, I just submitted it). I had to write something about the movie from Bentham's viewpoint (who was an utilitarian). From there, I came to the above question myself, which isn't directly needed for the assignment.
A:
Utilitarianism ethics aim for the most possible positive consequences, but do consequences like your own death count? Rationally thinking, this is a 'bad' consequence, thus making it less ethical, but somehow this feels illogical.
Some things that come to mind:
Chris' death did, at the very least, cause grief to his family and produced no utility for anyone, which makes it clearly bad from a utilitarian point of view
Happiness/utility of yourself and others in the future counts as part of the utilitarian consideration. If you will be happy in the future (or cause happiness in others), then your death decreases total utility and is unethical. If you will be unhappy in the future (or cause unhappiness to others), then the opposite is true. Predictions about the future should always be suspect, but sometimes it is clear, such as in the case of an incurable painful illness. Utilitarianism supports a right to die.
However, an important factor in judging the actions of Chris is that he absolutely did not intend to die. He did something risky which he believed would increased his happiness (good utility) but instead he died (bad utility). Dealing with risk from a utilitarian point of view is pretty straightforward: sum up the net utility of each outcome multiplied it by its probability. Of course, that's very hard to do objectively in this case.
A final factor is that Chris probably misjudged the risk he took. I don't think you can argue that making mistakes is in itself unethical, but a Utilitarian should consider it his responsibility to make a reasonable effort to inform himself about and reduce the risks he takes. One could certainly argue that Chris acted unethical in letting his desire to be atonomous overrule this responsibility. |
SOUTH Africa’s high-tech mobile health train, the Transnet-Phelophepa Health Care Train, begins its annual 36-week delivery of health services to remote rural areas across the country next week. The train provides services in basic health care, health education, dental and eye care as well as psychological counselling. Spectacles and sophisticated eye tests cost only R30 and R10 respectively, while the dental clinic charges only R10 for cleaning, flossing and tooth extraction as well as filling two teeth. |
<filename>2085/2819144_AC_62MS_116K.cpp
#include<cstdio>
#include<cmath>
void main()
{
const int MAX_SIZE=50000;
int size,inversion,i;
while(scanf("%d%d",&size,&inversion))
{
if(size<0)
break;
int length=ceil((1.0+sqrt(1.0+8.0*inversion))/2.0);
int head=size-length+(inversion-(length-1)*(length-2)/2)+1;
for(i=1;i<=size-length;i++)
printf("%d ",i);
printf("%d ",head);
for(i=size;i>=size-length+1;i--)
if(i!=head)
printf("%d ",i);
printf("\n");
}
} |
Cody Boteler is a reporter covering breaking news and county government. He has previously covered Baltimore County, including Catonsville, Towson, and the surrounding areas, for the Baltimore Sun Media Group. He came to the company from Industry Dive, where he covered the waste and recycling industries. He graduated from Towson University in May 2017 with a degree in journalism and environmental science. |
Posted by PeterChan on February 4, 2012
The hype is strong with this one. Thankfully, I've been pacing myself nicely for the past 5 years so that I still have some energy to dodge all the negative comments and backlash Guild Wars 2 has come under in the last few months. However, that doesn't hold true to all gamers. For some reason, I am compelled to write about it. I am compelled to write in hopes that I could convince some people that the Guild Wars 2 beta is not worth hurting yourself over.
In the latest Guild Wars 2 news, ArenaNet has invited a number of press members to their closed beta. This is wonderful news! To consumers, it is also extremely vague which is all the more reason to just be patient. I am positive that everyone wants to be a part of the closed beta and help test while giving feedback (for SCIENCE, of course!). At the same time I have no doubt that there is not a single person outside of ArenaNet who, in the back of their mind, just cannot wait to get their hands on the game for the sake of being entertained and exploring the world ArenaNet has created. This includes members of the press and we have to appreciate that. Although this month's beta phase is limited to mostly press members, this is the sign of ArenaNet opening up to the public after only testing internally for quite a long time.
Some players have asked the question, "why invite the press when clearly I'm an expert at Guild Wars?" That is probably the biggest reason they don't need these players to test! ArenaNet is going where other developers haven't dared take their testing phases. They want to introduce the best MMORPG that is a product and evolution of multiple ideas from several games to make the most immersive title ever. They also want to introduce the game to people who have never played a game before and do their best to keep these people. It can be a tough balance between the old and new, but to give players what ArenaNet wants to deliver, they need unbiased feedback from more than just fans.
Members of the press sometimes have to play games they're not initially interested in while established fans can be biased with their opinions. That makes them a good crowd to introduce an evolved game to that requires learning new mechanics as well. Guild Wars 2 just happens to be the best thing since sliced bread. Ignoring that fact though, the press are often required to play games to an extent where they can legitimately write a detailed review and earn their pay. With that said, these people have more reason to uphold the NDA presented to them. A NDA (non-disclosure agreement) is serious business which both parties (ArenaNet and members of the press, each with their own position) have invested too much time to risk tarnishing their reputation. By no means will ArenaNet show off the entire game to the press simply for review. What the press really needs to see is how well ArenaNet is can carry an unreleased product. While people have passed Guild Wars 2 as vapourware or as an over-hyped title, press coverage will repair this small wound, contributing to the now exponentially growing interest.
Naturally, ArenaNet wants good press for their game. No one is trying to hide that fact. However, ArenaNet is doing their testing differently than conventional modern MMOs. We will get our turn, eventually, and so will all the other potential players who came across a "first impression" post on a media website. Surely we want to see our most anticipated game make history and break sales records! Reaching to the press also suggests that ArenaNet is confident enough with their product to show it to the mass media before it's released. The developers behind Guild Wars 2 have been doing this all along and I'm sure that February is going to be one huge press-push before launch later in the year. A less confident developer would not allow for as many press members to review their game before it goes on sale.
What is an invite to closed beta anyway? We often think of this term as a product that still has bugs to be worked out. This term is loosely defined by ArenaNet though. ArenaNet wouldn't invite press to review a bug infested game. When we do get to play, finding a bug later than sooner would probably be more self rewarding than catching the early obvious ones. If anxious gamers were expecting anything more than "self-rewarding" then they're fighting too hard for the wrong reason to get into the beta, in my opinion.
One could argue that the press is not going to test the game as much as other gamers, yet, as a consumer, it's not our second job either. I don't know anyone else who would quit their job just to work full-time on a fansite for a game. Who is that crazy? While I am indeed crazy, there were more pressing factors that lead me to where I stand now with Guild Wars 2 Live, and I did it to prove to myself that I can do something amazing! I don't need a beta invite as proof that I've been successful on this tangent of my life. Of course, in the back of my mind and deep down in my soul, I admit it would be nice. If being a successful, die-hard fan meant getting into a closed beta at the end, then being such a fan is a pretty risky business! I don't believe the turnover is that significant in the long run anyway.
I can tell others who won't be getting an invitation to test, the folks who have jumped the gun. To the left are people who have broadcasted misleading information in order to gain attention while on the right are people who believe everything posted by every major and minor news source. Get the facts right, fellow gamers! Read it from the ArenaNet blog and only out of mouths of ArenaNet staff! Take the information that has been given by the developers and we will realize that ArenaNet has never lied about Guild Wars 2 or its development. It doesn't matter if someone gets their hands on the game sooner than later, there will always be another person ahead of them. This begs the question, "Why all the hate?" We should spend that energy to properly inform others who might have missed a blog post or a tweet from ArenaNet. And maybe, just knowing that we're doing the right thing by respecting ArenaNet's time and effort can be enough.
Up to this point, ArenaNet has been very supportive of all the fansites who have poured their hearts into Guild Wars 2 alongside the developers. I did it not for Guild Wars 2 or ArenaNet, but for me to grow and learn. It all starts with a simple idea of having fun and then the idea grows into creating something new and unique. I have been doing that with GuildWars2Live and have learned a lot on the way. ArenaNet has rewarded me with their kindness and support by allowing me to continue my fansite and acknowledging me, but even more rewarding is my own work. It has kept me so busy that I could care less of when Guild Wars 2 is going to release. The release date can take its sweet time and be announced "when it's ready."
The whole point one should take away from this rant is, "be patient." There's many things we could be doing other than waiting for the game, or fighting to contribute to the game via a beta invite. Consider starting a blog, a fansite, or join new fansites and encourage others to share their excitement without hinder. There are potentially hundreds of thousands who are about to flock over to Guild Wars 2. We don't need a beta invite to show new folks what a great community Guild Wars has. We know that Guild Wars 2 is going to change gamers, so why not prepare ourselves?
----
Written by @peter_chan
Much thanks to editors including @theLazyGeek, Bakamono and @MalchiorDeven.
Note: ArenaNet does not necessarily condone the things I've said here, though as fans we have this privilege to talk this way. |
Perspectives on the COVID-19 Pandemic Response in a Forensic Psychiatric Hospital: Informing Future Planning ABSTRACT The COVID-19 pandemic has resulted in rapid and unprecedented public policy and legislative interventions to reduce the global spread of the virus. The scope of these challenges has been particularly broad and pressing within health care settings. Individuals with severe mental illness hospitalized in psychiatric facilities are at greater risk of infection than the general population due to both the characteristics of the population (e.g., mentally ill individuals may find the physical distancing measures difficult to understand) and the nature of the settings (e.g., communal living, frequent admissions and discharges). Therefore, it is essential that preventative measures are taken to minimize the chance of nosocomial outbreak in long-term psychiatric facilities; yet minimal information specific to forensic contexts is available. This paper reviews the system-wide strategies that have been put in place across a large Canadian forensic facility and offers recommendations on how to respond to a pandemic or other outbreak in a secure psychiatric setting. Taking the response to COVID-19 in the context of a forensic psychiatric setting, we discuss a wide range of essential aspects of pandemic planning and provide examples of innovative practices that should be considered for retention, future research and broader implementation. |
Mass estimation of loose parts in nuclear power plant based on multiple regression According to the application of the HilbertHuang transform to the non-stationary signal and the relation between the mass of loose parts in nuclear power plant and corresponding frequency content, a new method for loose part mass estimation based on the marginal HilbertHuang spectrum (MHS) and multiple regression is proposed in this paper. The frequency spectrum of a loose part in a nuclear power plant can be expressed by the MHS. The multiple regression model that is constructed by the MHS feature of the impact signals for mass estimation is used to predict the unknown masses of a loose part. A simulated experiment verified that the method is feasible and the errors of the results are acceptable. |
class CarStatusPacket:
""" Process car status packets """
def process(self, packet, session):
"""
We get tyre info from this packet
Store it continuously
"""
return self.update_session(packet, session)
def update_session(self, packet, session):
car_status = packet.carStatusData[packet.header.playerCarIndex]
if session.lap_number_current is not None and session.lap_list.get(session.lap_number_current):
session.lap_list[session.lap_number_current]['tyre_compound_visual'] = car_status.visualTyreCompound
return session |
An Overview of Knowledge Level of the Farmers about Recommended Cultural Practices for Vegetable Production in North India Vegetables are more valuable due to the presence of important mineral, vitamins, carbohydrates, iron protein and other important body nutrients in these. Vegetables play an important role in our daily diet. Vegetables promote our body growth and development and also protect our body from various disease and deficiencies. We all know about the importance of vegetable but knowledge levels of the farmers about vegetable production are still very low. During the surveying of literature on the knowledge level of the farmers about recommended cultural practices for vegetable production studied that majority of farmers had a medium level of knowledge followed by the low level of knowledge. Only a few farmers had a high level of knowledge about recommended cultural practices for vegetable production. |
Advance care planning in Japanese nursing homes Background Japan has the highest rate of ageing worldwide. In Japan, the Ministry of Health, Labour and Welfare is concerned that there are insufficient doctors to care for the elderly in nursing homes and also insufficient advance care planning (ACP). In Japan, the study of the ACP at the nursing home just still started. Purpose The study aim was to evaluate the utility of ACP at the nursing home, supported by an ACP clinician trained in the National Centre for Geriatrics and Gerontology in Japan. Methods Out of 80 families of nursing home residents, 51 participated in this study with all attending a lecture about ACP. We investigated the following: Do the families of the residents prefer the nursing home as the place for the residents' last moments? Would the families like the residents to use mechanical ventilators, gastric feeding tubes, blood transfusion, intravenous infusion or sedative drugs in their terminal states? Do the families prefer to care for the residents on their deathbeds at the nursing home, even if death certificates are not promptly issued because the nursing home does not have any full-time doctors? Results Intervention by a trained ACP clinician significantly increased the number of families who chose terminal care at the nursing home, and the nursing home as the place for the residents' last moments, even if death certificates are not promptly issued (p<0.01). Conclusion A trained ACP clinician is useful for ACP in the nursing home. |
Laminates for MEMS and BioMEMS This paper describes a new way to build MEMS and BioMEMS devices using laminate technologies borrowed from the packaging industry. This approach to building MEMS departs from the traditional silicon approach, and offers many advantages, including the ability to design and build the package at the same time as the device itself. The use of MEMS fabrication techniques, combined with microelectronics manufacturing technology allows a host of integrated devices to be built using novel materials and processes. This frees the MEMS designer from the significant limitations imposed by silicon and its related materials and processes. Devices can be built that are intended for high power applications, optical applications, or biomedical applications. |
<reponame>claymore2015/bamder<filename>src/main/java/com/claymore/bamder/cybersecurity/city/entity/MCityEntity.java
package com.claymore.bamder.cybersecurity.city.entity;
import javax.persistence.*;
/**
* @author li.zhuo
* @create 2018/11/12 11:15 PM
* @since 1.0.0
*/
@Entity
@Table(name = "m_city")
public class MCityEntity {
private int cityId;
private String cityName;
private Long zipCode;
private Long provinceId;
private Byte rn;
@Id
@Column(name = "CityID")
public int getCityId() {
return cityId;
}
public void setCityId(int cityId) {
this.cityId = cityId;
}
@Basic
@Column(name = "CityName")
public String getCityName() {
return cityName;
}
public void setCityName(String cityName) {
this.cityName = cityName;
}
@Basic
@Column(name = "ZipCode")
public Long getZipCode() {
return zipCode;
}
public void setZipCode(Long zipCode) {
this.zipCode = zipCode;
}
@Basic
@Column(name = "ProvinceID")
public Long getProvinceId() {
return provinceId;
}
public void setProvinceId(Long provinceId) {
this.provinceId = provinceId;
}
@Basic
@Column(name = "RN")
public Byte getRn() {
return rn;
}
public void setRn(Byte rn) {
this.rn = rn;
}
@Override
public boolean equals(Object o) {
if (this == o) return true;
if (o == null || getClass() != o.getClass()) return false;
MCityEntity that = (MCityEntity) o;
if (cityId != that.cityId) return false;
if (cityName != null ? !cityName.equals(that.cityName) : that.cityName != null) return false;
if (zipCode != null ? !zipCode.equals(that.zipCode) : that.zipCode != null) return false;
if (provinceId != null ? !provinceId.equals(that.provinceId) : that.provinceId != null) return false;
if (rn != null ? !rn.equals(that.rn) : that.rn != null) return false;
return true;
}
@Override
public int hashCode() {
int result = cityId;
result = 31 * result + (cityName != null ? cityName.hashCode() : 0);
result = 31 * result + (zipCode != null ? zipCode.hashCode() : 0);
result = 31 * result + (provinceId != null ? provinceId.hashCode() : 0);
result = 31 * result + (rn != null ? rn.hashCode() : 0);
return result;
}
}
|
International Liability Issues for Software Quality Abstract : This report focuses on international law related to cybercrime, international information security standards, and software liability issues as they relate to information security for critical infrastructure applications. Each area is explored and implications for U.S. policy and efforts to create cyber security policy worldwide are discussed. Recommendations are made for U.S. government participation and leadership. This report is one of a series of reports on U.S. policy by the CERT Coordination Center. Prior reports focused on international infrastructure for global security incident response and the technical challenges and global policy issues of tracking and tracing cyber attacks. |
Regional differences in acute stroke admission and thrombolysis rates in the German federal state of Hesse. BACKGROUND Using data from the mandatory quality assurance program for stroke care in Hesse, we analyze regional differences in thrombolysis rates and infer some ways in which care can be improved. METHODS We identified 7707 patients with acute ischemic stroke who were admitted to hospital within 3 hours of symptom onset in 2007 and 2008, and we determined the local thrombolysis rate district by district. In order to exclude the possibility that the observed local differences in thrombolysis rates might be accounted for, in large part, by off-label thrombolysis procedures, we further narrowed down the subgroup of patients who underwent thrombolysis to the 1108 patients admitted within 2 hours of symptom onset. We also analyzed the local thrombolysis rates for patients who were primarily referred to stroke units. RESULTS The overall thrombolysis rate among patients admitted within 3 hours of symptom onset was 19%, varying locally from 6% to 35%. Among patients admitted within 2 hours of symptom onset, the local thrombolysis rate ranged from 13% to 85%. Even in patients primarily referred to stroke units, the local thrombolysis rate ranged from 8% to 44% in the 3-hour group and from 16% to 62% in the 2-hour group. CONCLUSION Local thrombolysis rates vary unexpectedly widely across the state of Hesse. The care of patients with acute stroke after they reach the hospital urgently needs critical reappraisal and improvement. |
A former local newspaper reporter, Tanveer is a student at the Medill School of Journalism learning all things digital and entrepreneurial. He also writes about political figures for WhoRunsGov.com and hopes to own the high score on multiple Ms. Pac-Man machines one day.
While many government agencies still tend to employ the "broadcast" model when using social media, some are engaging through hashtags, community building initiatives, and geo-location analysis. These efforts are helping to better inform the public and alert them to public safety emergencies in real-time.
A good recent example of this is how the team of energy companies and government agencies responding to the oil spill in the Gulf of Mexico are putting these strategies to use.
Here are ways other government agencies, from local law enforcement to the National Weather Service, are seizing on these tools to improve their services.
At the most basic level, social media is about community building. Government agencies have adopted this mindset to varying degrees as a way to foster trust and dialogue with people. "It is truly a national town hall that has never been attempted during a disaster," said Commander James Hoeft of the U.S. Navy, who oversees the cleanup effort's social media team.
The idea has been implemented in parts of the U.S. government to varying degrees. In 2008, Admiral Thad Allen of the U.S. Coast Guard sent out a service-wide message saying, “[To] modernize the Coast Guard we must learn how to effectively use social media tools to enhance our ability to perform as a more transparent, change-centric organization."
The Coast Guard has since deployed a series of Flickr, YouTube and Twitter accounts, both at the headquarters and regional levels, as a part of The Coast Guard Compass. Some are better than others, with many serving simply as multimedia RSS feeds. But there are stars, like the Twitter feed for the Portsmouth, VA-based District Five, which discusses their latest coastal rescue operations.
Much like the Coast Guard, the Federal Emergency Management Agency also has a multichannel scheme in social media. Its FEMA in Focus Twitter feed serves as a way to disseminate information in a timely way. The agency has a series of regional accounts, as well.
At the local law enforcement level, Web 2.0 technology has been implemented in some departments to give people details about what officers have been up to. At the Bellevue Police Department in Nebraska, Twitter is used to solicit help from the public and Facebook is used as a comment and complaint board for residents. In Great Britain, the Merseyside Police website personalizes information according to neighborhood, also appealing to the public for help as needed.
While the more traditional means of sharing information with people, such as press conferences or releases, will always be necessary to brief the public in detail about events, agencies are turning to social media to keep the public informed in real-time.
On April 3rd, Detective Chief Inspector Mark Payne of the West Midlands Police in the United Kingdom used Tweetdeck to keep an eye on demonstrations involving two controversial and politically opposed groups; the English Defence League and Unite Against Fascism. He checked out Facebook rumors of stabbings and vandalism, and posted on Twitter when the information was found to be false — potentially quelling violent backlash. "This is groundbreaking stuff for policing in the UK. We have used social media as a broadcast platform during protests in the past, but we have not had immediate updates from officers on the ground, enabling two-way conversations," Payne wrote after the event.
Lauri Stevens, a Massachusetts-based social media consultant for law enforcement, said that such tools have been key in enhancing the reputation of agencies. "Cops are just getting the interactive engagement thing," Stevens said. "I think the law enforcement's policy should state you have to be competent with this stuff."
Law enforcement and emergency response agencies alike are also becoming more sophisticated in how they use Twitter. While monitoring hashtags is commonplace, some agencies are creating them to denote specific social media priorities — particularly, getting users to document certain events.
During the 2009 PRIDE Parade, the Toronto police encouraged the use of the #PrideTO hashtag to keep an eye on any suspected crimes related to the LGBT community’s event.
The National Weather Service is enlisting the help of Twitter users, asking them to use the hashtag #wxreport to share significant weather reports. The website gives precise instructions for how to report damaging winds, snow, hail, tornadoes and other potentially serious weather events. As the website notes, the project is "experimental."
Researchers at government agencies are experimenting with social media to try and spot possible issues and trends before more scientific measurements can be taken.
During last year’s H1N1 outbreak, the Internet famously took a starring role in illustrating how the swine flu epidemic was spreading across the globe. Now, a group including researchers from City University London, the European Centre for Disease Prevention and Control (ECDC) and Britain's National Health Service are teaming up ahead of the 2012 London Olympics to develop ways to detect and respond to epidemics via Twitter.
Funded by the American Recovery and Reinvestment Act, the U.S. Geological Survey's Twitter Earthquake Detector (@USGSTed) is a prototype that gathers real-time Twitter updates during seismic activities faster than scientific equipment can be tapped for more precise measurements and alerts. It examines earthquakes at an anecdotal level, and complements scientific analysis, according to the project’s overseer, Paul Earle.
For government agencies, social media not only sends and gathers information instantaneously — it fosters relationships and trust, while encouraging users to share important information. While not all social media use needs to be creative, agency engagement with these platforms can help show people that government organizations are listening. |
/**
* Represents single To Do Action of the section of To Do List, using {@link CheckBox}.
* @param action the to do action of which it's set the View.
* @return a {@link CheckBox} representing the toDoAction in input
*/
private CheckBox createSingleToDoAction(final ToDoAction action) {
final CheckBox box;
box = new CheckBox(action.getAnnotation());
box.setSelected(action.isDone());
return box;
} |
C60 fullerene and its nanocomplexes with anticancer drugs modulate circulating phagocyte functions and dramatically increase ROS generation in transformed monocytes Background C60 fullerene-based nanoformulations are proposed to have a direct toxic effect on tumor cells. Previous investigations demonstrated that C60 fullerene used alone or being conjugated with chemotherapeutic agents possesses a potent anticancer activity. The main aim of this study was to investigate the effect of C60 fullerene and its nanocomplexes with anticancer drugs on human phagocyte metabolic profile in vitro. Methods Analysis of the metabolic profile of phagocytes exposed to C60 fullerene in vitro revealed augmented phagocytic activity and down-regulated reactive nitrogen species generation in these cells. Additionally, cytofluorimetric analysis showed that C60 fullerene can exert direct cytotoxic effect on normal and transformed phagocytes through the vigorous induction of intracellular reactive oxygen species generation. Results Cytotoxic action as well as the pro-oxidant effect of C60 fullerene was more pronounced toward malignant phagocytes. At the same time, C60 fullerenes have the ability to down-regulate the pro-oxidant effect of cisplatin on normal cells. These results indicate that C60 fullerenes may influence phagocyte metabolism and have both pro-oxidant and antioxidant properties. Conclusions The antineoplastic effect of C60 fullerene has been observed by direct toxic effect on tumor cells, as well as through the modulation of the functions of effector cells of antitumor immunity. pharmacokinetics and increases their therapeutic efficacy (;;. Previous investigations from our research group revealed that pristine C 60 fullerene possesses a potent anticancer activity per se (a, b;). In addition, we have demonstrated that conjugation of Dox and Cisplatin (Cis) with the C 60 fullerene led to significant increase in its toxicity toward various human tumor cell lines in vitro and greatly enhances the inhibitory effect of these drugs on the growth of Lewis lung carcinoma in vivo (a, b, 2017. It is well documented that the underlying mechanism of antineoplastic effect of C 60 fullerene-based nanoformulations is related with the combination of the direct effect on tumor cells (modulation of oxidative stress, apoptosis, necrosis, and autophagy), their effect on tumor microenvironment (reduction of the blood supply to tumor tissues), and activation of the host immune system (;;Shi and Li 2012). There are limited and controversial data concerning immunomodulatory effects of C 60 fullerenes. Liu et al. have shown that C 60 fullerene and its derivatives exert potent immunomodulatory pro-inflammatory activities: increase TNF production and enhance cell immune response, and show almost no adverse effect to the viability of immune cells. Fujita et al. revealed that genes involved in the inflammatory response, oxidative stress, metalloendopeptidase activity, as well as MHC expression were upregulated in immune cells after the exposure to C 60 fullerene. Zogovich et al. have reported a significant increase in splenocyte production of the immunoregulatory free radical nitric oxide (NO) after the treatment with C 60 fullerene. A special attention in the analysis of C 60 fullerene immunomodulatory properties is given to their influence on phagocytes. Phagocytes are credited with a crucial role in the development of the immune response. Professional phagocytes (monocytes, dendritic cells, macrophages, neutrophils, and mast cells) constitute the first line of cellular immune defense. These cells are important players in the antitumor protection and are the link between the innate and adaptive immunity (Murray and Wynn 2011;;). Monocytes/macrophages and neutrophils activate, orient, and regulate adaptive responses. These cells are characterized by high plasticity and can possess different metabolic profiles, depending on the activating stimuli. There are two main diametrally opposite polarizational state of macrophages/neutrophils (not excepting numerous intermediate metabolic states): classical (M1 and N1, respectively) and alternative (M2 and N2, respectively). The main distinctive features of classically polarized phagocytes are the generations of effector molecules (reactive oxygen and nitrogen species) and cytokines (TNF,IL1,IL6,etc.) participating in the inflammation and promotion of Th1 immune response. These cells mediate immune defense against intracellular pathogens and malignant tumors. Characteristic features of alternatively polarized phagocytes are their participations in the resolution of inflammation, tissue repair, and switching of adaptive immune response to Th2 profile. These cells express high levels of receptors involved in endocytosis, produce anti-inflammatory cytokines, and metabolize arginine to ornithine and polyamines. Alternatively, polarized phagocytes can possess immunosuppressive activity, participate in tissue remodeling, and promote tumor growth and angiogenesis (;Okabe and Medzhitov 2016;). Regardless of the route of C 60 fullerene administration chosen, phagocytes are the first cells of the body immune system, which are influenced by these nanoformulations. Therefore, a greater understanding of interactions between C 60 fullerene and phagocyte will enable the creation of better predictive models of their therapeutic efficacy as well as adverse outcomes following exposure. Russ et al. suggested that C 60 fullerene's effect on phagocyte signaling is achieved through endocytosis/pinocytosis as well as passive diffusion of this nanostructure. According to the limited literature data available, the character of the C 60 fullerene's influence on phagocytes depends on their geometric structure, dose, duration of exposure, and the initial functional state of the cells. C 60 fullerene and their derivatives at low concentrations exerted mainly negative effect on the functions of nonsensitized macrophages in vitro by inhibiting the myeloperoxidase activity, and suppressed the expression of CD54 involved in the adhesion (;). Inhibitory effect of C 60 fullerene on pro-inflammatory (classically)-activated macrophages was described in rats with experimental adjuvantinduced arthritis (). Many authors are unison in their opinions that C 60 fullerenes dramatically affect the phagocyte oxidative metabolism and can even induce apoptosis of macrophage by changing the mitochondrial membrane potential (;;). Based on our previous experience, in vivo anticancer effects of C 60 fullerene and C 60 + Dox nanocomplexes are accompanied by the increase of macrophage oxygen-dependent cytotoxicity toward authologic tumor cells. In addition, our previous results revealed direct toxic effects of C 60 fullerene and its nanocomplexes with anticancer drugs on transformed leukocytes including myelocytic cells (;;). The main aim of this study was to investigate the effect of C 60 fullerene and its nanocomplexes with anticancer drugs on human phagocyte metabolic profile in vitro. Potential mechanisms of this effect have been analyzed using docking experiments of direct interactions between the C 60 fullerene and phagocyte pattern-recognition receptors as well as cytochrome p450 (CYP). A study of the effect of C 60 fullerene and its nanocomplexes with anticancer preparations on pro-monocytic, human myeloid leukemia cell line U937 was also performed to explore the potential application of C 60 fullerene in the therapy of acute myelocytic leukemia. Preparation of C 60 fullerene aqueous colloid solution The pristine C 60 fullerene aqueous colloid solution (C 60 FAS; final concentration 0.15 mg/ ml) used in the experiments was prepared according to the protocols developed previously (;;). Preparation of C 60 + Cis and C 60 + Dox nanocomplexes Cis solution (Cisplatin-TEVA, Pharmachemie B.V., 0.5 mg/ml) was immobilized on the C 60 fullerene according to the protocol developed by our research group. The initial solution of C 60 FAS (final concentration 0.15 mg/ml) and Cis (final concentration 0.15 mg/ml) was mixed in 1:1 volume ratio. Afterward, the mixture was treated for 20 min in ultrasonic disperser and, next, subjected to overnight magnetic stirring at the room temperature. Dox (Doxorubicin-TEVA, Pharmachemie B.V., lyophilized powder, 10 mg) was dissolved in saline to obtain a final concentration of 0.15 mg/ml. It was immobilized on the C 60 fullerene according to a previously described protocol (. Specifically, C 60 FAS (final concentration 0.15 mg/ml) and Dox (final concentration 0.15 mg/ml) were mixed in 1:2 volume ratio, and the resulting mixture was treated for 20 min in the ultrasonic disperser, and then, it was subjected to overnight magnetic stirring at the room temperature. Simulation The geometric structures for CYP (3tis-crystal structure of the complex between human CYP 3A4 and desthiazolylmethyloxycarbonyl ritonavir) and Toll-like receptors (TLRs) (2z7x-crystal structure of the TLR1-TLR2 heterodimer, Homo sapiens, Eptatretus burger; 4g8a-crystal structure of human TLR4 polymorphic variants, D299G and T399I) from the PDB base were used in calculations. Computational methods such as docking and molecular dynamics (MD) simulations were applied. We used an algorithm of systematic docking (SDOCK+) implemented in QXP docking software, which has shown a higher reproducibility of compound conformation with a minimum Root mean square deviation (Rmsd) in comparison with the crystallographic data (). The maximum number of SDOCK+ routine steps was set to 200, and the 10 best protein (target)-C 60 fullerene complexes, based on built-in QXP scoring function (McMartin and Bohacek 1997), were retained for each compound. The optimal position of obtained complexes was selected based on their overall energy. Molecular dynamics simulation was performed for up to 25 ps to evaluate the stability of protein (target)-C 60 fullerene complexes. For the calculation, we used NPA algorithm (Sturgeon and Laird 2000). The above method is one of the most accurate and sensitive methods, and it generates true ensemble trajectories (). Within the calculation, the next main parameters used were temperature (in K)-300; pressure (in kPa)-100; water molecules treated as rigid bodies. Cell isolation Four healthy adult men aged 21 ± 2 years were recruited to participate in the study. Approval was obtained from the ethical committee of Taras Shevchenko National University of Kyiv, and informed consent was obtained from all subjects before the commencement of the study. Fresh blood (20 ml) was obtained from volunteers and mixed with 50 l of preservative-free heparin (Hospira, UK). Sterile dextran was added to a final concentration of 0.6%, and the cells were allowed to settle at 37 °C until the red cells had sedimented. The buffy coat was then removed, washed in Hanks' buffered salt solution, and then resuspended in RPMI 1640 medium containing 20% fetal calf serum(FCS), penicillin (50 U/ ml), and streptomycin (50 g/mL) (). Cell line Human myeloid lineage cells U937 were kindly supplied by the Bank of Cell Cultures and Transplantable Experimental Tumors of R.E. Kavetsky Institute of Experimental Pathology, Oncology and Radiobiology of NAS of Ukraine (Kyiv, Ukraine). Cells were cultured in vitro in Dulbecco-modified Eagle medium (DMEM, Sigma, USA) supplemented with 10% fetal calf serum (FCS), penicillin (100 U/ml), and streptomycin (100 g/ml) at 37 °C in 5% CO 2. Cell incubation with C 60 fullerene, Cis, Dox, and their nanocomplexes Prior to the metabolism assays, 200 l of cell suspension in RPMI 1640 medium or 200 l of heparinized whole blood was treated with C 60 fullerene (final concentration 0.15 mg/ml), Cis (final concentration 0.15 mg/ml), Dox (final concentration 0.15 mg/ml), and nanocomplex of C 60 fullerene with Cis (mixed in 1:1 volume ratio) or Dox (mixed in 1:2 volume ratio) for 30 min. Nitrite assay Nitrite-level determination was performed to evaluate NO release into the conditioned media of human peripheral blood monocytes and granulocytes as described earlier (Neil 2009;). In brief, after 24 h of cultivation, the culture supernatants were collected, and the nitrite concentration in each supernatant was assayed by the Griess reaction. Equal volumes of 2% sulfanilamide in 10% phosphoric acid and 0.2% naphthylethylene diamine dihydrochloride were mixed to prepare the Griess reagent. The reagent (100 l) was added to equal volumes of the supernatant, and the mixture was then incubated for 30 min at room temperature in the dark. The A550 of the formed chromophore was measured using a plate reader. The nitrite content was calculated with sodium nitrite as a standard. Each sample was assayed for nitrite in triplicate. Each value was divided by the number of viable cells and expressed as nitrite level per 10 6 cells. The mean value and SD were calculated with normalized values. Intracellular ROS assay Reactive oxygen species (ROS) levels were measured using 27-dichlorodihydro-fluorescein diacetate (carboxy-H2DCFDA, Invitrogen) as previously described (). In brief, heparinized whole blood was incubated with PBS containing 10 M carboxy-H2DCFDA for 30 min at 37 °C to measure ROS production by peripheral blood monocytes and granulocytes. A short recovery time was allowed for the cellular esterases to hydrolyze the acetoxymethyl ester or acetate groups and render the dye responsive to oxidation. Erythrocytes were lysed with lysis buffer. The cells were then transferred to polystyrene tubes with cell-strainer caps (Falcon, Becton-Dickinson, USA) and analyzed with flow cytometry (excitation: 488 nm; emission: 525 nm). Only living cells, gated according to scatter parameters, were used for the analysis. Granulocytes or monocytes were gated according to forward and side scatters. Phorbol 12-myristate 13-acetate (PMA) (Sigma-Aldrich) was used to evaluate nonspecific reactivity reserve in phagocytes (;). Reactivity reserve was characterized by the modulation coefficient (MC) that was calculated by the following formula: where S is the ROS value in probes stimulated with PMA in vitro, and B is ROS value in unstimulated probes (basal value). Phagocytosis assay The flow cytometry phagocytosis assay was performed as previously described (). In brief, the fluorescein isothiocyanate(FITC)-labeled heat-inactivated Staphylococcus aureus Cowan I bacteria (collected by the Department of Microbiology and General Immunology of Taras Shevchenko National University of Kyiv) at the concentration of 1 10 7 cells/ml in the volume of 5 l were added to heparinized whole blood. All samples were incubated at 37 °C for 30 min. At the end of the assay, phagocytosis was arrested by the addition of cold stop solution (PBS with 0.02% EDTA and 0.04% paraformaldehyde). Erythrocytes were lysed with lysis buffer. Results were assessed using FACSCalibur flow cytometer and CellQuest software (Becton-Dickinson, USA). Granulocytes or monocytes were gated according to forward and side scatters. The results were registered as the percentage of cells emitting fluorescence after a defined culture period (phagocytosis percentage, PP) and as phagocytosis index (PI) that represents the mean fluorescence per one phagocytic cell (engulfed bacteria by one cell). Determination of arginase activity Arginase activity was measured in cell lysates by standard colorimetric method with some modifications (). In brief, 100 l of 0.1% Triton X-100 and 100 l of 50 mMTris-HCl (pH 7.5), containing 10 mM MnCl 2, were sequentially added to cell samples. Phagocyte arginase was then activated by heating of the mixture at 56 °C for 7 min. The reaction of l-arginine hydrolysis by arginase was carried out by incubation of the mixture containing activated arginase, with 100 l of l-arginine (0.5 M; pH 9.7) at 37 °C for 2 h, and was stopped by the addition of 800 l of the mixture of acidic solution (H 2 SO 4 :H 3 PO 4 :H 2 O = 1:3:7). For colorimetric determination of urea, -isonitrosopropiophenone (40 l, 9% solution in ethanol) was added, and the mixture was incubated at 95 °C for 30 min and then at 4 °C for 30 min. The urea concentration was determined spectrophotometrically at 540 nm with the use of a microplate reader. Each condition was tested in triplicate and the experiments were repeated at least three times. Each value was divided by the number of viable cells and expressed as urea level/h per 10 6 cells. The mean value and SD were calculated with normalized values. Cell death assay Apoptosis was assessed by staining cells with Annexin V-FITC and counterstaining with propidium iodide (PI) with the use Annexin V-FITC Apoptosis Detection Kit (Dojin-doEUGmbH, Munich, Germany) according to the manufacturer's instructions. In brief, 2 10 5 cells were placed into wells of a 96-well flat-bottomed plate and were either treated with C 60 fullerene, Cis, Dox and their nanocomplexes at maximum concentration (0.15 mg/ml) for 24 h. Untreated cells were used as a control. Afterward cells were washed twice with PBS and stained with 5 l Annexin V-FITC and 5 l PI in binding buffer for 10 min at room temperature in the dark. Cells from each sample were then analyzed by FacsCalibur flow cytometer (BD Biosciences). The data were analyzed using CELLQuest software (BD). PI detects cells that have lost CPM integrity (i.e., necrotic and secondary necrotic cells), whereas Annexin V detects early apoptotic cells. Statistical analysis All experimental results are reported as mean ± SD. Statistical significance of the results was determined by t test (unpaired, two-tailed) and the nonparametric Mann-Whitney test, comparing two groups of independent samples. Means were compared, and differences were considered significant at p values of 0.05 or less. Results and discussion Phagocytes play a crucial role in antitumor immunity. Modulation of their metabolism is involved in anticancer effect of conventional anticancer preparations (;) and can be considered as one of the mechanisms of C 60 fullerene antineoplastic action. C 60 fullerene can affect phagocyte metabolism by activation of membrane receptor followed (or no) by the endocytosis and/or by interaction with intracellular receptive structures after passive diffusion. Entry pathways of C 60 fullerenebased nanoformulations shift from passive diffusion to receptor-mediated endocytosis with increasing particle size (;). As we reported previously ((Prylutskyy et al., 2014), the probe microscopy revealed in our preparations randomly arranged individual C 60 molecules with a diameter of ~ 0.7 nm and their bulk sphere-like aggregates with a height of 2-100 nm in C 60 FAS. Therefore, C 60 fullerenes and their nanocomplexes with anticancer drugs can exert additive membrane-dependent and direct intracellular effect on phagocyte metabolism. It was important to investigate the effect of C 60 fullerenes and their nanocomplexes with anticancer drugs on different metabolic reactions of nonsensitized human peripheral blood phagocytes as well as myeloid leukemia cells U937, and revealed potential receptive membrane and intracellular structures, which could be involved in such effect. Molecular docking analysis The molecular docking approach allows simulating the behavior of small molecules such as C 60 fullerenes and their nanocomplexes with anticancer drugs in the binding sites of target cells. It also gives us an opportunity to predict the effect of such small molecules on metabolic processes in cells (;). Majority of metabolic phagocyte functions depends on the activation of pattern-recognition receptors, among which TLRs are most important. TLR signaling is involved in activation/regulation such fundamental phagocyte functions as phagocytosis, oxygendependent and oxygen independent cytotoxicity, arginine metabolism, antigen presentation, cytokine synthesis etc. (;McCoy and O'Neill 2008). Some TLRs are considered as a binding site for C 60 fullerene (). Taking into account the profound influence of C 60 fullerene-based nanoformulations on cell oxidative metabolism, one of its significant intracellular targets can be the CYP-a key component of the monooxygenase system. Unlike other hemoproteins having in cell usually one activity and well-defined function, CYP alongside monooxygenase activity can exhibit oxidase one, generating such ROS as superoxide and hydroxyl radicals, hydrogen peroxide. Poor coupling of the CYP catalytic cycles results in continuous ROS generation. CYP is reported to produce 12(S)-hydroxyheptadeca-5Z,8E,10E-trienoic acid (12-HHT) that is overexpressed in classically activated macrophages (;). Cytochrome p450 The active site for the majority of the monoxidation reactions, which CYP enzymes catalyze, contains a heme group during the catalytic processes (). It is known that CYP3A4 can accommodate multiple substrates owing to the large size of the active site (). So, in both cases C 60 fullerene can fill in the binding site, interact with a large number of amino acids and create cation -system with cofactor. For example, in the case of PDB 3tis structure, C 60 fullerene involved in the stacking interaction with Phe 304 and Phe 108 (Fig. 1). The docking and MD results indicate that C 60 fullerene forms stable complex with CYP: the respective free energy of complexation is − 54.4 kJ/mol after docking and − 47.5 kJ/mol after MD simulation. The energy of steric clashes for C 60 fullerene is higher after MD simulation than that after docking: it means that C 60 fullerene-CYP complex after MD simulation is more rigid, despite that its overall energy is less (18.8 kJ/ mol after docking and 25.7 kJ/mol after MD simulation). One can assume that this is due to changes in the binding site. Furthermore, the distance between cofactor and C 60 fullerene decreased: it equals to 5.4 after docking and 4.52 after MD simulation. At the same time, the Rmsd value for C 60 fullerene is not significant: it equals to 0.56 (in turn the Rmsd value of protein is 2.53 ). As a result C 60 fullerene deepens into the binding site and gets stuck among some amino acids (e.g., Phe 108 and Phe 304 (Fig. 1); Rmsd values are 3.5 and 1.2, respectively). TLRs Based on previous data (), docking and MD simulation were carried out using TLR1/TLR2 and TLR4 ( Fig. 3; we investigated the binding of C 60 fullerene in MD2 domain). Docking and MD results showed that numerous Phe and Tyr residues easily create - interactions, while other amino acids (e.g., Leu, Ile, Ala, Val, and Pro) form lipophilic contacts with the sidewalls of C 60 fullerene (Figs. 2, 3, 4, 5). In the case of binding of C 60 fullerene with TLR1/TLR2/ECD and TLR4/MD-2, significant mobilities of protein are observed (much larger than that for the previous object): 5.2 and 4.6, respectively. The obtained models were also characterized by a large energies of C 60 fullerene-TLR complexes both after docking (free energies of complexation: − 19.2 and − 50.9 kJ/mol, respectively) and MD simulation (free energy of complexation − 26.8 and − 50.6 kJ/mol, respectively). The binding of C 60 fullerene with TLR1/TLR2/ECDs occurs in the outer part of ECDs (Rmsd 6.7 ) and, consequently, there is formation of new stronger stacking interactions with Phe 209 and Tyr 300 (Figs. 2, 4). The binding of C 60 fullerene with TLR4/MD-2 is characterized by complete filling of the hydrophobic pocket of MD-2 (Figs. 3, 5) and the formation of a significant number of stacking interactions (e.g., with Phe 119, Phe 76 and Phe 104; Fig. 5B). This change of binding site is associated with significant mobility of interacting components: Rmsd value for protein is 4.6, and for C 60 fullerene-5.3. Thus, one can suggest that the formation of complexes of C 60 fullerene, both with CYP and different parts of TLR1/TLR2 and TLR4, is potentially possible. At the same time, the calculated energy and geometrical parameters for the C 60 fullerene-CYP complex indicate its greater stability compared with the probable C 60 fullerene-TLR complexes. Taking into account the results of the performed docking analysis, it was reasonable to investigate the ability of C 60 fullerene and its nanocomplexes with anticancer drugs to modulate phagocyte metabolic processes associated with CYP and TLR signaling: ROS and reactive nitrogen species (RNS) generations, arginase activity, and phagocytosis. Peripheral blood phagocyte ROS generation C 60 fullerenes are most frequently positioned as "free radical sponges" and antioxidants due to their ability to absorb electrons (;). However, depending on the circumstances, C 60 fullerenes are capable of both quenching ROS and induce their generation with or without photoexcitation (;;Markovic and Trajkovic 2008;). Moreover, spontaneous induction of ROS generation by C 60 fullerenes plays a pivotal role in their toxic effect on eukaryotic normal and malignant cells. C 60 fullerenes and their derivatives stimulate oxidative metabolism in erythrocytes and fibroblasts, malignant lymphocytes and epithelial cells, and endothelial cells and macrophages (). Ershova et al. reported about complex time-dependent changes in ROS levels in cells treated with C 60 fullerene derivatives. They differentiate between "early" (from 15 min to 3 h) and "late" (24 h) responses to C 60 fullerene exposure. Fluorescence microscopy in these experiments revealed that the early response is not associated with C 60 fullerene penetration into the cytoplasm and is accompanied by the increase of ROS generation within the first 15-30 min on the cell surface. The late response in turn is associated with the accumulation of nanoparticles inside the cells, and includes transitory (within 1-2 h after the treatment) decrease of ROS generation followed by the secondary increase of ROS levels after the 24 h exposure. There are two main sources of ROS in phagocytes: NADPH oxidase (NOX) and a number of cellular enzymes such as CYP and xanthine oxidase that can localize in cytosol, mitochondria, peroxisomes etc. (). ROS production can be induced by the activation of TLRs 1, 2, 4 and 9. TLR-associated ROS generation primarily depends on NOX signaling pathway, and is associated with the cell surface. In addition, engagement of TLRs 1, 2 and 4 leads to the recruitment of mitochondria to phagosome followed by the activation of ROS generation in these organelles (;). In our experiments, we investigated the effects of C 60 fullerenes and their nanocomplexes with anticancer drugs Cis and Dox on intracellular ROS generation of nonsensitized human peripheral blood phagocytes in buffy coat. The treatment exposure time was 30 min. It suggests the effect of nanoformulations on ROS generation mainly through the interaction with membrane-associated receptive structures. In addition to the assessment of the effect of studied preparations on basic level of phagocyte intracellular oxidative metabolism, we also examined the so-called nonspecific reactivity reserve of the investigated function. For this purpose, cells were additionally treated with PMA that is reported to stimulate ROS generation (). PMA activates NOX assembly in the plasma membrane, and can potentially synergize with the effect of nanoformulations (). Flow cytometry method allowed us to analyze intracellular oxidative metabolism of granulocytes and monocytes separately by gating according to forward and side scatters. The treatment of peripheral blood phagocytes with all studied preparations resulted in sharp increase of intracellular ROS generation in these cells (Fig. 6). Cis and Dox caused more than 15-fold increase in the production of ROS by phagocytes. Our results are consistent with numerous literature data. Approximately half of FDA-approved anticancer drugs (including Dox and Cis) are known to produce ROS that are critically involved in toxic side effects of these drugs (). Reactivity reserve in cells treated with anticancer drugs in response to PMA was absent. It indicates maximal degrees of this function activation and oxidative stress development that threaten the cell viability (Fig. 6). C 60 fullerenes caused eightfold increase in ROS generation by phagocytes. We registered the reactivity reserve in these cell samples. It indicates the less ROS-mediated toxicity of C 60 fullerenes as compared to cytotoxic agents. Reactivity reserve of intracellular oxidative metabolism after the treatment with C 60 fullerenes was greater in granulocytes than in monocytes as evidenced by MC values 89.7 and 25.8, respectively ( Fig. 6a and b). This suggests that the granulocytes better tolerated treatment with C 60 fullerenes. The granulocytes are terminally differentiated cells. Unlike neutrophils, circulating monocytes are less mature progenitor cells whose differentiation is completed in tissues (;). In addition, granulocytes have only few Fig. 6 The effect of C 60 fullerenes and their nanocomplexes with anticancer drugs Cis and Dox on intracellular ROS generation in nonsensitized human peripheral blood granulocytes (a) and monocytes (b). Whole blood samples were treated with mentioned preparations for 30 min, then were stained with carboxy-H2DCFDA (see "Methods"). The fluorescence intensities of 10.000 cells were then analyzed by BD-FACS Calibur. All results are expressed as mean ± SD of three independent experiments. *p < 0.05 compared to basal level of ROS generation in control; **p < 0.01 compared to basal level of ROS generation in control; ***p < 0.001 compared to basal level of ROS generation in control; # p < 0.05 compared to corresponding cells without treatment with PMA; § p < 0.05 compared to cells treated with Cis alone, as analyzed by unpaired t test mitochondria (;Dan ), and their contribution to the overall ROS production is minimal. Therefore, restricted mitochondria network can be another reason of better granulocyte tolerability. In mononuclear phagocytes, mitochondrial ROS generations are an important part of total intracellular oxidative metabolism. Probably, this fact along with immature state makes monocytes to be more vulnerable to the effect of C 60 fullerenes. The complexation of Dox with C 60 fullerene did not affect its stimulatory action on phagocyte oxidative metabolism. Whereas the complexation of Cis with C 60 fullerene leads to the downregulation of its stimulatory effect on intracellular ROS generation. Thus, C 60 fullerenes both when used alone and in nanocomplexes with anticancer drugs strongly induce intracellular ROS generation in peripheral blood phagocytes, and are capable to downregulate pro-oxidant activity of Cis. Peripheral blood phagocyte NO production Generation of RNS is TLR-dependent metabolic process and is associated with NFB activation (;McCoy and O'Neill 2008). In addition, RNS synthesis is adversely regulated by oxidative stress () and therefore is related to CYP activation. We characterized the RNS generation by the level of nitrites (Griess reaction) in culture medium of cells treated with C 60 fullerenes and their nanocomplexes with Cis and Dox. The Griess reaction is the most frequently used analytical approach to quantitate the major metabolites of NO, i.e., nitrite and nitrate, in a variety of biological fluids (Tsikas 2007). Nitrite production was analyzed in total pool of peripheral blood phagocytes contained in buffy coat. In our experiments, treatment of human phagocytes with Cis and Dox resulted in moderate decrease of nitrite level (Fig. 7). Cis and Dox are reported to downregulate inducible nitric oxide synthase (iNOS) expression followed by the decrease in RNS production by affected cells. As the Fig. 7 The effect of C 60 fullerenes and their nanocomplexes with anticancer drugs Cis and Dox on RNS generation in nonsensitized human peripheral blood phagocytes. Phagocytes in buffy coat were treated with mentioned preparations for 30 min, then RNS generation was analyzed in Griess reaction (see "Methods"). All results are expressed as mean ± SD of three independent experiments. *p < 0.05 compared to control; **p < 0.01 compared to control; # p < 0.05 compared to cells treated with Dox alone; § § p < 0.01 compared to cells treated with Cis alone, as analyzed by unpaired t test induction of ROS generation is a one of the mechanism of action of these anticancer drugs (;), an oxidative stress can be one of the reasons of their negative effect on RNS production. C 60 fullerenes significantly downregulated RNS generation. Similar observations have been published by Huang et al.. This research group revealed that nonfunctionalized C 60 fullerene suppresses the release of NO by macrophages RAW 264.7. Complexation of Dox with C 60 fullerene leads to further moderate reduction of RNS production by phagocytes. In the case of C 60 + Cis nanocomplex, the RNS production was extremely low (by 7 times compared with cells treated with Cis alone and by 6 times compared with untreated cells). One of the probable reasons for the reduction of RNS synthesis by the treated phagocytes can be oxidative burst. Peripheral blood phagocyte arginase activity Arginase converts l-arginine into l-ornithine (precursor of proline and polyamines) and urea. iNOS and arginase can compete for the same substrate l-arginine. Overexpressed arginase can affect iNOS activity and vice versa. Increase in phagocyte arginase activity is one of the features of alternatively activated cells participating in tissue remodeling and the resolution of inflammation. Whereas the reduction of arginine metabolism through arginase activity along with iNOS activation is a sign of classical phagocyte metabolic profile, key aspect of the inflammation (;Okabe and Medzhitov 2016;). Arginine metabolism with arginase is associated with TLRs activation (;). The impact of C 60 fullerene (as well as antineoplastic drugs) on arginase activity is virtually not explored. In our experiments, C 60 fullerene when used alone did not affect this phagocyte metabolic process (Fig. 8). Fig. 8 The effect of C 60 fullerenes and their nanocomplexes with anticancer drugs Cis and Dox on arginase activity of nonsensitized human peripheral blood phagocytes. Phagocytes in buffy coat were treated with mentioned preparations for 30 min, then arginase activity was analyzed by colorimetric method (see "Methods"). All results are expressed as mean ± SD of three independent experiments. *p < 0.05 compared to control, as analyzed by unpaired t test Dox slightly downregulated phagocyte arginase activity. Complexation of Dox with C 60 fullerene abrogated its inhibitory effect on phagocyte arginase activity. Cis used alone as well as in the nanocomplex with C 60 fullerene did not affect phagocyte arginase activity. The ability of C 60 fullerene to abrogate pro-inflammatory polarization of arginine metabolism in phagocytes caused by Dox can facilitate the reduction of cytotoxic side effect of the drugs. Peripheral blood phagocyte endocytosis Endocytosis (phagocytosis of solid particles and pinocytosis) is an essential part of phagocyte metabolism. Endocytosis is regulated by TLRs activation and is associated with ROS generation (;McCoy and O'Neill 2008;). The engulfment machinery differs in mononuclear (monocytes, macrophages, dendritic cells) and polymorphonuclear (neutrophils or granulocytes) phagocytes. In addition, phagocytosis plays different roles for the metabolic profile of mono-and polymorphonuclear cells. In neutrophils, phagocytosis is often associated with netosis and inflammation (). In monocytes/macrophages, phagocytosis can be associated with inflammation resolution and anti-inflammatory (alternative) metabolic profile (Soehnlein and Lindbom 2010). Treatment with all investigated preparations did not affect significantly the number of phagocytizing granulocytes in tested probes (phagocytosis percentage) (Fig. 9a). Fig. 9 The effect of C 60 fullerenes and their nanocomplexes with anticancer drugs Cis and Dox on phagocytic activity of nonsensitized human peripheral blood granulocytes (a, c) and monocytes (b, d). Whole blood samples were treated with mentioned preparations for 30 min, then phagocytic activity was analyzed by flow cytometry (see "Methods"). All results are expressed as mean ± SD of three independent experiments. *p < 0.05 compared to control, **p < 0.01 compared to control; ***p < 0.001 compared to control; # p < 0.05 compared to cells treated with Dox alone; ## p < 0.01 compared to cells treated with Dox alone; § p < 0.05 compared to cells treated with Cis alone, as analyzed by unpaired t test Whereas the proportion of phagocytizing monocytic cells in analyzed pool was increased after the treatment with Dox and C 60 fullerene used alone as well as with C 60 + Dox. All tested substance influenced phagocytosis intensity in monocytes and granulocytes. C 60 fullerene used alone stimulated phagocyte engulfment intensity. PIs in treated phagocytes were significantly higher than that in untreated cells: 3.7 times for granulocytes and 1.6 times for monocytes ( Fig. 9c and d). It is noteworthy that phagocytosis intensity in monocytes and granulocytes was characterized by significant individual variability that may result from nanoparticle size heterogeneity. Dox used alone also increased intensities of monocyte and granulocyte phagocytosis. However, its complexation with C 60 fullerene abolished this effect. Cis used alone and in the nanocomplex with C 60 fullerene did not affect phagocyte endocytosis intensity. U937 ROS generation As mentioned above, we observed the dramatic effect of C 60 fullerenes and their nanocomplexes with anticancer drugs on intracellular ROS generation by peripheral blood phagocytes. Induction of ROS generation is considered as one of the mechanism of cytotoxic effect of C 60 fullerenes (;;). Recently, activating ROS generation has become a promising approach for selective cancer treatment (Fruehauf and Meyskens 2007;). Tumor cells exhibit higher basal levels of ROS than normal cells. The intrinsic ROS stress is also characteristic for leukemic cells (). This metabolic feature makes them more vulnerable to damage by further ROS insults induced by exogenous agents, whereas nonmalignant cells better tolerate the oxidative stress. It gave us the reason to estimate the effect of C 60 fullerenes and their nanocomplexes with anticancer drugs on intracellular ROS generation phagocytic cells. To this end, we used U937 cell line. U937 is one of the most widely used myeloid cell lines (). U937 cells of histiocytic lymphoma origin are arrested in a promonocyte/monocyte stage of differentiation (). A genetic analysis by Strefford et al. showed that U937 bears the t(10;11)(p13;q14) translocation. This results in a fusion between the MLLT10 (myeloid/lymphoid or mixed-lineage leukemia) gene and the Ap-3-like clathrin assembly protein PICALM (Clathrin assembly lymphoid myeloid leukemia), which is likely important for the tumorous nature of the cell line. Treatment exposure was 24 h. It suggests that the effect of nanoformulation could be mediated through the interaction with both membrane-associated and intracellular receptive structures (). Intracellular ROS levels in U937 cells, as indicated by fluorescence intensity, significantly increased in response to the treatment with C 60 fullerenes and their nanocomplexes with anticancer drugs (Fig. 10). Oxidative stress caused by the treatment with mentioned preparations was more pronounced in transformed phagocytic cells (U937) than in normal peripheral blood phagocytes (see Fig. 6). The level of intracellular ROS in U937 cells after the treatment with C 60 fullerenes used alone was 1.8 times higher than that in treated circulating monocytes. As shown in Fig. 6, the complexation of Cis with C 60 fullerene caused the downregulation of its stimulatory effect on intracellular ROS production in circulating phagocytes. Unlike, the complexation both anticancer drugs with C 60 fullerene did not influence their pro-oxidant effect on transformed phagocytes. Thus, C 60 fullerenes used alone and in nanocomplexes with anticancer drugs caused dramatic increase of intracellular ROS generation in transformed phagocytes. As mentioned above, U937 cells represent the malignantly transformed myeloid cells that are arrested in a promonocyte/monocyte stage. Thus, these cells have an even more immature phenotypes in comparison with circulating normal monocytes, which turned out being more sensitive to the pro-oxidant effect of nanoformulations than neutrophilsdifferentiated mature cells. These results allowed us to speculate that immature state and increased basal ROS level make malignantly transformed cells more receptive to prooxidant effects of C 60 fullerenes and their complexes, and potentially can be the reasons of increased cytotoxicity of nanoformulations toward transformed cells. Toxic effects of C 60 fullerene and its nanocomplexes with anticancer drugs Cis and Dox on normal and malignant phagocytes To investigate the cause-effect relationship between pro-oxidant activity and cytotoxicity of C 60 fullerenes and their nanocomplexes with anticancer drugs toward normal and transformed phagocytes, Annexin V/PI double staining of these cells treated with mentioned compounds was conducted. As shown in Fig. 11, the total number of dead normal monocytes after the treatment with C 60 fullerenes was 6.7% vs 11.8% in the case of U937 cells. Sensitivity of transformed promonocytes/monocytes to anticancer drugmediated death was also higher than that in normal monocytic cells. The complexation of Cis with C 60 fullerenes resulted in slight increase in the death rate in transformed phagocytes but not in normal phagocytic cells compared with the cells treated with Cis alone. C 60 fullerenes being used in the nanocomplex with anticancer drugs influenced the mode of cell death induced by Cis and Dox. Annexin V-FITC/PI assay allows differentiating between early apoptotic cells (An+ PI−), late apoptotic cells (An+ PI+), and necrotic cells (An− PI+). In phagocytes treated with anticancer drugs used alone, we Fig. 10 The effect of C 60 fullerenes and their nanocomplexes with anticancer drugs Cis and Dox on intracellular ROS generation in U937 cells. U937 cells were treated with mentioned preparations for 30 min, then were stained with carboxy-H2DCFDA (see "Methods"). The fluorescence intensities of 10,000 cells were then analyzed by BD-FACS Calibur. All results are expressed as mean ± SD of three independent experiments. *p < 0.05 compared to basal level of ROS generation in control; # p < 0.05 compared to corresponding cells without treatment with PMA, as analyzed by unpaired t test registered higher necrosis level compared with their counterparts treated with C 60 + Cis and C 60 + Dox nanocomplexes. Apoptosis:necrosis ratio in cell samples treated with anticancer drugs was 2:1 (on average), whereas in the cells treated with C 60 + Cis and C 60 + Dox nanocomplexes, this ratio was 10:1 (on average). Conclusions Thus, C 60 fullerenes modulate phagocyte functions: stimulate phagocytic activity and downregulate RNS generation. In addition, C 60 fullerenes can exert direct cytotoxic effect on phagocytes, more pronounced in the case of malignant cells. This cytotoxic effect is associated with the vigorous induction of intracellular ROS generation. More pronounced pro-oxidant effect of C 60 fullerenes was also observed in transformed phagocytes. We hypothesize three main reasons for the increased receptivity of transformed cells to pro-oxidant, and thus, cytotoxic effects of nanoformulations. Two of Fig. 11 U937 cells (b) are more vulnerable to cytotoxic effect of C 60 fullerenes and their nanocomplexes with Cis and Dox compared with normal monocytes (a). Cells were treated with mentioned compounds at the concentration of 0.15 mg/ml for 24 h. After culturing, cells were stained with annexinV (AnnV)/propidium iodide (PI) and analyzed by flow cytometry. Control: cells were incubated without any additional agents. The average values for 4 independent experiments are presented. *p < 0.05 compared with untreated cells; # p < 0.05 compared to cells treated with Dox alone; § p < 0.05 compared to cells treated with Cis alone, as analyzed by unpaired t test them are the increased basal level of ROS and the immature state of transformed cells. The third reason can be the overexpression of enzymes responsible for metabolism of xenobiotics, including CYP (that is characteristic for malignant cells), if one takes into account that CYP can be considered as one of the intracellular receptive structures for C 60 fullerenes and their nanocomplexes. On the other hand, C 60 fullerenes have the ability to downregulate pro-oxidant effect of Cis on normal cells. Our results indicate that C 60 fullerenes have both pro-oxidant and antioxidant properties. These results are consistent with the observations of other scientific group (Markovic and Trajkovic 2008). |
Incorporation of radioactive precursors into filarial larvae of Brugia developing in susceptible and refractory mosquitoes. The incorporation of tritiated precursors injected into mosquito hosts parasitized by developing filarial larvae of Brugia patei has been studied by autoradiography in 2 species of mosquito, Aedes togoi in which filarial development was normal and Anopheles labranchiae atroparvus in which filarial development was abnormal. In both mosquito hosts there was significant incorporation into 4--5-day-old developing larvae of uridine and amino acids (isoleucine, leucine, valine, arginine, lysine, cystine, methionine, phenylalanine, tyrosine, tryptophan, histidine, and proline), although lower incorporation of methionine, tyrosine, and tryptophan was found during abnormal development. No incorporation of thymidine, hydroxytryptophan, dopa, or carbohydrate was found at this stage of larval development. Some incorporation of glucose and dopa was found in or around earlier stages of development in An. l. atroparvus. Mosquito flight muscle showed lower incorporation of glucose, but not of amino acids, around the site of filarial parasite development. The flight muscle of An. l. atroparvus showed a higher level of incorporation of lysine compared to that in A. togoi and higher levels of lysine and valine were found in the abnormally developing filarial larvae in the refractory mosquito. |
<gh_stars>1-10
// 异步加载组件的方法
import IndexPage from './page/Index';
// 配置index页的路由信息
export default {
path: '/index',
enable: true,
component: IndexPage,
};
|
Change of Thermo-Mechanical Properties of Optical Fibers Aged in CTAC Aqueous Solution Fiber-optic sensors are mostly used for in situ measurements of diverse chemical composition of industrial surfactants employed in industry as detergents, emulsifying and dispersing agents, coatings, and pharmaceutical adjuvants. These optical sensors are often used in wet chemical environments in which the temperature can be high.The purpose of this work is to study the mechanical behaviour of optical fibers in contact with CetylTrimethylAmmonium Chloride in aqueous solution (CTAC) at different immersion durations and different temperatures.Optical fibers were submitted to dynamic bending test under different velocities.Result analysis demonstrates that immersion in CTAC drastically decreases the fiber strength particularly when immersed for long aging periods at high temperatures. |
What a week in the NFL.
Article continues below ...
We honor America, question the Jets and Packers, predict the AFC West, help the NFL with the Vikings schedule, and pay tribute to Brett Favre. Seriously.
Time for the SCHEIN 9, assuming Sal Alosi doesn’t trip us.
1. America’s Team
If you watch “Cosmic Schein” on FOXSports.com or listen to the Sirius Blitz, you know that we call the Jaguars “America’s Team” because they are always slotted 32 out of 32 in the annual Harris Interactive Poll ranking a team’s popularity. The Jaguars barely register as Jacksonville’s team, which is a true shame. “America’s Team” is a catchy underdog role that players like Maurice Jones-Drew and David Garrard have embraced when they come on the radio show, trying to get some rightful pub for the Jaguars.
And preseason, let’s be honest, the morale was low. Nobody picked the Jags to make the playoffs. Nobody picked the Jags to finish third! Most had the Colts, Texans and Titans all rightly ranked ahead. The owner and coach publically criticized the quarterback in the offseason. The unknown general manager seemingly reached for a player in the draft. There were talks about the team moving at some point to Los Angeles or London. And the big debates weren’t about win totals, they were about the dates the quarterback would get benched and when the head coach would get fired.
Well, something happened en route to unemployment, relocation and the cellar of the AFC South. The Jaguars, after another slow start, after looking flat-out wretched and hapless in gruesome losses, have won five of their last six. And with a win in Indy on Sunday, the Jaguars would clinch the AFC South.
And here’s the beauty: I fully expect the Jaguars to beat the Colts on Sunday.
Last season, the Jaguars marched into the Meadowlands and beat the Jets to control their own destiny for making the playoffs. And then the Jaguars collapsed. Both Garrard and Jones-Drew have told us that they were motivated all offseason by what happened, vowing to never let it take place again. While I most certainly believed them, never did I think it would translate to a winning season.
Monday on Sirius NFL Radio, Mike Sims-Walker explained, “Last year we didn’t finish. We had a lot of young guys who hit the wall. Those guys have returned and matured. We now have a different mind-set and attitude. Everyone is buying into Jack (Del Rio) and Gene (Smith). There is not a better coach in this league than Jack. He’s been so positive, keeping us together after the bad losses early, after the San Diego and Philly games. He’s never shied away from criticism. He is the coach of the year.”
Ah, yes — Jack and Gene.
Del Rio seemed to be a goner after those aforementioned losses earlier in the year. Del Rio, who recycles assistants at an alarming rate, seemed just to be buying time. But according to Sims-Walker, Del Rio convinced his 1-2 Jaguars to focus and avoid criticism for the Week 4 showdown against the Colts, winning it on a Josh Scobee bomb field goal at the gun. After the Jaguars’ losing way reared its ugly head again with consecutive losses to the Titans (in disgraceful fashion on Monday Night Football) and the Chiefs, Del Rio rallied the troops for a shocking win in Dallas before the bye week.
Have the Jags been lucky? Absolutely. Are you questioning how good they actually are? Of course you are. When the Jaguars are bad, they are an eyesore. But they shouldn’t apologize for Scobee’s field goal, or Mike Thomas’ catch on the Hail Mary against Houston, or Indy being ravaged by a stunning number of key injuries. And in the five losses, Jacksonville has looked really, really bad. But when they’ve been on, they’ve had sizzle, and they’ve also earned it.
Garrard has evened out his play, survived an early witch hunt to have him benched or cut, and has become the leader that the coach and owner begged him to be. Marcedes Lewis has been a touchdown machine at tight end.
But the real difference maker is Jones-Drew. When we asked Colts coach Jim Caldwell about what makes MJD so good, he quipped, “Besides the balance, power, speed, deceptiveness and the fact he can catch the ball out of the backfield?”
Or as Sims-Walker said, “I need new pads and I love it. We are a running team. Our running game is so good. Maurice is so small but so powerful. And our offensive line has gelled." At 1,278 rushing yards, doing it with injuries along the offensive line, Jones-Drew has been the single most valuable running back in the league this year and worthy of being named to the All-Pro team. Jones-Drew has carried the Jaguars. He leads the NFL in rushing yards and is a top-five candidate for league MVP.
The defense has been up and down all year but clutch during the recent stretch. Gene Smith’s “reach” of defensive tackle Tyson Alualu has played beyond expectations. Smith smartly traded for Kirk Morrison, who has anchored the linebackers. And the defense deserves a ton of credit for surviving the season-ending loss of pass rusher and team leader Aaron Kampman.
I get it. Nobody wants to see the Jaguars in the playoffs. You want Peyton Manning. He’s an all-time great.
Jacksonville is used to being disrespected. In Sims-Walker’s words, “It would mean everything to be the AFC South champs. Nobody on the outside thought it was possible. To bring a division title to Jacksonville, when everyone talked Los Angeles in the offseason, it would mean everything to the guys and everything to the city.”
It would. Now, unlike last year, seal the deal and make it happen in the big spot. Prove you have learned your lesson. It’s America’s Team. Everyone loves the underdog. The Jaguars are good for the soul, good for the country.
2. Road warriors
Great job by the NFL putting the Giants and Vikings in Detroit after the Metrodome roof collapsed. You couldn’t play outside at the University of Minnesota because the Giants packed for an indoor game.
But the Vikes and Giants felt like a preseason atmosphere on Monday night. With ample time to prepare, the Vikings should play the remainder of their home games, starting this Monday against the Bears, outside in Minnesota. That will give them a distinct advantage with the home crowd.
3. Wild, wild West
Sometimes stats lie. The Chargers outgaining the Chiefs 426 to 67 tells you all you need to know. And I don’t want to hear that K.C. was missing Matt Cassel. The Chargers were that good.
But here’s a strong take for you.
I think BOTH the Chargers and Chiefs run the table. San Diego is 7-6 and has the Niners at home on Thursday before playing at the pathetic Bengals and Broncos.
Kansas City is in St. Louis then has the Titans and Raiders at home.
Think about it.
Which puts the pressure on . . .
4. Same old Jets
I thought the culture was changed. This was a pathetic, wretched, fraudulent loss at home to Miami. How do you come out flat against the Dolphins after losing by 42 to the Patriots? How do you lose a game in which Chad Henne completed five total passes and was begging you to win the game?
Mark Sanchez was horrible for the third straight game. Rex Ryan’s in-game decisions were awful. The Jets offensive line and running attack were nonexistent.
The Jets are 9-4 with dates in Pittsburgh and in Chicago coming up. Uh-oh.
Panic in New York and panic in . . .
5. Pack it in, Packers?
Injuries happen. And Aaron Rodgers’ second concussion is a major concern. But the fact is, Green Bay was leaving points on the field in Detroit even before Rodgers went down.
At 8-5, the Packers have the Patriots, Giants and Bears. My preseason Super Bowl team is in trouble, especially after the Bucs survived Washington to improve to 8-5.
6. Schein’s anatomy
Perhaps lost in the Philly win in Dallas was the Eagles losing Stewart Bradley, the pulse of their defense. And the Eagles’ young pass rusher, Brandon Graham, is done for the campaign. These are major, devastating injuries for the Eagles heading into the showdown in New York on Sunday against the Giants.
7. My Guys
Tom Brady — Snow? Stingy Bears defense? No problem. Brady is playing some of the best football of his future Hall of Fame career and that speaks volumes about his current brilliance.
Brandon Fields — In a game in which field position was everything, the Miami punter was the game MVP, averaging 56 yards per punt.
Troy Polamalu — Another week, chalk up another splash play for the star safety. Polamalu is in the mix for Defensive Player of the Year and Comeback Player of the Year.
Malcolm Jenkins — Talking to the New Orleans defensive back on Sirius NFL Radio, he is very comfortable now playing safety. It certainly shows. Jenkins returned one of his two picks for a touchdown.
Jay Feely — The Cards kicker ran in a touchdown on a trick play, giving him more points than his old Jets team for the week. Oh, yeah, he kicked five field goals, too.
8. My Goats
Sal Alosi — Tripping a player while you stand on the sideline is a fireable offense. It is as bush league as you can imagine. Thank goodness that Nolan Carroll wasn’t seriously hurt. The Jets suspended the strength and conditioning coach for the rest of the season and fined him $25,000.
Tashard Choice — How do you ask division rival Mike Vick for an autograph on the field after a loss? That’s inexplicable!
Mike Shanahan — His game and clock management at the end of the first half was awful.
The Redskins field-goal team — Graham Gano missed two easy field goals. And how do you botch an extra point to tie the score at the end with nine seconds to go?
Tarvaris Jackson — My guy had a chance. My guy failed.
9. Three nuggets of wisdom
• I’m so sick of the annual Brett Favre drama, his self-love, the love his lackeys in the media give him. But he deserves so much credit for being an iron man all these years.
• The Ravens win a wild one on Monday night with Josh Wilson picking off Matt Schaub in overtime for a “walk-off” win. But Baltimore blew the game in the fourth quarter. If the defense doesn’t play better, the Ravens will be walking out of the playoffs early.
• Imagine this. An overtime decided on defense! Here’s a brilliant idea: Let’s change the overtime rules! |
China’s foreign minister has urged Cambodia to resolve its ongoing political deadlock and quickly form a new government.
Foreign Minister Wang Yi, who was on a visit to the country, made his remarks Wednesday in Phnom Penh following meetings with his counterpart, Hor Namhong, and Prime Minister Hun Sen, amid a political crisis following last month’s national elections.
China's official Xinhua news agency said Wang congratulated Hun Sen on an election victory for his Cambodian People’s Party, despite ongoing protests by the opposition, which has rejected preliminary results showing a CPP win. Reporters were not allowed questions during the briefing.
“China is a good friend and good partner of Cambodia,” Wang said. “We hope that all [the] parties of Cambodia will peacefully discuss [the situation] in order to quickly put in place a new National Assembly and new government."
The opposition Cambodia National Rescue Party says last month’s election voting was deeply flawed and marred by irregularities. This has raised the prospect of a political deadlock, with the opposition threatening to boycott the National Assembly and calling for mass demonstrations if a proper investigation is not held.
The Rescue Party has demanded that an investigation be overseen by the U.N. or other outside observer, a demand the CPP has rejected. Both sides on Tuesday agreed to form an investigative committee.
Wang told reporters Wednesday that China continues to support Cambodia, but does not condone “outside” interference in its political process.
Independent analyst Lao Monghay said Wang’s visit likely would help the ruling party, which has had less support from the West following the election.
“It’s strengthening the confidence of the Cambodian People’s Party, while there are few governments and foreign parties recognizing the election as free and fair,” he said.
He added that Cambodia’s election, which saw a surprising resurgence for the opposition, echoes the movements of the Arab Spring and has made it of interest to the West and the U.S.
“It can extensively spread to Vietnam, Laos and China, and so on,” he said. “This is only a matter of time."
Sok Touch, rector of Khemarak University, said China does not want to lose influence over Cambodia, especially with Burma making democratic reforms and looking toward the West.
This report was produced in collaboration with the VOA Khmer service |
Characterization of cytoskeleton features and maturation status of cultured human iPSC-derived cardiomyocytes Recent innovations in stem cell technologies and the availability of human induced pluripotent stem cell-derived cardiomyocytes (hiPSC-CMs) have opened new possibilities for studies and drug testing on human cardiomyocytes in vitro. Still, there are concerns about the precise nature of such reprogrammed cells. We have performed an investigation using immunocytochemistry and confocal microscopy on several cellular features using commercially available hiPSC-CMs. For some selected developmentally regulated or cardiac chamber-specific proteins, we have compared the results from hiPSC-derived cardiomyocytes with freshly isolated, ventricular cardiomyocytes from adult rats. The results show that all typical cardiac proteins are expressed in these hiPSC-CMs. Furthermore, intercalated disclike structures, calcium cycling proteins, and myofibrils are present. However, some of these proteins are only known from early developmental stages of the ventricular myocardium or the diseased adult heart. A heterogeneous expression pattern in the cell population was noted for some muscle proteins, such as for myosin light chains, or incomplete organization in sarcomeres, such as for telethonin. These observations indicate that hiPSC-CMs can be considered genuine human cardiomyocytes of an early developmental state. The here described marker proteins of maturation may become instrumental in future studies attempting the improvement of cardiomyocyte in vitro models. Introduction There is a demand in the life sciences for a constant supply of highly differentiat-ed cell types such as cardiomyocytes, hepatocytes, endothelial cells, and neuronal cells that often show a limited life span, minimal proliferation capacity, and rapid de-differentiation in primary cell culture. For primary human cardiomyocytes, there is practically no supply since healthy organs are rarely available and the cells quickly expire without perfusion. Embryonic stem cells (ESC) originating from the blastocyst stage of early human embryos raise ethical questions and cannot be obtained from the same patient for cell therapies. The few immortalized cardiomyocyte cell lines that are commercially available are of animal origin and provide little advantage over primary cardiomyocytes from neonatal rodents. However, since the first publication in 2006 with murine and 2007 with human cells by Yamanaka et al., 1 so-called induced pluripotent stem cells (iPSC) have rapidly become a popular tool. Cells differentiated from iPSC appear as a valid alternative to primary cells for laboratory studies, drug development, and regenerative medicine. In the last few years, there have been considerable advances in the methods of 're-programming' somatic cells to iPSCs using non-viral transfection and genome editing. 2 Subsequently, these cells can be used to produce differentiated human cell types such as cardiomyocytes (hiPSC-CMs) using established protocols. 3,4 Existing applications of hiPSC-CMs include drug safety testing, disease modeling, and tissue engineering. 2,5 With this method, it is possible to obtain cells from adult patients with cardiac disorders, with a well-documented disease and family history, to study the resulting cellular phenotype in detail. Additionally, hiPSC-derived cell types can be genetically modified, for example, to test a gene therapy approach in vitro demonstrating the attempted rescue in a relevant genetic setting. A remaining concern with hiPSC-CMs is their reportedly immature developmental status and diverse phenotypes. 6,7 We have investigated several cellular features in commercially available hiPSC-CMs by immunocytochemistry and confocal microscopy to characterize this type of model system for future use in toxicology and basic research. For some selected proteins, we have compared expression and organization with freshly isolated, adult cardiomyocytes. This study provides a 'snapshot' of several cytoskeleton proteins of this cell type in culture. We did not attempt to improve maturation of the cells outside of culturing for ten days to replicate a typical experimental situation where the non-proliferating cells are used from frozen stock, treated with test compounds, and analyzed. Cell culture Human iPSC-derived cardiomyocytes (hiPSC-CMs, ax2505) were obtained from Axol Bioscience (Little Chesterford, Cambridge, UK). The original hiPSC cell line was produced using an episomal vector and CD34+ cord blood cells from a male newborn donor. Cryopreserved cardiomyocytes were rapidly thawed, then diluted in maintenance medium with supplements (Axol Bioscience, ax2530) and 10% fetal bovine serum. The cells were then seeded into dishes with a glass-bottom well of 10 with maintenance medium without serum for the first time, and then changed every two days. Immunocytochemistry of cultured cells and confocal microscopy Cells cultured for 10 days on glass-bottom dishes (MatTek) were washed with phosphate buffered saline (PBS) then fixed with 3% para-formaldehyde in PBS for 15 min, permeabilized with 0.2% Triton-X100 (Sigma-Aldrich) in PBS for 10 min, incubated for 30 min with bovine serum albumin (Sigma-Aldrich) 1 mg/mL in PBS at room temperature, incubated overnight with primary antibodies at 4°C, washed 3 times with PBS and incubated for 1 h with secondary antibodies goat anti-rabbit or goat antimouse coupled to AlexaFluor488 or AlexaFluor546 fluorescent dyes (Thermofisher Scientific/Molecular Probes, Waltham, MA, USA) at room temperature in the dark. Primary antibodies used in this study are listed in Table 1. Antibodies to embryonic heart myomesin (EH-myomesin) and M-protein were a kind gift from Dr. Irina Agarkova (Federal Institute of Technology, ETHZ, Zurich, Switzerland). Preparations were embedded using non-hardening solutions with anti-fading protection with or without DAPI (SlowFade Diamond, Invitrogen, Carlsbad, CA, USA; or Vectashield/DAPI, Vector Laboratories, Burlingame, CA, USA). Preparations were examined with a Zeiss LSM 710 confocal microscope using EC Plan-Neofluar 40x/1.30 and Plan-Apochromat 60x/1.40 oil immersion lenses (Carl Zeiss, Oberkochen, Germany). Manual counting of the respective fraction of cells showing protein expression and sarcomeric organization was conducted using at least 200 cells from multiple fields of confocal images from 3 independent immunocytochemistry experiments and different batches of hiPSC-CMs. Measurement of morphologic features was performed using the software NIH-ImageJ (U.S. National Institutes of Health and available on the Internet at http://rsb. info.nih.gov/nihimage/) on confocal images obtained from cardiomyocytes immunostained for the sarcomeric M-line marker myomesin. A lineprofile was created along the myofibril and distances between peaks were measured. Only cells at resting length were evaluated. For the form factor, maximum lengths and widths of cells in low-density cultures were measured and expressed as a ratio. At least ten images showing several cells were used for the analysis of each feature and the data were statistically analyzed using the twotailed Student's t-test (Graphpad Prism). Isolation of adult rat cardiomyocytes and immunocytochemistry Adult rat ventricular cardiomyocytes (ARVM) were isolated as previously described 8,9 from 3 months old (350-400 g) male Wistar rats, seeded onto gelatine-coated chamber slides and immediately fixed with 4% p-formaldehyde in PBS for 20 minutes. The preparations were permeabilized with 1% Triton-X100 (Sigma-Aldrich) in PBS for 20 min, incubated for 1 h with bovine serum albumin (Sigma-Aldrich) 1 mg/mL in PBS at room temperature, incubated overnight with primary antibodies at 4°C, washed 3 times with PBS and incubated 2 h with secondary antibodies coupled to Alexa fluorescent dyes (Thermofisher/ Molecular Probes) at room temperature. Preparations were then washed, embedded and examined as described for the hiPSC-CMs. Only primary antibodies equally recognizing both human and rodent protein isoforms were selected for this experiment. All experiments were carried out according to the Swiss animal protection law and with the permission of the canton Bern veterinary office (license BE32/14). This investigation conforms to the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health (NIH Publication No. 85-23, revised 1996). Cytoskeleton and sarcomeric proteins Cryopreserved hiPSC-CMs were thawed and cultured for ten days as described in the materials and methods section. The cells were then fixed and immunostained for several cytoskeleton and sarcomeric proteins ( Figure 1). In several co-stainings, a well-described marker of the sarcomeric M-line, myomesin, 10,11 served to demonstrate cellular identity and myofibrillar content in the cultured cells (Figure 1 C2, E1 and Figure 2 C2,D2,E2). Essential muscle proteins comprise the family of myosin heavy chains. Their increase in expression has been used as a cardiogenic marker in early embryonic development. 12 Immunolabeling of alpha-or beta-myosin heavy chain proteins resulted in a nearly identical pattern that only slightly varied in intensity between cells ( Figure 1A). The myosin heavy chain isoforms were also found in the freshly isolated ARVM ( Figure 3F). Two myosin light chain isoforms have been described as ventricular and atrial-specific proteins in the adult heart (MLC2v, MLC2a). 13 We have found hiPSC-CMs expressing either one of these myosin light chain isoforms or both, to a different extent in the cultures ( Figures 1B and 4A). As demonstrated by the immunostaining of freshly isolated cells from adult rats, the atrial-specific isoform MLC2a does not occur in ARVM ( Figure 3A). Desmin, the predominant intermediate filament protein in striated muscle, has previously been demonstrated in hiPSC-CMs and ARVM in our lab. 14,15 Vimentin is an intermediate filament protein that is expressed in fetal cardiomyocytes and in non-myocytes in adult tissue and was reported to disappear from cardiomyocytes postnatally. 16 Here we have stained the cells for vimentin, and for myomesin to identify cardiomyocytes ( Figure 1C). Vimentin-positive filaments were found in virtually all hiPSC-CMs in the culture, but not in freshly isolated ARVM ( Figure 3C). In contrast to vimentin, desmin was observed only in a fraction of the cultured cells (not shown, for quantification see Figure 4A). The Z-line marker protein sarcomeric alpha-actinin was found in all hiPSC-CMs and was organized in sarcomeres ( Figure 1D). Microtubules were observed in all hiPSC-CMs in culture ( Figure 1D). The distribution pattern of microtubules was not correlated with other cytoskeleton elements as previously observed in cultured adult rat cardiomyocytes. 17 Cell-cell contact proteins We observed prominent labeling of adherens junction, desmosomal and gap junction proteins at cell-cell contacts in the cultured hiPSC-CMs (Figure 2 A,B). Betacatenin-positive junctions occurred at all cell-cell contacts in the monolayer culture ( Figure 2A). Connexin-43 was found in the form of gap junction plaques and a perinuclear distribution, as has previously been observed in other cardiomyocyte in vitro models. 18 Junctions positive for N-cadherin were found at cell-cell contacts partially colocalizing with desmoglein-2, a typical marker of desmosomes 19 ( Figure 2B). Immunostaining freshly isolated ARVM for N-cadherin demonstrated the strictly polarized localization of adherens junctions at the ends of the rod-shaped cells, the intercalated discs ( Figure 3G). We also investigated proteins of the so-called costamer complex that provides mechanical anchorage linking Zdiscs of sarcomeres via integrins to laminin and other extracellular matrix proteins along the long axis of the cells in tissue. 20 In the cultured hiPSC-CMs, laminin was found to enclose cultured cardiomyocytes ( Figure 1E), but no striated pattern along Z-discs was observed as it is present in the freshly isolated cardiomyocytes immunostained for laminin ( Figure 3H). Calcium-handling proteins Calcium-handling proteins are part of the excitation-contraction machinery of cardiomyocytes and are therefore of great interest for disease models and drug screening. Labeling of bridging integrator-1 (bin-1), a membrane scaffolding protein that is essential for the formation of T-tubules and their efficient function, 21 showed bin 1 positive ladders and striation along myofibrils Original Paper ( Figure 2C). Immunostaining of hiPSC-CMs for sarcoplasmic reticulum ATPase2a (SERCA2a) showed labeling in virtually all cells, although a striated pattern of the staining was mainly visible in elongated features with linear myofibrils ( Figure 2D). Immunolabeling for ryanodine receptors (RyR) showed the protein in all cells, although the level of organization differed in the cell population ( Figure 2E). The immunostaining of freshly isolated ARVM for junctophilin-2, a protein that is essential for the structural organization of the above mentioned sarcoplasmic reticulum/ryanodine-complex of calcium-handling proteins, 22 showed a striated pattern along the long axis of the cells ( Figure 3H). Developmentally regulated proteins In the following, we report about sarcomeric proteins that are known to undergo an isoform switch from fetal to the adult human myocardium or to show increasing levels of structural organization during development of the heart. We have investigated immunolabeled hiPSC-CMs ( Figure 5) and performed counting of the observed phenotypes ( Figure 4A). Telethonin (also known as T-cap) was described as a signaling and mechano-sensitive element in the heart. 23, 24 We have found a generally intense telethonin labeling of hiPSC-CMs in the 10-day old culture, although a sarcomeric striation pattern in the telethoninchannel was only visible in few places ( Figures 4A and 5A) that included myofibrils as demonstrated by the actin (Figure 5 A2) or myomesin labeling ( Figure 1B). Cardiac ankyrin repeat protein (CARP) (also known as cardiac-adriamycin-responsive protein or ankyrin repeat domain-1, ANKRD1) was found in the culture and showed a striation pattern in a sub-population ( Figures 4A and 5B). Additionally, a nuclear localization of CARP in hiPSC-CMs has been noted ( Figure 5 B1), as it has been described previously in primary cardiomyocytes. 25,26 In the freshly isolated adult cardiomyocytes, CARP was observed in all cells showing a distinct I-band striation pattern ( Figure 3D). Troponin-I isoforms have been reported to represent good markers for cardiomyocyte maturation because of the switch that occurs during development from the fetal form of slow skeletal TnI to the adult cardiac TnI. 27 We have used an antibody for the adult cardiac troponin-I isoform (Table 1) and found a patchy expression in the hiPSC-CM culture ( Figures 4A and 5C). That image shows a group of cells positive for troponin-I, but also at least one cardiomyocyte in the lower part of the image, that does not show any apparent labeling (Figure 5 C1). Cardiac troponin-T was found in all hiPSC-CMs in sarcomeric organization (not shown). Some sarcomeric and cytoskeleton proteins have been observed to re-appear in diseases of the adult heart as part of an adaptation and remodeling process of the myocardium. An example for this type of proteins is alpha-smooth muscle actin, that is only expressed in smooth muscle cells of the vasculature in the normal adult heart, but is found in the myocardium in patients with pathological cardiac hypertrophy. 28 In the hiPSC-CMS we found a patchy distribution of cells expressing alpha-smooth muscle actin ( Figures 4A and 5D). Cardiomyocytes were identified by sarcomeric striation in the all-actin image (Figure 5 D2). Freshly isolated ventricular rat cardiomyocytes Freshly isolated ventricular cardiomyocytes (ARVM) from healthy adult rats provide bona fide examples for the fully differentiated endpoint of cytoskeleton and sarcomeric development in the mammalian heart. We have used ARVM for immunocytochemistry of selected proteins that had been previously investigated in the cultured hiPSC-CMs (Figure 3). The same antibodies were used as for the hiPSC-CMs. Generally, immunolabeling was uniform in the population of isolated ARVM unlike the patchy, cell-autonomous distribution seen for some proteins in the cultured hiPSC-CMs. The results confirmed the specificity of the myosin light chain 2a antibodies that do not label ARVM ( Figure 3A). The labeling of telethonin showed a well-organized, distinct sarcomeric pattern for the titinbinding protein ( Figure 3B). The immunolabeling for cardiac troponin-T and vimentin confirmed, that vimentin is absent from adult ventricular cardiomyocytes ( Figure 3C). The labeling of CARP showed a distinct sarcomeric pattern ( Figure 3D). In fetal heart cardiomyocytes, all M-lines contain embryonic-heart myomesin (EHmyomesin), while this isoform is strongly down-regulated after birth. 29 On the other hand, M-protein is a typical protein of the adult M-line 30 as was confirmed here in ARVM ( Figure 3E). EH-myomesin was consistently found in all cultured hiPSC-CMs (Figure 2 C,D), but not in freshly isolated ARVM ( Figure 3E). Quantification of immunolabeling and sarcomeric organization A manual quantification was performed on populations of cells that showed either incomplete sarcomeric organization as was the case for telethonin (in 9.4% of all cells organized in sarcomers) and CARP (organized in 32%), or showed a patchy distribution of expression such as observed for cardiac troponin-I (found in 19% cells), alphasmooth muscle actin (in 23% cells), desmin (in 9.9% cells), and the atrial/ventricular light chain isoforms (72% MLC2v vs 25% MLC2a) ( Figure 4A). The quantitative assessment of morphological features can be instrumental to identify the maturation stage of cardiomyocytes 7 and therefore we evaluated sarcomere length and the overall shape of the cells in hiPSC-CMs compared to ARVM. The analysis of sarcomere length in cultured hiPSC-CMs compared to freshly isolated ARVM showed a significant difference between the two cell types ( Figure 4B) with a value of 1.7±0.13 m for hiPSC-CMs and 1.9±0.09 for ARVM (n=40, P<0.001). The form factor, i.e. the ratio of cell length divided by the width, demonstrates the extent of elongation ( Figure 4C). This parameter showed a significant difference between the cells, although there was a considerable amount of variability observed. The ratio was measured as 2.9±1.7 for hiPSC-CMs and 5.7±1.3 in ARVM (n=17, P<0.001). Discussion As iPSC-derived cells are growing in popularity in recent years for both basic research and drug development, there is a need for a thorough characterization and validation of these new tools. Still, questions are raised about the general limitations of iPSC-derived cardiomyocytes as compared to more traditional models. 31 In principle, all investigations published so far on the core properties of hiPSC-CMs have found typical features of cardiac muscle. The remaining issues are the immaturity of the cells and the resulting differences in gene expression when compared to the adult human heart. Also, considerable heterogeneity between production lots and within the cell population regarding cardiomyocyte subtypes has been observed. 2 For complex in vitro model systems with multiple cell types or tissue-like 3D-culture systems and tissue engineering approaches, survival, stability, and development over longer periods of time in culture are relevant. 14,32 These cytoskeleton markers and the corresponding immunoreagents that we have evaluated in this study provide effective tools for future investigations to assess the maturation status of tissue constructs, especially in the field of tissue engineering where the integration of the tissue in a fully developed host environment is crucial. Although a variety of techniques for characterizing hiPSC-CM has already been employed so far, immunocytochemistry provides a rapid and straightforward tool to obtain information both on the protein expression and differentiation level of cellular features. We have tested proteins that are either known to become organized later in development or represent associated proteins that may serve as signaling or mechano-sensitive elements. The intermediate filament proteins desmin and vimentin are both expressed in the embryonal and postnatal heart, but only desmin remains as the characteristic intermediate protein of both striated and smooth muscle in the adult. 16 The sarcomeric M-line protein myomesin serves as a bona fide marker for myofibrils and genuine cardiomyocytes in our experiments, although by electron microscopy, electron-dense M-lines as a further sign of maturation have not yet been found in standard cultures of hiPSC-CMs. 3 Regarding the regulatory myosin light chain proteins it has been reported, that almost all early hiPSC-CM's express MLC2a, and then at later time-points, the MLC2v expressing population becomes predominant, although some cells still expressed or co-expressed MLC2a 13 (and personal observations). All hiPSC-CMs were positive for cardiac troponin-T (cTnT) and the labeling was equal in the entire cell population (not shown). Regarding troponin-I, expression from the fetal TNNI1 gene (ssTnI) is completely replaced in the adult heart by TNNI3 (cTnI). 33 In cultured hiPSC-CMs, cTnI labeling demonstrated different levels in the culture with some cells showing no labeling at all, while others are strongly stained. Similarly, CARP labeling showed a patchy distribution, however with a distinctly visible sarcomeric structure in strongly expressing cells. CARP is sensitive to cardiotoxic cancer therapies and it is upregulated in left ventricular heart failure, 34 but it is also normally expressed in cardiomy- ocytes. Alpha smooth muscle actin (aSMA) is expressed in the fetal and postnatal heart. 28 A different picture was observed in telethonin-labeled cells. It has been proposed that telethonin (T-cap) is important for the maturation of sarcomeres, for sensing biomechanical stress, and that its expression rapidly reacts to stimuli that lead to muscular atrophy or growth. 32,35 Although the signal was generally intense in all hiPSC-CMs, a distinct sarcomeric pattern was visible only in some places, suggesting a robust expression of telethonin but incomplete organization of the protein into sarcomeres. These cytoskeleton proteins discussed above may become instrumental in the assessment of maturation in cardiac in vitro models, since these proteins demonstrate stages of a progression. Calcium handling proteins such as the sodium-calcium exchanger and ryanodine receptor have been detected in hiPSC-CMs previously by Western blotting in our laboratory. 14 Bin-1 is a membrane scaffolding protein in cardiomyocytes involved in the trafficking of calcium-handling proteins to the Ttubules via microtubules. 21 Its localization in an I band-associated pattern in the hiPSC-CMs suggests that it is either forming early T-tubules or a precursor protein complex. Intercalated disc proteins in the adult tissue are restricted to the ends of the cylindrical cardiomyocytes. 36 The pattern of adherens and gap junctions as well as desmosomes observed here in hiPSC-CMs appeared similar compared to other culture models of cultured cardiomyocytes. 18 It is therefore unlikely, that cell contact proteins are the first choice for measuring maturation of cardiomyocyte models unless the geometry of the cells could be modified to such extent that the strict polarization seen in tissue reappears. In addition to immunocytochemistry, additional methods such as evaluation of morphological parameters, protein biochemistry and gene expression studies will complete the toolbox for the development of improved in vitro models. Original Paper In conclusion, all proteins that are known to be expressed in primary cardiomyocytes have been found in the hiPSC-CMs on day 10 in culture, identifying them as genuine cardiomyocytes. However, the expression pattern and patchy distribution of some proteins suggest that the cell population is still developing. A summary of the qualitative assessment of expression and organization of the investigated proteins is presented in Table 2. Table 2. Qualitative summary of protein expression and organization as observed in this study (left column), and notes about the known pattern in the adult human heart as previously reported (right column). |
<reponame>usnistgov/WIPP-frontend
import {Component, OnInit, ViewChild} from '@angular/core';
import {BehaviorSubject, Observable, of as observableOf} from 'rxjs';
import {MatPaginator, MatSort} from '@angular/material';
import {catchError, map, switchMap} from 'rxjs/operators';
import {GenericDataService} from '../generic-data.service';
import {GenericData} from '../generic-data';
@Component({
selector: 'app-generic-data-list',
templateUrl: './generic-data-list.component.html',
styleUrls: ['./generic-data-list.component.css']
})
export class GenericDataListComponent implements OnInit {
displayedColumns: string[] = ['name', 'creationDate', 'owner', 'publiclyShared'];
genericDatas: Observable<GenericData[]>;
resultsLength = 0;
pageSize = 10;
pageSizeOptions: number[] = [10, 25, 50, 100];
paramsChange: BehaviorSubject<{index: number, size: number, sort: string, filter: string}>;
@ViewChild(MatPaginator) paginator: MatPaginator;
@ViewChild(MatSort) sort: MatSort;
constructor(
private genericDataService: GenericDataService) {
this.paramsChange = new BehaviorSubject({
index: 0,
size: this.pageSize,
sort: 'creationDate,desc',
filter: ''
});
}
sortChanged(sort) {
// If the user changes the sort order, reset back to the first page.
this.paramsChange.next({
index: 0, size: this.paramsChange.value.size,
sort: sort.active + ',' + sort.direction, filter: this.paramsChange.value.filter
});
}
pageChanged(page) {
this.paramsChange.next({
index: page.pageIndex, size: page.pageSize,
sort: this.paramsChange.value.sort, filter: this.paramsChange.value.filter
});
}
applyFilterByName(filterValue: string) {
// if the user filters by name, reset back to the first page
this.paramsChange.next({
index: 0, size: this.paramsChange.value.size, sort: this.paramsChange.value.sort, filter: filterValue
});
}
ngOnInit() {
console.log(this.genericDatas);
this.getGenericDatas();
}
getGenericDatas(): void {
const paramsObservable = this.paramsChange.asObservable();
this.genericDatas = paramsObservable.pipe(
switchMap((page) => {
const params = {
pageIndex: page.index,
size: page.size,
sort: page.sort
};
if (page.filter) {
return this.genericDataService.getByNameContainingIgnoreCase(params, page.filter).pipe(
map((paginatedResult) => {
this.resultsLength = paginatedResult.page.totalElements;
return paginatedResult.data;
}),
catchError(() => {
return observableOf([]);
})
);
}
return this.genericDataService.get(params).pipe(
map((paginatedResult) => {
this.resultsLength = paginatedResult.page.totalElements;
return paginatedResult.data;
}),
catchError(() => {
return observableOf([]);
})
);
})
);
}
}
|
Optical laser microphone for human-robot interaction: speech recognition in extremely noisy service environments Domestic robots are often required to understand spoken commands in noisy environments, including service appliances' operating sounds. Most conventional domestic robots use electret condenser microphones (ECMs) to record the sound. However, the ECMs are known to be sensitive to the noise in the direction of sound arrival. The laser Doppler vibrometer (LDV), which has been widely used in the research field of measurement, has the potential to work as a new speech-input device to solve this problem. The aim of this paper is to investigate the effectiveness of using the LDV as an optical laser microphone for human-robot interaction in extremely noisy service environments. Our robot irradiates an object near a speaker with a laser and measures the vibration of the object to record the sound. We conducted three experiments to assess the performance of speech recognition using the optical laser microphone in various settings and showed stable performance in extremely noisy conditions compared with a conventional ECM. GRAPHICAL ABSTRACT |
Blood plasma proteins serve a wide variety of functions in the human body such as the maintenance of blood volume, osmotic pressure, viscosity, and other important physical parameters of blood. A number of commercial processes have been implemented to separate and purify these proteins from blood plasma for therapeutic use. Some common methods for protein purification include precipitation with ammonium sulfate and similar salts, organic solvent precipitation with cold ethanol or acetone and other such alcohols and ketones, selective adsorption on gels, isoelectric precipitation, and chromatography by use of adsorbents. Still other processes for selectively fractionating and purifying blood proteins involve the use of amino acids, water-soluble organic polymers, and water-insoluble polyelectrolyte polymers containing basic amino groups.
Many protein extraction and purification techniques are based on altering the solubility of a desired protein in a biologic fluid such as blood plasma or a plasma solution by adjusting any of number of properties of the protein solution. Through the addition of salts or dilution of a solution, separation may be carried out in the range of low ionic strengths at which the interactions of proteins with electrolytes differ from each other, both in the isoelectric condition and when dissociated as acids or bases. The solubility of a protein may also be reduced by the addition of alcohols, acetone or other water miscible organic solvents to protein solutions. The balance between the precipitating action of alcohols and the salt interactions permits attainment of a variety of conditions under which the protein to be separated may be brought to a desired solubility. This balance may be altered based on the pH and temperature for each protein component; however, to avoid denaturing of the desired protein, sufficiently low temperatures should be maintained. Moreover, the pH may be controlled by adding a buffer such as an acetate or other buffers of known ionic strength, and adjusted so as to take advantage of the differences in the isoelectric points and the directions of the interactions with salts of the protein components to be separated. Finally, the concentration of protein may be maintained as low as possible to obtain a desired amount of protein precipitate to minimize protein-protein interactions. As the number of components in the solution or suspension increases or if multiple components have similar physical chemical properties, more of these variables must be accurately controlled to lower the solubility of a single protein type.
Because these extracted and/or purified plasma proteins may be used therapeutically in humans, such extraction and purification processes require rigorous quality control, such as monitoring and analytical testing of the biologic fluid to ensure that the end-product is both consistent and safe, and that the chemical properties of the mixture are kept consistent with the intended process design. For example, conventional monitoring techniques such as the use of enzyme-linked immunosorbent assays (“ELISAs”) or Surface Plasmon resonance may be used to determine the amount of antibody activity in a prepared sample. Other common monitoring methods such as reversed-phase, affinity or cation-exchange chromatography require the preparation of samples and controls and incubating for a certain period of time to monitor protein degradation. Additionally, SDS-Page electrophoresis techniques can be used to determine antibody impurities.
Unfortunately, many of these known monitoring and control techniques are costly and may require specialized antigens, reagents and equipment to perform the requisite analyses. In fact, many of these known techniques are unsuitable for monitoring and controlling separation processes in real-time, as they require sample preparation, specific incubation times and/or other time-consuming steps.
A need therefore exists for Process Analytical Technology (PAT) methods and systems for in-process monitoring of suspensions and solutions in an improved manner that may enhance protein process understanding, improve process control, and/or achieve consistent product quality. As used herein, PAT includes, for example, a system for designing, analyzing, and/or controlling manufacturing through timely measurements (i.e., during processing) of critical quality and performance attributes of raw and in-process materials and processes with the goal of ensuring final product quality. It would be beneficial if these desired attributes were obtainable using a method that is fast, reliable, low maintenance and involves an ease in the use of equipment, such as the use of equipment that does not need to be dismantled or the use of sensors that are easily cleaned. |
Unbiased.co.uk
History
Unbiased originated as a consumer hotline in 1989 to help consumers find an independent financial adviser (IFA). It was among the first organisations to promote IFAs to consumers and created a badge of ‘independence’ (or roundel) for IFAs to show they offer the highest quality of financial advice.
Online search capabilities were added in 2000. A postcode or regional search was introduced to provide users with a list of registered advisers in their area. Users were also granted the facility to filter their searches by areas of adviser specialism, qualifications, preferred method of payment and other criteria.
The company expanded its service in 2008 by launching a local whole of market mortgage adviser search, and again in 2009 with a local solicitor search. It was revealed that there was “a 25% jump in the number of people contacting advisers for help with a mortgage”. In 2009 unbiased.co.uk added a local solicitor search to its website followed in June 2011 by an accountant search feature. The website was also relaunched with a new hub of tools, calculators and information to help users manage their financial and legal affairs.
The search engine also began to feature on various partner websites including the Financial Times, Moneywise, the Telegraph and This is Money.
In 2015 unbiased began to introduce new features to its search to make it easier to find the right adviser. These included an automated matching service to connect the consumer with a suitable adviser, if the consumer's initially selected adviser(s) happened to be unavailable. Unbiased.co.uk also introduced facilities for advisers to be searchable across wider regions, to reflect those who service larger geographical areas, or who offer advice remotely. In 2016 it published a new mission statement: 'Connecting millions to great advice' and rebranded from 'unbiased.co.uk' to simply 'unbiased'.
Credentials
Unbiased has numerous media partners including the Financial Times, Moneywise, the Telegraph, This is Money and The Times. unbiased is listed with various trade bodies, such as AIFA, the AIC and SOLLA. Relevant UK government and regulatory bodies also refer consumers to this service and it is featured on websites such as the Citizens Advice Bureau, Directgov, Consumer Direct, Department of Work and Pensions, Financial Conduct Authority, Financial Ombudsman Service, Financial Services Compensation Scheme, Pensions Advisory Service, Pensions Ombudsman, Pensions Regulator and the Pensions Service. |
If the NFL Scouting Combine were the magical entity known as "the female form," then the 40-yard dash would be the boobs. (Actually, this is where we got the term, "(that thing) is the tits!") And when a player comes into the combine and runs a sub-4.30, scouts look at each other cautiously and whisper, "I think I see nipple."
Well, Kent State running back Dri Archer was wearing a low-cut tank top and no bra at this year's combine, and he "accidentally" bent over to pick up a pencil on his way to a 4.26.
It's not the fastest of all-time, as he had predicted, but according to the numbers I've culled from NFLCombineResults.com, it's tied with Jerome Mathis for the seventh-fastest. Archer becomes just the 19th prospect in Combine history to break 4.30 and if nothing else, it put him on the map for fans. And I'll repeat that:
For fans.
Any NFL scout worth Andy Reid's weight already knew who Dri Archer is. How do I know this? Because he was at the NFL combine. They already knew he was fast, probably had a good idea that he was the fastest in this year's draft pool, and some may have even suspected he would break the record. Not only that, but one scout did unofficially clock him at 4.18, which puts him as close to Bo Jackson as it does to Chris Johnson.
The question then becomes: So what?
Let's say that Archer had run an official 4.18, would it have helped his draft stock? By how much? Because the truth is that no matter how fast he was, it wasn't going to change the myriad of reasons that were always going to keep Archer out of the first round. Like that he's 5'8, 175 pounds, that he played at Kent State, that his senior season was far less productive than his junior season due to injuries.
There's an intimation that a tenth of a second means everything on the NFL level. If that's even true, it's only true for certain types of players. Perhaps it's the difference between drafting Champ Bailey or Dominique Rodgers-Cromartie (many years apart but their 40 times are neck-and-neck) but both of those guys already had all the other factors that make them a first round pick.
It wasn't going to get Trindon Holliday into the first round (or first five rounds) alone and you could make the argument that Holliday's stock was improved by zero factors because again, these scouts already know who's fast.
What I want to do today is take a closer look at the fastest players in combine history. Where were they drafted and how much better did they turn out than peers were were just a hair slower than them. Did speed matter nearly as much as draft position, and was draft position really determined by speed?
Or was it just Al Davis mucking up the order of things?
(It has a lot to do with Al Davis.)
First, let's look at the fastest of the fastest. Before Archer's time this year, 18 players have run a sub-4.30 forty-yard dash. Here they are, with Pro-Football-Reference's Career Adjusted-Value listed by their names, plus their overall position in the draft:
Under-4.30 Ordered by speed
Name Time AV Years AV/Yr Draft TEAM Trindon Holliday 4.21 4 3 1.33 197 HOU Jacoby Ford 4.22 12 4 3 108 OAK Chris Johnson 4.24 70 6 11.6 24 TEN Rondel Melendez 4.24 0 0 0 247 ATL Darrius Heyward-Bey 4.25 20 5 4 7 OAK Demarcus Van Dyke 4.25 3 3 1 81 OAK Jerome Mathis 4.26 4 3 1.33 114 HOU Stanford Routt 4.27 26 8 3.25 38 OAK CJ Spiller 4.27 30 4 7.5 9 BUF Marquise Goodwin 4.27 2 1 2 78 BUF Champ Bailey 4.28 154 15 10.26 7 WAS Stephen Hill 4.28 4 2 2 43 NYJ Mike Wallace 4.28 42 5 8.4 84 PIT Dominique Rodgers-Cromartie 4.29 33 6 5.5 16 AZ Johnny Knox 4.29 20 3 6.66 149 CHI Josh Robinson 4.29 6 2 3 66 MIN Jay Hinton 4.29 0 0 0
UDF Fabian Washington 4.29 27 6 4.5 23 OAK Average 4.27 25 4.22 4.18 85.61
Notables:
- The Oakland Raiders have drafted four of the ten fastest players in combine history. The Houston Texans and Buffalo Bills took four of the other six. These are three of the least-successful franchises of the combine era, with this data dating back to 1999.
- Three of the four fastest players in combine history went after the third round. Melendez was a "what the hell" pick of the Falcons in the late seventh round (though he went one pick ahead of Bryce Fisher, so it's not always a good idea to just "what the hell" any of your picks.)
- These are the fastest players in combine history, and the average draft position is middle of the third round.
- Chris Johnson didn't just run 0.02 seconds faster than Archer, he's also a 5'11, 195 lb back and though he went to East Carolina, he was coming off of his career-year (1,951 total yards, 23 touchdowns.) He was also in one of the strongest running back classes of recent memory: Darren McFadden, Jonathan Stewart, Felix Jones and Rashard Mendenhall all were taken ahead of him.
Matt Forte and Ray Rice went in the second round. Kevin Smith, Jamaal Charles and Steve Slaton went in the third. Tashard Choice, Ryan Torain, Tim Hightower and Justin Forsett were also in the same class.
So, how fast would Archer, a 5'8, 175 lb back, have to run in order to be a first round pick? I don't know... a three?
Can players do that? Can you run... a "three"?
A recent ESPN Insider article notes Football Outsiders' "Speed Score" index that doesn't just measure speed, but speed in combination with size.
Introduced on ESPN Insider back in 2008, Speed Score is Football Outsiders' metric for evaluating running back prospects. It's built on the simple idea that, because smaller backs tend to run faster than larger backs, we should be more impressed by a 4.5-second 40-yard dash from a 220-pound back than the same clock reading from a 170-pound back. As such, Speed Score incorporates a back's official time in the 40-yard dash with his weight to produce a measure of his speed given his size, using this formula: (Weight x 200)/(40 time^4).
Archer's 40-time was the fastest at the combine but his speed score ranked eighth. Which is still pretty good, but also a good indicator of why he could never really be a great running back prospect so... again, what is the point?
Archer had a speed score of 105.1 while Johnson had a speed score of 121.9. (The average being 100.)
The highest speed score this year was Damien Williams of Oklahoma, a 222-lb back that ran a 4.45.
Let's take a look at this list again, but order it by who has had the most valuable career and see if there's any correlation between the leaders and the losers:
Ordered by Value per Year
Name Time AV Years AV/Yr Draft TEAM Chris Johnson 4.24 70 6 11.66 24 TEN Champ Bailey 4.28 154 15 10.26 7 WAS Mike Wallace 4.28 42 5 8.4 84 PIT CJ Spiller 4.27 30 4 7.5 9 BUF Johnny Knox 4.29 20 3 6.66 149 CHI Dominique Rodgers-Cromartie 4.29 33 6 5.5 16 AZ Fabian Washington 4.29 27 6 4.5 23 OAK Darrius Heyward-Bey 4.25 20 5 4 7 OAK Stanford Routt 4.27 26 8 3.25 38 OAK Jacoby Ford 4.22 12 4 3 108 OAK Josh Robinson 4.29 6 2 3 66 MIN Marquise Goodwin 4.27 2 1 2 78 BUF Stephen Hill 4.28 4 2 2 43 NYJ Trindon Holliday 4.21 4 3 1.33 197 HOU Jerome Mathis 4.26 4 3 1.33 114 HOU Demarcus Van Dyke 4.25 3 3 1 81 OAK Rondel Menendez 4.24 0 0 0 247 ATL Jay Hinton 4.29 0 0 0 250 UDF Averages 4.26 25.38 4.22 4.15 85.61
Notables:
- Instead of not having a number for undrafted free agents, I just assign them a draft value of 250 and team name of "UDF"
(UDF is currently drafting better than the Raiders)
- Of the eight most valuable players to run a sub-4.30, six of them were first round draft picks. You could make a case for a player like Johnson upping his value from high-second round to low-first round, but one 40-yard run at the combine didn't propel him up much further than that. There were only a couple of true value picks: Mike Wallace and the unfortunately-injured Johnny Knox.
- The worst bust was Heyward-Bey, a receiver taken by Oakland that was only a top pick due to the fact that the Raiders exist. And even then, Heyward-Bey has been just slightly-below average compared to the other players on this list in terms of AV/Year.
Routt is another great example of a player taken as high as he was only because Raiders.
- Players like Hill and Goodwin really don't have enough of a career yet to truly judge.
How big of a difference is it to run a 4.30 or a sub-4.30?
4.30 to 4.31, by total value:
Notables:
- Notice again that the top three total AV scores belong to first round picks. Moss and Peterson were both seen as elite draft picks before the combine, while Joseph actually went a bit later than expected despite his 4.31.
- Of the next 13 players, not a single one was drafted in the first round.
- The three undrafted players have produced next-to-nothing. And you would also want to note that if a player like Moss did see his stock go up due to a 4.31, why couldn't Owusu, Gates or Lockett even get drafted?
(Because it didn't/doesn't matter)
- The highest-drafted disappointment was Taylor Mays of USC, native of Seattle, but let's not overlook the fact that his 4.31 did nothing for a draft stock that was once thought to be "elite." It wasn't enough to overcome his shortcomings as a safety.
- Compare sub-4.30 to 4.30-31:
The faster group had a higher average AV, more years in the league on average, a higher AV/year, but they were also taken ~22 picks ahead of players that were just a fraction of a second slower; 33-percent of the players that ran a sub-4.30 were drafted in the first round, compared to just 18.75-percent of the players that ran a 4.30 or 4.31.
One possible difference -- and this isn't just another joke -- is the Oakland Raiders.
They selected 27.7-percent of the players to run a sub-4.30, compared to just one of the 16 players to run a 4.30 or 4.31. Could that simple fraction of a second be enough to increase the ADP of a player, due to the fact that Al Davis is paying closer attention to you?
It still doesn't change this fact:
34 total players have run a 4.31 or better at the combine, 13 of those players posted an AV/year of at least 4.00 (roughly average), and nine of those 13 above-average players were taken in the first round. The four exceptions being Wallace, Knox, Mike Thomas and Darrent Williams.
Out of those four exceptions, Mike Wallace is the only "true" exception. The rest are human error: Williams sadly passed away in 2007, though he was on his way to being a very good player it would seem, we will never truly know. He played just two seasons.
Thomas had one great season but hasn't played in the NFL since 2012, recently signing with the Texans to try and resurrect his career.
Knox was injured in a game against the Seahawks and has since retired because of it. None of those three players had a large enough sample size to really do justice to the AV/year marker.
A ton of players have run a 4.32-4.34, so let's go deeper...
4.32-4.33, by Total Value
Notables:
Notice what happens when you place that "Average" market in the middle of the list? See how there are just six players above it and half of those players were top four picks?
Griffin's time at the combine did not matter. People knew what he could do and unless he ran a 5.00, it wasn't going to scare anyone off. We knew he was fast.
Same goes for Vick and McFadden.
Notice how many first round picks are below that line? A total of 14 players are below average on this list (see how heavy the top is and how empty the bottom is?) and only one of them -- Troy Williamson, considered one of the biggest reaches of recent NFL memory -- was a first round pick.
There is certainly some value there, most notably the other Chris Clemons, who is a ~average player that was available in the fifth round, but is that anything compared to the massive dropoff after Clemons, and to a lesser extent, Louis Murphy?
The rise and fall of players like Chad Jackson do show that fast players are over-valued, but Jackson is another example of a guy that was projected to be a "first-round lock" and yet his 4.32 didn't get him there. He fell in the draft.
These same trends continue if you keep going, and a bunch of guys have run a 4.34. If 25 guys are this fast, then how is anyone going to separate themselves?
The 4.34 Group, by Total Value:
Notables:
- The top four Total AV players were top 10 picks. There are no "steals" here, where a guy runs a 4.34 and increases his draft stock to the third round, then blows up. That has happened still just the one time: Mike Wallace. The fifth guy on this list to be drafted in the top 10, Tavon Austin, is well on his way to being one of the most valuable players to ever run a 4.34. His AV of 6 this year places him fourth in AV/year on this particular list.
- Interestingly, Best is another guy whose career is probably over due to injury.
- Tye Hill was a player who saw his stock rise after the combine, as the draft approached, and was a surprise pick at 15th overall by the Rams. Which sort of serves as the counter-example of the argument, doesn't it?
If a player sees his stock rise because of a 4.34, and doesn't possess the other tools, it's not going to make him "better" just because he ran a 4.34. Typically, guys like DeAngelo Hall and Dunta Robinson aren't better because they run a 4.34, they run a 4.34 because it's part of the whole that made them better.
Did that make sense?
The playmaking ability isn't necessarily a product of their 40-time, for guys like Vick, Griffin, Hall, their speed makes them better but not their speed alone. You can roll out of bed and run a 4.34, maybe, but you can't roll out of bed and play football.
There's a lot more to it than that. The reason that Usain Bolt isn't the best player in the NFL isn't just because he hasn't signed with a team yet:
Guys, it's not that easy to play professional football.
Good players are fast but fast players are not necessarily good.
The 4.35 Group, by Total Value:
Notables:
- Compare these five groups again and you'll see that there isn't even any real sliding scale of draft value or correlation between speed and ADP. The fastest group is drafted considerably higher than the guys who ran a 4.30-4.31, but the guys who ran a 4.32-4.33 are nearly drafted in the same position as the guys who ran faster than 4.30. The group that ran a 4.34 again dip down in their ADP, but the group that ran a 4.35 is taken about nine spots higher than the 0.1-faster group.
There's also no real correlation between speed and success, at least not in these groups. Though you could argue that a player who runs a 4.70 is not going to make a good receiver, it's a lot harder to argue that running a 4.35 is that much different than running a 4.52.
Yes, Calvin Johnson is the best receiver in the league, but Josh Gordon is arguably giving him a run (run intended) for his money and ran that time of 4.52 I just referenced.
Heyward-Bey can run a 40-yard dash that is than three-tenths of a second better than A.J. Green, but that's not going to make him A.J. Green. Sometimes he's only marginally better at receiver than A.J. Feeley.
When Vernon Davis runs a 4.38 at over 250 pounds, it certainly draws attention -- and for those people that enjoy freaks like the ones you see at Ripley's Believe It Or Not, it does so deservingly -- but Rob Gronkowski ran a 4.65 and Matt Jones ran a 4.37. You could just as easily have watched Davis's games at Maryland or sat in on a practice and known just as well that he was a freak athlete.
Here are some more players that aren't among the 96 fastest in combine history.
Who's missing:
- Adrian Peterson, 4.40
- Marshawn Lynch, 4.46
- LeSean McCoy, 4.50
- Matt Forte, 4.46
- Jamaal Charles, 4.36
- Josh Gordon, 4.52 (unofficially)
- Antonio Brown, 4.47
- A.J. Green, 4.50
- Demaryius Thomas, 4.38
- Alshon Jeffery, 4.48
- Dez Bryant, 4.52
- Richard Sherman, 4.54
- Darrelle Revis, 4.38
- Earl Thomas, 4.43
Plenty of speed here, but again, I have to ask you: Did their 40 time even make a single difference? Charles is the closest on that list to being one of the fastest 96 times recorded at the combine, and yet he wasn't drafted in the first two rounds.
Green ran a 4.50, and nobody downgraded him from being an elite wide receiver prospect worthy of the top five.
Why are we actually doing an NFL Scouting Combine? Why are we actually treating 40-times like they mean anything?
More
- There were a total of 96 players in this study, all of whom ran a 4.35 or better at the NFL scouting combine from 1999 to 2013. Out of those 96 players:
- Five of the top six were drafted in the top seven of their respective drafts. (Exception: Chris Johnson)
- Eight of the top nine were drafted in the first round. (Exception: DeSean Jackson.)
- 12 of the top 14 were taken in the first round. (Exceptions: Jackson, Mike Wallace.)
- The highest-rated undrafted free agent was Ricardo Lockette.
- He's not rated very high.
- The only players to have an AV/year above 4.00 that weren't taken in the top 100 picks were Johnny Knox and Mike Thomas, both of whom had careers that lasted four years or less.
- The biggest bust was Troy Williamson.
- Of the 64-lowest rated players on this list, the only first round picks were: Ahmad Carroll, Tye Hill, R.Jay Soward, Williamson. That's 6.2-percent of those guys.
- Of the top 32 players on this list by AV/year, four of the five lowest-ranked first round players were drafted by the Raiders: Darren McFadden, Michael Huff, Fabian Washington and Darrius Heyward-Bey. All of whom had a AV/year of 4+.
Subsequently...
There are those that argue that the combine is overrated and that workouts like the 40-yard dash aren't a great measure of evaluating player talent, but I would take it a step further than that:
The combine and the 40-yard dash aren't even important to NFL teams, with the exception of the Oakland Raiders in one era that is likely over.
The 40-yard dash is important to players, to the media, and to fans, but it has exhibited almost no correlation between fast times (at combine high) and a better ADP. Surely some of these players went a bit higher than they were expected to go and some were probably drafted when they shouldn't have been, but that didn't stop players like Chad Jackson or Johnathan Joseph also going lower than expected. One could even argue that Jackson's "slide" in the 2006 NFL draft was entirely due to mock drafters and the media seeing his 40-yard dash of "4.32 seconds" and assuming he was now a lock for the top 15.
But no teams took a bite in the first round.
The sample size above of the 96 fastest players in combine history really looks no different than any random grouping of NFL draft prospects. Some elite players, a lot of average players, some very bad prospects and players. The best correlation above of anything to success is their draft position, which we already knew to be true. Despite the fact that there is always going to be your "Tom Brady" exceptions, higher-drafted players are always more likely to be successful players.
And this was even true decades ago, before any combine, before any 40-yard dash, hell... before most fans and teams could even watch that much game tape of college athletes. There was a Hall of Fame player drafted in the top five of every draft from 1975 to 1981. You didn't need to see Walter Payton at a "scouting combine" to know you were getting someone special. You didn't need to know how many reps of 225 that Anthony Munoz could do in order to draft him third overall in 1980, or Lawrence Taylor second overall in 1981.
Two more Hall of Famers went in the top 10 in 1982, and two Hall of Famers were the top two picks in 1983. Overall, six Hall of Famers were taken in the first round of the '83 draft and zero Hall of Famers over the rest of the draft; which back then lasted 12 rounds.
The 40-yard dash is absolutely and utterly meaningless.
It is often a byproduct of what we already knew: NFL players are pretty exceptional athletes. Whether they run a 4.30 or a 4.50, doesn't just mean little to teams and to their future success, but also means almost nothing for their draft position.
Sorry, but I guess you could say that I forgot about Dri. |
Student fears of oral presentations and public speaking in higher education: a qualitative survey ABSTRACT Oral presentations and public speaking are an important aspect of the student experience in the United Kingdom higher education. Many modules (self-contained units normally within a programme of study) use presentations as a form of assessment and require students to verbally engage in small and large group settings to enhance learning. Previous research evidence has indicated that many students experience fear in public speaking. The aims of this qualitative survey were two-fold. First, it sought to gather further insight into the fears experienced and strategies used by students who fear public speaking, including oral presentations. The second objective was to determine whether their fear affected their experience of higher education. A qualitative survey comprising four open-ended questions was completed by 46 undergraduate and postgraduate students with a fear of public speaking from the University of the West of England (UWE), Bristol. All participants were attending one of the Stand Up and Be Heard (SUBH) UWE library-based workshops for fear of public speaking. Thematic analysis was used to identify the following six themes, namely: fear of being judged, physical symptoms, uncertainty about the topic, negative effect on university experience, practice and preparation, and more practical support needed. The results of this survey identify the specific fears students have in public speaking and provide evidence of the overall negative effect on their higher education experience. This survey provides further evidence that higher education institutions should acknowledge public speaking fear among some students and provide more support in oral presentation assessments. Introduction Varying terms are used in the literature to describe a fear of public speaking and are often used interchangeably, such as stage fright (), communication apprehension (CA) () or public speaking anxiety (Bodie 2010). More specifically related to this qualitative survey, public speaking anxiety is defined by Bodie 'as a situation specific social anxiety that arises from the real or anticipated enactment of an oral presentation.' Another commonly used term is glossophobia, which is the fear of public speaking or speaking in general (). The term glossophobia comes from the Greek glssa, meaning tongue, and phobos, fear or dread. Oprescu 2016). Although there is an awareness of student anxiety in oral presentations and public speaking, more research evidence is needed regarding specific fears and the strategies that students use to address them. Further evidence is also required on how fear of public speaking and oral presentations affect students' university experience. This information would help educators in the planning of assessed oral presentations, enabling them to gain a deeper understanding of students' fears and support their needs. This qualitative survey had two aims. The first was to gather further insight into students' fears about public speaking, including oral presentations, and the strategies they used to overcome them. The second objective was to determine whether their fear of public speaking affected their experience of higher education Design A qualitative survey comprising four open-ended questions. Participants and recruitment All participants recruited for this study were undergraduate and postgraduate students from the University of the West of England (UWE), with a fear of public speaking, who attended a three-hour UWE SUBH workshops at either the Glenside, Frenchay or City Campuses. All students who registered for the SUBH workshops online via the Library study skills events page were invited to participate in this qualitative survey. All potential participants were issued with a participant information sheet (PIS) and informed consent form to complete. Participants were requested to complete the informed consent form and email it with their completed data collection sheet (comprising four open-ended questions) or to bring completed hard copies to the SUBH workshop. Participants were further reminded at the start of the SUBH workshop about the research project and PIS and data collection sheets were supplied on request. It was also made clear to all participants in the online SUBH workshop registration information that participation in this research was voluntary that they had the opportunity to withdraw from the study at any time and this would not affect their SUBH workshop attendance. Purposive sampling was used in the recruitment as this enables recruitment of participants who have the best knowledge concerning the research topic (). The saturation of data was used to indicate the optimal sample size for this qualitative survey as this is the most widely used principle for determining sample size and evaluating its sufficiency () The UWE Faculty Research Ethics Committee reviewed and approved this study. Data analysis Qualitative thematic analysis (Braun and Clarke 2006) What are your main issues/fears in public speaking (including presentations)? What strategies have you used to reduce your fear of public speaking (including presentations)? Does your fear of public speaking affect your student experience of higher education? How could the university support you and other students with a fear of public speaking? The thematic analysis method used was not based on a philosophical premise but guided by the six stages identified by Braun and Clarke, namely: familiarising yourself with your data; generating initial codes; searching for themes; reviewing themes; defining and naming themes; and producing the report. Although thematic analysis has become the most commonly used qualitative data analysis method, one of the criticisms is that researchers rarely report in depth on the process of doing the analysis (). This criticism has been partly addressed by describing and following the six stages (Braun and Clarke 2006) above in conducting the thematic analysis in this qualitative survey. The lead researchers (RG) and JW initially coded the responses to the four open-ended questions and searched for themes independently to ensure inter-code reliability (Cook 2012). The two researchers then met up and discussed the coding, reviewed, defined, and named the final themes from the data. Results Forty-six undergraduate and postgraduate students attending the SUBH workshops on three of the university campuses participated in this study; 33 females (72%), 7 males (15%), and 6 participants (13%) did not report their gender. Most participants answered all four open questions although some individual open questions were not completed. The following six final main themes were developed after thematic analysis and discussion between two of the researchers (RG and JW), namely: fear of being judged; physical symptoms; uncertainty about the topic; negative effect on university experience; practice and preparation; more practical support needed (see Figure 1). Fear of being judged This theme was a direct response to the first open question asking what the main issues/fears in public speaking and oral presentations were that students experienced. The fear of being judged is an external fear that related strongly to the fact that many students felt uncomfortable standing up and speaking in front of an audience: Related to standing up in front of an audience the overwhelming response was that students felt they were being judged: Fear of being judged. (This comment was expressed by many students in the study) Closely related to being judged was the concern about audience reaction to the student speaking in public: 'Worried of what people will think'. 'Worry that people are not interested'. 'The audience may not be interested in what I say'. 'That people will laugh at me'. This feeling of 'being judged' may limit a student's ability in oral assessments when they are assessed and graded on their presentation. This overriding fear may limit the ability to demonstrate knowledge and understanding, intellectual skills and fully address the oral presentation assessment guidelines. Further to formal oral assessments, the concern of 'being judged' may limit active learning with respect to asking questions and interacting with peers during seminars and other learning opportunities. Being perceived favourably becomes a greater priority than participating in a learning opportunity or communicating with the group. Physical symptoms This theme was in direct response to the first open question asking what the main issues/fears in public speaking and oral presentations were that students experienced. This theme is a combination of internal and external fears, and a clear example of the physical signs/symptoms and 'flight or fight' response to an external stimulus/fear: 'Physical clues I am nervous e.g. shaking hands, tongue-tied speech'. 'Panic attack that would stop me communicating'. 'Going red/blushing'. " Throat seems dry, hands sweaty and emotional experience is overwhelming enough to cause me to become tearful." 'Going red when I start to talk.' "Physical symptoms of stress These physical symptoms would directly affect a student's experience of public speaking and would negatively affect their learning. The above described physical symptoms are related to the above theme 'fear of being judged' as the innate fear would be on display and clearly identifiable by others. Uncertainty about the topic This theme was also closely related to the first open question on what the main issues, and, or fears in public speaking and oral presentations were that students experienced. This theme is an internal fear about the topic, although also related to external fears, audience reaction, and a prevalent response in this survey: Uncertainty about the topic is an aspect of public speaking that appears to relate to how the audience may respond negatively to a student who has a lack of subject knowledge. This theme`s focus appears to be related to making mistakes and getting information wrong. This desire to not get anything wrong, striving for perfection may increase stress and anxiety levels in relation to public speaking. Managing expectations away from a perfect delivery and towards an increased knowledge and understanding of the topic may be a key method to decrease fear and is a simple strategy in relation to reducing public speaking fear. Negative effect on university experience This theme was in response to the third open question relating to how public speaking affects student experience in higher education. The overwhelming response was one of the negativity and an adverse effect on the individual student experience: This overall negative effect on university experience clearly shows that public speaking fear pervades all aspects of the student experience and is not just related to oral presentations. The data suggests that students' public speaking fear affects learning in respect of interaction in class, asking questions and anxiety to speak, which are all key aspects to enhance learning. Further to fear of public speaking affecting learning, it appears to influence fundamental decisions about participating in education, reducing confidence in attending university and impacting progress. Practicing and preparation This theme was related to the second open question asking students what strategies they used to reduce their fear of public speaking including oral presentations. This theme confirms that although there may be issues with respect to how best to practice, the majority of the students we surveyed realise the importance of practicing public speaking: 'Talking presentation throughout loud'. 'Breathing exercises beforehand'. 'Practice before giving presentation e.g. talking to the cats!' 'Recording a practice presentation to view myself'. 'Practice to peers'. 'Present the presentation to a small selection of people, whom I find intimidating to rehearse the presentation and get ready to when it comes to presenting the presentation in front of a large group of highly knowledgeable experts and professionals'. The data revealed that many students had a clear idea of the breadth of practice techniques including breathing techniques, recording of presentations, practicing to peers and speaking out loud. All these methods of practice have been advocated in the literature to increase public speaking proficiency. This awareness of varying practice techniques is encouraging but does not fully equate with the high levels of anxiety and fear of public speaking that students overwhelmingly felt in this survey and link to the final theme identified: practical support. More practical support needed This final theme was in response to the fourth and final open question on how the University could support students with a fear of public speaking. It was acknowledged that the SUBH workshops were of benefit to many students, although other suggestions were made for student public speaking support: 'Show us how! Physical workshops are effective and essential'. 'Opportunities to practice public speaking in a relaxing atmosphere'. 'Practice sessions and small groups'. "Hold more sessions linked to courses to help students overcome their fears". 'Teaching presentation skills as part of the course'. 'More opportunities for practice'. 'Tutorial support.' "Maybe incorporate classes into the timetables, not as mandatory but will let students be more aware about them". The responses to this question were informative with respect to learning, teaching and student support. Specifically, suggestions were related to tutorial support, teaching presentation skills as part of the course and incorporating public speaking classes into the timetable. Students gave practical and useful public speaking support suggestions that could be adopted at a module, programme, and university-wide level. There was limited data with respect to the quality of support provided, although this was not a specific open question. Discussion The main findings from this qualitative survey have clearly indicated that for those students with a fear of public speaking and oral presentations, public speaking tasks have an overall negative effect on learning and the student experience. Specifically, the findings have indicated that many students' main fears are associated with being judged, uncertainty about the topic and physical symptoms. This survey also clearly indicated that most of the students in this study, although fearful of public speaking, were aware of the importance of practice and preparation. The findings clearly identified the lack of and need for further comprehensive support for students with a fear of public speaking. A qualitative study on self-described fears related to public speaking from university and college students in the US (LeFebvre, LeFebvre, and Allen 2018) identified 12 common themes from students on an introductory communication course that included both internal and external fears. An important finding from this research was that internal fears accounted for 25% and external fears 75% of student public speaking anxiety. Overall, the most commonly reported fear by students (30%) was the external fear 'audience responses', related to perceived attitudes from the audience towards the speaker. These perceived attitudes that students found challenging included judgement from the audience, being the focus of attention and no interaction from the audience. The next most commonly reported fear (23%) in that study was an ability to self-regulate. The researchers described this as students' fear about their own performance during a speech, which included fears about their own recall of information or their inability to remember presentation content during a speech (LeFebvre, LeFebvre, and Allen 2018). In this qualitative survey, two of the key findings or identified themes, 'fear of being judged' (external fear related to the audience) and 'uncertainty about the topic' (internal fear) are closely related and clearly supported in the above student self-described fears study (LeFebvre, LeFebvre, and Allen 2018). The 'fear of being judged', was clearly related to audience response and most participants expressed fears related to standing up in front of an audience. More recently, LeFebvre et al. conducted a study addressing student public speaking anxiety through an introductory speaking course and found that 'memory glitches' were the most cited overall public speaking fear. This further supports the key theme in this survey 'uncertainty about the topic' where forgetting material in public speaking was a predominant fear and occurrence. Further to internal fears, in a questionnaire concerning fear of public speaking of undergraduate students in higher education, the main findings showed an association between students with negative self-perceptions of their voice and public-speaking fear (Ferreira ). In this qualitative survey, 'physical symptoms' were identified as one of the six main themes and LeFebvre, LeFebvre, and Allen also found physiological problems described as 'excessive activation' to be an issue among students. In respect of physical symptoms, public speaking fear may affect the speaker physically with a dry mouth, increased blood pressure, blushing, sweating, irregular breathing, and emotionally in the form of feelings of humiliation and concerns about looking foolish (Kushner 2004). Most of these physical symptoms were described by the participants in this survey and could be related to the 'fight or flight' response. The 'fight or flight' response, is usually short-lived and the restoration of balance or homoeostasis is regulated by the parasympathetic nervous system; however, it can become damaging if initiated repeatedly or over prolonged time periods in the absence of a true threat (Chamberlain and Meuret 2017). The majority of the students attending the SUBH library workshops and participating in this survey reported experiencing related 'physical symptoms' with public speaking, some over a long period of time. Apart from the physical symptoms associated with 'fight or flight' response, the positioning of arms and hands may also be associated with anxiety in oral presentations (Tsang 2020). In interview findings from a mixed methods study of tertiary-level students, presenters were most concerned with their arms and hands (hand trembling) as signs of physical behaviour showing nervousness that could be seen by the audience (Tsang 2020). In respect of the theme related to "'more practical support needed', university students in this study reported that they required more support for public speaking from the university. Suggestions were made in respect of more workshops, opportunities to practice public speaking, and teaching presentations as part of a course and incorporating public speaking classes into the timetable. These findings support previous evidence that found 89% of the students were interested in public speaking training and would appreciate this as an addition to their curriculum (Ferreira ). There is clear evidence on the benefits of support for students with a fear of public speaking that confirms one of the main themes identified in this survey related to 'more practical support needed'. Previous evidence relating to two UK universities, from Russell and Topham, found that some students have identified various barriers to support, grouped under the themes of invisibility, stigmatisation and lack of confidence. A Personal Report of Public Speaking Anxiety questionnaire was administered at the beginning and end of a course in public speaking, related to student feelings towards giving a speech (Tse 2012). The findings revealed that teaching affective strategies to students appears to be an effective way of reducing anxiety in public speaking (Tse 2012). Previous research by Nash, Crimmins, and Oprescu, indicated that the first-year students who completed pre-and post-public speaking exercises and assessments identified greater feelings of satisfaction and less fear, indecision, and confusion in relation to public speaking and public speaking assessments. More recently, a study of undergraduate students at a large Midwestern university enrolled in an introductory public speaking course which included skills in public speaking supports the evidence for introductory communication courses (). The main findings indicated that almost half of the students reported less anxiety and diminished fears between an initial and subsequent final speech (). The above identified research findings (;Nash, Crimmins, and Oprescu 2016;Tse 2012) were all positive with respect to introductory courses providing much-needed public speaking skills development for students during their time at university. However, findings from an earlier survey of employers and a compulsory first-year communication course (Clokie and Fourie 2016) found that the communication course may not provide students with sufficient employability skills with respect to adequate business communication, including oral and visual presentation skills. Clokie and Fourie suggest that this gap in communication skills could be addressed by increasing the industry relevance of courses and by including communication skills at all levels of university learning. Further to the proposed benefits of practical support for public speaking was the theme related to 'practicing and preparation', where students reported which strategies they used to reduce their fear of public speaking. The importance of practice was frequently reported in this theme by many students. The importance of practice is widely known and advocated in respect of public speaking/ oral presentations. However, a recently conducted mixed-method study (questionnaire (n = 211), and semistructured interviews (n = 6) found opposing views on the importance of practice for improving public speaking skills and reducing fears (Tsang 2020). According to Tsang, the two main reasons put forward by the interviewees were the audience effect and the effectiveness of the approaches adopted during practice/rehearsals. The audience response may override the efforts of the practice/rehearsal and the approach to practice may be ineffective as the student may be unaware of appropriate practice skills Tsang. These are important findings, but as acknowledged by Tsang, more interview participants were needed, including those from various backgrounds (e.g. secondary-school level; different ethnic backgrounds) to ensure findings were more generalisable to students in other contexts. With respect to the recent COVID-19 pandemic and a move to online learning (Madzlan, Seng, and Kesevan 2020) found that the use of video blogs brought significant positive outcomes in reducing public-speaking anxiety among English language learners. However, a questionnaire of university students found that although online education has positive aspects, the negative elements were mainly related to a lack of communication and cooperation, as well as the general restriction of social contact in the academic context (Karalis and Raikou 2020). Finally, this survey clearly identified through one of the key themes that apart from oral presentations, public speaking in general had a 'negative effective on students` university experience.' Students expressed concern relating to oral presentations, but also public speaking fear related to participating in discussions, asking questions, and fear obstructing their learning. These public speaking fears are supported by previously discussed research evidence that indicated student social anxiety and prevalence in university learning situations (Prhl, Almonkari, and Kunttu 2019;;Russell and Shaw 2009). According to the Institute for Public Policy Research, student levels of mental distress, and low wellbeing are worsening in UK higher education and are high relative to other sections of the population (Thorley 2017). This negative effect of public speaking found in participants in this survey, may therefore be a factor or exacerbate issues related to student mental health and wellbeing. Related to the overall findings of this qualitative survey and participants' fear of public speaking, was the overall negative response that participants gave to most of the open questions. Recent evidence by Tsang on tertiary students' fear of oral presentations found a clear relationship between students' self-perceived competence in delivering oral presentations and their levels of anxiety in public speaking. Therefore, further to the above identified evidence for student public speaking support, is the immense potential value in boosting learners' perceptions of their own presentation delivery skills as a means of lowering their anxiety (Tsang 2020). Reassurance will be crucial to mitigate the fears identified and help students self-regulate when they engage in public speaking. Linked to student self-perceived competence are excessive perfectionist tendencies which may also contribute to public speaking fear and negativity. An analysis of findings from a perfectionism scale completed by 41, 641 American, Canadian, and British university/college students (Curran and Hill 2017) indicated that young people are particularly demanding of themselves. This striving for perfection manifests itself in public speaking and is often the default position of many students, who focus on style and slickness of presentation over substance, which in turn may increase the pressure on themselves and increase their public speaking fear (Grieve 2020). In respect of public speaking and the negative effect on the student experience, we have run university-wide library based SUBH workshops, for students with a fear of public speaking since 2015 (Grieve 2017). The main emphasis of these SUBH workshops and subsequently published book (Stand Up and Be Heard: Taking the Fear Out of Public Speaking at University; Grieve 2020), which differ from performance-based public speaking training, is the focus on authenticity. In the workshops, we focussed on the following components that enabled students to become authentic public speakers and reduce their public speaking fear, namely: being present in the moment; being yourself; awareness of vulnerability and letting go of perfection (Grieve 2020). An evaluation (n = 82) of the SUBH workshops and support for students was very positive with 87% of students feeling confident in reviewing and using the authentic public speaking strategies to help manage their fear of presentations/public speaking (Grieve 2018). Limitations As there were a disproportionate number of female (72%) compared to male (15%) students participating in this study, this may indicate a gender bias in recruitment. However, some of the evidence does indicate that there is a higher prevalence of public speaking fear among females compared to males (Perveen, Yamna, and Aleemi 2018;Ferreira ). Linked to inclusivity in higher education, there is evidence that second language or international students face anxieties in oral presentations and public speaking (Kao and Craigie 2018). Given the high percentage of international students in UK higher education and the number attending the SUBH workshops, this group of students were not purposely included in this qualitative survey. Further limitations might be associated with only recruiting students with a fear of public speaking as opposed to a broader random sample of students in higher education, which may reduce transferability. An indication of reduced transferability in this study could be related to data analysis in the open question format as the data may not have been sufficiently rich to enable comparison with other related evidence (). The data for this study was collected prior to the COVID-19 pandemic lockdown and subsequent changes to learning and teaching in the form of online and blended learning. Therefore, further research is required on student experiences of online public speaking and oral assessment in this new pedagogic landscape. Conclusion and future directions Overall, this qualitative survey has increased the evidence base with respect to furthering the understanding of student fears of public speaking and adds to the growing evidence that a proportion of university students have a fear of public speaking, specifically oral presentations. This survey has identified six themes that ultimately highlight and include the central point that public speaking may have a negative effect on students` university experience. Importantly, this negative effect on student experience related to public speaking, may be a contributing factor in student mental health and wellbeing issues. A key conclusion from this survey and in support of previous evidence, is the lack of and need for specific public speaking support for students with a public speaking fear in higher education. The data collected suggest that students may prefer more practical support in smaller group workshops in a supportive environment, with a focus on authenticity, letting go of perfection and substance over style as in the previously discussed SUBH workshops. Importantly, in respect of the need for public speaking support, the question was not asked in this survey on how students evaluated existing support, this warrants further investigation. Adequate student support in public speaking would require communication and collaboration across an institution and involve input from wellbeing services, library services, study skills units, academic tutors and individual programmes. Further considerations should include the need for academics and module leaders to ensure they are aware of the fears that many students have around public speaking in general and assessed oral presentations. Further research might include addressing student wellbeing with respect to including public speaking learning and teaching embedded in the curriculum. Further to the recruitment limitations of this survey in only including students with a fear of public speaking, future research might include a wider sample of students with and without public speaking fear. In respect of inclusivity and recognised in the evidence, international students with a fear of public speaking who attend UK higher education institutions should also be actively included in the future public speaking research. |
/**
* Wait for new threads to terminate
* @param initialThreads set of old threads
* @param max time to wait (in milliseconds)
* @return {@code true} if they all terminated within the limit;
* {@code false} otherwise
*/
public static boolean wait4NewThreads(Set<Thread> initialThreads, long max) {
long start = System.currentTimeMillis();
do {
int count = 0;
for(Thread thread: getCurrentThreads())
if(unexpectedThread(thread,initialThreads))
count++;
if(count > 0) {
try {
Thread.sleep(100L);
} catch (InterruptedException e) {
e.printStackTrace();
}
} else
return true;
max --;
} while(System.currentTimeMillis() - start < max);
for(Thread thread: getCurrentThreads())
if(unexpectedThread(thread,initialThreads))
System.out.println(thread.getName());
return false;
} |
Detection and classification of raw action potential patterns in human Muscle Sympathetic Nerve Activity The Muscle Sympathetic Nerve Activity (MSNA) consists of synchronous neural discharges separated by periods of neural silence dominated by heavy background noise. During measurement with electrodes, the raw MSNA signal is amplified, band-pass filtered, rectified and integrated. This integration process removes much neurophysiological information. In this paper a method for detecting a raw action potential (before the pre-amplifier) and filtered action potential (after the bandpass filter) is presented. This method is based on stationary wavelet transform (SWT) and a peak detection algorithm. Also, the detected action potentials were clustered using the k-means method and the cluster averages were calculated. The action potential detector and classification algorithm are evaluated using real MSNA recorded from three healthy subjects. |
<reponame>fwenqiang/jfinal-admin
package com.pintuan.service;
import java.util.Date;
import com.pintuan.common.Constants;
import com.pintuan.common.Fields;
import com.pintuan.model.Dict;
import com.supyuan.jfinal.base.BaseService;
import com.supyuan.util.extend.UuidUtils;
/**字典**/
public class DictService extends BaseService {
/**
* 添加
*
*/
public Dict add(String key,String val) {
Dict dict = new Dict();
dict.set(Fields.DIC_ID, UuidUtils.getUUID2());
dict.set(Fields.DIC_KEY, key);
dict.set(Fields.DIC_VAL, val);
dict.set(Fields.STATE, Constants.ACCESS_STATE);
dict.set(Fields.CREATE_TIME, new Date());
dict.save();
return dict;
}
public void update(Dict dict) {
dict.update();
}
public Dict findByKey(String key) {
return Dict.dao.findFirstByWhere("where dic_key=? and state=?",key,Constants.ACCESS_STATE);
}
}
|
Left Forum Zizek Controversy: Activists & Organizers Speak Out
Eoin Higgins Blocked Unblock Follow Following May 24, 2016
The controversial decision by the organizers of Left Forum 2016 to feature Slavoj Zizek as the closing plenary speaker was met with protest, a walk out, and social media activism.
One of the most outspoken critics of the decision to include the Slovenian philosopher at the annual conference was activist and journalist Taryn Fivek.
Fivek is the creator of No Platform, a project that will send her around the country in a mission to explore the disparate views of the American electorate that are undercovered by the mainstream media. Fivek describes her project as “a collaborative multimedia project that aims to explore a side of the US not covered in our 24-hour news cycle…. in the tradition of Studs Terkel.”
Fivek’s opposition to Zizek’s speaking to close Left Forum stemmed from a number of offensive comments the philosopher has made over the past decade. The poster below features some of the more egregious, but it appears that for Fivek- and for many on the left- his comments on refugees in April of this year were the last straw.
In April, Zizek made remarks Britain’s Channel 4 that were interpreted by many as racist and bigoted. His comments appeared to frame the refugee question as a problem of integration and multiculturalism on the part if the refugees- and furthermore, his comments suggested that the west shared no responsibility for the refugees plight.
That’s the way Fivek interpreted it in an email to Left Forum program coordinator Marcus Graetsch:
Zizek explains that opening our hearts to welcome refugees from imperial wars across the Middle East and North Africa will “make things worse for the refugees”, that we should “not just enlighten” them, that his idea is not that we should “live together with all different races, cultures, we all love each other,” but rather he wants a “polite ignorance.”
That email was part of a longer chain in which the two debated whether or not Zizek should be disinvited from speaking at the closing plenary. Fivek argued that the philosopher’s presence made it “not difficult to understand why women, people of color, refugees, LGBT and other oppressed populations would feel excluded and threatened by his presence at Left Forum.”
The exchange ended with Fivek repeatedly attempting to follow up on Graetsch’s April 18 acknowledgement of her concerns and promise to forward the information to the Left Forum board. In the email exchange forwarded to me by Fivek, she followed up on the 20 and the 27 with Graetsch, to no avail.
According to Fivek, the next time she spoke to Graetsch was the evening of May 22, as she awaited Zizek’s speech. Fivek says Graetsch told her he had informed her to submit a formal letter to the board in April.
“When he said ‘I told you to write a formal complaint to get a comment from the board,’” Fivek told me, “I said: ‘No. You didn’t. Otherwise I wouldn’t be here.’”
When I spoke to Graetsch on the 24, he admitted he had made a mistake with that reply.
“I made a mistake [with what I said to her],” Graetsch said. “I sent her emails with her links to the board, I made a mistake.”
Greatsch told me he had forwarded the information along to the board and they decided to add a Q and A section after the speech. The Q and A was expected to be chaired by Amy Goodman of Democracy Now!, but she abruptly left after her short speech, less than ten minutes into Zizek’s talk (I’ve repeatedly reached out to DN for comment and will update this article when and if they get back to me).
“I don’t know why she left,” Graetsch said when I asked him about Goodman. He added that he did not believe the walk out once Zizek took the stage was necessarily a reaction ot the speaker.
Graetsch said the question before the board on Zizek concerned whether or not he is a racist. If he is, Graetsch said, he would have been disinvited. But it was the opinion of the board, and Graetsch himself, that Zizek uses a clownish entertainer’s personality to critique the right.
“Zizek is using a specific European form of comedy,” Graetsch said. “He is going into different personality and using the language of the right wing to state his critique. Sometimes people don’t get this.”
Fivek isn’t buying it. Zizek’s use of racist and misogynistic language is not the language of the left. Her activism in response to his speaking was a reaction to his racism.
“I joined others — mainly women and people of color — who had come to protest his presence by handing out flyers and heckling him from the audience.” Fivek said in a statement. “I like to think that Left Forum is a space where women, people of color, refugees, Reds — and, of course, debate — are welcome.”
Fivek’s activism- the flyers and her heckling of Zizek- culminated with her approaching the microphone for the Q and A period. Instead of directly addressing Zizek, though, she aimed her comments at the Left Forum organizers. As she began to speak, the livestream on the Left Forum website was cut.
“My question, directed at the organizers of Left Forum, was how much they paid a man who publicly said and wrote such racist, misogynist, and xenophobic things,” Fivek told me. “We would all like to know the real story — why Left Forum used community money to bring a man to New York to say such things.”
Graetsch said that Left Forum does not pay speakers.
“Some people believe Left Forum is swimming in money,” he said, “But we don’t pay. Maybe a plane ticket, maybe a hotel room. And not all of them.”
The Left Forum will have to wrestle with the after effects of Zizek’s talk for some time. Fivek was far from the only person who was disgusted by the Slovenian’s inclusion in the talk- social media was alight Sunday night with people criticizing both Zizek for his comments and Left Forum for inviting him.
The debate, as much as there is one, over Zizek can be boiled down to your opinion on the philosopher’s motivation for his racist comments.
On the one hand, there are people who believe as Graetsch does; that Zizek is not a racist and that his influence and work for the left supersedes the feelings of those who oopse him.
Graetsch pointed to the many people he said have told him they became interested in Marxist thought and communism because of exposure to Zizek over the past decade.
“If he is on the left and promoting these communist ideals, as a controversial critical intellectual,” Graetsch said, “Then to say he is a racist take him out of the game. Is this a decision somebody here should make? I don’t know.”
Fivek believes it’s a decision Left Forum should make, and should have made a month ago.
“If they think right-wing demagogues sell tickets, then we should all look forward to seeing other speakers such as Alex Jones, David Duke, and Tom Metzger closing Left Forum in the future,” Fivek said. “All they need to do is call themselves Marxists first.” |
Disinfection of Sewage Sludge Using Microwave Enhanced Advanced Oxidation Process The microwave enhanced advanced oxidation process (MW/H2O2-AOP) was used to treat municipal sewage sludge for pathogen destruction. Two levels of microwave heating temperatures of 55degC and 70degC, and six levels of hydrogen peroxide dosages (0% to 0.1%) were used. Fecal coliform concentrations were found below detection limit (1000 CFU/L) immediately after treatment when sludge was treated at 70degC with more than 0.04% of hydrogen peroxide. Significant regrowth of fecal coliforms was observed after the treated samples were stored at ambient temperature for 72 hours. However, no regrowth was observed with hydrogen peroxide dosages higher than 0.08%, suggesting complete elimination of fecal coliforms. The MW/H 2 O 2 -AOP might be a very promising technology in sewage sludge treatment; not only can it solubilize nutrients for subsequent recovery, but also destroy pathogens, for producing Class A sludge. |
<filename>qizhi-management/deploy.py
#!/usr/bin/env python
# Copyright (c) Microsoft Corporation
# All rights reserved.
#
# MIT License
#
# Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated
# documentation files (the "Software"), to deal in the Software without restriction, including without limitation
# the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and
# to permit persons to whom the Software is furnished to do so, subject to the following conditions:
# The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
#
# THE SOFTWARE IS PROVIDED *AS IS*, WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING
# BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND
# NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM,
# DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
# OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.
#
#
# Copyright (c) Peking University 2018
#
# The software is released under the Open-Intelligence Open Source License V1.0.
# The copyright owner promises to follow "Open-Intelligence Open Source Platform
# Management Regulation V1.0", which is provided by The New Generation of
# Artificial Intelligence Technology Innovation Strategic Alliance (the AITISA).
from __future__ import print_function
import yaml
import os
import sys
import subprocess
import jinja2
import argparse
import logging
import logging.config
from paiLibrary.clusterObjectModel import objectModelFactory
logger = logging.getLogger(__name__)
def write_generated_file(file_path, content_data):
with open(file_path, "w+") as fout:
fout.write(content_data)
def load_yaml_config(config_path):
with open(config_path, "r") as f:
cluster_data = yaml.load(f)
return cluster_data
def loadClusterObjectModel(config_path):
objectModel = objectModelFactory.objectModelFactory(config_path)
ret = objectModel.objectModelPipeLine()
return ret["service"]
def read_template(template_path):
with open(template_path, "r") as f:
template_data = f.read()
return template_data
def generate_from_template(template_data, cluster_config):
generated_file = jinja2.Template(template_data).render(
{
"clusterinfo": cluster_config['clusterinfo'],
"machineinfo": cluster_config["machineinfo"],
"machinelist": cluster_config["machinelist"]
}
)
return generated_file
def execute_shell_with_output(shell_cmd, error_msg):
try:
res = subprocess.check_output( shell_cmd, shell=True )
except subprocess.CalledProcessError:
logger.error(error_msg)
sys.exit(1)
return res
def execute_shell(shell_cmd, error_msg):
try:
subprocess.check_call( shell_cmd, shell=True )
except subprocess.CalledProcessError:
logger.error(error_msg)
sys.exit(1)
def login_docker_registry(docker_registry, docker_username, docker_password):
shell_cmd = "docker login -u {0} -p {1} {2}".format(docker_username, docker_password, docker_registry)
error_msg = "docker registry login error"
execute_shell(shell_cmd, error_msg)
logger.info("docker registry login successfully")
def generate_docker_credential(docker_info):
username = str(docker_info[ "docker_username" ])
passwd = str(docker_info[ "docker_password" ])
if username and passwd:
credential = execute_shell_with_output(
"cat ~/.docker/config.json",
"Failed to get the docker's config.json"
)
else:
credential = "{}"
docker_info["credential"] = credential
def generate_secret_base64code(docker_info):
domain = str(docker_info[ "docker_registry_domain" ])
username = str(docker_info[ "docker_username" ])
passwd = str(docker_info[ "docker_password" ])
if domain == "public":
domain = ""
if username and passwd:
login_docker_registry( domain, username, passwd )
base64code = execute_shell_with_output(
"cat ~/.docker/config.json | base64",
"Failed to base64 the docker's config.json"
)
else:
logger.info("docker registry authentication not provided")
base64code = "{}".encode("base64")
docker_info["base64code"] = base64code.replace("\n", "")
def generate_image_url_prefix(docker_info):
domain = str(docker_info["docker_registry_domain"])
namespace = str(docker_info["docker_namespace"])
if domain != "public":
prefix = "{0}/{1}/".format(domain, namespace)
else:
prefix = "{0}/".format(namespace)
docker_info["prefix"] = prefix
def clean_up_generated_file(service_config):
service_list = service_config['servicelist']
for serv in service_list:
template_list = service_list[serv]['templatelist']
if 'None' in template_list:
continue
for template in template_list:
if os.path.exists("bootstrap/{0}/{1}".format(serv,template)):
shell_cmd = "rm -rf bootstrap/{0}/{1}".format(serv,template)
error_msg = "failed to rm bootstrap/{0}/{1}".format(serv,template)
execute_shell(shell_cmd, error_msg)
logger.info("Successfully clean up the generated file")
def generate_template_file_service(serv, cluster_config, service_config):
service_list = service_config['servicelist']
template_list = service_list[serv]['templatelist']
if 'None' in template_list:
return
for template in template_list:
template_data = read_template("bootstrap/{0}/{1}.template".format(serv, template))
generate_data = generate_from_template(template_data, cluster_config)
write_generated_file("bootstrap/{0}/{1}".format(serv, template), generate_data)
def generate_template_file(cluster_config, service_config):
service_list = service_config['servicelist']
for serv in service_list:
generate_template_file_service(serv, cluster_config, service_config)
def single_service_bootstrap(serv, service_config):
if serv == 'None':
return
if 'stopscript' not in service_config['servicelist'][serv]:
return
shell_cmd = 'chmod u+x bootstrap/{0}/{1}'.format(serv, service_config['servicelist'][serv]['stopscript'])
error_msg = 'Failed to grant permission to stopscript'
execute_shell(shell_cmd, error_msg)
shell_cmd = './bootstrap/{0}/{1}'.format(serv, service_config['servicelist'][serv]['stopscript'])
error_msg = 'Failed stopscript the service {0}'.format(serv)
execute_shell(shell_cmd, error_msg)
if 'startscript' not in service_config['servicelist'][serv]:
return
shell_cmd = 'chmod u+x bootstrap/{0}/{1}'.format(serv, service_config['servicelist'][serv]['startscript'])
error_msg = 'Failed to grant permission to startscript'
execute_shell(shell_cmd, error_msg)
shell_cmd = './bootstrap/{0}/{1}'.format(serv, service_config['servicelist'][serv]['startscript'])
error_msg = 'Failed startscript the service {0}'.format(serv)
execute_shell(shell_cmd, error_msg)
def dependency_bootstrap(serv, service_config, started_service):
if serv == 'None':
return
if serv in started_service:
return
for pre_serv in service_config['servicelist'][serv]['prerequisite']:
dependency_bootstrap(pre_serv, service_config, started_service)
shell_cmd = './bootstrap/{0}/{1}'.format(serv, service_config['servicelist'][serv]['startscript'])
error_msg = 'Failed start the service {0}'.format(serv)
execute_shell(shell_cmd, error_msg)
started_service[serv] = True
def bootstrap_service(service_config):
started_service = {}
for serv in service_config['servicelist']:
if 'startscript' not in service_config['servicelist'][serv]:
continue
dependency_bootstrap(serv, service_config, started_service)
def copy_arrangement_service(serv, service_config):
service_list = service_config['servicelist']
if 'copy' not in service_list[serv]:
return
for target in service_list[serv]['copy']:
dst = "bootstrap/{0}/{1}".format(serv, target['dst'])
src = target['src']
if os.path.exists(dst) == False:
shell_cmd = "mkdir -p {0}".format(dst)
error_msg = "failed to mkdir -p {0}".format(dst)
execute_shell(shell_cmd, error_msg)
shell_cmd = "cp -r {0} {1}".format(src, dst)
error_msg = "failed to copy {0}".format(src)
execute_shell(shell_cmd, error_msg)
def copy_arrangement(service_config):
service_list = service_config['servicelist']
for srv in service_list:
copy_arrangement_service(srv, service_config)
def generate_configuration_of_hadoop_queues(cluster_config):
#
hadoop_queues_config = {}
#
total_weight = 0
for vc_name in cluster_config["clusterinfo"]["virtualClusters"]:
vc_config = cluster_config["clusterinfo"]["virtualClusters"][vc_name]
weight = float(vc_config["capacity"])
hadoop_queues_config[vc_name] = {
"description": vc_config["description"],
"weight": weight
}
total_weight += weight
hadoop_queues_config["default"] = {
"description": "Default virtual cluster.",
"weight": max(0, 100 - total_weight)
}
if total_weight > 100:
logger.warning("Too many resources configured in virtual clusters.")
for hq_name in hadoop_queues_config:
hq_config = hadoop_queues_config[hq_name]
hq_config["weight"] /= (total_weight / 100)
#
cluster_config["clusterinfo"]["hadoopQueues"] = hadoop_queues_config
"""
def generate_configuration_of_hadoop_queues_by_num_gpus(cluster_config):
#
hadoop_queues_config = {}
#
total_num_gpus = 0
for machine_name in cluster_config["machinelist"]:
machine_config = cluster_config["machinelist"][machine_name]
if "yarnrole" not in machine_config or machine_config["yarnrole"] != "worker":
continue
machine_type = machine_config["machinetype"]
machine_type_config = cluster_config["machineinfo"][machine_type]
num_gpus = 0
if "gpu" in machine_type_config:
num_gpus = machine_type_config["gpu"]["count"]
total_num_gpus += num_gpus
#
total_weight = 0
for vc_name in cluster_config["clusterinfo"]["virtualClusters"]:
vc_config = cluster_config["clusterinfo"]["virtualClusters"][vc_name]
num_gpus_configured = vc_config["numGPUs"]
weight = float(num_gpus_configured) / float(total_num_gpus) * 100
hadoop_queues_config[vc_name] = {
"description": vc_config["description"],
"weight": weight
}
total_weight += weight
hadoop_queues_config["default"] = {
"description": "Default virtual cluster.",
"weight": max(0, 100 - total_weight)
}
if total_weight > 100:
print("WARNING: Too many GPUs configured in virtual clusters.")
for hq_name in hadoop_queues_config:
hq_config = hadoop_queues_config[hq_name]
hq_config["weight"] /= (total_weight / 100)
#
cluster_config["clusterinfo"]["hadoopQueues"] = hadoop_queues_config
"""
def setup_logging():
"""
Setup logging configuration.
"""
configuration_path = "sysconf/logging.yaml"
logging_configuration = load_yaml_config(configuration_path)
logging.config.dictConfig(logging_configuration)
def main():
setup_logging()
parser = argparse.ArgumentParser()
parser.add_argument('-p', '--path', required=True, help="cluster configuration's path")
parser.add_argument('-c', '--clean', action="store_true", help="clean the generated script")
parser.add_argument('-d', '--deploy', action="store_true", help="deploy all the service")
parser.add_argument('-s', '--service', default='all', help="bootStrap a target service")
args = parser.parse_args()
# step 1: load configuration from yaml file.
config_path = args.path
cluster_config = loadClusterObjectModel(config_path)
service_config = load_yaml_config("service.yaml")
# step 2: generate base64code for secret.yaml and get the config.json of docker after logining
generate_secret_base64code(cluster_config[ "clusterinfo" ][ "dockerregistryinfo" ])
generate_docker_credential(cluster_config[ "clusterinfo" ][ "dockerregistryinfo" ])
# step 3: generate image url prefix for yaml file.
generate_image_url_prefix(cluster_config[ "clusterinfo" ][ "dockerregistryinfo" ])
if 'docker_tag' not in cluster_config['clusterinfo']['dockerregistryinfo']:
cluster_config['clusterinfo']['dockerregistryinfo']['docker_tag'] = 'latest'
# step 4: generate configuration of hadoop queues
generate_configuration_of_hadoop_queues(cluster_config)
# step 5: generate templatefile
if args.service == 'all':
copy_arrangement(service_config)
generate_template_file(cluster_config, service_config)
else:
copy_arrangement_service(args.service, service_config)
generate_template_file_service(args.service, cluster_config, service_config)
# step 6: Bootstrap service.
# Without flag -d, this deploy process will be skipped.
if args.deploy:
if args.service == 'all':
# dependency_bootstrap will auto-start all service in a correctly order.
bootstrap_service(service_config)
else:
# Single service startup will ignore the dependency. User should ensure the operation is in a correctly order. This option is mainly designed for debug.
single_service_bootstrap(args.service, service_config)
# Optional : clean all the generated file.
if args.clean:
clean_up_generated_file(service_config)
if __name__ == "__main__":
main()
|
<gh_stars>1-10
/*
* Copyright (c) 2018-2020 ARM Limited.
*
* SPDX-License-Identifier: MIT
*
* Permission is hereby granted, free of charge, to any person obtaining a copy
* of this software and associated documentation files (the "Software"), to
* deal in the Software without restriction, including without limitation the
* rights to use, copy, modify, merge, publish, distribute, sublicense, and/or
* sell copies of the Software, and to permit persons to whom the Software is
* furnished to do so, subject to the following conditions:
*
* The above copyright notice and this permission notice shall be included in all
* copies or substantial portions of the Software.
*
* THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
* IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
* FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
* AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
* LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
* OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
* SOFTWARE.
*/
#include "arm_compute/core/CL/kernels/CLStackLayerKernel.h"
#include "arm_compute/core/CL/CLHelpers.h"
#include "arm_compute/core/CL/CLKernelLibrary.h"
#include "arm_compute/core/CL/CLValidate.h"
#include "arm_compute/core/CL/ICLTensor.h"
#include "arm_compute/core/CL/OpenCL.h"
#include "arm_compute/core/Error.h"
#include "arm_compute/core/Helpers.h"
#include "arm_compute/core/IAccessWindow.h"
#include "arm_compute/core/TensorInfo.h"
#include "arm_compute/core/Utils.h"
#include "arm_compute/core/Window.h"
#include "arm_compute/core/utils/misc/ShapeCalculator.h"
#include "support/StringSupport.h"
using namespace arm_compute::misc::shape_calculator;
namespace arm_compute
{
namespace
{
Status validate_arguments(const ITensorInfo *input, unsigned int axis, unsigned int idx_input, unsigned int num_tensors, const ITensorInfo *output)
{
ARM_COMPUTE_RETURN_ERROR_ON_NULLPTR(input, output);
ARM_COMPUTE_RETURN_ERROR_ON_F16_UNSUPPORTED(input);
ARM_COMPUTE_RETURN_ERROR_ON(input->data_type() == DataType::UNKNOWN);
ARM_COMPUTE_RETURN_ERROR_ON(idx_input >= num_tensors);
ARM_COMPUTE_RETURN_ERROR_ON(axis > input->num_dimensions());
ARM_COMPUTE_RETURN_ERROR_ON(input->num_dimensions() > 4);
if(output->total_size() != 0)
{
ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_DIMENSIONS(output->tensor_shape(), compute_stack_shape(*input, axis, num_tensors));
ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_DATA_TYPES(input, output);
ARM_COMPUTE_RETURN_ERROR_ON_MISMATCHING_QUANTIZATION_INFO(input, output);
}
return Status{};
}
std::pair<Status, Window> validate_and_configure_window(ITensorInfo *input, unsigned int axis, unsigned int num_tensors, ITensorInfo *output)
{
// Output auto inizialitation if not yet initialized
auto_init_if_empty(*output, input->clone()->set_tensor_shape(compute_stack_shape(*input, axis, num_tensors)));
// Configure kernel window
Window win = calculate_max_window(*input);
return std::make_pair(Status{}, win);
}
} // namespace
CLStackLayerKernel::CLStackLayerKernel()
: _input(nullptr), _output(nullptr)
{
}
void CLStackLayerKernel::configure(const ICLTensor *input, unsigned int axis, unsigned int idx_input, unsigned int num_tensors, ICLTensor *output)
{
configure(CLKernelLibrary::get().get_compile_context(), input, axis, idx_input, num_tensors, output);
}
void CLStackLayerKernel::configure(const CLCompileContext &compile_context, const ICLTensor *input, unsigned int axis, unsigned int idx_input, unsigned int num_tensors, ICLTensor *output)
{
ARM_COMPUTE_ERROR_ON_NULLPTR(input, output);
ARM_COMPUTE_ERROR_THROW_ON(validate_arguments(input->info(), axis, idx_input, num_tensors, output->info()));
_input = input;
_output = output;
// Configure kernel window
auto win_config = validate_and_configure_window(input->info(), axis, num_tensors, output->info());
// Add build options
CLBuildOptions build_opts;
build_opts.add_option("-DDATA_TYPE=" + get_underlying_cl_type_from_data_type(input->info()->data_type()));
build_opts.add_option("-DAXIS=" + support::cpp11::to_string(axis));
build_opts.add_option("-DSRC_DIM2=" + support::cpp11::to_string(input->info()->dimension(2)));
build_opts.add_option("-DDST_DIM3=" + support::cpp11::to_string(output->info()->dimension(3)));
// Create kernel
_kernel = create_kernel(compile_context, "stack_layer", build_opts.options());
ARM_COMPUTE_ERROR_THROW_ON(win_config.first);
ICLKernel::configure_internal(win_config.second);
const unsigned int idx = 2 * num_arguments_per_4D_tensor();
_kernel.setArg<cl_uint>(idx, idx_input);
}
Status CLStackLayerKernel::validate(const ITensorInfo *input, unsigned int axis, unsigned int idx_input, unsigned int num_tensors, const ITensorInfo *output)
{
ARM_COMPUTE_RETURN_ON_ERROR(validate_arguments(input, axis, idx_input, num_tensors, output));
ARM_COMPUTE_RETURN_ON_ERROR(validate_and_configure_window(input->clone().get(), axis, num_tensors, output->clone().get()).first);
return Status{};
}
void CLStackLayerKernel::run(const Window &window, cl::CommandQueue &queue)
{
ARM_COMPUTE_ERROR_ON_UNCONFIGURED_KERNEL(this);
ARM_COMPUTE_ERROR_ON_INVALID_SUBWINDOW(ICLKernel::window(), window);
Window window_out;
window_out.use_tensor_dimensions(_output->info()->tensor_shape());
Window collapsed = window.collapse(ICLKernel::window(), Window::DimZ);
Window slice_in = collapsed.first_slice_window_4D();
Window slice_out = window_out.first_slice_window_4D();
unsigned int idx = 0;
add_4D_tensor_argument(idx, _input, slice_in);
add_4D_tensor_argument(idx, _output, slice_out);
enqueue(queue, *this, slice_in, lws_hint());
}
} // namespace arm_compute
|
1. Field
Apparatuses and methods consistent with exemplary embodiments relate to acquiring a magnetic resonance (MR) image, and more particularly, to acquiring an MR image that allow setting of a plurality of regions and imaging of an independent MR image for each of the plurality of regions.
2. Description of the Related Art
Magnetic resonance images are obtained by using information determined through resonance of atomic nuclei exposed to a magnetic field. The resonance of atomic nuclei is a phenomenon where an atomic nucleus in a low energy state absorbs radio frequency (RF) energy and is excited to a higher energy state when a specific RF is incident on the atomic nucleus magnetized by an external magnetic field. Atomic nuclei have different resonant frequencies depending on their types, and resonance thereof is affected by the intensity of an external magnetic field. The human body includes a large number of atomic nuclei, and hydrogen nuclei are commonly used for magnetic resonance imaging (MRI).
An MRI system includes a magnet creating a main magnetic field in an imaging space, an RF coil generating an RF signal into the imaging space, and gradient coils creating magnetic field gradients for selecting a region of an object to be imaged from the imaging space. The MRI system applies pulse sequences designed for imaging the object to the RF coil and the gradient coils and acquires an echo signal corresponding to the RF signal generated into the imaging space. Signals output from the RF coil and the gradient coils have errors due to several factors, and these errors in the output signals may degrade a signal-to-noise ratio (SNR) of an MR image and cause artifacts in the MR image.
MRI systems use a noninvasive imaging technique and provide images with superior tissue contrast. In addition, unlike the computed tomography (CT) imaging, MRI images are not affected by bone artifacts. Furthermore, the MRI systems can produce various cross-sectional images in desired directions without changing a position of an object, and thus, they are widely used in many medical areas alone or in conjunction with other imaging diagnostic tools.
When a scout image of an object cannot be obtained by imaging a wide range of the object, it may be difficult to set a precise field of view (FOV). Furthermore, when MR images have to be acquired for different lesions in one object and the lesions are far away from each other although they are in the same local range, multiple MRI scans are required, which is not only time consuming and expensive but also causes adverse health effects due to a high specific absorption rate (SAR). |
Rapper and Lil Wayne protege Lil Twist was reportedly driving Justin Bieber’s car when a paparazzo was struck and killed by an SUV.
Although the car belonged to the Canadian pop star, Lil Twist, the 19-year-old rapper was behind the wheel when a California Highway Patrol officer pulled the vehicle over. According to TMZ, Bieber and Lil Twist had been hanging out at the Four Seasons prior to the accident.
At some point on Tuesday afternoon, the rapper decided to leave the hotel with Bieber’s car. At around 6 pm, the vehicle was pulled over by the California Highway Patrol. It was around this time that the tragic incident took place.
The Los Angeles Times reports that a photographer attempted to get a shot of the singer’s Ferrari after the officer had stopped the car. After getting the pics, the man then attempted to cross the street to return to his car. He was struck by an SUV in the process.
The photographer later died from his injuries at the Ronald Reagan UCLA Medical Center.
The Inquisitr previously reported that Justin Bieber is hoping the incident will help inspire new laws to help control the actions of the paparazzi. The singer recently issued a statement about the accident to the public.
Sean Burke, the founder of the Paparazzi Reform Initiative, is also interested in getting new laws in place to help keep incidents like this from happening in the future.
Are you a fan of Lil Twist? Do you think new laws would help control photographers looking to get pictures of celebrities? |
This invention relates, generally, to compositions and methods useful for sealing or plugging porous or fractured subterranean formations encountered during the drilling of vertical or horizontal earthen boreholes, or during the excavation of earthen trenches, while fluid is flowing into or out of the borehole or trench. The invention also generally relates to compositions and methods useful for sealing or plugging porous or fractured formations associated with any subterranean cavity, which may have fluids flowing into or out of it. More specifically the invention relates to compositions and methods involving coated mixtures of bentonite and water-swellable polymer which, when mixed with an aqueous fluid, result in a pumpable suspension useful for sealing or plugging porous or fractured subterranean formations.
Rotary drilling methods have long been used to drill vertical boreholes to reach subterranean aquifers, oil and gas reservoir zones and mineral deposits. In most of these boreholes a drilling fluid comprised of water, clay, especially bentonite, polymers, weighting materials and other additives is used to lubricate the drill pipe as it rotates in the bore, cool the drill head as it cuts through the formations and carry the soil and rock cuttings produced during the process back to the surface. If the circulating drilling fluid were to be rapidly lost to porous formations encountered during drilling, so that it could not be quickly and fully replaced, the integrity of the bore might be compromised. For this reason the development of a wide variety of lost circulation control agents has long been a priority in the drilling industry.
More recently vertical drilling techniques have been adapted for the drilling of relatively shallow, horizontal bores using a process known as horizontal directional drilling. The typical depth for most horizontal directional drill bores is less than 50 feet below the surface. This process is used to install such things as oil, gas, water and sewer pipelines, and power and telecommunications cables without the need for extensive surface excavations. The limited impact that this method has on streets, highways and other surface infrastructure makes it particularly useful in metropolitan areas.
The horizontal orientation of the borehole in this method results in a large amount of friction between the drill pipe and the borehole. This makes it impractical to rapidly rotate the drill pipe to power the drill head and enable it to bore the hole, as is typically done in vertical bores. As a result, alternate methods of drilling have been developed. The simplest and most often used of these methods relies on pumping a fluid under high pressure through the drill pipe and out through small openings in a xe2x80x9cduck-billxe2x80x9d shaped drill bit where it acts to hydraulically cut through the soil that the bore is penetrating. When this technique is employed it is standard practice for the drilling fluid to be composed of water, or water and bentonite clay. The drill pipe is then advanced through the soil as it is cut and flushed away by the hydraulic action of the drilling fluid at the drill bit. It is quite common in this type of drilling for all of the drilling fluid, and the drill cuttings that it contains, to be lost into the formation that is being drilled. This method is very effective in relatively soft, loose and unconsolidated soil, sand or gravel zones. It is not effective in hard rock situations, however.
When hard rock is encountered a drilling method is used that employs a rotating tri-cone bit similar to that used in most vertical drilling operations. In vertical drilling the rotation of the drill pipe also rotates the drill bit allowing the teeth on the cones to cut the formation into small pieces that can be flushed out of the hole. In horizontal directional drilling the friction of the drill pipe against the bore hole, especially along the bottom of the hole where the pipe usually lies, makes it impractical to rapidly rotate the drill pipe to power the drill bit. Instead, a xe2x80x9cmud motorxe2x80x9d located within the drill pipe, that is powered by the hydraulic movement of the drilling fluid through the pipe, is used to rotate the head of the drill bit.
In order to effectively use a mud motor a sufficient volume of drilling fluid must pass through the motor to turn it. If the volume of drilling fluid is insufficient then the motor will stop working. This makes it critical to immediately replace any drilling fluid lost to highly porous zones encountered during drilling in order to ensure continued operation of a mud motor system. Unfortunately, most horizontal directional drilling equipment has only limited ability to mix and store fresh drilling fluids. As a result, the use of mud motor-driven equipment is often difficult or impossible in highly fractured rock or in zones of large rock cobbles where significant loss of drilling fluids to the formation is common. For this reason it is particularly important that the loss of drilling fluids to the formation be controlled so that the drilling fluid can be returned to the surface where it can be cleaned and reused.
Drilling fluid technology has also been adapted for use in the excavation of earthen trenches for foundation and subterranean hydraulic barrier construction. In this application the excavation fluid is primarily used to stabilize the trench walls. When properly practiced the technique is sufficiently successful to enable excavation through unstable soils, such as loose sands, to depths of 50 to 100 feet or more. Where excavation fluid is rapidly lost from the trench into the formation the integrity of the trench walls can be compromised often leading to wall collapse.
Many attempts have been made over the years to devise methods of controlling the loss of fluids to porous subterranean formations during all types of drilling and excavation operations. The references cited here provide a listing of many of the patented technologies that have been invented for this purpose. These technologies suffer from being too complex, too difficult to use, too imprecise and difficult to properly position in the borehole so as to adequately and consistently seal the desired zone in the formation, or too prone to premature gellation causing problems in mixing, pumping or placement. This is particularly true for small drilling operations such as are typical for water, mineral exploration and environmental monitoring wells or for horizontal directional drilling. In these cases it is necessary to have a simple method of easily placing a sealant and plugging agent that will not plug the mixing, pumping or conveying equipment while still effectively sealing the porous zone in the borehole. The composition and method of the present invention succeed in doing this.
It is, therefore, the primary object of the present invention to provide an improved composition and method for the sealing and plugging of porous subterranean formations.
It is also an object of the present invention to provide a composition that is capable of controlling the loss of drilling fluids into porous or fractured subterranean formations.
It is another object of the present invention to provide a method of drilling a well into the earth whereby the phenomenon of loss of the drilling fluid into porous or fractured formations is substantially or entirely eliminated.
It is another object of the present invention to provide a composition and method that is capable of plugging and sealing porous zones within subterranean formations by introducing a plug or pill into the borehole that is useful for sealing discrete, periodic high porosity zones within a formation.
It is yet another object of the present invention to provide a composition and method that is capable of plugging and sealing frequently encountered porous zones or continuous porous zones within a formation penetrated by a well or borehole by using the composition as a drilling fluid.
It is a further object of the present invention to provide a composition and method that is capable of stabilizing boreholes, especially horizontal boreholes in unstable formations such as gravel or loose sand formations, particularly where the borehole must be temporarily abandoned or left idle, so that it can remain intact and can be re-entered at a later time and the drilling can be continued without interference from hole collapse.
It is also an object of the present invention to provide a composition that is capable of controlling the loss of excavation fluids into porous subterranean formations.
It is another object of the present invention to provide a method of excavation into the earth whereby the phenomenon of loss of excavation fluid into porous or fractured formations is substantially alleviated or eliminated entirely.
It is another object of the present invention to provide a composition and method that is capable of sealing and plugging porous zones within subterranean formations by introducing a plug or pill into excavation trench, pit or hole that is useful for sealing discrete, periodic high porosity zones within a formation.
It is yet another object of the present invention to provide a composition and method that is capable of sealing and plugging frequently encountered porous zones or continuous porous zones in a formation intersected by an excavation trench, pit or hole by using the composition as an excavation fluid.
It is a further object of the present invention to provide a composition and method that is capable of stabilizing excavated trenches, pits or holes in unstable formations such as gravel or loose sand formations, particularly where the excavation must be temporarily abandoned or left idle, so that it can remain intact and can be re-entered at a later time and the excavation can be continued without interference from wall collapse.
Other objects and advantages of the present invention will become apparent as a description thereof proceeds.
The present invention is directed toward a unique composition and method useful for the sealing and plugging of zones of highly porous, subterranean formations that may be encountered during the drilling of holes using the technique of horizontal directional drilling. The invention may also be used for plugging and sealing permeable formations encountered during the drilling of vertical bore holes, such as those drilled for water, mineral exploration and environmental monitoring wells, oil and gas exploration wells or for any other type of well. The invention may further be used for sealing and plugging zones of high permeability in formations encountered during excavation of trenches using the technique known as slurry trench excavation, or any other subterranean hole, void or gap encountered during any boring or excavation in order to limit or prevent the movement of water through it.
In particular it has been found that the injection into the bore hole of a suspension comprised of water, or an aqueous, clay-based drilling fluid, mixed with granular bentonite and granular water-insoluble, water-swellable, cross-linked polyacrylamide that have been coated with a non-cross linked, water-soluble, high molecular weight polymer substantially and unexpectedly seals and plugs high permeability formations that are encountered during drilling without causing plugging of the mixing, pumping or conveying equipment or, without forming a troublesome plug in the borehole that might impede further drilling or excavation. |
The hundreds of millions of voters whose collective actions will determine the course of India’s 16th general elections represent an electorate that is changing in many complex ways which we are only beginning to understand.
To better understand these changes, the Center for the Advanced Study of India (CASI) at the University of Pennsylvania, in conjunction with the Carnegie Endowment, partnered with the Lok Foundation to conduct a survey of a cross-section of the population. CMIE, on behalf of the Lok Foundation, conducted face-to-face interviews of 68,500 randomly selected Indians across 24 states and union territories between September and December 2013.
The “Lok Surveys” aim to track the attitudes of individuals from these households over the next several years, as part of a new effort to understand the social and political reconfigurations taking place across India today. Because the majority of our sample is urban, we use 2011 Census data to reweight our sample to ensure representativeness of various social groups.
The Lok Survey was conducted in the final quarter of 2013, and things have obviously evolved since then. Last September, when it was launched, AAP was still a curiosity and Telangana had not yet become a reality.
While our aim is to understand the deeper trends that will shape the forthcoming polls, we cannot ignore the big questions around partisan performance.
Turning to the electoral horse race, we focus our attention on 15 large states comprising roughly 51,000 respondents (accounting for 87% of LS seats). Our data found the BJP-led NDA occupying a much-improved position.
Our analyses suggest that, as of the start of 2014, the NDA was poised to capture nearly 31% of the all-India vote. This is a significant improvement upon the alliance’s 21.5% vote share from the 2009 elections and considerably greater than the BJP’s own best performance in 1996 when it garnered 25.6% of the vote. The ruling Congress-led UPA, on the other hand, was set to win 23% of the vote—a sharp fall from its 31.5% vote share in 2009.
Our numbers for the NDA are broadly in line with other leading surveys. The Centre for the Study of Developing Societies (CSDS) estimated vote shares of 29 and 36% for the NDA in their July 2013 and January 2014 tracker polls, respectively. Given the timing of our survey, our estimate is squarely in line with these projections. Our estimate for the UPA is somewhat lower than CSDS’ but virtually identical to the January 2014 ABP-Nielsen survey.
The data suggests NDA has gained significant ground in West and North India—places where it has already had a presence (below). In Bihar, we estimate that the BJP’s vote share shot up by as much as 23 percentage points by end 2013. In UP, the BJP has the most support with 29% vote share, five and nine percentage points over the Bahujan Samaj Party and Samajwadi Party, respectively. In other regions, the BJP has made impressive strides although usually from a relatively low base. Without prepoll alliances, these gains may not translate into a large share of seats.
The BJP has made a strategic bet that Modi would energize three key demographics: youth, urban dwellers, and individuals belonging to the Other Backward Classes (OBCs)—a group to which Modi belongs. Our data suggest the BJP has not only succeeded in mobilizing these groups but has gone beyond. The NDA holds a lead over the UPA in cities and towns (30 to 24%) as well as in rural areas (31 to 23%). The gains to the NDA appear to be driven largely by the better off: while the NDA is leading UPA among literate voters in both urban and rural areas, the race is much closer among illiterates (chart on top).
And what about the vaunted youth vote? Young voters between 18 and 34 years favor the NDA over the UPA but not significantly more than older voters. In the Hindi heartland states where the BJP vote swing is the most pronounced, the party has succeeded in bringing voters across all age groups into its net.
In terms of social groupings, the NDA continues to struggle attracting support from Muslims, who prefer the UPA by a margin of nearly 3 to 1. UPA retains its traditional lead over the NDA when it comes to Scheduled Tribes (31 to 23%) but trails the NDA when it comes to Dalits (21 to 25%) (chart on top). When it comes to the pivotal OBC category, however, the NDA lead widens; it is polling nearly 15 percentage points above the UPA (35 to 21%). The support of the Upper Castes remains solidly with the NDA: 42% versus just 19% for the UPA.
Even if the BJP does as well as this and other surveys are predicting, getting the coalition math right remains a challenge. The BJP will have to win over at least a few of the most significant regional players, several of whom have opened up commanding leads in their strongholds. These include Mamata Banerjee’s Trinamool Congress, which is dominating the Left Front (41 to 25%) in West Bengal and Jayalalaitha’s AIADMK in Tamil Nadu, which is leading the opposition DMK 27 to 20%.
The groundswell of support for the BJP and the corresponding weakening of the Congress are two unmistakable trends emanating from our data. This change in the overall balance of power begs the question why Indian voters are reassessing their loyalties in such a stark manner. This is the subject of our next piece.
Kapur and Sircar are with CASI at the University of Pennsylvania. Vaishnav is with the Carnegie Endowment for International Peace. |
Validation of a swallowing disturbance questionnaire for detecting dysphagia in patients with Parkinson's disease Underreporting of swallowing disturbances by Parkinson's disease (PD) patients may lead to delay in diagnosis and treatment, alerting the physician to an existing dysphagia only after the first episode of aspiration pneumonia. We developed and validated a swallowing disturbance questionnaire (SDQ) for PD patients and compared its findings to an objective assessment. Fiftyseven PD patients (mean age 69 ± 10 years) participated in this study. Each patient was queried about experiencing swallowing disturbances and asked to complete a selfreported 15item yes/no questionnaire on swallowing disturbances (24 replied no). All study patients underwent a physical/clinical swallowing evaluation by a speech pathologist and an otolaryngologist. The 33 patients who complained of swallowing disturbances also underwent fiberoptic endoscopyic evaluation of swallowing (FEES). According to the ROC test, the optimal score (where the sensitivity and specificity curves cross) is 11 (sensitivity 80.5%, specificity 81.3%). Using the SDQ questionnaire substantially reduced Type I errors (specifically, an existing swallowing problem missed by the selected cutoff point). On the basis of the SDQ assessment alone, 12 of the 24 (50%) noncomplaining patients would have been referred to further evaluation that they otherwise would not have undergone. The SDQ emerged as a validated tool to detect early dysphagia in PD patients. © 2007 Movement Disorder Society |
<filename>duang/src/main/java/com/robot/service/common/requests/set/SetrOutRequest.java
package com.robot.service.common.requests.set;
import com.robot.service.common.ActionRequest;
/**
* 下发路径指令
*
* @author Laotang
*/
public class SetrOutRequest extends ActionRequest {
public SetrOutRequest(String deviceId, String param) {
super(deviceId, param);
}
@Override
public String cmd() {
return "setrout";
}
}
|
<reponame>ayyoob/component-dep
/*******************************************************************************
* Copyright (c) 2015-2016, WSO2.Telco Inc. (http://www.wso2telco.com) All Rights Reserved.
*
* WSO2.Telco Inc. licences this file to you under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
******************************************************************************/
package com.wso2telco.dep.mediator.entity.cep;
import java.util.List;
import javax.xml.bind.annotation.XmlElement;
import javax.xml.bind.annotation.XmlRootElement;
@XmlRootElement(name="Group")
public class Group {
private String groupName;
private String operator;
private String dayAmount;
private String monthAmount;
private List<ServiceProvider> serviceProviderList;
public String getGroupName() {
return groupName;
}
@XmlElement(name="GroupName")
public void setGroupName(String groupName) {
this.groupName = groupName;
}
public String getOperator() {
return operator;
}
@XmlElement(name="Operator")
public void setOperator(String operator) {
this.operator = operator;
}
public String getDayAmount() {
return dayAmount;
}
@XmlElement(name="DayAmount")
public void setDayAmount(String dayAmount) {
this.dayAmount = dayAmount;
}
public String getMonthAmount() {
return monthAmount;
}
@XmlElement(name = "MonthAmount")
public void setMonthAmount(String monthAmount) {
this.monthAmount = monthAmount;
}
public List<ServiceProvider> getServiceProviderList() {
return serviceProviderList;
}
@XmlElement(name="ServiceProvider")
public void setServiceProviderList(List<ServiceProvider> serviceProviderList) {
this.serviceProviderList = serviceProviderList;
}
}
|
The present invention relates generally to novel herbicidal compounds and methods for their use in controlling unwanted plant species in agriculture. In particular, the present invention relates to novel 3-(pyrazolylphenyl)propanoic acids and derivatives and their use as herbicides.
There is a continuing demand for new herbicides. Herbicides are useful for controlling unwanted vegetation which may otherwise cause significant damage to crops such as wheat, corn, soybeans and cotton, to name a few. For crop protection, so-called "selective" herbicides are desired which can control the weeds without damaging the crop. Such crops are said to exhibit tolerance to the herbicide. In certain other situations, it is desirable to use herbicides that provide complete vegetation control such as in areas around railroad tracks and other structures. While many commercial products are available that provide selective or complete vegetation control, the demand exists for new, safe herbicides that are more effective and less costly.
German Patent Application 4419517-A1 discloses, among others, the following compound having herbicidal activity: ##STR2## |
The Relationship Between Family-of-Origin Experience and Current Family Violence: A Test of Mediation by Attachment Style and Mental Health Symptom Distress The purpose of this study was to examine whether the presence of substance abuse, physical, sexual, and emotional abuse, and mental illness in the home or family-of-origin is predictive of variance in current family violence perpetration. Additionally, a secondary purpose of this study was to examine whether mental health symptom distress and attachment style mediated the relationship between the presence of traumatic experiences in one's family-of-origin and current family violence perpetration. The results suggested that difficult family-of-origin experiences may predict variance in current family violence indirectly through mental health symptom distress and anxious attachment. |
An overview on the aquatic fern Azolla spp. as a sustainable source of nutrients and bioactive compounds with resourceful applications At present, there is intensive attention for using natural materials as sustainable resources for bioactive compounds. An interesting aquatic fern that is extensively studied is Azolla spp.. Azolla is a genus of aquatic ferns and smallleafed floating plants, native to the tropics, subtropics, and warm temperate regions of Africa, Asia, and the Americas. It is a floating pteridophyte that is symbiotic with the cyanobacterium Anabaena Azollae (). Azolla species were successfully used as a biofertilizer, animal feed, water purifier, biological herbicide and for concentration of nutrients and heavy metals from flood waters (). Azolla was also shown to be a source of bioactive secondary metabolites such as phenols, tannins, sugar, anthraquinone glycosides and steroids, and also rich in protein, vitamin and minerals (). Concerning Azolla spp. the current study overviewed: classification, distribution and important applications in different fields in many countries and in Egypt in particular. ARTICLE INFO ABSTRACT Article History: Received: Jan. 29, 2021 Accepted: Feb. 15, 2021 Online: Feb. 20, 2021 _______________ |
<reponame>WormLabCaltech/alaska
"""Script to run various analyses.
This script is used by AlaskaServer to perform quality control, read
alignment/quantification and differential expression analysis.
"""
__author__ = '<NAME> (<NAME>'
__copyright__ = 'Copyright 2017 WormLabCaltech'
__credits__ = ['<NAME>', '<NAME>', '<NAME>']
__license__ = "MIT"
__version__ = "alpha"
__maintainer__ = "<NAME> (J<NAME>"
__email__ = "<EMAIL>"
__status__ = "alpha"
import os
import sys
import time
import json
import queue
import tarfile
from threading import Thread
import multiprocessing as mp
from multiprocessing import Process
import subprocess as sp
def print_with_flush(s='', **kwargs):
"""
Prints the given string and passes on additional kwargs to the builtin
print function. This function flushes stdout immediately.
Arguments:
s -- (str) to print
kwargs -- additional arguments to pass to print()
Returns: None
"""
print(s, **kwargs)
sys.stdout.flush()
def load_proj(_id):
"""
Loads the project json into dictionary object.
Arguments:
_id
"""
# Read json.
with open(_id, 'r') as f:
loaded = json.load(f)
return loaded
def run_sys(cmd, prefix=''):
"""
Runs a system command and echos all output.
This function blocks until command execution is terminated.
"""
print_with_flush('## ' + ' '.join(cmd))
output = ''
with sp.Popen(cmd, stdout=sp.PIPE, stderr=sp.STDOUT, bufsize=1, universal_newlines=True) as p:
# while p.poll() is None:
# try:
# line, _err = p.communicate(timeout=30)
#
# if not line.isspace() and len(line) > 1:
# output += line
# print_with_flush('{}: {}'.format(prefix, line), end='')
# except sp.TimeoutExpired:
# sys.stdout.flush()
# if p.poll() is None:
# continue
# else:
# break
#
output = ''
while p.poll() is None:
line = p.stdout.readline()
if not line.isspace() and len(line) > 1:
output += line
print(prefix + ': ' + line, end='')
sys.stdout.flush()
p.stdout.read()
# p.stderr.read()
p.stdout.close()
# p.stderr.close()
if p.returncode != 0:
sys.exit('command terminated with non-zero return code {}!'.format(p.returncode))
return output
def archive(out, source_dir):
"""
Archive given source directory into output file.
"""
with tarfile.open(out, 'w:gz') as tar:
if isinstance(source_dir, str):
tar.add(source_dir, arcname=os.path.sep)
elif isinstance(source_dir, list):
for d in source_dir:
tar.add(d)
######### These functions must be here to allow multiprocessing.
def read_distribution(_id, bed_path):
"""
Helper function to run read_distribution.py
"""
args = ['read_distribution.py']
args += ['-i', '{}_sorted.bam'.format(_id)]
args += ['-r', bed_path]
# print_with_flush(args)
output = run_sys(args, prefix=_id)
# output file
with open('{}_distribution.txt'.format(_id), 'w') as out:
out.write(output)
def geneBody_coverage(_id, bed_path):
"""
Helper function to run geneBody_coverage.py
"""
args = ['geneBody_coverage.py']
args += ['-i', '{}_sorted.bam'.format(_id)]
args += ['-r', bed_path]
args += ['-o', '{}_coverage'.format(_id)]
run_sys(args, prefix=_id)
def tin(_id, bed_path):
"""
Helper function to run tin.py
"""
args = ['tin.py']
args += ['-i', '{}_sorted.bam'.format(_id)]
args += ['-r', bed_path]
output = run_sys(args, prefix=_id)
# output file
with open('{}_tin.txt'.format(_id), 'w') as out:
out.write(output)
def fastqc(_id):
"""
Helper function to run fastqc.
"""
args = ['fastqc', '{}_sorted.bam'.format(_id)]
run_sys(args, prefix=_id)
def mp_helper(f, args, name, _id):
"""
Helper function for multiprocessing.
"""
print_with_flush('# starting {} for {}'.format(name, _id))
f(*args)
print_with_flush('# finished {} for {}'.format(name, _id))
######### These functions must be here to allow multiprocessing.
def run_qc(proj, nthreads):
"""
Runs read quantification with RSeQC, FastQC and MultiQC.
"""
def bowtie2(_id, path, bt2_path, reads, t):
"""
Helper function to call bowtie2 alignment.
"""
upto = 2 * (10 ** 5)
args = ['bowtie2', '-x', bt2_path]
# single/paired-end
if t == 1:
args += ['-U', ','.join(reads)]
elif t == 2:
m1 = []
m2 = []
for pair in reads:
m1.append(pair[0])
m2.append(pair[1])
args += ['-1', ','.join(m1)]
args += ['-2', ','.join(m2)]
args += ['-S', '{}/{}_alignments.sam'.format(path, _id)]
args += ['-u', str(upto)]
args += ['--threads', str(nthreads)]
args += ['--verbose']
output = run_sys(args, prefix=_id)
# Write bowtie stderr output.
first = '{} reads; of these'.format(upto)
last = 'overall alignment rate'
found = False
bt2_info = ''
for line in output.split('\n'):
if first in line:
found = True
if found:
bt2_info += line + '\n'
if last in line:
break
# Write the file.
with open('{}/{}_sorted.txt'.format(path, _id), 'w') as f:
f.write(bt2_info)
def samtools_sort(_id):
"""
Helper function to call samtools to sort .bam
"""
sam_path = '{}_alignments.sam'.format(_id)
args = ['samtools', 'sort', sam_path]
sorted_bam = '{}_sorted.bam'.format(_id)
args += ['-o', sorted_bam]
args += ['-@', str(nthreads-1)]
args += ['-m', '2G']
run_sys(args, prefix=_id)
def samtools_index(_id):
"""
Helper function to call samtools to index .bam
"""
args = ['samtools', 'index', '{}_sorted.bam'.format(_id)]
args += ['-@', str(nthreads-1)]
run_sys(args, prefix=_id)
def multiqc(_id=''):
"""
Helper function to run multiqc.
"""
args = ['multiqc', '.', '--ignore', '*.sam', '--ignore', 'qc_out.txt']
run_sys(args, prefix=_id)
########## HELPER FUNCTIONS END HERE ###########
# First, make sure that environment variables are set.
os.environ['LC_ALL'] = 'C.UTF-8'
os.environ['LANG'] = 'C.UTF-8'
print_with_flush('{} samples detected...'.format(len(proj['samples'])), end='')
for _id in proj['samples']:
print_with_flush('{}({})'.format(_id, proj['samples'][_id]['name']), end=' ')
print_with_flush()
# run kallisto to get pseudobam
# run_kallisto(proj, nthreads, qc=True, nbootstraps=0, ver=235)
wdir = os.getcwd()
for _id in proj['samples']:
# define necessary variables
name = proj['samples'][_id]['name']
path = '1_qc/{}'.format(name)
org = proj['samples'][_id]['organism']
genus = org['genus']
species = org['species']
version = org['version']
bed_path = '/alaska/root/organisms/{}/{}/{}/reference'.format(genus,
species,
version)
bed_path += '/{}_{}_{}.bed'.format(genus[0], species, version)
bt2_idx = '{}_{}_{}'.format(genus[0], species, version)
bt2_path = '/alaska/root/organisms/{}/{}/{}/index/{}'.format(genus,
species,
version,
bt2_idx)
# Align with bowtie2
if (proj['samples'][_id]['type'] == 1):
reads = proj['samples'][_id]['reads'].keys()
elif (proj['samples'][_id]['type'] == 2):
reads = proj['samples'][_id]['pairs']
else:
print_with_flush('unrecognized sample type!')
bowtie2(name, path, bt2_path, reads, proj['samples'][_id]['type'])
os.chdir(path)
print_with_flush('# changed working directory to {}'.format(path))
_id = name
# Sort and index reads with samtools first.
samtools_sort(_id)
samtools_index(_id)
# sambamba_sort(_id)
# If nthreads > 1, we want to multithread.
if nthreads > 1:
pool = mp.Pool(processes=nthreads)
print_with_flush('# multithreading on.')
args = [
[read_distribution, [_id, bed_path], 'read_distribution', _id],
[geneBody_coverage, [_id, bed_path], 'geneBody_coverage', _id],
[tin, [_id, bed_path], 'tin', _id],
[fastqc, [_id], 'fastqc', _id],
]
# start processes
pool.starmap(mp_helper, args)
else:
# read_distribution.py
read_distribution(_id, bed_path)
# geneBody_coverage.py
geneBody_coverage(_id, bed_path)
# tin.py
tin(_id, bed_path)
# FastQC
fastqc(_id)
# # MultiQC
# multiqc(_id)
os.chdir(wdir)
print_with_flush('# returned to {}'.format(wdir))
print_with_flush('# running multiqc for all samples')
path = '1_qc'
os.chdir(path)
multiqc()
os.chdir(wdir)
print_with_flush('# qc finished, archiving')
archive(path + '.tar.gz', path)
print_with_flush('# done')
def run_kallisto(proj, nthreads):
"""
Runs read quantification with Kallisto.
Assumes that the indices are in the folder /organisms
"""
print_with_flush('{} samples detected...'.format(len(proj['samples'])), end='')
for _id in proj['samples']:
print_with_flush('{}({})'.format(_id, proj['samples'][_id]['name']), end=' ')
print_with_flush()
for _id in proj['samples']:
name = proj['samples'][_id]['name']
path = '2_alignment/{}'.format(name)
args = ['kallisto', 'quant']
# first find the index
org = proj['samples'][_id]['organism']
genus = org['genus']
species = org['species']
version = org['version']
idx_path = '/alaska/root/organisms/{}/{}/{}/index'.format(genus,
species,
version)
idx_path += '/{}_{}_{}.idx'.format(genus[0], species, version)
args += ['-i', idx_path]
# find the output directory
args += ['-o', path]
# find the number of bootstraps
nbootstraps = proj['samples'][_id]['bootstrap_n']
args += ['-b', str(nbootstraps)]
# number of threads
args += ['-t', str(nthreads)]
# single/paired end
if (proj['samples'][_id]['type'] == 1):
length = proj['samples'][_id]['length']
stdev = proj['samples'][_id]['stdev']
arg = ['--single', '-l', str(length), '-s', str(stdev)]
args += arg
# finally, add the sample reads
for read in proj['samples'][_id]['reads']:
args.append(read)
elif (proj['samples'][_id]['type'] == 2):
for pair in proj['samples'][_id]['pairs']:
args.append(pair[0])
args.append(pair[1])
else:
print_with_flush('unrecognized sample type!')
# Add bias correction flag
args += ['--bias']
_id = name
run_sys(args, prefix=_id)
path = '2_alignment'
print_with_flush('# kallisto finished, archiving')
archive(path + '.tar.gz', path)
print_with_flush('# done')
def run_sleuth(proj):
"""
Runs differential expression analysis with sleuth.
Assumes that the design matrix is already present in the directory.
Once sleuth is finished, runs TEA.
"""
def run_tea(parent, sleuth, out, q_threshold=0.05):
"""
Runs TEA on sleuth output.
"""
try:
import tissue_enrichment_analysis as tea
except ImportError as e:
print_with_flush('# TEA is not installed...skipping')
sys.exit(0)
try:
import pandas as pd
except ImportError as e:
print_with_flush('# pandas is not installed...skipping')
sys.exit(0)
analyses = ['tissue', 'phenotype', 'go']
# Load sleuth results.
wdir = os.getcwd()
print_with_flush('# entering 3_diff_exp')
os.chdir(parent)
print_with_flush('# creating {} directory'.format(out))
os.makedirs(out, exist_ok=True)
for file in os.listdir(sleuth):
if file.endswith('.csv'):
df = pd.read_csv(os.path.join(sleuth, file), index_col=0)
gene_list = df[df.qval < q_threshold].ens_gene
name = os.path.splitext(os.path.basename(file))[0]
if len(gene_list) == 0:
print_with_flush(('# there are no genes with q < {} in '
+ '{}!').format(q_threshold, file))
print_with_flush('# this means there are no significantly '
+ 'differentially-expressed genes for '
+ 'this set of conditions.')
continue
for analysis in analyses:
print_with_flush(('# performing {} enrichment analysis '
+ 'for {}').format(analysis, file))
fname = '{}_{}'.format(name.replace('betas_wt', out), analysis)
title = os.path.join(out, fname)
df_dict = tea.fetch_dictionary(analysis)
df_results = tea.enrichment_analysis(gene_list, df_dict,
aname=title + '.csv',
save=True,
show=False)
tea.plot_enrichment_results(df_results, analysis=analysis,
title=title, save=True)
os.chdir(wdir)
print_with_flush('# returned to root')
args = ['Rscript']
args += ['./sleuth.R']
args += ['-d', './3_diff_exp']
args += ['-k', './2_alignment']
args += ['-o', './3_diff_exp/betas']
# args += ['--shiny']
run_sys(args)
# Run tissue enrichment only if it is flagged to do so
parent, sleuth, out = '3_diff_exp', 'betas', 'enrichment'
if proj['enrichment']:
print_with_flush('# sleuth finished, starting enrichment analysis')
run_tea(parent, sleuth, out)
print_with_flush('# enrichment analysis finished, archiving')
else:
print_with_flush('# this organisms is not whitelisted for enrichment '
+ 'analysis, archiving')
archive(parent + '.tar.gz', parent)
# Archive all
print_with_flush('# all analyses finished, archiving entire project')
dirs_to_archive = []
for d in os.listdir():
if os.path.isdir(d) and d not in ['_temp', '0_raw_reads']:
dirs_to_archive.append(d)
archive(proj['id'] + '.tar.gz', dirs_to_archive)
print_with_flush('# done')
if __name__ == '__main__':
import argparse
parser = argparse.ArgumentParser(description='Perform analysis.')
parser.add_argument('type', choices=['qc', 'kallisto', 'sleuth'])
parser.add_argument('--threads', type=int, default=1)
args = parser.parse_args()
nthreads = args.threads
# Assume only one json file exists in current directory.
files = os.listdir()
# Find that json.
for f in files:
if f.endswith('.json'):
data = f
break
# Load project json.
proj = load_proj(data)
print_with_flush('{} loaded'.format(data))
if args.type == 'qc':
run_qc(proj, nthreads)
elif args.type == 'kallisto':
run_kallisto(proj, nthreads)
elif args.type == 'sleuth':
run_sleuth(proj)
|
Alternative Credentials for Workforce Development. In a highly transformative environment like health care, continued professional development and learning is essential. For individuals without degrees, stackable credentials offer another pathway to a degree, and for those with multiple and/or advanced degrees, microcredentials hold promise for retooling and updating knowledge, skills, and abilities. J Contin Educ Nurs. 2018;49:449-450. |
""" Module implementing GAN which will be trained using the Progressive growing
technique -> https://arxiv.org/abs/1710.10196
"""
import datetime
import os
import time
import timeit
import copy
import numpy as np
import torch as th
class Generator(th.nn.Module):
""" Generator of the GAN network """
def __init__(self, depth=7, latent_size=512, use_eql=True):
"""
constructor for the Generator class
:param depth: required depth of the Network
:param latent_size: size of the latent manifold
:param use_eql: whether to use equalized learning rate
"""
from torch.nn import ModuleList, Conv2d
from MSG_GAN.CustomLayers import GenGeneralConvBlock, \
GenInitialBlock, _equalized_conv2d
super().__init__()
assert latent_size != 0 and ((latent_size & (latent_size - 1)) == 0), \
"latent size not a power of 2"
if depth >= 4:
assert latent_size >= np.power(2, depth - 4), "latent size will diminish to zero"
# state of the generator:
self.use_eql = use_eql
self.depth = depth
self.latent_size = latent_size
# register the modules required for the Generator Below ...
# create the ToRGB layers for various outputs:
if self.use_eql:
def to_rgb(in_channels):
return _equalized_conv2d(in_channels, 3, (1, 1), bias=True)
else:
def to_rgb(in_channels):
return Conv2d(in_channels, 3, (1, 1), bias=True)
# create a module list of the other required general convolution blocks
self.layers = ModuleList([GenInitialBlock(self.latent_size, use_eql=self.use_eql)])
self.rgb_converters = ModuleList([to_rgb(self.latent_size)])
# create the remaining layers
for i in range(self.depth - 1):
if i <= 2:
layer = GenGeneralConvBlock(self.latent_size, self.latent_size,
use_eql=self.use_eql)
rgb = to_rgb(self.latent_size)
else:
layer = GenGeneralConvBlock(
int(self.latent_size // np.power(2, i - 3)),
int(self.latent_size // np.power(2, i - 2)),
use_eql=self.use_eql
)
rgb = to_rgb(int(self.latent_size // np.power(2, i - 2)))
self.layers.append(layer)
self.rgb_converters.append(rgb)
def forward(self, x):
"""
forward pass of the Generator
:param x: input noise
:return: *y => output of the generator at various scales
"""
outputs = [] # initialize to empty list
y = x # start the computational pipeline
for block, converter in zip(self.layers, self.rgb_converters):
y = block(y)
outputs.append(converter(y))
return outputs
@staticmethod
def adjust_dynamic_range(data, drange_in=(-1, 1), drange_out=(0, 1)):
"""
adjust the dynamic colour range of the given input data
:param data: input image data
:param drange_in: original range of input
:param drange_out: required range of output
:return: img => colour range adjusted images
"""
if drange_in != drange_out:
scale = (np.float32(drange_out[1]) - np.float32(drange_out[0])) / (
np.float32(drange_in[1]) - np.float32(drange_in[0]))
bias = (np.float32(drange_out[0]) - np.float32(drange_in[0]) * scale)
data = data * scale + bias
return th.clamp(data, min=0, max=1)
class Discriminator(th.nn.Module):
""" Discriminator of the GAN """
def __init__(self, depth=7, feature_size=512,
use_eql=True, gpu_parallelize=False):
"""
constructor for the class
:param depth: total depth of the discriminator
(Must be equal to the Generator depth)
:param feature_size: size of the deepest features extracted
(Must be equal to Generator latent_size)
:param use_eql: whether to use the equalized learning rate or not
:param gpu_parallelize: whether to use DataParallel on the discriminator
Note that the Last block contains StdDev layer
So, it is not parallelized.
"""
from torch.nn import ModuleList
from MSG_GAN.CustomLayers import DisGeneralConvBlock, \
DisFinalBlock, _equalized_conv2d
from torch.nn import Conv2d
super().__init__()
assert feature_size != 0 and ((feature_size & (feature_size - 1)) == 0), \
"latent size not a power of 2"
if depth >= 4:
assert feature_size >= np.power(2, depth - 4), \
"feature size cannot be produced"
# create state of the object
self.gpu_parallelize = gpu_parallelize
self.use_eql = use_eql
self.depth = depth
self.feature_size = feature_size
# create the fromRGB layers for various inputs:
if self.use_eql:
def from_rgb(out_channels):
return _equalized_conv2d(3, out_channels, (1, 1), bias=True)
else:
def from_rgb(out_channels):
return Conv2d(3, out_channels, (1, 1), bias=True)
self.rgb_to_features = ModuleList()
self.final_converter = from_rgb(self.feature_size)
# create a module list of the other required general convolution blocks
self.layers = ModuleList()
self.final_block = DisFinalBlock(self.feature_size * 2, use_eql=self.use_eql)
# create the remaining layers
for i in range(self.depth - 1):
if i > 2:
layer = DisGeneralConvBlock(
int(self.feature_size // np.power(2, i - 3)),
int(self.feature_size // np.power(2, i - 3)),
use_eql=self.use_eql
)
rgb = from_rgb(int(self.feature_size // np.power(2, i - 2)))
else:
layer = DisGeneralConvBlock(self.feature_size * 2, self.feature_size,
use_eql=self.use_eql)
rgb = from_rgb(self.feature_size)
self.layers.append(layer)
self.rgb_to_features.append(rgb)
# handle the case where the depth is less than or equal to 4
if self.depth > 4:
self.rgb_to_features[self.depth - 2] = \
from_rgb(self.feature_size // np.power(2, i - 3))
else:
self.rgb_to_features[self.depth - 2] = \
from_rgb(self.feature_size * 2)
# parallelize the modules from the module-lists if asked to:
if self.gpu_parallelize:
for i in range(len(self.layers)):
self.layers[i] = th.nn.DataParallel(self.layers[i])
self.rgb_to_features[i] = th.nn.DataParallel(
self.rgb_to_features[i])
# Note that since the FinalBlock contains the StdDev layer,
# it cannot be parallelized so easily. It will have to be parallelized
# from the Lower level (from CustomLayers). This much parallelism
# seems enough for me.
def forward(self, inputs):
"""
forward pass of the discriminator
:param inputs: (multi-scale input images) to the network list[Tensors]
:return: out => raw prediction values
"""
assert len(inputs) == self.depth, \
"Mismatch between input and Network scales"
y = self.rgb_to_features[self.depth - 2](inputs[self.depth - 1])
y = self.layers[self.depth - 2](y)
for x, block, converter in \
zip(reversed(inputs[1:-1]),
reversed(self.layers[:-1]),
reversed(self.rgb_to_features[:-1])):
input_part = converter(x) # convert the input:
y = th.cat((input_part, y), dim=1) # concatenate the inputs:
y = block(y) # apply the block
# calculate the final block:
input_part = self.final_converter(inputs[0])
y = th.cat((input_part, y), dim=1)
y = self.final_block(y)
# return calculated y
return y
class MSG_GAN:
""" Unconditional MSG-GAN
args:
depth: depth of the GAN (will be used for each generator and discriminator)
latent_size: latent size of the manifold used by the GAN
use_eql: whether to use the equalized learning rate
use_ema: whether to use exponential moving averages.
ema_decay: value of ema decay. Used only if use_ema is True
device: device to run the GAN on (GPU / CPU)
"""
def __init__(self, depth=7, latent_size=512,
use_eql=True, use_ema=True, ema_decay=0.999,
device=th.device("cpu")):
""" constructor for the class """
from torch.nn import DataParallel
self.gen = Generator(depth, latent_size, use_eql=use_eql).to(device)
# Parallelize them if required:
if device == th.device("cuda"):
self.gen = DataParallel(self.gen)
self.dis = Discriminator(depth, latent_size,
use_eql=use_eql, gpu_parallelize=True).to(device)
else:
self.dis = Discriminator(depth, latent_size, use_eql=True).to(device)
# state of the object
self.use_ema = use_ema
self.ema_decay = ema_decay
self.use_eql = use_eql
self.latent_size = latent_size
self.depth = depth
self.device = device
if self.use_ema:
from MSG_GAN.CustomLayers import update_average
# create a shadow copy of the generator
self.gen_shadow = copy.deepcopy(self.gen)
# updater function:
self.ema_updater = update_average
# initialize the gen_shadow weights equal to the
# weights of gen
self.ema_updater(self.gen_shadow, self.gen, beta=0)
# by default the generator and discriminator are in eval mode
self.gen.eval()
self.dis.eval()
if self.use_ema:
self.gen_shadow.eval()
def generate_samples(self, num_samples):
"""
generate samples using this gan
:param num_samples: number of samples to be generated
:return: generated samples tensor: list[ Tensor(B x H x W x C)]
"""
noise = th.randn(num_samples, self.latent_size).to(self.device)
generated_images = self.gen(noise)
# reshape the generated images
generated_images = list(map(lambda x: (x.detach().permute(0, 2, 3, 1) / 2) + 0.5,
generated_images))
return generated_images
def optimize_discriminator(self, dis_optim, noise,
real_batch, loss_fn,
accumulate=False, zero_grad=True,
num_accumulations=1):
"""
performs one step of weight update on discriminator using the batch of data
:param dis_optim: discriminator optimizer
:param noise: input noise of sample generation
:param real_batch: real samples batch
should contain a list of tensors at different scales
:param loss_fn: loss function to be used (object of GANLoss)
:param accumulate: whether to accumulate or make a step
:param zero_grad: used to control the behaviour of grad buffers
:param num_accumulations: number of accumulation steps performed
(required to scale the loss function)
:return: current loss (accumulation scaled value)
"""
# generate a batch of samples
fake_samples = self.gen(noise)
fake_samples = list(map(lambda x: x.detach(), fake_samples))
# scale the loss by the number of accumulation steps performed
# (if not performed, it is 1)
loss = loss_fn.dis_loss(real_batch, fake_samples) / num_accumulations
# optimize discriminator according to the accumulation dynamics
# zero the grad of the discriminator weights if required
if zero_grad:
dis_optim.zero_grad()
# perform the backward pass to accumulate the gradients
loss.backward()
# if not running in accumulate mode, (make a step)
if not accumulate:
dis_optim.step()
return loss.item()
def optimize_generator(self, gen_optim, noise,
real_batch, loss_fn,
accumulate=False, zero_grad=True,
num_accumulations=1):
"""
performs one step of weight update on generator using the batch of data
:param gen_optim: generator optimizer
:param noise: input noise of sample generation
:param real_batch: real samples batch
should contain a list of tensors at different scales
:param loss_fn: loss function to be used (object of GANLoss)
:param accumulate: whether to accumulate or make a step
:param zero_grad: used to control the behaviour of grad buffers
:param num_accumulations: number of accumulation steps performed
(required to scale the loss function)
:return: current loss (accumulation scaled value)
"""
# generate a batch of samples
fake_samples = self.gen(noise)
loss = loss_fn.gen_loss(real_batch, fake_samples) / num_accumulations
# optimize the generator according the accumulation dynamics
if zero_grad:
gen_optim.zero_grad()
# perform backward pass for gradient accumulation
loss.backward()
# perform an update step if not running in the accumulate mode
if not accumulate:
gen_optim.step()
# if self.use_ema is true, apply the moving average here:
# Note that ema update will also be done only during the update
# pass of the function (not during accumulation).
if self.use_ema:
self.ema_updater(self.gen_shadow, self.gen, self.ema_decay)
return loss.item()
def create_grid(self, samples, img_files,
sum_writer=None, reses=None, step=None):
"""
utility function to create a grid of GAN samples
:param samples: generated samples for storing list[Tensors]
:param img_files: list of names of files to write
:param sum_writer: summary writer object
:param reses: resolution strings (used only if sum_writer is not None)
:param step: global step (used only if sum_writer is not None)
:return: None (saves multiple files)
"""
from torchvision.utils import save_image, make_grid
from torch.nn.functional import interpolate
from numpy import sqrt, power, ceil
# dynamically adjust the colour range of the images:
samples = [Generator.adjust_dynamic_range(sample) for sample in samples]
# resize the samples to have same resolution:
for i in range(len(samples)):
samples[i] = interpolate(samples[i],
scale_factor=power(2, self.depth - 1 - i))
# save the images:
for sample, img_file in zip(samples, img_files):
save_image(sample, img_file, nrow=int(sqrt(sample.shape[0])),
normalize=True, scale_each=True, padding=0)
if sum_writer is not None:
for sample, res in zip(samples, reses):
image = make_grid(sample, nrow=int(ceil(sqrt(sample.shape[0]))),
normalize=True, scale_each=True, padding=0)
sum_writer.add_image(res, image, step)
def _downsampled_images(self, images):
"""
private utility function to compute list of downsampled images
:param images: Original sized images
:return: images => list of downsampled images
"""
from torch.nn.functional import avg_pool2d
# create a list of downsampled images from the real images:
images = [images] + [avg_pool2d(images, int(np.power(2, i)))
for i in range(1, self.depth)]
images = list(reversed(images))
return images
def _get_images_and_latents(self, data_store, normalize_latents):
"""
private utility function to obtain random latent_points and
downsampled images from the datastore
:param data_store: object containing the data
:param normalize_latents: boolean for hyper-sphere normalization
:return: images, latents => images and random latent points
"""
# extract current batch of data for training
batch = next(data_store)
images = batch.to(self.device)
extracted_batch_size = images.shape[0]
# list of downsampled versions of images
images = self._downsampled_images(images)
# sample some random latent points
gan_input = th.randn(
extracted_batch_size, self.latent_size).to(self.device)
# normalize them if asked
if normalize_latents:
gan_input = ((gan_input
/ gan_input.norm(dim=-1, keepdim=True))
* (self.latent_size ** 0.5))
return images, gan_input
def train(self, data, gen_optim, dis_optim, loss_fn, normalize_latents=True,
start=1, num_epochs=12, spoofing_factor=1,
feedback_factor=10, checkpoint_factor=1,
data_percentage=100, num_samples=36,
log_dir=None, sample_dir="./samples",
log_fid_values=False, num_fid_images=50000,
save_dir="./models", fid_temp_folder="./samples/fid_imgs/",
fid_real_stats=None, fid_batch_size=64):
"""
Method for training the network
:param data: pytorch dataloader which iterates over images
:param gen_optim: Optimizer for generator.
please wrap this inside a Scheduler if you want to
:param dis_optim: Optimizer for discriminator.
please wrap this inside a Scheduler if you want to
:param loss_fn: Object of GANLoss
:param normalize_latents: whether to normalize the latent vectors during training
:param start: starting epoch number
:param num_epochs: total number of epochs to run for (ending epoch number)
note this is absolute and not relative to start
:param spoofing_factor: number of actual batches used to spoof a bigger batch
for instance, actual batch size is 16 and
spoofing factor is 4, then virtual batch_size is 64
:param feedback_factor: number of logs generated and samples generated
during training per epoch
:param checkpoint_factor: save model after these many epochs
:param data_percentage: amount of data to be used
:param num_samples: number of samples to be drawn for feedback grid
:param log_dir: path to directory for saving the loss.log file
:param sample_dir: path to directory for saving generated samples' grids
:param log_fid_values: boolean for whether to log fid values during training or not
:param num_fid_images: number of images to generate for calculating the FID
:param save_dir: path to directory for saving the trained models
:param fid_temp_folder: path to save the generated images
:param fid_real_stats: path to the npz stats file for real images
:param fid_batch_size: batch size used for generating fid images
Same will be used for calculating the fid too.
:return: None (writes multiple files to disk)
"""
from tensorboardX import SummaryWriter
from shutil import rmtree
from tqdm import tqdm
from scipy.misc import imsave
from MSG_GAN.FID import fid_score
from math import ceil
from MSG_GAN.utils.iter_utils import hn_wrapper
# turn the generator and discriminator into train mode
self.gen.train()
self.dis.train()
assert isinstance(gen_optim, th.optim.Optimizer), \
"gen_optim is not an Optimizer"
assert isinstance(dis_optim, th.optim.Optimizer), \
"dis_optim is not an Optimizer"
print("Starting the training process ... ")
# create the summary writer
sum_writer = SummaryWriter(os.path.join(log_dir, "tensorboard"))
# create a grid of samples and save it
reses = [str(int(np.power(2, dep))) + "_x_"
+ str(int(np.power(2, dep)))
for dep in range(2, self.depth + 2)]
# create fixed_input for debugging
fixed_input = th.randn(num_samples, self.latent_size).to(self.device)
if normalize_latents:
fixed_input = (fixed_input
/ fixed_input.norm(dim=-1, keepdim=True)
* (self.latent_size ** 0.5))
viter_samples = 2 * data.batch_size * spoofing_factor
total_imgs = len(data.dataset)
total_batches = int(total_imgs / viter_samples)
limit = int(ceil((data_percentage / 100) * total_batches))
# create a global time counter
global_time = time.time()
global_step = 0
for epoch in range(start, num_epochs + 1):
start_time = timeit.default_timer() # record time at the start of epoch
print("\nEpoch: %d" % epoch)
# setup the dataloader (where the real images are sampled from)
real_data_store = iter(hn_wrapper(data))
batch_counter = 0 # counter for number of batches completed
# this includes the two Generator passes and spoofing adjusted
# batch_sizes
# Note that this might lose the last batch (samples less than batch_size)
# but the subsequent epochs perform random shuffling prior to starting the
# training for that epoch
while real_data_store.hasnext() and batch_counter < limit:
# perform batch spoofing via gradient accumulation
dis_loss, gen_loss = 0, 0 # losses initialized to zeros
# =================================================================
# discriminator iterations:
# =================================================================
for spoofing_iter in range(spoofing_factor - 1):
# =============================================================
# Discriminator spoofing pass
# =============================================================
# sample images and latents for discriminator pass
images, gan_input = self._get_images_and_latents(
real_data_store, normalize_latents)
# accumulate gradients in the discriminator:
dis_loss += self.optimize_discriminator(
dis_optim, gan_input,
images, loss_fn,
accumulate=True,
zero_grad=(spoofing_iter == 0),
num_accumulations=spoofing_factor)
# =============================================================
# Discriminator update pass
# (note values for accumulate and zero_grad)
# =============================================================
# sample images and latents for discriminator pass
images, gan_input = self._get_images_and_latents(
real_data_store, normalize_latents)
# accumulate final gradients in the discriminator and make a step:
dis_loss += self.optimize_discriminator(
dis_optim, gan_input,
images, loss_fn,
accumulate=False, # perform update
zero_grad=spoofing_factor == 1, # make gradient buffers zero if spoofing_factor is 1
num_accumulations=spoofing_factor)
# =================================================================
# =================================================================
# generator iterations:
# =================================================================
for spoofing_iter in range(spoofing_factor - 1):
# =============================================================
# Generator spoofing pass
# =============================================================
# re-sample images and latents for generator pass
images, gan_input = self._get_images_and_latents(
real_data_store, normalize_latents)
# accumulate gradients in the generator
gen_loss += self.optimize_generator(
gen_optim, gan_input,
images, loss_fn,
accumulate=True,
zero_grad=(spoofing_iter == 0),
num_accumulations=spoofing_factor)
# =============================================================
# Generator update pass
# (note values for accumulate and zero_grad)
# =============================================================
# sample images and latents for generator pass
images, gan_input = self._get_images_and_latents(
real_data_store, normalize_latents)
# accumulate final gradients in the generator and make a step:
gen_loss += self.optimize_generator(
gen_optim, gan_input,
images, loss_fn,
accumulate=False, # perform update
zero_grad=spoofing_factor == 1, # make gradient buffers zero if spoofing_factor is 1
num_accumulations=spoofing_factor)
# =================================================================
# increment the global_step and the batch_counter:
global_step += 1
batch_counter += 1
# provide a loss feedback
if batch_counter % int(limit / feedback_factor) == 0 or \
batch_counter == 1:
elapsed = time.time() - global_time
elapsed = str(datetime.timedelta(seconds=elapsed))
print("Elapsed [%s] batch: %d d_loss: %f g_loss: %f"
% (elapsed, batch_counter, dis_loss, gen_loss))
# add summary of the losses
sum_writer.add_scalar("dis_loss", dis_loss, global_step)
sum_writer.add_scalar("gen_loss", gen_loss, global_step)
# also write the losses to the log file:
if log_dir is not None:
log_file = os.path.join(log_dir, "loss.log")
os.makedirs(os.path.dirname(log_file), exist_ok=True)
with open(log_file, "a") as log:
log.write(str(global_step) + "\t" + str(dis_loss) +
"\t" + str(gen_loss) + "\n")
# create a grid of samples and save it
gen_img_files = [os.path.join(sample_dir, res, "gen_" +
str(epoch) + "_" +
str(batch_counter) + ".png")
for res in reses]
# Make sure all the required directories exist
# otherwise make them
os.makedirs(sample_dir, exist_ok=True)
for gen_img_file in gen_img_files:
os.makedirs(os.path.dirname(gen_img_file), exist_ok=True)
# following zero_grads are required to allow pytorch
# adjust buffers properly on the GPU.
# This causes lesser GPU memory consumption
dis_optim.zero_grad()
gen_optim.zero_grad()
with th.no_grad(): # this makes the training faster.
self.create_grid(
self.gen(fixed_input) if not self.use_ema
else self.gen_shadow(fixed_input),
gen_img_files,
sum_writer,
reses,
global_step)
# calculate the time required for the epoch
stop_time = timeit.default_timer()
print("Time taken for epoch: %.3f secs" % (stop_time - start_time))
if epoch % checkpoint_factor == 0 or epoch == 1 or epoch == num_epochs:
os.makedirs(save_dir, exist_ok=True)
gen_save_file = os.path.join(save_dir, "GAN_GEN_" + str(epoch) + ".pth")
dis_save_file = os.path.join(save_dir, "GAN_DIS_" + str(epoch) + ".pth")
gen_optim_save_file = os.path.join(save_dir,
"GAN_GEN_OPTIM_" + str(epoch) + ".pth")
dis_optim_save_file = os.path.join(save_dir,
"GAN_DIS_OPTIM_" + str(epoch) + ".pth")
th.save(self.gen.state_dict(), gen_save_file)
th.save(self.dis.state_dict(), dis_save_file)
th.save(gen_optim.state_dict(), gen_optim_save_file)
th.save(dis_optim.state_dict(), dis_optim_save_file)
if self.use_ema:
gen_shadow_save_file = os.path.join(save_dir, "GAN_GEN_SHADOW_"
+ str(epoch) + ".pth")
th.save(self.gen_shadow.state_dict(), gen_shadow_save_file)
print("log_fid_values:", log_fid_values)
if log_fid_values: # perform the following fid calculations during training
# if the boolean is set to true
# ==================================================================
# Perform the FID calculation during training for estimating
# the quality of the training
# ==================================================================
# setup the directory for generating the images
if os.path.isdir(fid_temp_folder):
rmtree(fid_temp_folder)
os.makedirs(fid_temp_folder, exist_ok=True)
# generate the images:
print("generating images for fid calculation ...")
pbar = tqdm(total=num_fid_images)
generated_images = 0
while generated_images < num_fid_images:
b_size = min(fid_batch_size, num_fid_images - generated_images)
points = th.randn(b_size, self.latent_size).to(self.device)
if normalize_latents:
points = (points / points.norm(dim=1, keepdim=True)) \
* (self.latent_size ** 0.5)
imgs = self.gen(points)[-1].detach()
for i in range(len(imgs)):
imgs[i] = Generator.adjust_dynamic_range(imgs[i])
pbar.update(b_size)
for img in imgs:
imsave(os.path.join(fid_temp_folder,
str(generated_images) + ".jpg"),
img.permute(1, 2, 0).cpu())
generated_images += 1
pbar.close()
# compute the fid now:
fid = fid_score.calculate_fid_given_paths(
(fid_real_stats, fid_temp_folder),
fid_batch_size,
True if self.device == th.device("cuda") else False,
2048 # using he default value
)
# print the compute fid value:
print("FID at epoch %d: %.6f" % (epoch, fid))
# log the fid value in tensorboard:
sum_writer.add_scalar("FID", fid, epoch)
# note that for fid value, the global step is the epoch number.
# it is not the global step. This makes the fid graph more informative
# ==================================================================
print("Training completed ...")
# return the generator and discriminator back to eval mode
self.gen.eval()
self.dis.eval()
|
/// Parse a set of ASCII strings each terminated by NUL, or, an empty (zero byte) value.
pub fn parse_ascii_nul_string_values<D: Default, F: Fn(&mut D, &[u8]) -> Result<(), &'static str>>(bytes: &[u8], add_ascii_string: &F) -> Result<D, &'static str>
{
let length = bytes.len();
let mut collection = D::default();
if unlikely!(length == 0)
{
return Ok(collection)
}
const AsciiNul: u8 = b'\0';
let final_byte_index = length - 1;
let final_byte = bytes.get_unchecked_value_safe(final_byte_index);
if unlikely!(final_byte != AsciiNul)
{
return Err("bytes must end with an Ascii NUL");
}
for ascii_string in (&bytes[0 .. final_byte_index]).split_bytes(AsciiNul)
{
add_ascii_string(&mut collection, ascii_string)?
}
Ok(collection)
} |
def remove_item(self, item):
if item not in self._dock_items:
return
item._manager = None
for container in self.dock_containers():
if container.dockItem() is item:
if not container.isWindow():
container.unplug()
container.hide()
self._free_container(container)
break |
Fake news, real impact: Changing practices among public broadcasting organizations in the Asia-Pacific region This study is an attempt to understand fake news from the unique perspective of public broadcasting organizations in the Asia-Pacific region; the focus is on how and how well they have adapted to the growing incidence of various forms of disinformation, and how they see their role in educating diverse audiences about the phenomenon. This research provides public broadcasting organizations and (news) media practitioners with up-to-date, evidence-based insights on how to combat fake news and disinformation effectively. In collaboration with the Asia-Pacific Institute for Broadcasting Development, a census survey was conducted among their members, resulting in 57 completed questionnaires, representing 24 public broadcasting organizations in eighteen countries. A major finding is that public broadcasting organizations have to get used to revisiting, revising and refining what it is they do (practices), how it is being done (systems, operating procedures, technologies, financial resources) and by whom (human but also IT resources). The analysis also points up the need for collaboration so that scarce resources can be utilized more efficiently. |
/**
* Attempt to delete a variable from the workspace. If a {@link VariableCallback} is set
* {@link VariableCallback#onDeleteVariable} will be called to check if deletion is
* allowed.
*
* @param variable The variable to delete.
* @return True if the variable existed and was deleted, false otherwise.
*/
public boolean requestDeleteVariable(final String variable) {
final boolean[] resultSuccess = {false};
groupAndFireEvents(new Runnable() {
@Override
public void run() {
resultSuccess[0] = deleteVariableImpl(variable, false);
}
});
return resultSuccess[0];
} |
Impact of Electromobility to the Power Distribution System For many years, the global automotive industry has been systematically redirecting the development of conventional vehicle propulsion systems to new ones. Electric vehicles (EV) represents a new concept of urban mobility and is one of the most efficient and environmentally friendly individual transport modes, especially if electricity is generated from renewables (RES). The greater use of EVs in transport requires investment in the development and construction of infrastructure - the electricity grid and charging stations, as well as the development of regulatory and market policies. One of the most important areas of technology research and development is an advanced power grid, which ultimately boils down to integrating RES and EVs into the power system to maintain its reliability and stability. Due to all of the above, it is necessary to consider the impact of EV charging stations on the electricity distribution network in the light of different scenarios of electromobility. |
/*
* Copyright 2015-2020 the original author or authors.
*
* Licensed under the Apache License, Version 2.0 (the "License");
* you may not use this file except in compliance with the License.
* You may obtain a copy of the License at
*
* http://www.apache.org/licenses/LICENSE-2.0
*
* Unless required by applicable law or agreed to in writing, software
* distributed under the License is distributed on an "AS IS" BASIS,
* WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
* See the License for the specific language governing permissions and
* limitations under the License.
*/
package org.appng.tomcat.session.mongo;
import org.apache.catalina.core.StandardContext;
import org.apache.catalina.core.StandardEngine;
import org.apache.catalina.core.StandardHost;
import org.apache.catalina.core.StandardService;
import org.apache.catalina.loader.WebappLoader;
import org.apache.catalina.session.StandardSession;
import org.junit.Assert;
import org.junit.Test;
/**
* Integration test for {@link MongoStore}
*/
public class MongoStoreIT {
@Test
public void test() throws Exception {
StandardContext context = new StandardContext();
context.setName("foo");
WebappLoader loader = new WebappLoader() {
@Override
public ClassLoader getClassLoader() {
return WebappLoader.class.getClassLoader();
}
};
context.setLoader(loader);
StandardHost host = new StandardHost();
StandardEngine engine = new StandardEngine();
engine.setService(new StandardService());
host.setParent(engine);
context.setParent(host);
loader.setContext(context);
MongoPersistentManager manager = new MongoPersistentManager();
manager.setContext(context);
MongoStore store = new MongoStore();
store.setManager(manager);
store.setHosts("localhost:27017");
store.setDbName("mongo_session_test");
store.setCollectionName("mongo_session_test");
manager.setStore(store);
store.start();
StandardSession session = new StandardSession(manager);
session.setId("4711");
session.setNew(true);
session.setValid(true);
session.setCreationTime(System.currentTimeMillis());
session.setAttribute("foo", "test");
store.save(session);
StandardSession loaded = store.load(session.getId());
Assert.assertEquals(session.getAttribute("foo"), loaded.getAttribute("foo"));
Assert.assertEquals(1, store.getSize());
Assert.assertArrayEquals(new String[] { "4711" }, store.keys());
}
}
|
Statistical Evaluation of Conventional and Portable Instrumentations for Cr(VI) Analysis on Chemistry Laboratory Waste Water The development of portable instrumentation for heavy metals analysis was increased rapidly. However, the quality of data from portable methods has so far been questioned when compared to conventional instrumentation. In this research, a comparative study of conventional and portable instrumentations for Cr(VI) analysis on liquid waste water of Chemistry Laboratory at Universitas Gadjah Mada (UGM) was conducted. This research started with validation and statistical evaluation of instrumentation methods which are UV-Visible spectrophotometer, portable spectrophotometer (PiCOEXPLORER) and Inductively Coupled Plasma Atomic Emission Spectrometer (ICP-AES). The validation methods consist of determination of linearity, sensitivity, limit of detection and limit of quantification. The results showed that the validation methods of ICP-AES were better than PiCOEXPLORER and UV-Vis spectrophotometer. Based on t-test, it was obtained that the result of Cr(VI) analyses with the comparison of UV-Vis and PiCOEXPLORER, ICP-AES and PiCOEXPLORER, and UV-Vis and ICP-AES; there were no significant difference (tcount< ttable). The ANOVA test (F test) results showed that the Fcount value is less than Ftable so that the results of Cr(VI) analysis in the standard solution and liquid waste samples measured by three instrumentations. Thus, it was concluded that portable instrumentations was good and gives the same quality as conventional instrumentations (UV-Vis and ICP AES). |
The present invention relates generally to semiconductor memory designs, and, more particularly, to a memory word-line driver design.
The core of a semiconductor memory comprises at least one two-dimensional memory cell array, where information is stored. Traditionally, word-lines select rows, which activate cells, and bit-lines select columns, which access, i.e., read or write, the cells. When a word-line and a bit line are activated, a particular memory cell connected to them is selected.
To activate a word-line, its voltage is normally set to a high voltage, which is equal to a positive supply voltage in a CMOS circuitry. Setting word-line to a low voltage, which is a voltage complimentary to the positive supply voltage, de-activates the word-line. While the low voltage is customarily set to ground, or 0 V, the value for the high voltage can be different for various semiconductor manufacturing technologies. For instance, in a deep-sub-micron technology, a high voltage can be 1.2 V or even lower, while in a sub-micron technology the high voltage can be 2.5 V. But for a given memory chip and a given technology, the high voltage is normally designed to a fixed value, and this is particularly true for a complimentary-metal-oxide-semiconductor (CMOS) memory circuitry.
Since there are multiple memory cells connected to a single word-line, and word-line itself can be very long depending on the memory array size and technology used, so the word-line can be quite a load for its corresponding decoder, then a driver is needed to drive the word-line. The word-line driver couples, on one end, to a word-line decoder output, and on the other end, to a word-line. When a memory chip is in an active mode, i.e., the memory chip is ready for being actively read or written, the word-line driver functions just as a regular driver, following the word-line decoder, and providing a current source to pull up the word-line to a high voltage when the word-line is selected, and pull down the word-line to a low voltage when the word-line is not selected. When the memory chip is in a standby mode, i.e., the memory can not be actively read or written, and the power consumption is maintained at a minimum just to retain the information stored in the memory cell arrays, then the word-line driver clamps the word-line voltage to low. Besides, it is desirable for the word-line driver to have lower stand-by power consumption.
What is needed is an improved word-line driver design with reduced leakage and reduced layout area. |
Primary health care personnel faced with cadaveric organ donation: a multicenter study in southeastern Spain Abstract: Introduction: Primary health care (PHC) is the first point of contact between the public and the health system and it is an important channel for the communication and promotion of organ donation and transplantation. The objective of this study was to analyze the attitude of PHC personnel toward donation and to determine the psychosocial variables affecting this attitude. |
High pressure gas flow, storage, and displacement in fractured rock-Experimental setup development and application. This paper presents the design, development, and application of a laboratory setup for the experimental investigations of gas flow and reactions in a fractured rock. The laboratory facility comprises (i) a high pressure manometric sorption apparatus, where equilibrium and kinetic phenomena of adsorption and desorption can be examined, (ii) a high pressure triaxial core flooding system where the chemical reactive transport properties or processes can be explored, and (iii) an ancillary system including pure and mixed gas supply and analysis units. Underground conditions, in terms of pore pressure, confining pressure, and temperature, can be replicated using the triaxial core flooding system developed for depths up to 2 km. Core flooding experiments can be conducted under a range of gas injection pressures up to 20 MPa and temperatures up to 338 K. Details of the design considerations and the specification for the critical measuring instruments are described. The newly developed laboratory facility has been applied to study the adsorption of N2, CH4, and CO2 relevant to applications in carbon sequestration in coal and enhanced coalbed methane recovery. Under a wide range of pressures, the flow of helium in a core sample was studied and the evolution of absolute permeability at different effective stress conditions has been investigated. A comprehensive set of high resolution data has been produced on anthracite coal samples from the South Wales coalfield, using the developed apparatus. The results of the applications provide improved insight into the high pressure flow and reaction of various gas species in the coal samples from the South Wales coalfield. |
Introduction: The dynamics of ideas The suggestion that we now live in a knowledge society implies that knowledge is being organized differently. How compelling is this suggestion? And what implications might it have for the way one conceptualizes ideas, intellectuals and universities? Is there still a role for the intellectual who generates big ideas in todays knowledge society? What is the role of universities and the academics and disciplines within them? More generally, do we have any clear understanding of the relations between ideas, institutions and their social context? The contributors to this special issue on The Dynamics of Ideas engage with these questions in diverse, but engagingly complementary, ways. In his article Thomas Osborne contends that a mutation in the predominant type of ideas, and corresponding modes of intellectual work, may be occurring in ostensibly knowledge-driven societies. This involves a shift from oracular to vehicular ideas, where the relevant model of the intellectual is that of ideas mediation or brokerage. Rejecting epochal models of shift between intellectual functions, Osborne does, nonetheless, suggest that there may be an elective affinity between the intellectual as mediator and broader features immanent in contemporary society. Adopting Osbornes conceptual framework, Gregor McLennan explores the possibilities of understanding the Third Way as a vehicular idea. He suggests that the Third Way appears to confirm the vehicularity notion: more popular than many other academic political ideas, it splashes on to the discursive scene, is taken up by diverse audiences and users, and progressively weakens in specific content and urgency. This claim raises a prior question about the nature of ideas themselves: are they transcendental entities with intrinsic properties? Do they have their own trajectories? Or are they constituted through the processes of their articulation? These are the questions raised by Steve Woolgar, who suggests that, while these different perspectives represent evaluative discourses that provide particular resources in distinct argumentative contexts, we should abandon the romantic |
May 7, 2014, 1:29 p.m.
Let’s say you’re in an industry that’s facing the prospect of technological disruption. What could you learn from the news business, which has — maybe you noticed — had a rough time of late?
That was a question posed to Raju Narisetti, the senior vice president for strategy at News Corp, and Margaret Sullivan, The New York Times public editor, in an interview with Reuters TV at the recent International Journalism Festival in bucolic Perugia, Italy.
Hopefully, they won’t make the same mistake the news industry has made, which is to wait until you’re pushed to the wall before you start to innovate. I think the ability to innovate in advance of changes is important for these industries. The other thing is that as an industry, newspapers were never able to attract good business talent. We attracted the best journalism talent. And I think that’s been a big shortcoming as we’ve needed to adapt to business models, and hopefully some of the other industries are learning from that and are gathering their talent in all aspects of their business.
Watch the full interview above; Narisetti and Sullivan touch on a few other topics, including paywalls, increased segmentation in the media, and more.
Lichterman, Joseph. "A lesson from the disruption of the news business? Don’t wait until your backs are against the wall to innovate." Nieman Journalism Lab. Nieman Foundation for Journalism at Harvard, 7 May. 2014. Web. 20 Apr. 2019.
Lichterman, Joseph. "A lesson from the disruption of the news business? Don’t wait until your backs are against the wall to innovate." Nieman Journalism Lab. Last modified May 7, 2014. Accessed April 20, 2019. https://www.niemanlab.org/2014/05/a-lesson-from-the-disruption-of-the-news-business-dont-wait-until-your-backs-are-against-the-wall-to-innovate/. |
Gag Cytotoxic T Lymphocyte Escape Mutations Can Increase Sensitivity of HIV-1 to Human TRIM5, Linking Intrinsic and Acquired Immunity ABSTRACT Although laboratory-adapted HIV-1 strains are largely resistant to the human restriction factor TRIM5 (hTRIM5), we have recently shown that some viruses carrying capsid (CA) sequences from clinical isolates can be more sensitive to this restriction factor. In this study we evaluated the contribution to this phenotype of CA mutations known to be associated with escape from cytotoxic T lymphocyte (CTL) responses. Recombinant viruses carrying HIV-1 CA sequences from NL4-3 and three different clinical isolates were prepared, along with variants in which mutations associated with CTL resistance were modified by site-directed mutagenesis, and the infectivities of these viruses in target cells expressing hTRIM5 and cells in which TRIM5 activity had been inhibited by overexpression of TRIM5 were compared. For both hTRIM5-sensitive viruses studied, CTL-associated mutations were found to be responsible for this phenotype. Both CTL resistance mutations occurring within HLA-restricted CA epitopes and compensatory mutations occurring outside CTL epitopes influenced hTRIM5 sensitivity, and mutations associated with CTL resistance selected in prior hosts can contribute to this effect. The impact of CTL resistance mutations on hTRIM5 sensitivity was context dependent, because mutations shown to be responsible for the TRIM5-sensitive phenotype in viruses from one patient could have little or no impact on this parameter when introduced into another virus. No fixed relationship between changes in hTRIM5 sensitivity and infectivity was discernible in our studies. Taken together, these findings suggest that CTL mutations may influence HIV-1 replication by modifying both viral infectivity and sensitivity to TRIM5. |
Problem: You've used quantitative risk analysis to quantify the likelihood, and impact, of a climate change disaster. What's the next step?
Tip: Now it's time to model your response to climate change — create a data set with a range of potential adaptation options or responses, which can be analyzed for cost-efficiency and likelihood of effectiveness. Adaptation measures can have high capital and ongoing maintenance costs, so businesses and organizations need to demonstrate value for money and a return on their investment. Thus, optimization techniques are crucial — they allow you to determine the timing and scale of mitigation measures, ensuring that any commitments to capital expenditures will have a likely maximum return on investment. |
WASHINGTON—Both candidates in the 2012 presidential election ran on a corporate tax cut.
Five years later, a unified Republican government made that a reality, and a bipartisan idea has turned into a deeply partisan brawl.
President Donald Trump signed the 21% corporate tax rate into law Friday and Democrats are talking about which pieces of the bill they will keep and which they will toss aside should they assume power.
If there are problems with the law, Republicans are unlikely to have willing partners in Democrats to fix it, in the same way Republicans have stood against making many repairs to the Affordable Care Act.
The political backdrop to the tax overhaul is a warning sign. The new tax system encoded by Republicans is fragile. There are many challenges on the horizon that are likely to warrant changing the law—expiring provisions, tax competition from other countries, the risk of revenue shortfalls, the potential for flaws in the law itself that might be gamed by households or businesses.
Advertisement
As has been the case with the ACA, addressing them won’t be easy.
“It’s like training for a marathon. And you finally run a marathon,” said Jon Traub, a former Republican staff director at the House Ways and Means Committee who is now a managing principal at Deloitte LLP. “You can’t go back the next day and run another marathon.”
WSJ's Richard Rubin takes us to a weird, wacky Santa's workshop to explain who's getting Christmas presents and who's getting coal with the GOP tax bill. Photo/Illustration: Adam Falk/The Wall Street Journal
The new law was designed, at least partly, with that idea in mind. Rep. Kevin Brady (R., Texas), one of its chief authors, talked frequently about the rate cut to 21% from 35% as a “leapfrog,” a way to get ahead of moves by other countries, in recognition that the U.S. system can move slowly.
But less than two hours after the bill was formally enrolled by congressional leaders on Thursday, Mr. Brady said the U.S. isn’t necessarily done jumping.
Advertisement
“For our foreign competitors, I want to make it really clear, we are not going to fall behind like we have as a country. We’re not waiting another 31 years,” he said, referring to other countries’ rate cuts since the previous U.S. reduction in 1986. “So we’ll do what it takes to compete and give our workers and businesses a fighting chance around the world.”
Mr. Brady has also talked about technical corrections, especially in international tax rules, as companies get familiar with the new system. He mentioned changes to retirement savings and education tax policies as possible areas for bipartisanship.
His Senate counterpart, Orrin Hatch (R., Utah), on Wednesday proposed extending a long list of tax provisions that expired at the end of 2016, an issue that lawmakers had hoped to address in the broader bill they just finished.
Republicans could try again to use the legislative process that let them pass this year’s tax law with a simple majority, but that is far from certain at this point. Anything else will require 60 votes in the Senate, and that means Democrats. The minority, sensing electoral opportunity in the bill’s low poll numbers, isn’t ready to engage.
Advertisement
“This bill is so bad that it’s very hard to fix in small little tweaks,” said Senate Minority Leader Chuck Schumer (D., N.Y.). “We’ll see what they propose.”
Democrats seem unlikely to roll back the increased child tax credit, larger standard deduction or lower tax rates in low- and middle-income brackets. But other provisions are in play. Rep. Richard Neal (D., Mass.) would become Ways and Means chairman if Democrats retake the House in 2018. The first two items on his list were pushing the top individual tax rate back from 37% to 39.6% and reversing the doubled exemption for the estate tax.
Asked whether Democrats would help plug revenue holes if the new GOP law spurs more tax avoidance than expected, Mr. Neal demurred.
“We have to wait and see after there is an acknowledgment on their side that it didn’t work,” he said.
Advertisement
The current level of partisanship is a departure from many past tax laws. Tax cuts of 1964 proposed by Democrat John F. Kennedy and signed into law by Democrat Lyndon Johnson won the support of 21 Republicans in the Senate. Republican Ronald Reagan’s 1986 tax overhaul won the support of 33 Democrats in the Senate and his 1981 cuts were backed by 26 Senate Democrats. George W. Bush’s 2001 tax cut won support from 12 Democrats in the Senate. President Bill Clinton’s 1993 tax increase, however, passed with no Republican support in the House or Senate.
The path to the latest corporate tax cut is instructive in just how hard it might be to replicate anything like it.
For a long time, Republicans and Democrats agreed that U.S. corporate tax rates were too high, encouraged tax avoidance and made the U.S. uncompetitive globally.
In 2005, President George W. Bush’s tax overhaul commission proposed setting the rate as low as 30%. Then Rep. Charles Rangel (D., N.Y.) offered a 30.5% rate in his 2007 “mother of all tax reforms.” Sen. John McCain (R., Ariz.) included a rate cut in his presidential campaign in 2008.
When Republicans took the House in 2010, then-Rep. Dave Camp (R., Mich.) took up the charge for a 25% rate, joined by a coalition of companies that sensed an opportunity.
“There was some point in the late 2000s where everybody started to realize that the United States was just behind the rest of the world. We were sticking out like a sore thumb,” said Rachelle Bernstein, tax counsel at the National Retail Federation. “It’s a dozen years that we’ve sort of been in the wilderness of getting to this point.”
President Barack Obama joined in 2012, seeking a 28% rate. His rival that year, Mitt Romney, pushed for 25%. At times, a deal looked possible, whether between Mr. Obama and then-House Speaker John Boehner or between Mr. Camp and then-Sen. Max Baucus (D., Mont.).
By the time Mr. Camp released his plan in February 2014, GOP leadership showed little inclination to work with Mr. Obama on a compromise and gave Mr. Camp little support.
In the end, the only way to move forward was to look inside the Republican Party itself.
Beyond the worsened Washington gridlock, the corporate rate cut took a long time because Republicans had trouble deciding what to package with it. Removing business tax breaks would erode support. Leaving out pass-through businesses—partnerships, S corporations, sole proprietorships—would fracture the GOP. Ignoring individual taxes would make the idea politically treacherous.
Something had to give and that was the stated Republican commitment to reducing budget deficits. When they agreed to allow adding $1.5 trillion to deficits over the next decade, the door opened to drive the corporate rate all the way down to 21% without some of the other painful trade-offs they had considered.
There were still last-minute efforts this month to get a bipartisan deal, after Mr. Trump signaled that the 20% corporate tax rate he had insisted on might go to 22% in the final version.
Sen. Bob Corker (R., Tenn.) said he told Mr. Trump and his advisers that an agreement with Democrats was possible to create a more durable bill, a legacy for the president that wouldn’t be ripe for immediate reversal. Mr. Corker said he pushed for the president to meet with 10 to 12 Senate Democrats to see what was possible. It didn’t happen because a process was in train that Republican leaders didn’t want to derail.
“Their concern was: They had the votes,” Mr. Corker said in an interview this week. “If they brought in Democrats and began adjusting anything, it could cause the Republican side of this to fall apart.”
—Siobhan Hughes contributed to this article.
Write to Richard Rubin at richard.rubin@wsj.com |
OneThird Tubular Plate Remains a Clinically Good Option in DanisWeber Type B Distal Fibular Fracture Fixation Objective To compare the clinical outcomes of locking plate (LP) and nonlocking onethird tubular plate (TP) fixation, and to provide guidance on plate selection for DanisWeber type B distal fibular fracture treatment. Methods In total, 83 patients who underwent plate fixation for DanisWeber type B distal fibular fractures between March 2013 and July 2018 were retrospectively reviewed: 41 (49.0%) received LPs and 42 (51.0%) received TPs. Patients' demographic data, followup durations, the proportion of comminuted fractures, and ankle range of motion were investigated. The American Orthopaedic Foot and Ankle Society (AOFAS) anklehindfoot scale, Karlsson scale, Foot and Ankle Ability Measure (FAAM), and Lower Extremity Functional Scale (LEFS) scores were assessed. The radiographic union progression and implant removal time were evaluated, along with postoperative complications. Data from the LP and TP groups were compared statistically. Results The mean patient ages were 53.3 ±17.5years (range, 1680years) and 47.6 ±17.0years (range, 1468years) in the LP and TP groups, respectively (P > 0.05). The gender distribution did not differ significantly between groups (P > 0.05). Other demographic data also did not differ significantly between groups (P > 0.05). The mean followup durations were 16.8 ±7.7 months (range, 13.019.0 months) in the LP group and 16.1 ±6.2 months (range, 12.020.0 months) in the TP group (P > 0.05). Comminuted fractures were observed in 18 of 41 (43.9%) patients with LP and 10 of 42 (23.8%) patients with TP (P > 0.05). Forward bending ankle dorsiflexion was possible at the final followup in 82.9% and 85.7% of LP and TP patients, respectively (P > 0.05). The AOFAS anklehindfoot scale, Karlsson scale, FAAM, and LEFS scores did not differ significantly between groups at the final followup (P > 0.05). The prefracture and final postoperative scores on these four instruments did not differ significantly in the LP or TP group (P > 0.05). The mean times to radiographic union progression were 13.5 ±7.1weeks and 15.1 ±10.2weeks in the LP and TP groups, respectively (P > 0.05). The mean times to implant removal surgery reaffirming solid union were 15.6 ±5.5 months and 14.8 ±4.9 months in the LP and TP groups, respectively (P > 0.05). Hardware irritation was detected in five patients in the LP group (12.2%) and three in the TP group (7.1%) (P > 0.05). One patient in the LP group and two in the TP group developed superficial wound infections, which resolved without further surgical intervention. Conclusion Conventional TP remains a good option for the fixation of DanisWeber type B distal fibular fractures, regardless of the biomechanical properties. Introduction A nkle fracture is one of the most common injuries requiring surgical treatment 1. Of the various types, distal fibular fractures are the most prevalent (57.6%) 2. Danis-Weber type B fractures are the most common form of distal fibular fractures; they are mainly caused by a supinationexternal rotation injury and are characterized by an oblique fracture line 3,4. Of the various treatment options for distal fibular fractures, open reduction and internal fixation is the most preferred option to restore fibular length and rotation 13,14. Open reduction and internal fixation surgical methods for distal fibular fractures include lateral plating, posterior antiglide plating, and the lag screw-only technique 10,15,16. Conventional plating techniques depend on compression between the plate and cortical bone for stability. Fixation of a distal fibular fragment relies on unicortical cancellous bone to avoid intra-articular penetration of the distal screws. The most difficult aspect is achieving adequate distal fixation in cases with a short segment or comminution of the distal fibula or osteoporotic bone. Various kinds of implants have been developed to compensate for this difficulty, including a locking plate (LP) system, tibial pro-fibular screw 17, Kirschner wire cage, and intramedullary fibular nails 18,19. Among them, the LP system has become popular in recent decades, and various types of LPs have been introduced into the orthopaedic implant market (Fig. 1). A LP may be advantageous for osteoporotic distal fibular fractures because its fixed-angle construct precludes the need for contact or compression between the plate and bone, such that there is less reliance on bone mineral density (BMD) to stabilize the fracture 20. Several cadaver studies have indicated that LP is more biomechanically advantageous than the conventional non-locking one-third tubular plate (TP) for fixing distal fibular fractures 20,21. According to Kim et al. 21, the LP construct with two distal unicortical screws is mechanically equivalent to a standard plate with three distal screws in cadavers. Following a biomechanical report, the clinical use of LP fixation for distal fibular fractures has been increasing, along with its use for other anatomical sites, such as the proximal humerus, distal radius, distal femur, and proximal tibia, despite the cost of an LP implant being higher than that of a TP. Eckel et al. 22 conducted a comparative biomechanical study to identify the plate types associated with improved distal fibular fracture outcomes. However, the authors concluded that their experimental results were insufficient to identify a superior plate type. Some studies have reported clinical outcomes of distal fibular fractures treated with and without LPs. However, the results were variable and there was heterogeneity in fracture classification and implant type. There is still a lack of clinical evidence in the literature regarding the most suitable plate type for distal fibular fractures, particularly Danis-Weber type B fractures. Therefore, the purposes of this study are as follows: (i) to compare clinical outcomes between LP Fig. 1 The various types of locking plate (LP) for fixing distal fibular fractures have been introduced into the orthopaedic implant market. The LP system has been developed in recent decades. The reason for the popularity of the LP system is based on previous studies demonstrating the biomechanical superiority of the LP design, especially in osteoporotic or comminuted periarticular fractures. and TP fixation for Danis-Weber type B distal fibular fractures; (ii) to provide guidance on plate selection for Danis-Weber type B distal fibular fracture fixation. Inclusion and Exclusion Criteria The inclusion criteria were as follows: (i) a Danis-Weber type B fibular fracture, with >2 mm displacement, shortening, or rotation; (ii) patients who underwent surgical fixation using either LP or TP system which was randomly selected by one surgeon (foot and ankle orthopaedic specialist); (iii) patients with clinical records as well as perioperative radiographs; (iv) a minimum 12-months follow-up. The exclusion criteria were as follows: (i) open fractures; (ii) pilon fractures; (iii) concomitant ankle dislocation; (iv) bilateral ankle fractures; and (v) severe syndesmotic or medial injury that required additional transsyndesmotic screw fixation or deltoid ligament repair. Patients This was a retrospective cohort study comparing clinical outcomes between LP and TP fixation for Danis-Weber type B distal fibular fractures. Patients were enrolled consecutively from March 2013 to July 2018. All patients provided written informed consent before surgery, as well as consent for publication. All data were extracted from electronic medical records. This study was approved by the institutional review board of the corresponding author's university hospital (No. VC21RISI0182). Finally, a total of 83 out of 132 patients treated with either LP or TP fixation were included. Patients were divided into two groups according to the type of implant used: LP group (n = 41): 2.7-mm/3.5-mm Locking Distal Fibular Plate (Arthrex, Inc., Naples, FL, USA). Surgical Procedures All procedures were performed by the same surgeon who is the corresponding author of this article. The specific process is described in the following steps. Anesthesia and Position The patients were placed in the lateral recumbent position on a radiolucent operating table under general or spinal anesthesia. A pressure pneumatic tourniquet was placed at the proximal thigh, and inflated to 280 mmHg after exsanguination. Approach, Exposure, Reduction, and Fixation Techniques All operations were performed through a lateral approach. After manual traction for fracture reduction, we compressed the fracture site using a reduction clamp, and fixed two Kirschner wires temporarily across the reduced oblique fracture, instead of using an interfragmentary lag screw; then, the LP or TP was placed on the lateral aspect of the distal fibula after bending. In the LP group, three proximal 3.5-mm bicortical non-locking screws and at least three distal 2.7-mm unicortical locking screws were used ( Fig. 2A). In the TP group, three proximal 3.5-mm bicortical non-locking A B 56-year-old male patient was treated with one-third tubular plate (TP) fixation. Three 3.5-mm bicortical non-locking screws were used for proximal holes, while two 4.0-mm unicortical cancellous screws were used for distal holes. Two technical tips are recommended to provide a good bone purchase of the distal fragment during TP fixation. First, the plate should be pre-bent anatomically along the lateral aspect of distal fibula. Second, the most distal unicortical cancellous screw fixation should be slightly angled to prevent pulling out (white arrow). screws and two distal 4.0-mm unicortical cancellous screws were used. To provide good bone purchase of distal fragment in the TP group, we used "two technical tips" from AO surgery reference 28. First, the TP should be pre-bent anatomically along the lateral aspect of distal fibula. Second, the most distal unicortical cancellous screw fixation should be slightly angled to prevent pulling out (Fig. 2B). We removed the Kirschner wires after adequate plate fixation. Fracture reduction, stable fixation, and ankle joint congruency were confirmed during the operation using C-arm imaging. Incision Closure A drainage tube was placed and the wound was closed in the same manner in both groups; the subcutaneous fascia was closed with synthetic absorbable interrupted sutures, and the skin was closed with a horizontal mattress pattern using non-absorbable monofilament sutures. Postoperative Management The ankle joint was maintained in a neutral position postoperatively using a short-leg splint. From the second day after surgery, the patients started active ankle range of motion (ROM) exercises, then were allowed to walk with crutches without weight-bearing for 2-4 weeks in a walking boot or splint. After 4-6 weeks, the patients were allowed to walk with partial weight-bearing without a walking boot or splint and start passive ankle ROM exercises at the orthopaedic rehabilitation unit. After 8 weeks, the patients were allowed to walk with full weight-bearing, and after 10-12 weeks they were allowed to run slowly on a flat floor. All sports activities, with progressively heavier loads, were allowed 4-6 months postoperatively after confirming bony union on plain radiographs. Data Collection and Assessments Baseline Data Demographic data of patients including age, gender, body mass index (BMI), diabetes mellitus (DM), and smoking history were collected. The duration of follow-up was investigated in both groups. The proportion of comminuted fracture patterns in the two groups was investigated during surgery. Range of Motion (ROM) The ankle ROM at the final follow-up visit was also determined. The ROM of each ankle was measured as follows. Dorsiflexion was assessed in a standing position in terms of the patient's ability to perform forward bending (Fig. 3), and plantar flexion was assessed in a sitting position with the foot hanging freely over the end of an examination bed. The degree of plantar flexion was measured as far as the patient could reach with the help of the examiner, not with the help of gravity. The measuring device was a standard (30-cm) goniometer, and two measurements were made per patient by the corresponding author. We obtained the mean value of those results. If the final ROM was below the normal side ankle ROM or at least text reference ROM (approximately 10 -15 of dorsiflexion to 45 -50 of plantar flexion) 29, the patient was considered to have limited ROM. Four Instruments to Assess Functional Ankle Joint Outcomes We also obtained the American Orthopaedic Foot and Ankle Society (AOFAS) ankle-hindfoot scale, Karlsson scale, Foot and Ankle Ability Measure (FAAM), and Lower Extremity Functional Scale (LEFS) scores. These four instruments were completed before surgery and at the final follow-up visit, and the scores were compared. American Orthopaedic Foot and Ankle Society (AOFAS) Ankle-Hindfoot Scale. The AOFAS ankle-hindfoot scale 30 is among the most commonly used instruments for measuring outcome of treatment in patients who sustained a complex ankle or hindfoot injury. This scale grades ankle, subtalar, talonavicular, and calcaneocuboid joint levels. It consists of a patient-reported and a physician-reported part. The AOFAS scale is a 100-point score system with three categories: 40 points to pain, 50 points to function, and 10 points to alignment. A total score of 100 points is possible in a patient with no pain, full range of sagittal and hindfoot motion, no ankle or hindfoot instability, good alignment, ability to walk more than six blocks, ability to ambulate on any walking surface, no limp, no limitation of daily or recreational activities, and no assistive devices needed for ambulation. Karlsson Scale. The Karlsson scale 31 consists of eight items. These items are a subjective evaluation of functional stability, pain, swelling, and stiffness. Tasks of daily life, such as running and stair climbing, working ability, as well as sports activity and leisure activities are evaluated. The maximum score is 100 points and a higher score represents a higher level of ankle function. Subjective or functional stability is given 25 points. Absence of pain is given 20 points. Swelling is given 10 points. Stiffness is given 5 points. Activities such as running and stair climbing are given a score of 10 points each. Overall assessment of work, sports activities, and leisure time activities is given a score of 15 points in combination. The last item, the need for wearing an ankle support, is given a score of 5 points. The Karlsson scale scoring was originally developed to assess ankle joint function after ligament reconstruction for chronic lateral instability. However, it can also be used to evaluate ankle joint function before and after treatment for injury. The Karlsson scale does not only relate to symptoms, but also to the functional level of the individual. Lower Extremity Functional Scale (LEFS). The LEFS 33 is easy to administer and score and is applicable to a wide range of disability levels and conditions and all lower-extremity sites. It is more interpretable with respect to understanding error associated measurement and for determining minimally clinically important score changes and is a sufficient measure of reliability, validity, and sensitivity to change, at a level that is commensurate with utilization at an individual patient level. The LEFS can be used by clinicians as a measure of patients' initial function, ongoing progress, and outcome as well as to set functional goals. The LEFS consists of 20 items, with scores ranging from 0 (extreme difficulty/unable to perform activity) to 4 (no difficulty). The total score can be obtained by summing the scores of the individual items. The maximum score of 80 indicates no functional limitations and the minimum score of 0 indicates extreme limitations. Radiographic Assessment for Time to Fracture Union The progression of radiographic union was evaluated. All pre-and postoperative radiographs of the ankles were assessed by two independent observers, and the interval to radiographic union was noted. If no consensus was reached, the senior author made the final decision. The progression of union was defined as evidence of bridging bone in three of four cortices in the anteroposterior and lateral planes on simple radiographs 23,34. Intraoperative Assessment for Fracture Union Completion The implant removal time was investigated. We reaffirmed the "solid" union intraoperatively at the time of implant removal. Postoperative Complications Complications related mainly to the surgical site were investigated. Statistical Analysis For continuous variables, the student's t-test was used for the comparison of mean age, BMI, and the four instrument scores at the final follow-up between the two groups. The paired t-test was used to compare pre-fracture and postoperative (final visit) scores on the four instruments detailed above in the same patient. The Wilcoxon rank sum test with continuity correction was used for the comparison of mean follow-up durations and mean times to fracture union between the two groups. For categorical variables, the chi-square test was used for comparisons of gender distribution, DM and smoking history, the proportion of comminuted fracture and the ankle ROM at the final follow-up visit between the two groups. The Fisher's exact test was used for the comparison of postoperative complications. Statistical analyses were performed using R software (ver. 3.5.3; R Development Core Team, R Foundation for Statistical Computing, Vienna, Austria). The mean and standard deviation were computed for normally distributed continuous variables, and the median and interquartile range (25th-75th percentile) for non-normally distributed continuous data. A P-value <0.05 was considered statistically significant. General Results As summarized in Table 1 Range of Motion (ROM) The proportions of patients who achieved full ROM at the final follow-up visit were 82.9% (34 of 41 patients) in the LP group and 85.7% (36 of 42 patients) in the TP group (P = 0.962) ( Table 1). Four Instruments to Assess Functional Ankle Joint Outcomes The AOFAS ankle-hindfoot score was not significantly different between the LP (83.9 AE 14. Times to Fracture Union The mean time to radiographic union progression was 13.5 AE 7.1 weeks in the LP group and 15.1 AE 10.2 weeks in the TP group (P = 0.717). The mean time to implant removal surgery reaffirming intraoperative solid union was 15.6 AE 5.5 months in the LP group and 14.8 AE 4.9 months in the TP group (P = 0.728). BMI, body mass index; LP, locking plate; TP, non-locking one-third tubular plate.; The P-value was calculated using a Student's t-test; not statistically significant between the two groups.; The P-value was calculated using a chi-square test; not statistically significant between the two groups. Postoperative Complications Hardware irritation was detected in five of 41 patients (12.2%) in the LP group and three of 42 patients (7.1%) in the TP group (P = 0.483). One of 41 patients (2.4%) in the LP group and two of 42 patients (4.8%) in the TP group (P = 1.000) developed a superficial wound infection, which was controlled by intravenous antibiotics and meticulous wound management without further intervention. No patient in either group required removal of an implant before complete fracture union because of hardware irritation or infection. No other complications were encountered, such as nonunion, malunion, or neurovascular injury. Main Findings We retrospectively assessed the medical records of patients treated using plate fixation for Danis-Weber B distal fibular fractures, which is the most common clinical pattern. We hypothesized that LP fixation would provide better clinical outcomes due to its enhanced biomechanical stability relative to the TP. However, we did not observe any significant difference between the LP and TP groups in demographic or clinical data, including proportion of comminuted fracture, ankle ROM, scores on the AOFAS ankle-hindfoot scale, Karlsson scale, FAAM, and LEFS, time to fracture union, or postoperative complications. Comparison of Clinical Outcomes between LP and TP Fixation Plate selection for distal fibular fractures remains controversial among clinicians; some favor the newer LP and others favor a conventional TP or other non-LP. Although many biomechanical and clinical comparative studies have been reported, there remains a paucity of guidelines for clinicians to follow regarding plate selection to fix a distal fibular fracture. The LPs used for periarticular fractures have evolved steadily over the past decades. However, despite several cadaver studies demonstrating the biomechanical superiority of the LP design for cases of osteoporosis or comminuted fractures 20,21,35, the clinical benefits of the LP have not been established. The in vivo (patient) environment is more suitable than the in vitro (cadaver) environment for assessing clinical outcomes. In our study, there were no significant differences between the LP and TP groups in terms of comparing clinical outcomes. A biomechanical study by Eckel et al. 22 using the Weber B fracture model to compare LPs and non-LPs suggested that in vivo loading magnitudes are unknown and will vary depending on patient compliance, weight, and activity. Nguyentat et al. 36 compared the biomechanical properties of LP and TP fixation for the AO Foundation/ Orthopaedic Trauma Association (AO/OTA) 44-B3.3 distal fibular fractures in cases of trimalleolar ankle injury. They suggested that a distal fibular LP does not provide a mechanical advantage for trimalleolar ankle injuries in individuals with normal BMD and an absence of fracture comminution. Zahn et al. 20 and Kim et al. 21 reported that the biomechanical attributes of LP are independent of BMD, such that it is a more functional treatment for osteoporotic and comminuted distal fibular fractures. However, the BMD was measured at different sites among cadaver studies, including the tibia, fibula, distal metaphysis, diaphysis, and even the calcaneus or proximal femur. This makes direct comparison of the results of these studies difficult. Also, it is difficult to model comminuted fractures that mimic those in vivo in cadaver specimens. Davis et al. 37 questioned a previous cadaver study by Kim et al. 21 comparing TPs with LPs. They concluded that the LP does not outperform its non-locking counterparts, and that TPs provide adequate fixation for AO/OTA 44-B2.1 fractures. Several clinical studies have compared surgical outcomes between LPs and non-LPs. Huang et al. 25 reviewed 147 patients divided into TP, locking compression metaphyseal plate, and locking anatomical distal fibular plate groups. A Danis-Weber subgroup was also analyzed and compared. No significant differences were observed among the three plates in the Weber C subgroup in terms of ankle ROM, radiographic reduction accuracy, complication rate, Olerud and Molander score, or the AOFAS scale score. In contrast, the LP groups showed better functional scores than the TP group in the Weber A and B subgroups. Tsukada et al. 38 conducted a randomized controlled trial comparing the clinical effectiveness of locking and non-locking neutralization plates for AO/OTA 44-B fractures. They observed no differences in terms of the radiographic fracture union rate, 36-Item Short Form Survey score, or wound complication rate between the plate groups. These results were consistent with our study, in which no clinical advantage of the LP over the TP was seen. We investigated the proportion of comminuted fracture, ankle ROM recovery, and four instruments of functional scores as clinical outcomes, and observed no significant group difference. The radiographic fracture union outcomes and postoperative complications were also not different between the groups, similar to other clinical comparative studies 25,27,38,39. Patient-reported outcome measurements (PROMs) are considered to be more important than physician-reported outcome measurements. In clinical studies of orthopaedic surgery, many types of validated PROMs are increasingly being used to better understand the effectiveness of surgical treatment. In the current study, we used not only the physician-reported AOFAS scale, but also three PROMs, namely the Karlsson scale, FAAM, and LEFS, to assess functional ankle joint outcomes in a comprehensive manner. Moreover, we compared pre-fracture and postoperative (at the final follow-up visit) scores on the four instruments. Although the pre-fracture scores on all four instruments were numerically higher, the differences were not significant in either group. Therefore, the functional performance of patients with Danis-Weber B distal fibular fractures can be expected to recover well regardless of whether LP or TP fixation is used. Technical Tips for TP Fixation to Obtain Comparable Outcomes to LP Fixation In our study, we did not use interfragmentary lag screws. A recent prospective randomized study demonstrated that interfragmentary lag screws are not essential for precontoured lateral LP fixation in non-comminuted supination-external rotation lateral malleolar fractures, provided that intraoperative compression with a reduction clamp is performed 40. The lag screw head and purchase impede proper plate placement. Hence, temporary fixation in our study involved two Kirschner wires instead of interfragmentary lag screws. In both groups, plate fixation was performed under compression of the fracture site using a reduction clamp in combination with temporary Kirschner wire fixation. Moreover, we did not follow the biomechanical report of Kim et al. 21, and used only two distal unicortical cancellous screws in the TP group. We obtained comparable surgical outcomes in the LP and TP groups, and found that TP construct with two distal unicortical screws was clinically acceptable. We attribute these findings to the successful placement of the reduction clamp and Kirschner wires, which function to compress the fracture and maintain this compression, along with our "two technical tips" for TP fixation described above. Other Considerations for Clinical Comparisons between LP and TP Although not assessed in our study, the cost-effectiveness of a LP should be considered. The cost may be high depending on the surgery time, complications, length of hospital stay, and insurance system (which differs among countries), but newer LPs are generally more expensive than conventional plates. Moss et al. 41 demonstrated that treating Weber B distal fibular fractures with a TP can be significantly less expensive than using a contoured LP, without any increase in the likelihood of complications, failure, or loss of reduction. Nguyentat et al. 36 reported that overuse of LPs can be detrimental to healthcare systems, by consuming resources that could have been better used elsewhere. The cost issue has also been emphasized in cases of fracture at other anatomical sites 42. Limitation of the Study This study had some limitations that should be considered. First, it used a non-randomized, retrospective design. Second, the bone density of the patients was not measured. Our cohort included young patients, and we are unable to routinely perform dual-energy X-ray absorptiometry scans for patients aged <65 years who use the national insurance system. Third, intra-and inter-observer reliability were not evaluated for the ankle ROM and radiographic union assessments. Further randomized controlled trials with larger sample sizes or meta-analyses are required to verify which plate is superior in various clinical conditions and patient populations. Conclusion In conclusion, the LP is not a prerequisite for fixing Danis-Weber type B distal fibular fractures, especially when the fracture pattern is not complicated. Surgeons can choose a TP at their discretion and still achieve a good clinical outcome. The conventional TP remains a good option for the fixation of Danis-Weber type B distal fibular fractures, regardless of the biomechanical properties. |
It's possible to find leads on Instagram, but most successful real estate users are focusing on its relationship-building capabilities.
Instagram Stories and Instagram Live are excellent tools for promoting open houses; consider how those "extras" fit into your branding.
When Facebook bought Instagram for almost $1 billion in 2012, some people raised their eyebrows at what seemed like a ridiculous amount of money for a social media network. But five years later, Facebook is laughing all the way to the bank. Most real estate agents (and brokerages) have embraced Facebook as an important part of a well-rounded marketing strategy, but a lot fewer have jumped on Instagram’s bus to date. That might be a mistake. |
This article is a highly speculative account of how our distant ancestors evolved the capacity for speech, together with the evolved capacities that would later be exapted to the behavioural functions of reading and writing.
I shall not pretend to be an infallible guide. Whenever you have a chance, take counsel with other travelers who have passed along the same route before. Compare their observations with mine and if this leads you to different conclusions, I shall certainly not be angry with you.
Did our human ancestors actually 'invent' writing, or did spoken language and writing co-evolve?
Consider that in reading we can generally read the intent of a writer rather than be baulked by the writer's errors. Evolution has provided us with a largely subconscious error-correction mechanism which operates for both speech and writing.
I suggest that the human invention of writing 'out of thin air' is an impossibility. When writing first began to be used it must have had a firm underlying evolutionary basis in language, or some component of language, in order to be used as a system of visible tokens for language. I suggest that there must have been components of speech linked or linkable to eye-hand coordination without which writing could not have been developed.
Humans and other primates have much in common. In particular, we are vocalisers and tool users. At some time in the past when hominids diverged from other primates there was only one feature common to all hominids and lacking in other primates: all hominids stood upright. This, I suggest, is a crucial factor in the evolution of our use of language.
It has become unremarkable to suggest that evolution of the upright stance freed the hands to use tools. Nevertheless, I suggest that, since various primates use tools without the benefit of an upright stance, the upright stance must have evolved from other causes and then permitted a more skillful use of tools, especially with the evolution of opposable thumbs.
I suggest that upright stance co-evolved with ever more refined vocalising abilities. An upright stance would appear to allow a greater degree of voluntary control over breathing, perhaps related to swimming. Such breathing control is a pre-requisite for human-style, precisely-controlled, smooth vocalisation. Other refinements to the 'organs of speech', such as the descended larynx, may have become possible only after the upright posture was fully, or nearly fully evolved.
Before the evolution of language, there had to be a role for language to fill, an evolutionary advantage to be gained. On the reasonable assumption that our remote ancestors were hunters, an advantage suggests itself: cooperation in the hunt. However, a species that coordinates the hunt by making noises is going to be less successful than a species that can hunt in silence.
It appears reasonable to suppose that, in a sufficiently open area, first body language and then overt gestures could have evolved as a means of coordinating the hunt. Co-evolution of gestural abilities and of intelligence seems a plausible evolutionary path. Hunting requires an ability to find, track, and hunt down prey. A species which can out-think the prey has an evolutionary advantage. A sufficiently evolved gestural language, together with the necessary mental skills, would eventually allow a hunting species to plan ahead for the hunt as a social group.
A sufficiently evolved gestural language would eventually permit planning ahead in general, and the more precise allocation of tasks. This would enhance the survivability of a group whilst concurrently contributing to the evolution of greater cohesion. I suggest that the evolution of language was in parallel with the evolution of a social grouping based on a greater division of labour than occurs in animal societies generally: the precursor of human society.
Once a facility evolved to interpret gestural signs as having ever more precise meaning, it would take little brain evolution to provide a capacity for the abstraction of signs. An ability to associate marks in the sand first with the creatures that made them, and then with gestures, would assist in planning a hunt.
I suggest that once the association of marks with gestures became possible, the groundwork was laid for an association of sounds with marks and gestures. I suggest that a gestural language, a language of marks, and a sonic language all co-evolved, with speech perhaps lagging behind. Later, in an environment where line-of-sight communication became difficult, spoken language would have had greater survival value. It could have rapidly overtaken gestural language in its utility and versatility, perhaps by exapting gestural brain functions to speech.
Through co-evolution, our distant ancestors could have evolved the capacity for speech, together with the capacities that would later be exapted to the functions of reading and writing. Our remote ancestors hunted using gestures during the hunt. This evolved to abstraction: gestures to plan the hunt, later accompanied by and even later replaceable by both marks in the sand and sounds. Later, the sounds and marks (sketches) came to be signs for objects, actions and finally, ideas.
It is highly unlikely that we may find the fossilised remains of marks deliberately made in the sand by an ancestor. However, it is not difficult to imagine an ancestor showing a child how to track an animal by making approximations of animal tracks and human footprints in the sand.
The 'drawing' skill would later be evidenced first in cave art, then in pottery and carvings, finally in recognisable visual abstractions of sounds, i.e. writing.
If you have enjoyed reading this, you may also enjoy New Hominid Discovered : Anoiapithecus brevirostris and other articles on language and linguistics in my blog: The Chatter Box. |
There’s a summary copy of Frank’s bailout bill up at TPM. It’s essentially a Wall Street giveaway plan, with only some fig leaves to try and pretend that it isn’t.
Why? Because the language about taking warrants in exchange for buying up toxic assets is only for direct purchases and not for reverse auction puchases, which will be the majority of the purchases. As Soros points out, in any reverse auction, the government will get stuck with the most toxic of toxic waste because of information asymetries. In exchange they should at least get stock, equal not to what they paid, but to the face of the crap they are buying.
There is quite a bit of language about helping mortgage holders, but it is almost all qualified with words like encourage and request, rather than require. Since the Treasury is bailing mortgage holders out, the idea that the Secretary must "encourage" and "request" is just BS. The correct response is to make help for mortgage holders a requirement of participating in the program at all. If financial institutions don’t like that they don’t need to participate. Good way to make sure that companies that don’t really need help don’t swill at the trough.
Unlike the Dodd bill, this is not a copy of the actual language of the bill, but a summary gloss. Without seeing the language we don’t know what’s actually in there. Dodd was straight up with us. Frank is hiding his legislative language. Why?
This isn’t to say there isn’t some good stuff in here, the best part of it is that it includes the "Helping Families Save Their Homes in Bankruptcy Act", which allows restructuring of mortgages so that bankrupt individuals can keep their houses. Not as good as helping people not go bankrupt in the first place, like is being done for banks, but still an excellent provision.
Nonetheless, overall, this is a stab in the back, an attempt to give public money to private concerns with almost zero upside for ordinary citizens. If Reid doesn’t back Dodd, hardcore, or if the House doesn’t revolt, this is probably the bill we’re going to wind up with.
(Update: Dave Johnson nails the scam pefectly.) |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.