id
string | submitter
string | authors
string | title
string | comments
string | journal-ref
string | doi
string | report-no
string | categories
string | license
string | abstract
string | versions
list | update_date
timestamp[s] | authors_parsed
sequence |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
2504.07552 | Federico Bertacco | Federico Bertacco, Martin Hairer | Uniqueness of supercritical Gaussian multiplicative chaos | 21 pages. v2: minor fixes | null | null | null | math.PR math-ph math.MP | http://creativecommons.org/licenses/by/4.0/ | We show that, for general convolution approximations to a large class of log-correlated Gaussian fields, the properly normalised supercritical Gaussian multiplicative chaos measures converge stably to a nontrivial limit. This limit depends on the choice of regularisation only through a multiplicative constant and can be characterised as an integrated atomic measure with a random intensity expressed in terms of the critical Gaussian multiplicative chaos. | [
{
"version": "v1",
"created": "Thu, 10 Apr 2025 08:29:12 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 16:58:40 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Bertacco",
"Federico",
""
],
[
"Hairer",
"Martin",
""
]
] |
2504.07679 | William Wolf | William J. Wolf and Carlos Garc\'ia-Garc\'ia and Theodore Anton and Pedro G. Ferreira | Assessing cosmological evidence for non-minimal coupling | v2. Added references and minor changes to the discussion | null | null | null | astro-ph.CO gr-qc hep-ph hep-th | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The recent observational evidence of deviations from the $\Lambda$-Cold Dark Matter ($\Lambda$CDM) model points towards the presence of evolving dark energy. The simplest possibility consists of a cosmological scalar field $\varphi$, dubbed quintessence, driving the accelerated expansion. We assess the evidence for the existence of such a scalar field. We find that, if the accelerated expansion is driven by quintessence, the data favour a potential energy $V(\varphi)$ that is concave, i.e., $m^2=d^2V/d\varphi^2<0$. Furthermore, and more significantly, the data strongly favour a scalar field that is non-minimally coupled to gravity (Bayes factor $\log(B) = 7.34 \pm 0.6$), leading to time variations in the gravitational constant on cosmological scales, and the existence of fifth forces on smaller scales. The fact that we do not observe such fifth forces implies that either new physics must come into play on non-cosmological scales or that quintessence is an unlikely explanation for the observed cosmic acceleration. | [
{
"version": "v1",
"created": "Thu, 10 Apr 2025 12:07:47 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 14:08:48 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Wolf",
"William J.",
""
],
[
"García-García",
"Carlos",
""
],
[
"Anton",
"Theodore",
""
],
[
"Ferreira",
"Pedro G.",
""
]
] |
2504.07732 | Qifan Huang | Qifan Huang, Li Zhou, Wang Fang, Mengyu Zhao, Mingsheng Ying | Efficient Formal Verification of Quantum Error Correcting Programs | 41 pages, 10 figures, 4 tables; v2: Extended version of the paper in PLDI 2025; Evaluated artifact at https://doi.org/10.5281/zenodo.15267327 | null | 10.1145/3729293 | null | cs.PL quant-ph | http://creativecommons.org/licenses/by/4.0/ | Quantum error correction (QEC) is fundamental for suppressing noise in quantum hardware and enabling fault-tolerant quantum computation. In this paper, we propose an efficient verification framework for QEC programs. We define an assertion logic and a program logic specifically crafted for QEC programs and establish a sound proof system. We then develop an efficient method for handling verification conditions (VCs) of QEC programs: for Pauli errors, the VCs are reduced to classical assertions that can be solved by SMT solvers, and for non-Pauli errors, we provide a heuristic algorithm. We formalize the proposed program logic in Coq proof assistant, making it a verified QEC verifier. Additionally, we implement an automated QEC verifier, Veri-QEC, for verifying various fault-tolerant scenarios. We demonstrate the efficiency and broad functionality of the framework by performing different verification tasks across various scenarios. Finally, we present a benchmark of 14 verified stabilizer codes. | [
{
"version": "v1",
"created": "Thu, 10 Apr 2025 13:28:49 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 02:20:09 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Huang",
"Qifan",
""
],
[
"Zhou",
"Li",
""
],
[
"Fang",
"Wang",
""
],
[
"Zhao",
"Mengyu",
""
],
[
"Ying",
"Mingsheng",
""
]
] |
2504.07749 | Erik Velldal | Vladislav Mikhailov, Tita Enstad, David Samuel, Hans Christian Farseth{\aa}s, Andrey Kutuzov, Erik Velldal, Lilja {\O}vrelid | NorEval: A Norwegian Language Understanding and Generation Evaluation Benchmark | Accepted for Findings of the Association for Computational Linguistics: ACL 2025 | null | null | null | cs.CL cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper introduces NorEval, a new and comprehensive evaluation suite for large-scale standardized benchmarking of Norwegian generative language models (LMs). NorEval consists of 24 high-quality human-created datasets -- of which five are created from scratch. In contrast to existing benchmarks for Norwegian, NorEval covers a broad spectrum of task categories targeting Norwegian language understanding and generation, establishes human baselines, and focuses on both of the official written standards of the Norwegian language: Bokm{\aa}l and Nynorsk. All our datasets and a collection of over 100 human-written prompts are integrated into LM Evaluation Harness, ensuring flexible and reproducible evaluation. We describe the NorEval design and present the results of benchmarking 19 open-source pre-trained and instruction-tuned LMs for Norwegian in various scenarios. Our benchmark, evaluation framework, and annotation materials are publicly available. | [
{
"version": "v1",
"created": "Thu, 10 Apr 2025 13:44:55 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 10:51:26 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Mikhailov",
"Vladislav",
""
],
[
"Enstad",
"Tita",
""
],
[
"Samuel",
"David",
""
],
[
"Farsethås",
"Hans Christian",
""
],
[
"Kutuzov",
"Andrey",
""
],
[
"Velldal",
"Erik",
""
],
[
"Øvrelid",
"Lilja",
""
]
] |
2504.08472 | Matteo Abbondati | Matteo Abbondati (LIRMM | ECO), Eleonora Guerrini (LIRMM | ECO), Romain Lebreton (LIRMM | ECO) | Simultaneous Rational Number Codes: Decoding Beyond Half the Minimum Distance with Multiplicities and Bad Primes | null | null | null | null | cs.IT cs.SC math.IT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we extend the work of (Abbondati et al., 2024) on decoding simultaneous rational number codes by addressing two important scenarios: multiplicities and the presence of bad primes (divisors of denominators). First, we generalize previous results to multiplicity rational codes by considering modular reductions with respect to prime power moduli. Then, using hybrid analysis techniques, we extend our approach to vectors of fractions that may present bad primes. Our contributions include: a decoding algorithm for simultaneous rational number reconstruction with multiplicities, a rigorous analysis of the algorithm's failure probability that generalizes several previous results, an extension to a hybrid model handling situations where not all errors can be assumed random, and a unified approach to handle bad primes within multiplicities. The theoretical results provide a comprehensive probabilistic analysis of reconstruction failure in these more complex scenarios, advancing the state of the art in error correction for rational number codes. | [
{
"version": "v1",
"created": "Fri, 11 Apr 2025 12:01:41 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 12:01:47 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Abbondati",
"Matteo",
"",
"LIRMM | ECO"
],
[
"Guerrini",
"Eleonora",
"",
"LIRMM | ECO"
],
[
"Lebreton",
"Romain",
"",
"LIRMM | ECO"
]
] |
2504.08625 | Maciej Matys | Maciej Matys, Atsushi Yamada, Yoichi Kamada, and Toshihiro Ohki | A simple method for detection and quantitative estimation of deep levels in a barrier layer of AlGaN/GaN HEMT structures by analysis of light induced threshold voltage shift | Updated version after reviewer's comments | null | null | null | cond-mat.mtrl-sci | http://creativecommons.org/licenses/by/4.0/ | The characterization of deep levels in AlGaN/GaN heterostructures is one of the most important problems in GaN high electron mobility transistors (HEMTs) technology. This work reports on a technique for determination of deep level concentration in AlGaN/GaN HEMT structures. The proposed method is relatively simple, and it is based on the detection of free holes created by optically induced transitions of electrons from the deep levels to the conduction band. The developed method can detect and provide quantitative estimation of deep level traps in a barrier layer of AlGaN/GaN HEMT structures. Furthermore, it provides a framework for analysis of light induced threshold voltage shift, which includes an important experimental criterion of determination whether the holes are generated or not in the AlGaN/GaN HEMT structures by sub-band gap illumination. The method was verified by applications it to a study of the deep levels in GaN HEMTs grown on various substrates, i.e. SiC and GaN. | [
{
"version": "v1",
"created": "Fri, 11 Apr 2025 15:26:39 GMT"
},
{
"version": "v2",
"created": "Mon, 14 Apr 2025 03:46:05 GMT"
},
{
"version": "v3",
"created": "Thu, 05 Jun 2025 06:37:28 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Matys",
"Maciej",
""
],
[
"Yamada",
"Atsushi",
""
],
[
"Kamada",
"Yoichi",
""
],
[
"Ohki",
"Toshihiro",
""
]
] |
2504.08648 | Jacques Magnaudet | Jacques Magnaudet, Hadrien Bruhier, Samuel Mer and Thomas Bonometti | Rational constitutive law for the viscous stress tensor in incompressible two-phase flows: Derivation and tests against a 3D benchmark experiment | 19 pages, 9 figures | null | null | null | physics.flu-dyn | http://creativecommons.org/licenses/by-nc-nd/4.0/ | We analyze the representation of viscous stresses in the one-fluid formulation of the two-phase Navier-Stokes equations, the model on which all computational approaches making use of a fixed mesh to discretize the flow field are grounded. Recognizing that the Navier-Stokes-like equations that are actually solved in these approaches imply a spatial filtering, we show by considering a specific two-dimensional flow configuration that the proper representation of the viscous stress tensor requires the introduction of two distinct viscosity coefficients, owing to the different behaviors of shear and normal stresses in control volumes straddling the interface. Making use of classical results of continuum mechanics for anisotropic fluids, we derive the general form of the constitutive law linking the viscous stress tensor of the two-phase medium to the filtered strain-rate tensor, and take advantage of the findings provided by the above two-dimensional configuration to close the determination of the fluid-dependent coefficients involved. Predictions of the resulting anisotropic model are then assessed and compared with those of available \textit{ad hoc} models against original experimental results obtained in a reference flow in which some parts of the interface are dominated by shear while others are mostly controlled by stretching. The selected configuration corresponds to a viscous buoyancy-driven exchange flow in a closed vertical pipe, generated by unstably superimposing two immiscible fluids with a large viscosity contrast and negligible interfacial tension and molecular diffusivity. \textcolor{black}{Using different levels of grid refinement}, we show that the anisotropic model is the only one capable of predicting correctly the evolution of the front of the ascending and descending fingers \textcolor{black}{at a reasonable computational cost. | [
{
"version": "v1",
"created": "Fri, 11 Apr 2025 15:48:02 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 17:39:18 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Magnaudet",
"Jacques",
""
],
[
"Bruhier",
"Hadrien",
""
],
[
"Mer",
"Samuel",
""
],
[
"Bonometti",
"Thomas",
""
]
] |
2504.09200 | Darius Saif | Darius Saif and Ashraf Matrawy | A Datagram Extension to DNS over QUIC: Proven Resource Conservation in the Internet of Things | null | null | null | null | cs.NI | http://creativecommons.org/licenses/by-sa/4.0/ | In this paper, we investigate the Domain Name System (DNS) over QUIC (DoQ) and propose a non-disruptive extension, which can greatly reduce DoQ's resource consumption. This extension can benefit all DNS clients - especially Internet of Things (IoT) devices. This is important because even resource-constrained IoT devices can generate dozens of DNS requests every hour. DNS is a crucial service that correlates IP addresses and domain names. It is traditionally sent as plain-text, favoring low-latency results over security and privacy. The repercussion of this can be eavesdropping and information leakage about IoT devices. To address these concerns, the newest and most promising solution is DoQ. QUIC offers features similar to TCP and TLS while also supporting early data delivery and stream multiplexing. DoQ's specification requires that DNS exchanges occur over independent streams in a long-lived QUIC connection. Our hypothesis is that due to DNS's typically high transaction volume, managing QUIC streams may be overly resource intensive for IoT devices. Therefore, we have designed and implemented a data delivery mode for DoQ using QUIC datagrams, which we believe to be more preferable than stream-based delivery. To test our theory, we analyzed the memory, CPU, signaling, power, and time of each DoQ delivery mode in a setup generating real queries and network traffic. Our novel datagram-based delivery mode proved to be decisively more resource-friendly with little compromise in terms of functionality or performance. Furthermore, our paper is the first to investigate multiple queries over DoQ, to our knowledge. | [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 12:59:37 GMT"
},
{
"version": "v2",
"created": "Wed, 04 Jun 2025 18:35:11 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Saif",
"Darius",
""
],
[
"Matrawy",
"Ashraf",
""
]
] |
2504.09304 | Mikhail Volkov | Romain Gervalle and Mikhail S. Volkov | Black holes with electroweak hair -- the detailed derivation | 106 pages, many figures, many appendices. The type of the magnetic vortices contained in the black hole hair is clarified -- they are similar to the Nambu strings; Appendix F describing electroweak vortices is considerably expanded; references added | null | null | null | hep-th astro-ph.HE gr-qc hep-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present a very detailed derivation of solutions describing hairy black holes within the gravity-coupled Weinberg-Salam theory, which were previously reported in \href{https://doi.org/10.1103/PhysRevLett.133.171402}{Phys.Rev.Lett. 133 (2024) 171402}. These black holes support a strong magnetic field that polarizes the electroweak vacuum and creates a condensate of massive fields carrying superconducting currents along the black hole horizon. The currents, in turn, generate a ``corona'' of magnetic vortex segments attached to the horizon at both ends. The condensate and corona together constitute the black hole hair. The extremal solutions approach, in the far field, the magnetic Reissner-Nordstr\"om configuration, with a total mass that is {\it lower} than the total charge, $M<|Q|$, due to the negative Zeeman energy of the condensate. This makes the removal of the hair energetically unfavorable. The maximally hairy black holes exhibit masses comparable to terrestrial values, with approximately 11\% of their total mass stored in the hair. Given that these solutions arise within a well-tested theoretical framework, they are likely to have physical relevance. | [
{
"version": "v1",
"created": "Sat, 12 Apr 2025 18:32:16 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 16:24:20 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Gervalle",
"Romain",
""
],
[
"Volkov",
"Mikhail S.",
""
]
] |
2504.10505 | Nikolay G. Kuznetsov | Nikolay Kuznetsov and Oleg Motygin | On sloshing in containers with porous baffles | 10 pages, 5 figures | null | null | null | physics.flu-dyn | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Sloshing eigenvalues are studied for containers with porous baffles extending throughout the constant (possibly infinite) depth. The fluid transmission across the baffles is described by Darcy's law, and so the spectral problem is nonself-adjoint. In particular, much attention is paid to vertical cylinders with circular walls and radial baffles. Explicit solutions are obtained, whose crucial feature is that the corresponding pressure difference vanishes across the baffle. These real solutions coincide with those that describe sloshing in the presence of rigid baffles, thus demonstrating that at certain frequencies the damping efficiency of porous baffles is the same as that of the rigid ones having the same configuration. | [
{
"version": "v1",
"created": "Wed, 09 Apr 2025 13:31:50 GMT"
},
{
"version": "v2",
"created": "Wed, 04 Jun 2025 19:13:57 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Kuznetsov",
"Nikolay",
""
],
[
"Motygin",
"Oleg",
""
]
] |
2504.10746 | Xiulong Liu | Xiulong Liu and Anurag Kumar and Paul Calamia and Sebastia V. Amengual and Calvin Murdock and Ishwarya Ananthabhotla and Philip Robinson and Eli Shlizerman and Vamsi Krishna Ithapu and Ruohan Gao | Hearing Anywhere in Any Environment | CVPR 2025; Project Page: https://dragonliu1995.github.io/hearinganywhereinanyenvironment/ | null | null | null | cs.CV cs.AI cs.LG cs.SD eess.AS | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In mixed reality applications, a realistic acoustic experience in spatial environments is as crucial as the visual experience for achieving true immersion. Despite recent advances in neural approaches for Room Impulse Response (RIR) estimation, most existing methods are limited to the single environment on which they are trained, lacking the ability to generalize to new rooms with different geometries and surface materials. We aim to develop a unified model capable of reconstructing the spatial acoustic experience of any environment with minimum additional measurements. To this end, we present xRIR, a framework for cross-room RIR prediction. The core of our generalizable approach lies in combining a geometric feature extractor, which captures spatial context from panorama depth images, with a RIR encoder that extracts detailed acoustic features from only a few reference RIR samples. To evaluate our method, we introduce ACOUSTICROOMS, a new dataset featuring high-fidelity simulation of over 300,000 RIRs from 260 rooms. Experiments show that our method strongly outperforms a series of baselines. Furthermore, we successfully perform sim-to-real transfer by evaluating our model on four real-world environments, demonstrating the generalizability of our approach and the realism of our dataset. | [
{
"version": "v1",
"created": "Mon, 14 Apr 2025 22:37:52 GMT"
},
{
"version": "v2",
"created": "Wed, 04 Jun 2025 19:59:42 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Liu",
"Xiulong",
""
],
[
"Kumar",
"Anurag",
""
],
[
"Calamia",
"Paul",
""
],
[
"Amengual",
"Sebastia V.",
""
],
[
"Murdock",
"Calvin",
""
],
[
"Ananthabhotla",
"Ishwarya",
""
],
[
"Robinson",
"Philip",
""
],
[
"Shlizerman",
"Eli",
""
],
[
"Ithapu",
"Vamsi Krishna",
""
],
[
"Gao",
"Ruohan",
""
]
] |
2504.11199 | Min Jung Lee | Min Jung Lee, Dayoung Gong, Minsu Cho | Video Summarization with Large Language Models | Accepted to CVPR 2025 | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The exponential increase in video content poses significant challenges in terms of efficient navigation, search, and retrieval, thus requiring advanced video summarization techniques. Existing video summarization methods, which heavily rely on visual features and temporal dynamics, often fail to capture the semantics of video content, resulting in incomplete or incoherent summaries. To tackle the challenge, we propose a new video summarization framework that leverages the capabilities of recent Large Language Models (LLMs), expecting that the knowledge learned from massive data enables LLMs to evaluate video frames in a manner that better aligns with diverse semantics and human judgments, effectively addressing the inherent subjectivity in defining keyframes. Our method, dubbed LLM-based Video Summarization (LLMVS), translates video frames into a sequence of captions using a Muti-modal Large Language Model (M-LLM) and then assesses the importance of each frame using an LLM, based on the captions in its local context. These local importance scores are refined through a global attention mechanism in the entire context of video captions, ensuring that our summaries effectively reflect both the details and the overarching narrative. Our experimental results demonstrate the superiority of the proposed method over existing ones in standard benchmarks, highlighting the potential of LLMs in the processing of multimedia content. | [
{
"version": "v1",
"created": "Tue, 15 Apr 2025 13:56:14 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 05:18:38 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Lee",
"Min Jung",
""
],
[
"Gong",
"Dayoung",
""
],
[
"Cho",
"Minsu",
""
]
] |
2504.11632 | Joel Ivan Bobadilla | Joel Bobadilla, Marcelo J. Rozenberg and Alberto Camjayi | Magnetoresistivity in the Antiferromagnetic Hubbard Model | null | null | null | null | cond-mat.str-el | http://creativecommons.org/licenses/by/4.0/ | We investigate the magnetotransport properties of the half-filled antiferromagnetic (AF) one-band Hubbard model under an external magnetic field using the single-site dynamical mean-field approximation (DMFT). Particular attention is paid to the mechanisms driving the magnetoresistivity behavior. We analyze the dependence of magnetoresistivity on temperature and the strength of the applied magnetic field, providing insights into the interplay between magnetic fluctuations and transport properties in AF systems. | [
{
"version": "v1",
"created": "Tue, 15 Apr 2025 21:43:25 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 13:49:28 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Bobadilla",
"Joel",
""
],
[
"Rozenberg",
"Marcelo J.",
""
],
[
"Camjayi",
"Alberto",
""
]
] |
2504.12483 | Maxim Gritskov | Maxim Gritskov and Andrey Losev | Beta function without UV divergences | Appendix A.3 added. References on page 3 have been corrected. Corrected typos, added some clarifications | null | null | null | math-ph hep-th math.MP | http://creativecommons.org/licenses/by/4.0/ | In this paper, we construct the beta function in the functorial formulation of two-dimensional quantum field theories (FQFT). A key feature of this approach is the absence of ultraviolet divergences. We show that, nevertheless, in the FQFT perturbation theory, the local observables of deformed theories acquire logarithmic dimension, leading to a conformal anomaly. The beta function arises in the functorial approach as an infinitesimal transformation of the partition function under the variation of the metric's conformal factor, without ultraviolet divergences, UV cutoff, or the traditional renormalization procedure. | [
{
"version": "v1",
"created": "Wed, 16 Apr 2025 20:38:23 GMT"
},
{
"version": "v2",
"created": "Sun, 20 Apr 2025 15:11:16 GMT"
},
{
"version": "v3",
"created": "Wed, 04 Jun 2025 19:01:32 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Gritskov",
"Maxim",
""
],
[
"Losev",
"Andrey",
""
]
] |
2504.12880 | Lukas Rauch | Lukas Rauch, Ren\'e Heinrich, Ilyass Moummad, Alexis Joly, Bernhard Sick, Christoph Scholz | Can Masked Autoencoders Also Listen to Birds? | under review @TMLR | null | null | null | cs.LG cs.SD eess.AS | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Masked Autoencoders (MAEs) have shown competitive results in audio classification by learning rich semantic representations through an efficient self-supervised reconstruction task. However, general-purpose models fail to generalize well when applied directly to fine-grained audio domains. Specifically, bird-sound classification requires distinguishing subtle inter-species differences and managing high intra-species acoustic variability, thereby revealing the performance limitations of general-domain Audio-MAE models. This work demonstrates that bridging this domain gap requires more than domain-specific pretraining data; adapting the entire training pipeline is crucial. We systematically revisit and adapt the pretraining recipe, fine-tuning methods, and frozen feature utilization to bird sounds using BirdSet, a large-scale bioacoustic dataset comparable to AudioSet. Our resulting Bird-MAE achieves new state-of-the-art results in BirdSet's multi-label classification benchmark. Additionally, we introduce the parameter-efficient prototypical probing, enhancing the utility of frozen MAE representations and closely approaching fine-tuning performance in low-resource settings. Bird-MAE's prototypical probes outperform linear probing by up to 37%$_\text{p}$ in MAP and narrow the gap to fine-tuning to approximately 3.3%$_\text{p}$ on average across BirdSet downstream tasks. Bird-MAE also demonstrates robust few-shot capabilities with prototypical probing in our newly established few-shot benchmark on BirdSet, highlighting the potential of tailored self-supervised learning pipelines for fine-grained audio domains. | [
{
"version": "v1",
"created": "Thu, 17 Apr 2025 12:13:25 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 13:20:02 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Rauch",
"Lukas",
""
],
[
"Heinrich",
"René",
""
],
[
"Moummad",
"Ilyass",
""
],
[
"Joly",
"Alexis",
""
],
[
"Sick",
"Bernhard",
""
],
[
"Scholz",
"Christoph",
""
]
] |
2504.13671 | Mihai Tibar | Piotr Migus, Lauren\c{t}iu P\u{a}unescu, and Mihai Tib\u{a}r | On higher Lipschitz invariants | light revision | null | null | null | math.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We find new bi-Lipschitz invariants for functions of two complex variables. | [
{
"version": "v1",
"created": "Fri, 18 Apr 2025 13:06:08 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 07:52:58 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Migus",
"Piotr",
""
],
[
"Păunescu",
"Laurenţiu",
""
],
[
"Tibăr",
"Mihai",
""
]
] |
2504.14170 | Akash Vardhan | Akash Vardhan, Ram Avinery, Hosain Bagheri, Velin Kojohourav, Shengkai Li, Hridesh Kedia, Tianyu Wang, Daniel Soto, Kurt Wiesenfeld, Daniel I. Goldman | Collision Induced Binding and Transport of Shape Changing Robot Pairs | 7 pages, 6 figures, submitted to PRL | null | null | null | cs.RO nlin.AO | http://creativecommons.org/publicdomain/zero/1.0/ | We report in experiment and simulation the spontaneous formation of dynamically bound pairs of shape changing robots undergoing locally repulsive collisions. These physical `gliders' robustly emerge from an ensemble of individually undulating three-link two-motor robots and can remain bound for hundreds of undulations and travel for multiple robot dimensions. Gliders occur in two distinct binding symmetries and form over a wide range of angular oscillation extent. This parameter sets the maximal concavity which influences formation probability and translation characteristics. Analysis of dynamics in simulation reveals the mechanism of effective dynamical attraction -- a result of the emergent interplay of appropriately oriented and timed repulsive interactions. Tactile sensing stabilizes the short-lived conformation via concavity modulation. | [
{
"version": "v1",
"created": "Sat, 19 Apr 2025 04:17:27 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 14:59:48 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Vardhan",
"Akash",
""
],
[
"Avinery",
"Ram",
""
],
[
"Bagheri",
"Hosain",
""
],
[
"Kojohourav",
"Velin",
""
],
[
"Li",
"Shengkai",
""
],
[
"Kedia",
"Hridesh",
""
],
[
"Wang",
"Tianyu",
""
],
[
"Soto",
"Daniel",
""
],
[
"Wiesenfeld",
"Kurt",
""
],
[
"Goldman",
"Daniel I.",
""
]
] |
2504.14522 | Liudmila Zavolokina | Liudmila Zavolokina, Kilian Sprenkamp, Zoya Katashinskaya, Daniel Gordon Jones | Biased by Design: Leveraging Inherent AI Biases to Enhance Critical Thinking of News Readers | European Conference on Information Systems (ECIS) | null | null | null | cs.HC cs.AI | http://creativecommons.org/licenses/by-nc-nd/4.0/ | This paper explores the design of a propaganda detection tool using Large Language Models (LLMs). Acknowledging the inherent biases in AI models, especially in political contexts, we investigate how these biases might be leveraged to enhance critical thinking in news consumption. Countering the typical view of AI biases as detrimental, our research proposes strategies of user choice and personalization in response to a user's political stance, applying psychological concepts of confirmation bias and cognitive dissonance. We present findings from a qualitative user study, offering insights and design recommendations (bias awareness, personalization and choice, and gradual introduction of diverse perspectives) for AI tools in propaganda detection. | [
{
"version": "v1",
"created": "Sun, 20 Apr 2025 07:39:00 GMT"
},
{
"version": "v2",
"created": "Wed, 04 Jun 2025 17:15:23 GMT"
},
{
"version": "v3",
"created": "Thu, 05 Jun 2025 06:39:10 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Zavolokina",
"Liudmila",
""
],
[
"Sprenkamp",
"Kilian",
""
],
[
"Katashinskaya",
"Zoya",
""
],
[
"Jones",
"Daniel Gordon",
""
]
] |
2504.15046 | Shilin Zhang | Shilin Zhang, Zican Hu, Wenhao Wu, Xinyi Xie, Jianxiang Tang, Chunlin Chen, Daoyi Dong, Yu Cheng, Zhenhong Sun, Zhi Wang | Text-to-Decision Agent: Offline Meta-Reinforcement Learning from Natural Language Supervision | 18 pages, 8 figures | null | null | null | cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Offline meta-RL usually tackles generalization by inferring task beliefs from high-quality samples or warmup explorations. The restricted form limits their generality and usability since these supervision signals are expensive and even infeasible to acquire in advance for unseen tasks. Learning directly from the raw text about decision tasks is a promising alternative to leverage a much broader source of supervision. In the paper, we propose \textbf{T}ext-to-\textbf{D}ecision \textbf{A}gent (\textbf{T2DA}), a simple and scalable framework that supervises offline meta-RL with natural language. We first introduce a generalized world model to encode multi-task decision data into a dynamics-aware embedding space. Then, inspired by CLIP, we predict which textual description goes with which decision embedding, effectively bridging their semantic gap via contrastive language-decision pre-training and aligning the text embeddings to comprehend the environment dynamics. After training the text-conditioned generalist policy, the agent can directly realize zero-shot text-to-decision generation in response to language instructions. Comprehensive experiments on MuJoCo and Meta-World benchmarks show that T2DA facilitates high-capacity zero-shot generalization and outperforms various types of baselines. Our code is available at https://github.com/NJU-RL/T2DA. | [
{
"version": "v1",
"created": "Mon, 21 Apr 2025 12:00:20 GMT"
},
{
"version": "v2",
"created": "Tue, 22 Apr 2025 05:56:57 GMT"
},
{
"version": "v3",
"created": "Thu, 29 May 2025 06:17:36 GMT"
},
{
"version": "v4",
"created": "Thu, 05 Jun 2025 05:40:34 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Zhang",
"Shilin",
""
],
[
"Hu",
"Zican",
""
],
[
"Wu",
"Wenhao",
""
],
[
"Xie",
"Xinyi",
""
],
[
"Tang",
"Jianxiang",
""
],
[
"Chen",
"Chunlin",
""
],
[
"Dong",
"Daoyi",
""
],
[
"Cheng",
"Yu",
""
],
[
"Sun",
"Zhenhong",
""
],
[
"Wang",
"Zhi",
""
]
] |
2504.15265 | Yash Saxena | Yash Saxena, Sagnik Chatterjee, Tharrmashastha Sapv | Realization of maximally-entangling two-qutrit gates using the Cross-Resonance scheme | null | null | null | null | quant-ph | http://creativecommons.org/licenses/by-sa/4.0/ | In this letter, we present a comprehensive theoretical framework for realizing maximally entangling two-qutrit gates on fixed-frequency transmons that generalize the qubit-centric cross-resonance (CR) interaction beyond the 0-1 subspace, which we call as generalized cross-resonance (GCR) scheme. We use the GCR scheme to design parametric two-qutrit gates, namely, $U_{CR}^{01}$ and $U_{CR}^{12}$, that act on the $0{-}1$ and $1{-}2$ energy transitions of transmons, respectively. Our gates improve upon the existing works in two aspects. First, our gates \emph{directly} allow for entanglement on the $1{-}2$ levels instead of relying on the entangling ability on the $0{-}1$ levels to perform entanglement on the subspace spanned by $|1\rangle $ and $|2\rangle $, as done in previous works. Second, our gates are parametric in nature, enabling us to construct multiple entangling gates of interest, while the prior works that use cross-resonance for qutrits focus on individual gates. Further, we use Qiskit Dynamics to perform a simulation-based demonstration of two-qutrit generalized controlled-$X$ and controlled-$\sqrt{X}$ gates, namely, $U_{CX}^{01}$ and $U_{CSX}^{12}$, which are instances of proposed $U_{CR}$ gates. Numerical simulations in Qiskit Dynamics, calibrated to experimental parameters that mimic the actual backend, demonstrate gate fidelities of $99.73\pm0.01\%$ and $98.10\pm0.01\%$ for $U_{CX}^{01}$ and $U_{CSX}^{12}$ gates, respectively, obtained using Quantum Process Tomography. A two-qutrit Bell state $|\psi\rangle = \frac{1}{\sqrt{3}}(|00\rangle + |11\rangle + |22\rangle)$ is prepared with fidelity $98.33 \pm 0.01\%$ obtained using Quantum State Tomography with a $\sim 514$ ns pulse sequence. We note that the complete time taken for Bell state preparation in our setup is less than the gate time of cross-Kerr-based entangling gates. | [
{
"version": "v1",
"created": "Mon, 21 Apr 2025 17:47:16 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 17:43:01 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Saxena",
"Yash",
""
],
[
"Chatterjee",
"Sagnik",
""
],
[
"Sapv",
"Tharrmashastha",
""
]
] |
2504.15268 | J.D. Opdyke | JD Opdyke | Beating the Correlation Breakdown: Robust Inference, Flexible Scenarios, and Stress Testing for Financial Portfolios | null | null | null | null | q-fin.RM q-fin.PM q-fin.ST stat.AP | http://creativecommons.org/licenses/by/4.0/ | We live in a multivariate world, and effective modeling of financial portfolios, including their construction, allocation, forecasting, and risk analysis, simply is not possible without explicitly modeling the dependence structure of their assets. Dependence structure can drive portfolio results more than the combined effects of other parameters in investment and risk models, but the literature provides relatively little to define the finite-sample distributions of dependence measures in useable and useful ways under challenging, real-world financial data conditions. Yet this is exactly what is needed to make valid inferences about their estimates, and to use these inferences for essential purposes such as hypothesis testing, dynamic monitoring, realistic and granular scenario and reverse scenario analyses, and mitigating the effects of correlation breakdowns during market upheavals. This work develops a new and straightforward method, Nonparametric Angles-based Correlation (NAbC), for defining the finite-sample distributions of any dependence measure whose matrix of pairwise associations is positive definite (e.g. Pearsons, Kendalls, Spearmans, Tail Dependence Matrix, and others). The solution remains valid under marginal asset distributions characterized by notably different and varying degrees of serial correlation, non-stationarity, heavy-tailedness, and asymmetry. Importantly, NAbCs p-values and confidence intervals remain analytically consistent at both the matrix level and the pairwise cell level. Finally, NAbC maintains validity even when selected cells in the matrix are frozen for a given scenario or stress test, thus enabling flexible, granular, and realistic scenarios. NAbC stands alone in providing all of these capabilities simultaneously, and should prove to be a very useful means by which we can better understand and manage financial portfolios in our multivariate world. | [
{
"version": "v1",
"created": "Mon, 21 Apr 2025 17:52:36 GMT"
},
{
"version": "v2",
"created": "Sat, 03 May 2025 15:34:21 GMT"
},
{
"version": "v3",
"created": "Mon, 19 May 2025 18:57:59 GMT"
},
{
"version": "v4",
"created": "Thu, 05 Jun 2025 17:33:08 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Opdyke",
"JD",
""
]
] |
2504.16062 | Jiaxu Xing | Hardik Shah, Jiaxu Xing, Nico Messikommer, Boyang Sun, Marc Pollefeys, Davide Scaramuzza | ForesightNav: Learning Scene Imagination for Efficient Exploration | null | IEEE/CVF Conference on Computer Vision and Pattern Recognition Workshops (CVPRW), Nashville, 2025 | null | null | cs.RO cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Understanding how humans leverage prior knowledge to navigate unseen environments while making exploratory decisions is essential for developing autonomous robots with similar abilities. In this work, we propose ForesightNav, a novel exploration strategy inspired by human imagination and reasoning. Our approach equips robotic agents with the capability to predict contextual information, such as occupancy and semantic details, for unexplored regions. These predictions enable the robot to efficiently select meaningful long-term navigation goals, significantly enhancing exploration in unseen environments. We validate our imagination-based approach using the Structured3D dataset, demonstrating accurate occupancy prediction and superior performance in anticipating unseen scene geometry. Our experiments show that the imagination module improves exploration efficiency in unseen environments, achieving a 100% completion rate for PointNav and an SPL of 67% for ObjectNav on the Structured3D Validation split. These contributions demonstrate the power of imagination-driven reasoning for autonomous systems to enhance generalizable and efficient exploration. | [
{
"version": "v1",
"created": "Tue, 22 Apr 2025 17:38:38 GMT"
},
{
"version": "v2",
"created": "Tue, 29 Apr 2025 14:31:24 GMT"
},
{
"version": "v3",
"created": "Mon, 05 May 2025 06:53:10 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Shah",
"Hardik",
""
],
[
"Xing",
"Jiaxu",
""
],
[
"Messikommer",
"Nico",
""
],
[
"Sun",
"Boyang",
""
],
[
"Pollefeys",
"Marc",
""
],
[
"Scaramuzza",
"Davide",
""
]
] |
2504.17264 | Zhaolu Kang | Zhaolu Kang and Hongtian Cai and Xiangyang Ji and Jinzhe Li and Nanfei Gu | JurisCTC: Enhancing Legal Judgment Prediction via Cross-Domain Transfer and Contrastive Learning | Accepted in International Joint Conference on Neural Networks (IJCNN) 2025 | null | null | null | cs.CL cs.AI cs.CY | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In recent years, Unsupervised Domain Adaptation (UDA) has gained significant attention in the field of Natural Language Processing (NLP) owing to its ability to enhance model generalization across diverse domains. However, its application for knowledge transfer between distinct legal domains remains largely unexplored. To address the challenges posed by lengthy and complex legal texts and the limited availability of large-scale annotated datasets, we propose JurisCTC, a novel model designed to improve the accuracy of Legal Judgment Prediction (LJP) tasks. Unlike existing approaches, JurisCTC facilitates effective knowledge transfer across various legal domains and employs contrastive learning to distinguish samples from different domains. Specifically, for the LJP task, we enable knowledge transfer between civil and criminal law domains. Compared to other models and specific large language models (LLMs), JurisCTC demonstrates notable advancements, achieving peak accuracies of 76.59% and 78.83%, respectively. | [
{
"version": "v1",
"created": "Thu, 24 Apr 2025 05:48:57 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Kang",
"Zhaolu",
""
],
[
"Cai",
"Hongtian",
""
],
[
"Ji",
"Xiangyang",
""
],
[
"Li",
"Jinzhe",
""
],
[
"Gu",
"Nanfei",
""
]
] |
2504.17493 | Tareq Si Salem | Luca-Andrei Fechete, Mohamed Sana, Fadhel Ayed, Nicola Piovesan, Wenjie Li, Antonio De Domenico, Tareq Si Salem | Goal-Oriented Time-Series Forecasting: Foundation Framework Design | null | null | null | null | cs.LG cs.AI | http://creativecommons.org/licenses/by/4.0/ | Traditional time-series forecasting often focuses only on minimizing prediction errors, ignoring the specific requirements of real-world applications that employ them. This paper presents a new training methodology, which allows a forecasting model to dynamically adjust its focus based on the importance of forecast ranges specified by the end application. Unlike previous methods that fix these ranges beforehand, our training approach breaks down predictions over the entire signal range into smaller segments, which are then dynamically weighted and combined to produce accurate forecasts within a region of interest. We tested our method on standard datasets, including a new wireless communication dataset, and found that not only it improves prediction accuracy but also enhances the performance of end application employing the forecasting model. This research provides a basis for creating forecasting systems that better connect prediction and decision-making in various practical applications. | [
{
"version": "v1",
"created": "Thu, 24 Apr 2025 12:34:43 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 15:31:17 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Fechete",
"Luca-Andrei",
""
],
[
"Sana",
"Mohamed",
""
],
[
"Ayed",
"Fadhel",
""
],
[
"Piovesan",
"Nicola",
""
],
[
"Li",
"Wenjie",
""
],
[
"De Domenico",
"Antonio",
""
],
[
"Salem",
"Tareq Si",
""
]
] |
2504.17644 | Ronggang Shi | Qianlin Huang, Ronggang Shi | Bounded diagonal orbits in homogeneous spaces over function fields | null | null | null | null | math.DS | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper is about topological rigidity of diagonal group actions on the homogeneous $\SL_4\big(\F(\!(t^{-1})\!)\big)/\SL_4(\F[t])$ where $\F$ is a finite field of characteristic $3$. We show that there is a non-closed relatively compact orbit of the diagonal group. | [
{
"version": "v1",
"created": "Thu, 24 Apr 2025 15:14:10 GMT"
},
{
"version": "v2",
"created": "Wed, 21 May 2025 08:36:06 GMT"
},
{
"version": "v3",
"created": "Thu, 05 Jun 2025 03:39:09 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Huang",
"Qianlin",
""
],
[
"Shi",
"Ronggang",
""
]
] |
2504.17781 | Jyotirmoy Bhattacharya | Jyotirmoy Bhattacharya, Nibedita Padhi, Aditya Sharma and Sourav Singha | Thermal Product Formula for Shear Modes | 1+31 pages, 7 figures; V2: typos corrected, new references added | null | null | null | hep-th gr-qc | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We investigate the validity of the thermal product formula proposed in [arXiv:2304.12339], for the shear channel fluctuations of R-charged black branes in five-dimensional AdS where the shear mode is coupled with charge diffusion mode at non-zero momentum. When these modes are suitably decoupled, we are able to obtain an exact formula for the two point functions of the boundary current and energy-momentum tensor in terms of the quasinormal modes of this channel. This exact formula is a simple modification of the previous version of the product formula. We also obtain a similar formula for the case involving a boundary global R-symmetry anomaly, when we have a bulk Chern-Simons term which introduces additional couplings in the shear channel. Also based on insights from the quasinormal mode spectrum, we report on an instability as well as the presence of high momentum long-lived modes associated with large values of the anomaly coefficient. | [
{
"version": "v1",
"created": "Thu, 24 Apr 2025 17:58:16 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 16:12:58 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Bhattacharya",
"Jyotirmoy",
""
],
[
"Padhi",
"Nibedita",
""
],
[
"Sharma",
"Aditya",
""
],
[
"Singha",
"Sourav",
""
]
] |
2504.17934 | Chaoran Chen | Chaoran Chen, Zhiping Zhang, Ibrahim Khalilov, Bingcan Guo, Simret A Gebreegziabher, Yanfang Ye, Ziang Xiao, Yaxing Yao, Tianshi Li, Toby Jia-Jun Li | Toward a Human-Centered Evaluation Framework for Trustworthy LLM-Powered GUI Agents | null | null | null | null | cs.HC cs.CL cs.CR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The rise of Large Language Models (LLMs) has revolutionized Graphical User Interface (GUI) automation through LLM-powered GUI agents, yet their ability to process sensitive data with limited human oversight raises significant privacy and security risks. This position paper identifies three key risks of GUI agents and examines how they differ from traditional GUI automation and general autonomous agents. Despite these risks, existing evaluations focus primarily on performance, leaving privacy and security assessments largely unexplored. We review current evaluation metrics for both GUI and general LLM agents and outline five key challenges in integrating human evaluators for GUI agent assessments. To address these gaps, we advocate for a human-centered evaluation framework that incorporates risk assessments, enhances user awareness through in-context consent, and embeds privacy and security considerations into GUI agent design and evaluation. | [
{
"version": "v1",
"created": "Thu, 24 Apr 2025 20:51:20 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 12:40:53 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Chen",
"Chaoran",
""
],
[
"Zhang",
"Zhiping",
""
],
[
"Khalilov",
"Ibrahim",
""
],
[
"Guo",
"Bingcan",
""
],
[
"Gebreegziabher",
"Simret A",
""
],
[
"Ye",
"Yanfang",
""
],
[
"Xiao",
"Ziang",
""
],
[
"Yao",
"Yaxing",
""
],
[
"Li",
"Tianshi",
""
],
[
"Li",
"Toby Jia-Jun",
""
]
] |
2504.18019 | Min Yuan | Min Yuan, Yin Huang | A search for $B_{s0}^{*}$ and $B^{*}_{s1}$ through the $K^{-}p$ interaction | 8 pages, 5 figures, 1 table | null | null | null | hep-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The formation and properties of hadrons, particularly those involving heavy quarks, remain a central topic in particle physics due to the nonperturbative nature of Quantum Chromodynamics (QCD) at low energies. Heavy-quark symmetry (HQS) provides a valuable framework for understanding the behavior of heavy quarks in hadronic systems, offering insights into spin and flavor symmetries. However, discrepancies between HQS predictions and experimental observations -- particularly the absence of predicted $B^{(*)}\bar{K}$ molecular states, which should be the bottom counterparts of the observed $\bar{D}^{(*)}\bar{K}$ molecular states (corresponding to $D_{s0}^*(2317)^-$ and $D^*_{s1}(2460)^{-}$) -- challenge its universality. This study investigates the production of the $B_{s0}^{*}(5725)$ and $B_{s1}^{*}(5778)$ states, interpreted as molecular states containing $B\bar{K}$ and $B^{*}\bar{K}$ components, through the reactions $K^{-}p \to \Lambda_{b}^{0}B_{s0}^{*}$ and $K^{-}p \to \Lambda_{b}^{0}B_{s1}^{*}$. Using effective Lagrangians, we analyze the cross sections, incorporating $t$-channel $B^{(*)}$ meson exchanges and $K^{-}p$ initial-state interactions (ISI). Our results reveal significant ISI enhancements and energy-dependent production thresholds, with $B_{s0}^{*}(5725)$ exhibiting higher cross sections than $B_{s1}^{*}(5778)$. We suggest that the $B_{s0}^{*}(5725)$ state can be optimally searched for at beam energies around $12.90\,\mathrm{GeV}$ (or $12.93\,\mathrm{GeV}$), while the $B_{s1}^{*}(5778)$ state requires significantly higher energies for its production. These findings provide valuable guidance for experimental searches at facilities like CERN and J-PARC, offering a pathway to test HQS and explore the nature of exotic hadrons. The differences in production cross sections between molecular states and conventional quark-antiquark structures are also compared. | [
{
"version": "v1",
"created": "Fri, 25 Apr 2025 02:10:17 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 05:12:36 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Yuan",
"Min",
""
],
[
"Huang",
"Yin",
""
]
] |
2504.18026 | Emiliano Penaloza | Emiliano Penaloza, Tianyue H. Zhan, Laurent Charlin, Mateo Espinosa Zarlenga | Addressing Concept Mislabeling in Concept Bottleneck Models Through Preference Optimization | null | null | null | null | cs.LG cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Concept Bottleneck Models (CBMs) propose to enhance the trustworthiness of AI systems by constraining their decisions on a set of human-understandable concepts. However, CBMs typically assume that datasets contain accurate concept labels-an assumption often violated in practice, which we show can significantly degrade performance (by 25% in some cases). To address this, we introduce the Concept Preference Optimization (CPO) objective, a new loss function based on Direct Preference Optimization, which effectively mitigates the negative impact of concept mislabeling on CBM performance. We provide an analysis of key properties of the CPO objective, showing it directly optimizes for the concept's posterior distribution, and contrast it against Binary Cross Entropy (BCE), demonstrating that CPO is inherently less sensitive to concept noise. We empirically confirm our analysis by finding that CPO consistently outperforms BCE on three real-world datasets, both with and without added label noise. We make our code available on Github. | [
{
"version": "v1",
"created": "Fri, 25 Apr 2025 02:43:10 GMT"
},
{
"version": "v2",
"created": "Tue, 27 May 2025 20:26:10 GMT"
},
{
"version": "v3",
"created": "Thu, 05 Jun 2025 03:06:29 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Penaloza",
"Emiliano",
""
],
[
"Zhan",
"Tianyue H.",
""
],
[
"Charlin",
"Laurent",
""
],
[
"Zarlenga",
"Mateo Espinosa",
""
]
] |
2504.18053 | Xingyuan Bu | Jianyu Liu, Hangyu Guo, Ranjie Duan, Xingyuan Bu, Yancheng He, Shilong Li, Hui Huang, Jiaheng Liu, Yucheng Wang, Chenchen Jing, Xingwei Qu, Xiao Zhang, Yingshui Tan, Yanan Wu, Jihao Gu, Yangguang Li, Jianke Zhu | DREAM: Disentangling Risks to Enhance Safety Alignment in Multimodal Large Language Models | [NAACL 2025] The first four authors contribute equally, 23 pages, repo at https://github.com/Kizna1ver/DREAM | null | null | null | cs.CL cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Multimodal Large Language Models (MLLMs) pose unique safety challenges due to their integration of visual and textual data, thereby introducing new dimensions of potential attacks and complex risk combinations. In this paper, we begin with a detailed analysis aimed at disentangling risks through step-by-step reasoning within multimodal inputs. We find that systematic multimodal risk disentanglement substantially enhances the risk awareness of MLLMs. Via leveraging the strong discriminative abilities of multimodal risk disentanglement, we further introduce \textbf{DREAM} (\textit{\textbf{D}isentangling \textbf{R}isks to \textbf{E}nhance Safety \textbf{A}lignment in \textbf{M}LLMs}), a novel approach that enhances safety alignment in MLLMs through supervised fine-tuning and iterative Reinforcement Learning from AI Feedback (RLAIF). Experimental results show that DREAM significantly boosts safety during both inference and training phases without compromising performance on normal tasks (namely oversafety), achieving a 16.17\% improvement in the SIUO safe\&effective score compared to GPT-4V. The data and code are available at https://github.com/Kizna1ver/DREAM. | [
{
"version": "v1",
"created": "Fri, 25 Apr 2025 03:54:24 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 16:13:05 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Liu",
"Jianyu",
""
],
[
"Guo",
"Hangyu",
""
],
[
"Duan",
"Ranjie",
""
],
[
"Bu",
"Xingyuan",
""
],
[
"He",
"Yancheng",
""
],
[
"Li",
"Shilong",
""
],
[
"Huang",
"Hui",
""
],
[
"Liu",
"Jiaheng",
""
],
[
"Wang",
"Yucheng",
""
],
[
"Jing",
"Chenchen",
""
],
[
"Qu",
"Xingwei",
""
],
[
"Zhang",
"Xiao",
""
],
[
"Tan",
"Yingshui",
""
],
[
"Wu",
"Yanan",
""
],
[
"Gu",
"Jihao",
""
],
[
"Li",
"Yangguang",
""
],
[
"Zhu",
"Jianke",
""
]
] |
2504.18954 | Marco Mezzina Ir. | Marco Mezzina, Pieter De Backer, Tom Vercauteren, Matthew Blaschko, Alexandre Mottrie, Tinne Tuytelaars | Surgeons vs. Computer Vision: A comparative analysis on surgical phase recognition capabilities | null | null | 10.1007/s11548-025-03383-4 | null | eess.IV cs.AI cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Purpose: Automated Surgical Phase Recognition (SPR) uses Artificial Intelligence (AI) to segment the surgical workflow into its key events, functioning as a building block for efficient video review, surgical education as well as skill assessment. Previous research has focused on short and linear surgical procedures and has not explored if temporal context influences experts' ability to better classify surgical phases. This research addresses these gaps, focusing on Robot-Assisted Partial Nephrectomy (RAPN) as a highly non-linear procedure. Methods: Urologists of varying expertise were grouped and tasked to indicate the surgical phase for RAPN on both single frames and video snippets using a custom-made web platform. Participants reported their confidence levels and the visual landmarks used in their decision-making. AI architectures without and with temporal context as trained and benchmarked on the Cholec80 dataset were subsequently trained on this RAPN dataset. Results: Video snippets and presence of specific visual landmarks improved phase classification accuracy across all groups. Surgeons displayed high confidence in their classifications and outperformed novices, who struggled discriminating phases. The performance of the AI models is comparable to the surgeons in the survey, with improvements when temporal context was incorporated in both cases. Conclusion: SPR is an inherently complex task for expert surgeons and computer vision, where both perform equally well when given the same context. Performance increases when temporal information is provided. Surgical tools and organs form the key landmarks for human interpretation and are expected to shape the future of automated SPR. | [
{
"version": "v1",
"created": "Sat, 26 Apr 2025 15:37:22 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Mezzina",
"Marco",
""
],
[
"De Backer",
"Pieter",
""
],
[
"Vercauteren",
"Tom",
""
],
[
"Blaschko",
"Matthew",
""
],
[
"Mottrie",
"Alexandre",
""
],
[
"Tuytelaars",
"Tinne",
""
]
] |
2504.19044 | Di Wu | Di Wu, Yibin Lei, Christof Monz | Calibrating Translation Decoding with Quality Estimation on LLMs | null | null | null | null | cs.CL | http://creativecommons.org/licenses/by/4.0/ | Neural machine translation (NMT) systems typically employ maximum a posteriori (MAP) decoding to select the highest-scoring translation from the distribution mass. However, recent evidence highlights the inadequacy of MAP decoding, often resulting in low-quality or even pathological hypotheses -- the decoding objective is not aligned with real-world translation quality. This paper proposes calibrating hypothesis likelihoods with translation quality from a distribution view by directly optimizing their Pearson correlation -- thereby enhancing the effectiveness of translation decoding. With our method, translation on large language models (LLMs) improves substantially after limited training (2K instances per direction). This improvement is orthogonal to those achieved through supervised fine-tuning, leading to substantial gains across a broad range of metrics and human evaluations -- even when applied to top-performing translation-specialized LLMs fine-tuned on high-quality translation data, such as Tower, or when compared to recent preference optimization methods, like CPO. Moreover, the calibrated translation likelihood can directly serve as a strong proxy for translation quality, closely approximating or even surpassing some state-of-the-art translation quality estimation models, like CometKiwi. Lastly, our in-depth analysis demonstrates that calibration enhances the effectiveness of MAP decoding, thereby enabling greater efficiency in real-world deployment. The resulting state-of-the-art translation model, which covers 10 languages, along with the accompanying code and human evaluation data, has been released to the community: https://github.com/moore3930/calibrating-llm-mt. | [
{
"version": "v1",
"created": "Sat, 26 Apr 2025 22:38:47 GMT"
},
{
"version": "v2",
"created": "Sat, 10 May 2025 21:53:02 GMT"
},
{
"version": "v3",
"created": "Thu, 05 Jun 2025 00:33:42 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Wu",
"Di",
""
],
[
"Lei",
"Yibin",
""
],
[
"Monz",
"Christof",
""
]
] |
2504.19354 | Erkan Karabulut | Erkan Karabulut, Paul Groth, Victoria Degeler | Neurosymbolic Association Rule Mining from Tabular Data | null | null | null | null | cs.AI cs.LG | http://creativecommons.org/licenses/by/4.0/ | Association Rule Mining (ARM) is the task of mining patterns among data features in the form of logical rules, with applications across a myriad of domains. However, high-dimensional datasets often result in an excessive number of rules, increasing execution time and negatively impacting downstream task performance. Managing this rule explosion remains a central challenge in ARM research. To address this, we introduce Aerial+, a novel neurosymbolic ARM method. Aerial+ leverages an under-complete autoencoder to create a neural representation of the data, capturing associations between features. It extracts rules from this neural representation by exploiting the model's reconstruction mechanism. Extensive evaluations on five datasets against seven baselines demonstrate that Aerial+ achieves state-of-the-art results by learning more concise, high-quality rule sets with full data coverage. When integrated into rule-based interpretable machine learning models, Aerial+ significantly reduces execution time while maintaining or improving accuracy. | [
{
"version": "v1",
"created": "Sun, 27 Apr 2025 20:43:33 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 13:58:39 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Karabulut",
"Erkan",
""
],
[
"Groth",
"Paul",
""
],
[
"Degeler",
"Victoria",
""
]
] |
2504.19442 | Size Zheng | Size Zheng, Wenlei Bao, Qi Hou, Xuegui Zheng, Jin Fang, Chenhui Huang, Tianqi Li, Haojie Duanmu, Renze Chen, Ruifan Xu, Yifan Guo, Ningxin Zheng, Ziheng Jiang, Xinyi Di, Dongyang Wang, Jianxi Ye, Haibin Lin, Li-Wen Chang, Liqiang Lu, Yun Liang, Jidong Zhai, Xin Liu | Triton-distributed: Programming Overlapping Kernels on Distributed AI Systems with the Triton Compiler | null | null | null | null | cs.DC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this report, we propose Triton-distributed, an extension of existing Triton compiler, to overcome the programming challenges in distributed AI systems. Triton-distributed is the first compiler that supports native overlapping optimizations for distributed AI workloads, providing a good coverage of existing optimizations from different frameworks. First, we integrate communication primitives compliant with the OpenSHMEM standard into the compiler. This enables programmers to utilize these primitives with a higher-level Python programming model. Second, we illustrate how to achieve complex joint optimization of computation, memory access, and communication with the assistance of the compiler. In particular, we show how to use overlapping techniques to hide latency and present our compiler-based programming methods in both single-node and multi-node scenarios. Finally, we showcase the performance of the code generated by our compiler. In a test environment with up to 64 devices, our compiler can fully utilize heterogeneous communication and computation resources to provide effective overlapping and high performance. In many cases, the performance of the generated code can even outperform hand-optimized code. Moreover, the development difficulty and the time cost for development using our compiler are far less than those of low-level programming such as CUDA/C++, which clearly demonstrates significant productivity advantages. | [
{
"version": "v1",
"created": "Mon, 28 Apr 2025 03:09:22 GMT"
},
{
"version": "v2",
"created": "Sun, 04 May 2025 08:05:49 GMT"
},
{
"version": "v3",
"created": "Thu, 05 Jun 2025 05:48:26 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Zheng",
"Size",
""
],
[
"Bao",
"Wenlei",
""
],
[
"Hou",
"Qi",
""
],
[
"Zheng",
"Xuegui",
""
],
[
"Fang",
"Jin",
""
],
[
"Huang",
"Chenhui",
""
],
[
"Li",
"Tianqi",
""
],
[
"Duanmu",
"Haojie",
""
],
[
"Chen",
"Renze",
""
],
[
"Xu",
"Ruifan",
""
],
[
"Guo",
"Yifan",
""
],
[
"Zheng",
"Ningxin",
""
],
[
"Jiang",
"Ziheng",
""
],
[
"Di",
"Xinyi",
""
],
[
"Wang",
"Dongyang",
""
],
[
"Ye",
"Jianxi",
""
],
[
"Lin",
"Haibin",
""
],
[
"Chang",
"Li-Wen",
""
],
[
"Lu",
"Liqiang",
""
],
[
"Liang",
"Yun",
""
],
[
"Zhai",
"Jidong",
""
],
[
"Liu",
"Xin",
""
]
] |
2504.19466 | Christopher Seaton | Hans-Christian Herbig, Daniel Herden, Harper Kolehmainen, and Christopher Seaton | The partial derivative of ratios of Schur polynomials and applications to symplectic quotients | 14 pages, 3 figures v2: 15 pages, 3 figures; minor revisions and addition of references | null | null | null | math.CO math.SG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We show that a ratio of Schur polynomials $s_{\lambda}/s_{\rho}$ associated to partitions $\lambda$ and $\rho$ such $\lambda\subsetneq\rho$ has a negative partial derivative at any point where all variables are positive. This is accomplished by establishing an injective map between sets of pairs of skew semistandard Young tableaux that preserves the product of the corresponding monomials. We use this result and the description of the first Laurent coefficient of the Hilbert series of the graded algebra of regular functions on a linear symplectic quotient by the circle to demonstrate that many such symplectic quotients are not graded regularly diffeomorphic. In addition, we give an upper bound for this Laurent coefficient in terms of the largest two weights of the circle representation and demonstrate that all but finitely many circle symplectic quotients of each dimension are not graded regularly diffeomorphic to linear symplectic quotients by $\operatorname{SU}_2$. | [
{
"version": "v1",
"created": "Mon, 28 Apr 2025 04:11:22 GMT"
},
{
"version": "v2",
"created": "Wed, 04 Jun 2025 22:18:15 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Herbig",
"Hans-Christian",
""
],
[
"Herden",
"Daniel",
""
],
[
"Kolehmainen",
"Harper",
""
],
[
"Seaton",
"Christopher",
""
]
] |
2504.19649 | Lei Xu | Lei Xu, Shanshan Wang, Emmanuel Casseau, Chenglong Xiao | Intelligent4DSE: Optimizing High-Level Synthesis Design Space Exploration with Graph Neural Networks and Large Language Models | null | null | null | null | cs.LG cs.AR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | High-level synthesis (HLS) design space exploration (DSE) is an optimization process in electronic design automation (EDA) that systematically explores high-level design configurations to achieve Pareto-optimal hardware implementations balancing performance, area, and power (PPA). To optimize this process, HLS prediction tasks often employ message-passing neural networks (MPNNs), leveraging complex architectures to achieve high accuracy. These predictors serve as evaluators in the DSE process, effectively bypassing the time-consuming estimations traditionally required by HLS tools. However, existing models often prioritize structural complexity and minimization of training loss, overlooking task-specific characteristics. Additionally, while evolutionary algorithms are widely used in DSE, they typically require extensive domain-specific knowledge to design effective crossover and mutation operators. To address these limitations, we propose CoGNNs-LLMEA, a framework that integrates a graph neural network with task-adaptive message passing and a large language model-enhanced evolutionary algorithm. As a predictive model, CoGNNs directly leverages intermediate representations generated from source code after compiler front-end processing, enabling prediction of quality of results (QoR) without invoking HLS tools. Due to its strong adaptability to tasks, CoGNNs can be tuned to predict post-HLS and post-implementation outcomes, effectively bridging the gap between high-level abstractions and physical implementation characteristics. CoGNNs achieves state-of-the-art prediction accuracy in post-HLS QoR prediction, reducing mean prediction errors by 2.8$\times$ for latency and 3.4$\times$ for resource utilization compared to baseline models. | [
{
"version": "v1",
"created": "Mon, 28 Apr 2025 10:08:56 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 01:16:48 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Xu",
"Lei",
""
],
[
"Wang",
"Shanshan",
""
],
[
"Casseau",
"Emmanuel",
""
],
[
"Xiao",
"Chenglong",
""
]
] |
2504.20305 | Edgar Solomonik | Edgar Solomonik | Fast LDL factorization for dense and sparse symmetric matrices over an arbitrary field | null | null | null | null | math.NA cs.NA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | While existing algorithms may be used to solve a linear system over a general field in matrix-multiplication time, the complexity of constructing a symmetric triangular factorization (LDL) has received relatively little formal study. The LDL factorization is a common tool for factorization of symmetric matrices, and, unlike orthogonal counterparts, generalizes to an arbitrary field. We provide algorithms for dense and sparse LDL nd LU factorization that aim to minimize complexity for factorization over a general field. For LDL of an $n\times n$ matrix, we give an algorithm with complexity $O(n^\omega)$, where the complexity of $n\times n$ matrix multiplication is assumed to be $O(n^\omega)$ with $\omega>2$. For sparse matrices corresponding to graphs with treewidth $\tau$, we give an algorithm with complexity $O(n\tau^{\omega-1})$, to compute an LDL an implicit form, or the explicit LDL if the matrix is near full rank. Our sparse LDL algorithm is based on an adaptation of the null-space method for solving saddle point systems of equations, which may be of independent interest. The sparse LDL factorization algorithm also extends to computing a sparse LU factorization. | [
{
"version": "v1",
"created": "Mon, 28 Apr 2025 23:26:50 GMT"
},
{
"version": "v2",
"created": "Wed, 30 Apr 2025 15:00:39 GMT"
},
{
"version": "v3",
"created": "Wed, 04 Jun 2025 23:54:35 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Solomonik",
"Edgar",
""
]
] |
2504.20606 | Kensuke Arakawa | Kensuke Arakawa | Monoidal Relative Categories Model Monoidal $\infty$-Categories | 17 pages, comments welcome! v2: improved exposition in Construction 1.6 | null | null | null | math.CT math.AT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We show that the homotopy theory of monoidal relative categories is equivalent to that of monoidal $\infty$-categories, as well as its symmetric monoidal version. As an application, we give a concise and complete proof of the fact that every presentably monoidal or presentably symmetric monoidal $\infty$-category is presented by a monoidal or symmetric monoidal model category, which, in the monoidal case, was sketched by Lurie, and in the symmetric monoidal case, was proved by Nikolaus--Sagave. | [
{
"version": "v1",
"created": "Tue, 29 Apr 2025 10:15:32 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 10:54:32 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Arakawa",
"Kensuke",
""
]
] |
2504.21059 | R\'emi Molinier | R\'emi Molinier | A remark on the number of automorphisms of some algebraic structures | 8 pages, fix of some mistakes | null | null | null | math.GR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In these notes we look at the following question. Given a category $\mathcal C$ of algebraic structure (e.g. the category of groups, monoids, partial groups, ...) and a rational $r\in \mathbb Q$, does there exists an element $x\in \mathcal C$ such that the size of its automorphism group $\text{Aut}_{\mathcal C} (x)$ divided by the size of $x$ (whatever that would means) is equal to $r$\,? To our knowledge, this question was introduced by T\u{a}rn\u{a}uceanu in the category of groups. Here, we answer positively to this question in the categories of graphs, monoids, partial groups and posets. | [
{
"version": "v1",
"created": "Tue, 29 Apr 2025 08:47:22 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 12:45:40 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Molinier",
"Rémi",
""
]
] |
2505.00514 | Daniele Nicoletti | M. Rosenberg, D. Nicoletti, M. Buzzi, A. Iudica, C. Putzke, Y. Liu, B. Keimer, and A. Cavalleri | Signatures of three-dimensional photo-induced superconductivity in YBa$_2$Cu$_3$O$_{6.48}$ | 32 pages, 10 figures | null | null | null | cond-mat.supr-con cond-mat.str-el | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Optical excitation of large-amplitude apical oxygen phonon oscillations has been shown to renormalize the electronic properties of YBa$_2$Cu$_3$O$_{6+x}$, inducing a superconducting-like optical response above equilibrium $T_C$. All of the evidence collected so far has been based on the changes of the THz frequency $c$-axis response. In these measurements, the capacitive interlayer coupling was seen to transform into a superconducting-like inductive response. This assignment was strengthened by recent measurements of ultrafast magnetic field expulsion. Here, we report the first experimental determination of the transient in-plane optical properties, which has so far been elusive due to the high equilibrium reflectivity and the need to evaluate minute changes in the optical response. We report the appearance of a photo-induced in-plane optical gap $2\Delta\simeq30$ cm$^{-1}$ and a divergent imaginary conductivity, both consistent with photo-induced superconductivity. A global fit to these data suggests that in- and out-of-plane electronic properties never completely equilibrate during the dynamics. | [
{
"version": "v1",
"created": "Thu, 01 May 2025 13:32:35 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 15:32:56 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Rosenberg",
"M.",
""
],
[
"Nicoletti",
"D.",
""
],
[
"Buzzi",
"M.",
""
],
[
"Iudica",
"A.",
""
],
[
"Putzke",
"C.",
""
],
[
"Liu",
"Y.",
""
],
[
"Keimer",
"B.",
""
],
[
"Cavalleri",
"A.",
""
]
] |
2505.01214 | Philippe Lacomme Pr | E. Bourreau, G. Fleury, P. Lacomme | A Warm-start QAOA based approach using a swap-based mixer for the TSP: theoretical considerations,implementation and experiments | null | null | null | null | quant-ph math-ph math.MP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper investigates quantum heuristics based on Mixer Hamiltonians, which allow the search to be restricted to a specific subspace and enable warm-start strategies for solving the Traveling Salesman Problem (TSP). Approaches involving Mixer Hamiltonians can be integrated into the Quantum Approximate Optimization Algorithm (QAOA), where the Mixer acts as a mapping function that transforms qubit strings into feasible solution sets. We first introduce a swap-based mixer tailored to the TSP, which ensures that only qubit strings representing valid TSP solutions are explored during the QAOA process. Second, we propose a warm-start technique that initializes QAOA with a solution generated by any classical heuristic, thereby promoting faster convergence. These two contributions are combined into a Warm-Start QAOA framework with a Swap-Based Mixer, leveraging both structural and initialization advantages. Experimental results on a custom TSP instance involving five customers demonstrate the effectiveness of this approach, providing, for the first time, a viable integration of warm-start and swap-based mixers for the TSP within a quantum optimization framework. | [
{
"version": "v1",
"created": "Fri, 02 May 2025 12:04:46 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 08:38:30 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Bourreau",
"E.",
""
],
[
"Fleury",
"G.",
""
],
[
"Lacomme",
"P.",
""
]
] |
2505.01329 | Shan Cheng | Wang Lin, Xiao-En Huang, Shan Cheng and De-Liang Yao | Semileptonic Decays of $D \to \rho l^+ \nu$ and $D_{(s)} \to K^\ast l^+ \nu$ from Light-Cone Sum Rules | 25 pages, 7 figures, and 6 tables, accepted for publication in Phys.Rev.D | null | null | null | hep-ph hep-ex | http://creativecommons.org/licenses/by/4.0/ | We investigate the semileptonic decays of charmed mesons to light vector mesons within the framework of light-cone sum rules. Our calculation is performed at leading order in QCD coupling, incorporating contributions up to twist-five accuracy from both two-particle and three-particle light-cone distribution amplitudes. Transition form factors are predicted twist by twist to assess the convergence property of the operator product expansion. It is verified that the twist-four and twist-five contributions are indeed negligible for all the decays under consideration. Twist-three dominance is observed for some of the form factors, subject to heavy quark effective field theory interpretation. Branching ratios for the decays $D^+ \to \rho^0 \ell^+\nu_\ell$ , $D_s^+ \to K^{\ast0} \ell^+\nu_\ell$, $D^0 \to K^{\ast-} \ell^+\nu_\ell$ and $D^+ \to \bar{K}^{\ast0} \ell^+\nu_\ell$ are obtained, and a $10\%$--$20\%$ discrepancy from experimental measurements is found. Our finding indicates that the resonant-width and non-resonant QCD backgrounds effects should be potentially significant, implying the necessity to further implement their contributions in future precision studies of the semileptonic charm decays. | [
{
"version": "v1",
"created": "Fri, 02 May 2025 15:02:11 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 03:14:28 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Lin",
"Wang",
""
],
[
"Huang",
"Xiao-En",
""
],
[
"Cheng",
"Shan",
""
],
[
"Yao",
"De-Liang",
""
]
] |
2505.01449 | Jiayu Wang | Jiayu Wang, Aws Albarghouthi, Frederic Sala | COSMOS: Predictable and Cost-Effective Adaptation of LLMs | null | null | null | null | cs.LG cs.AI | http://creativecommons.org/licenses/by/4.0/ | Large language models (LLMs) achieve remarkable performance across numerous tasks by using a diverse array of adaptation strategies. However, optimally selecting a model and adaptation strategy under resource constraints is challenging and often requires extensive experimentation. We investigate whether it is possible to accurately predict both performance and cost without expensive trials. We formalize the strategy selection problem for LLMs and introduce COSMOS, a unified prediction framework that efficiently estimates adaptation outcomes at minimal cost. We instantiate and study the capability of our framework via a pair of powerful predictors: embedding-augmented lightweight proxy models to predict fine-tuning performance, and low-sample scaling laws to forecast retrieval-augmented in-context learning. Extensive evaluation across eight representative benchmarks demonstrates that COSMOS achieves high prediction accuracy while reducing computational costs by 92.72% on average, and up to 98.71% in resource-intensive scenarios. Our results show that efficient prediction of adaptation outcomes is not only feasible but can substantially reduce the computational overhead of LLM deployment while maintaining performance standards. | [
{
"version": "v1",
"created": "Wed, 30 Apr 2025 02:06:26 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Wang",
"Jiayu",
""
],
[
"Albarghouthi",
"Aws",
""
],
[
"Sala",
"Frederic",
""
]
] |
2505.02270 | Seiji Zenitani | Seiji Zenitani and Tsunehiko N. Kato | Hyper Boris integrators for kinetic plasma simulations | To appear in Comput. Phys. Commun.; 27 pages, 4 figures | Computer Physics Communications, 315, 109695 (2025) | 10.1016/j.cpc.2025.109695 | null | physics.comp-ph physics.plasm-ph physics.space-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We propose a family of numerical solvers for the nonrelativistic Newton--Lorentz equation in kinetic plasma simulations. The new solvers extend the standard 4-step Boris procedure, which has second-order accuracy in time, in three ways. First, we repeat the 4-step procedure multiple times, using an $n$-times smaller timestep ($\Delta t/n$). We derive a formula for the arbitrary subcycling number $n$, so that we obtain the result without repeating the same calculations. Second, prior to the 4-step procedure, we apply Boris-type gyrophase corrections to the electromagnetic field. In addition to a well-known correction to the magnetic field, we correct the electric field in an anisotropic manner to achieve higher-order ($N=2,4,6 \dots$th order) accuracy. Third, combining these two methods, we propose a family of high-accuracy particle solvers, the hyper Boris solvers, which have two hyperparameters of the subcycling number $n$ and the order of accuracy, $N$. The $n$-cycle $N$th-order solver gives a numerical error of $\sim (\Delta t/n)^{N}$ at affordable computational cost. | [
{
"version": "v1",
"created": "Sun, 04 May 2025 21:56:15 GMT"
},
{
"version": "v2",
"created": "Wed, 28 May 2025 10:19:42 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Zenitani",
"Seiji",
""
],
[
"Kato",
"Tsunehiko N.",
""
]
] |
2505.02383 | Zhiming Huang | Bingshan Hu, Zhiming Huang, Tianyue H. Zhang, Mathias L\'ecuyer, Nidhi Hegde | Connecting Thompson Sampling and UCB: Towards More Efficient Trade-offs Between Privacy and Regret | Camera-ready Version for ICML 2025 | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | We address differentially private stochastic bandit problems from the angles of exploring the deep connections among Thompson Sampling with Gaussian priors, Gaussian mechanisms, and Gaussian differential privacy (GDP). We propose DP-TS-UCB, a novel parametrized private bandit algorithm that enables to trade off privacy and regret. DP-TS-UCB satisfies $ \tilde{O} \left(T^{0.25(1-\alpha)}\right)$-GDP and enjoys an $O \left(K\ln^{\alpha+1}(T)/\Delta \right)$ regret bound, where $\alpha \in [0,1]$ controls the trade-off between privacy and regret. Theoretically, our DP-TS-UCB relies on anti-concentration bounds of Gaussian distributions and links exploration mechanisms in Thompson Sampling-based algorithms and Upper Confidence Bound-based algorithms, which may be of independent interest. | [
{
"version": "v1",
"created": "Mon, 05 May 2025 05:48:52 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 03:07:12 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Hu",
"Bingshan",
""
],
[
"Huang",
"Zhiming",
""
],
[
"Zhang",
"Tianyue H.",
""
],
[
"Lécuyer",
"Mathias",
""
],
[
"Hegde",
"Nidhi",
""
]
] |
2505.02580 | Robin Barta | R. Barta, M.-C. Volk, C. Bauer, C. Wagner and M. Mommert | Temperature and pressure reconstruction in turbulent Rayleigh-B\'enard convection by Lagrangian velocities using PINN | This manuscript is also submitted to Measurments Science and Technology | null | null | null | physics.flu-dyn physics.comp-ph | http://creativecommons.org/licenses/by/4.0/ | Velocity, pressure, and temperature are the key variables for understanding thermal convection, and measuring them all is a complex task. In this paper, we demonstrate a method to reconstruct temperature and pressure fields based on given Lagrangian velocity data. A physics-informed neural network (PINN) based on a multilayer perceptron architecture and a periodic sine activation function is used to reconstruct both the temperature and the pressure for two cases of turbulent Rayleigh-B\'enard convection (Pr = 6.9, Ra = $10^9$). The first dataset is generated with DNS and it includes Lagrangian velocity data of 150000 tracer particles. The second contains a PTV experiment with the same system parameters in a water-filled cubic cell, and we observed about 50000 active particle tracks per time step with the open-source framework proPTV. A realistic temperature and pressure field could be reconstructed in both cases, which underlines the importance of PINNs also in the context of experimental data. In the case of the DNS, the reconstructed temperature and pressure fields show a 90\% correlation over all particles when directly validated against the ground truth. Thus, the proposed method, in combination with particle tracking velocimetry, is able to provide velocity, temperature, and pressure fields in convective flows even in the hard turbulence regime. The PINN used in this paper is compatible with proPTV and is part of an open source project. It is available on request at https://github.com/DLR-AS-BOA. | [
{
"version": "v1",
"created": "Mon, 05 May 2025 11:31:33 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 10:24:47 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Barta",
"R.",
""
],
[
"Volk",
"M. -C.",
""
],
[
"Bauer",
"C.",
""
],
[
"Wagner",
"C.",
""
],
[
"Mommert",
"M.",
""
]
] |
2505.02748 | Christopher Clark | Christopher J. R. Clark and Julia. C. Roman-Duval and Suzanne C. Madden Marc Mertens and Claire E. Murray and J\"urgen Stutzki and Elizabeth Tarantino and and Kirill Tchernyshyov | Measuring Interstellar Carbon Abundance via 158 um [CII] Absorption with SOFIA -- A Potential Detection, and Proof-of-Concept for Depletion Studies with Future Far-IR Facilities | Accepted for publication in the Astronomical Journal; machine-readable versions of data tables accessible with online open-access version of published article at the journal: https://iopscience.iop.org/article/10.3847/1538-3881/add40f | AJ 170 8 (2025) | 0.3847/1538-3881/add40f | null | astro-ph.GA astro-ph.IM | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Carbon plays key roles in the InterStellar Medium (ISM) -- as a constituent of dust, as the carrier of the dominant far-infrared cooling line, and as a component of various important molecules. But despite this, there are very few measurements of the abundance and depletion of carbon in the diffuse ISM. As with other elements, these measurements are traditionally performed in the ultraviolet. But for carbon, such measurements are extremely difficult, and less than 20 have been reported in the literature to date. Here, we present a novel method of measuring the abundance and depletion of carbon in the diffuse ISM: by observing absorption of the 158 $\mu$m [CII] line in the far-infrared. We present a catalog of 432 candidate sightlines that use bright nearby galaxies as background sources, and predict the [CII] absorption expected towards each. We conducted a pilot study using SOFIA, targeting sightlines towards the galaxies IC342 and Circinus. We report a potential detection of Galactic [CII] absorption along the IC342 sightline, although it requires disentangling [CII] emission from IC342 itself. The Circinus sightline had an insufficiently stable instrumental baseline to allow a detection. This SOFIA study informs the prospects for [CII] absorption measurements with future facilities. To that end, we explore the potential for four proposed future FIR telescopes -- PRIMA, FIRSST, SALTUS, and Origins -- to detect [CII] absorption. We find that all four facilities would be able to detect [CII] absorption along a significant number of sightlines. | [
{
"version": "v1",
"created": "Mon, 05 May 2025 15:58:33 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 16:26:12 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Clark",
"Christopher J. R.",
""
],
[
"Roman-Duval",
"Julia. C.",
""
],
[
"Mertens",
"Suzanne C. Madden Marc",
""
],
[
"Murray",
"Claire E.",
""
],
[
"Stutzki",
"Jürgen",
""
],
[
"Tarantino",
"Elizabeth",
""
],
[
"Tchernyshyov",
"and Kirill",
""
]
] |
2505.03264 | Ajay Kumar | Sunidhi Saxena, Rajiv Gupta, Gauri Devi, Ajay Kumar | Study of Heavy Hadron Production in Au + Au Collisions at a Center-of-Mass Energy of $\sqrt{s_{NN}}=200$ GeV | null | null | null | null | hep-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Using the Monte Carlo HYDJET++ model, the transverse momentum ($p_{T}$) spectra of heavy hadrons ($D^{0}$, $\overline{D}^{0}$, $D^{+}$, $D^{-}$ and $\Lambda_{c}$), as well as the nuclear modification factors of $D^{0}$ and $D^{\pm}$, produced in Au + Au collisions at $\sqrt{s_{NN}} = 200$ GeV RHIC energy across various centrality bins, are presented. This study is motivated by the need to understand the centrality dependence of the charm enhancement factor ($\gamma_{c}$) and the roles of different hadronization mechanisms such as coalescence and fragmentation, in charm hadron production. To achieve the best description of heavy hadron production, several input parameters in both the soft and hard components of the model are tuned. The study finds a decreasing trend of $\gamma_{c}$ from central to peripheral collisions and a mass dependence across charm hadrons. Moreover, the model effectively reproduces experimental data of $p_{T}$ spectra at low and intermediate $p_{T}$, capturing key features of charm hadron production in the quark-gluon plasma medium. However, it overpredicts the data at high $p_{T}$, indicating the need for improvements in modeling heavy quark energy loss mechanisms. Further, the nuclear modification factors ($R_{AA}$ and $R_{CP}$) for $D^{0}$ mesons exhibit significant suppression in central collisions, which matches with experimental observations. This highlights the roles of collisional and radiative energy loss due to collective effects, such as coalescence and radial flow. The antiparticle-to-particle and mixed particle ratios are also presented, showing good agreement with experimental data and revealing limitations in baryon production due to the absence of heavy quark coalescence in HYDJET++. | [
{
"version": "v1",
"created": "Tue, 06 May 2025 07:46:33 GMT"
},
{
"version": "v2",
"created": "Sat, 17 May 2025 06:27:32 GMT"
},
{
"version": "v3",
"created": "Thu, 05 Jun 2025 09:39:44 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Saxena",
"Sunidhi",
""
],
[
"Gupta",
"Rajiv",
""
],
[
"Devi",
"Gauri",
""
],
[
"Kumar",
"Ajay",
""
]
] |
2505.03447 | Fumi Ogihara | Fumi Ogihara | Asymptotic formula for the sum of a prime and a square-full number in short intervals shorter than $X^{1/2}$ | null | null | null | null | math.NT | http://creativecommons.org/licenses/by/4.0/ | Let $R(N)$ be the number of representations of $N$ as a sum of a prime and a square-full number weighted with logarithmic function. In $2024$, the author and Y. Suzuki obtained an asymptotic formula for the sum of $R(N)$ over positive integers $N$ in a short interval ($X$, $X+H$] for $X^{\frac{1}{2}+\varepsilon} \le H < X^{1-\varepsilon}$. In this article, we improve the range of $H$, that is, we prove the same asymptotic formula for $X^{\frac{32-4\sqrt{15}}{49}+\varepsilon} \le H \le X^{1- \varepsilon}$. | [
{
"version": "v1",
"created": "Tue, 06 May 2025 11:37:00 GMT"
},
{
"version": "v2",
"created": "Wed, 04 Jun 2025 02:40:38 GMT"
},
{
"version": "v3",
"created": "Thu, 05 Jun 2025 08:45:24 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Ogihara",
"Fumi",
""
]
] |
2505.03448 | Yifan Peng | Yifan Peng, Yuze Hong, Ziyang Hong, Apple Pui-Yi Chui, Junfeng Wu | AquaticVision: Benchmarking Visual SLAM in Underwater Environment with Events and Frames | null | null | null | null | cs.RO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Many underwater applications, such as offshore asset inspections, rely on visual inspection and detailed 3D reconstruction. Recent advancements in underwater visual SLAM systems for aquatic environments have garnered significant attention in marine robotics research. However, existing underwater visual SLAM datasets often lack groundtruth trajectory data, making it difficult to objectively compare the performance of different SLAM algorithms based solely on qualitative results or COLMAP reconstruction. In this paper, we present a novel underwater dataset that includes ground truth trajectory data obtained using a motion capture system. Additionally, for the first time, we release visual data that includes both events and frames for benchmarking underwater visual positioning. By providing event camera data, we aim to facilitate the development of more robust and advanced underwater visual SLAM algorithms. The use of event cameras can help mitigate challenges posed by extremely low light or hazy underwater conditions. The webpage of our dataset is https://sites.google.com/view/aquaticvision-lias. | [
{
"version": "v1",
"created": "Tue, 06 May 2025 11:37:27 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 16:30:06 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Peng",
"Yifan",
""
],
[
"Hong",
"Yuze",
""
],
[
"Hong",
"Ziyang",
""
],
[
"Chui",
"Apple Pui-Yi",
""
],
[
"Wu",
"Junfeng",
""
]
] |
2505.03758 | Naohiko Otuka Dr. | Naohiko Otuka, Vidya Devi, Osamu Iwamoto | EXFOR utility codes (ForEXy) and their application to neutron fission cross section evaluation | Submitted to Applied Radiation and Isotopes | null | 10.1016/j.apradiso.2025.111903 | null | physics.comp-ph nucl-ex | http://creativecommons.org/licenses/by/4.0/ | We developed a set of EXFOR utility codes (ForEXy) to process the information of the experimental nuclear reaction data stored in the EXFOR library. We designed a new JSON format (J4) for the EXFOR library, and developed a code converting the information in an EXFOR file to a J4 file (X4TOJ4) and another code converting it to an EXFOR file (J4TOX4) as a core part of this new code package. We also developed some other codes for managements of the EXFOR storage, bibliography and dictionary. As an application of the new code package, we constructed covariance matrices for the fast neutron induced fission cross sections of neptunium-237 in the EXFOR library by using the new codes, and applied them to evaluation of the cross section between 100 keV and 200 MeV. | [
{
"version": "v1",
"created": "Sun, 20 Apr 2025 00:33:10 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Otuka",
"Naohiko",
""
],
[
"Devi",
"Vidya",
""
],
[
"Iwamoto",
"Osamu",
""
]
] |
2505.03768 | Mansur Aliyu | Mansur Aliyu Masama, Niclas Kannengie{\ss}er, Ali Sunyaev | From Concept to Measurement: A Survey of How the Blockchain Trilemma Can Be Analyzed | We corrected authors' names (e.g., corrected order of first name and last name). Revised methods from grounded theory to thematic analysis as it is more suitable. We also updated the reference of the systematic literature search. However, results remain unchanged | null | null | null | cs.CR | http://creativecommons.org/licenses/by/4.0/ | To meet non-functional requirements, practitioners must identify Pareto-optimal configurations of the degree of decentralization, scalability, and security of blockchain systems. Maximizing all of these subconcepts is, however, impossible due to the trade-offs highlighted by the blockchain trilemma. We reviewed analysis approaches to identify constructs and their operationalization through metrics for analyzing the blockchain trilemma subconcepts and to assess the applicability of the operationalized constructs to various blockchain systems. By clarifying these constructs and metrics, this work offers a theoretical foundation for more sophisticated investigations into how the blockchain trilemma manifests in blockchain systems, helping practitioners identify Pareto-optimal configurations. | [
{
"version": "v1",
"created": "Sat, 26 Apr 2025 13:05:33 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 17:36:20 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Masama",
"Mansur Aliyu",
""
],
[
"Kannengießer",
"Niclas",
""
],
[
"Sunyaev",
"Ali",
""
]
] |
2505.03789 | Syoiti Ninomiya | Syoiti Ninomiya, Yuming Ma | A new architecture of high-order deep neural networks that learn martingales | 19 pages, 3 figures | null | null | null | cs.LG math.PR q-fin.CP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | A new deep-learning neural network architecture based on high-order weak approximation algorithms for stochastic differential equations (SDEs) is proposed. The architecture enables the efficient learning of martingales by deep learning models. The behaviour of deep neural networks based on this architecture, when applied to the problem of pricing financial derivatives, is also examined. The core of this new architecture lies in the high-order weak approximation algorithms of the explicit Runge--Kutta type, wherein the approximation is realised solely through iterative compositions and linear combinations of vector fields of the target SDEs. | [
{
"version": "v1",
"created": "Thu, 01 May 2025 04:49:53 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 13:22:50 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Ninomiya",
"Syoiti",
""
],
[
"Ma",
"Yuming",
""
]
] |
2505.04075 | Jack Sanderson | Jack Sanderson, Teddy Foley, Spencer Guo, Anqi Qu, Henry Josephson | Rethinking LLM Advancement: Compute-Dependent and Independent Paths to Progress | null | null | null | null | cs.LG cs.AI | http://creativecommons.org/licenses/by/4.0/ | Regulatory efforts to govern large language model (LLM) development have predominantly focused on restricting access to high-performance computational resources. This study evaluates the efficacy of such measures by examining whether LLM capabilities can advance through algorithmic innovation in compute-constrained environments. We propose a novel framework distinguishing compute-dependent innovations--which yield disproportionate benefits at high compute--from compute-independent innovations, which improve efficiency across compute scales. The impact is quantified using Compute-Equivalent Gain (CEG). Experimental validation with nanoGPT models confirms that compute-independent advancements yield significant performance gains (e.g., with combined CEG up to $3.5\times$) across the tested scales. In contrast, compute-dependent advancements were detrimental to performance at smaller experimental scales, but showed improved CEG (on par with the baseline) as model size increased, a trend consistent with their definition of yielding primary benefits at higher compute. Crucially, these findings indicate that restrictions on computational hardware, while potentially slowing LLM progress, are insufficient to prevent all capability gains driven by algorithmic advancements. We argue that effective AI oversight must therefore incorporate mechanisms for understanding, anticipating, and potentially guiding algorithmic research, moving beyond a singular focus on hardware. The proposed framework also serves as an analytical tool for forecasting AI progress. | [
{
"version": "v1",
"created": "Wed, 07 May 2025 02:26:17 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 17:09:08 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Sanderson",
"Jack",
""
],
[
"Foley",
"Teddy",
""
],
[
"Guo",
"Spencer",
""
],
[
"Qu",
"Anqi",
""
],
[
"Josephson",
"Henry",
""
]
] |
2505.04788 | Zhenjun Zhao | Bangyan Liao, Zhenjun Zhao, Haoang Li, Yi Zhou, Yingping Zeng, Hao Li, Peidong Liu | Convex Relaxation for Robust Vanishing Point Estimation in Manhattan World | Accepted to CVPR 2025 as Award Candidate & Oral Presentation. The first two authors contributed equally to this work. Code: https://github.com/WU-CVGL/GlobustVP | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Determining the vanishing points (VPs) in a Manhattan world, as a fundamental task in many 3D vision applications, consists of jointly inferring the line-VP association and locating each VP. Existing methods are, however, either sub-optimal solvers or pursuing global optimality at a significant cost of computing time. In contrast to prior works, we introduce convex relaxation techniques to solve this task for the first time. Specifically, we employ a "soft" association scheme, realized via a truncated multi-selection error, that allows for joint estimation of VPs' locations and line-VP associations. This approach leads to a primal problem that can be reformulated into a quadratically constrained quadratic programming (QCQP) problem, which is then relaxed into a convex semidefinite programming (SDP) problem. To solve this SDP problem efficiently, we present a globally optimal outlier-robust iterative solver (called GlobustVP), which independently searches for one VP and its associated lines in each iteration, treating other lines as outliers. After each independent update of all VPs, the mutual orthogonality between the three VPs in a Manhattan world is reinforced via local refinement. Extensive experiments on both synthetic and real-world data demonstrate that GlobustVP achieves a favorable balance between efficiency, robustness, and global optimality compared to previous works. The code is publicly available at https://github.com/WU-CVGL/GlobustVP. | [
{
"version": "v1",
"created": "Wed, 07 May 2025 20:30:08 GMT"
},
{
"version": "v2",
"created": "Wed, 21 May 2025 12:59:11 GMT"
},
{
"version": "v3",
"created": "Thu, 05 Jun 2025 13:05:31 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Liao",
"Bangyan",
""
],
[
"Zhao",
"Zhenjun",
""
],
[
"Li",
"Haoang",
""
],
[
"Zhou",
"Yi",
""
],
[
"Zeng",
"Yingping",
""
],
[
"Li",
"Hao",
""
],
[
"Liu",
"Peidong",
""
]
] |
2505.05227 | C. W. J. Beenakker | Giulia Venditti, Carlo Beenakker, and Louk Rademaker | Something about the Mechanism of Induction Phenomena: The forgotten work of Berta de Haas-Lorentz on diamagnetism in superconductors | 7 pages | null | null | null | physics.hist-ph cond-mat.supr-con | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In 1925, Dr. Geertruida Luberta "Berta" de Haas-Lorentz published the paper "Iets over het mechanisme van inductieverschijnselen" in the journal Physica. Her paper was the first to discuss perfect diamagnetism of superconductors, eight years before the discovery of the Meissner effect, when the essential difference between the two phenomena was not understood. Unfortunately, her work was almost forgotten by the scientific community. To counter this, we translate her seminal 1925 paper from Dutch into English. We provide an overview of the life of Dr. De Haas-Lorentz, and comment on her pioneering contribution to the theory of superconductivity. | [
{
"version": "v1",
"created": "Thu, 08 May 2025 13:18:38 GMT"
},
{
"version": "v2",
"created": "Fri, 09 May 2025 11:59:44 GMT"
},
{
"version": "v3",
"created": "Thu, 05 Jun 2025 17:22:24 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Venditti",
"Giulia",
""
],
[
"Beenakker",
"Carlo",
""
],
[
"Rademaker",
"Louk",
""
]
] |
2505.05432 | Jelle Hartong | Jelle Hartong, Emil Have, Vijay Nenmeli, Gerben Oling | Boundary Energy-Momentum Tensors for Asymptotically Flat Spacetimes | v2: citations added, typos fixed, missing term added to boost anomaly | null | null | null | hep-th gr-qc | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We consider 3D and 4D asymptotically flat spacetimes near future null infinity endowed with the most general allowed Carroll geometry. We define a boundary energy-momentum tensor by varying the on-shell action with respect to the Carroll metric data. This requires adding counterterms to the Einstein-Hilbert action. We show that, in 4D, the shear is on par with the Carroll metric data. Their combined response defines a boundary energy-momentum-news complex whose diffeomorphism Ward identity is equivalent to the Bondi mass and angular momentum loss equations. Weyl invariance leads to an identity for the trace of the energy-momentum tensor, and local Carroll boosts are anomalous in 3D and in 4D. | [
{
"version": "v1",
"created": "Thu, 08 May 2025 17:22:02 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 15:54:30 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Hartong",
"Jelle",
""
],
[
"Have",
"Emil",
""
],
[
"Nenmeli",
"Vijay",
""
],
[
"Oling",
"Gerben",
""
]
] |
2505.05849 | Zhun Wang | Zhun Wang, Vincent Siu, Zhe Ye, Tianneng Shi, Yuzhou Nie, Xuandong Zhao, Chenguang Wang, Wenbo Guo, Dawn Song | AGENTFUZZER: Generic Black-Box Fuzzing for Indirect Prompt Injection against LLM Agents | null | null | null | null | cs.CR cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The strong planning and reasoning capabilities of Large Language Models (LLMs) have fostered the development of agent-based systems capable of leveraging external tools and interacting with increasingly complex environments. However, these powerful features also introduce a critical security risk: indirect prompt injection, a sophisticated attack vector that compromises the core of these agents, the LLM, by manipulating contextual information rather than direct user prompts. In this work, we propose a generic black-box fuzzing framework, AgentFuzzer, designed to automatically discover and exploit indirect prompt injection vulnerabilities across diverse LLM agents. Our approach starts by constructing a high-quality initial seed corpus, then employs a seed selection algorithm based on Monte Carlo Tree Search (MCTS) to iteratively refine inputs, thereby maximizing the likelihood of uncovering agent weaknesses. We evaluate AgentFuzzer on two public benchmarks, AgentDojo and VWA-adv, where it achieves 71% and 70% success rates against agents based on o3-mini and GPT-4o, respectively, nearly doubling the performance of baseline attacks. Moreover, AgentFuzzer exhibits strong transferability across unseen tasks and internal LLMs, as well as promising results against defenses. Beyond benchmark evaluations, we apply our attacks in real-world environments, successfully misleading agents to navigate to arbitrary URLs, including malicious sites. | [
{
"version": "v1",
"created": "Fri, 09 May 2025 07:40:17 GMT"
},
{
"version": "v2",
"created": "Wed, 21 May 2025 05:34:08 GMT"
},
{
"version": "v3",
"created": "Thu, 05 Jun 2025 06:44:46 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Wang",
"Zhun",
""
],
[
"Siu",
"Vincent",
""
],
[
"Ye",
"Zhe",
""
],
[
"Shi",
"Tianneng",
""
],
[
"Nie",
"Yuzhou",
""
],
[
"Zhao",
"Xuandong",
""
],
[
"Wang",
"Chenguang",
""
],
[
"Guo",
"Wenbo",
""
],
[
"Song",
"Dawn",
""
]
] |
2505.05946 | Povilas Daniu\v{s}is | Vytenis \v{S}liogeris, Povilas Daniu\v{s}is, Art\=uras Nakvosas | Full-Parameter Continual Pretraining of Gemma2: Insights into Fluency and Domain Knowledge | 9 pages, 3 figures, 1 table | null | null | null | cs.CL cs.AI cs.LG | http://creativecommons.org/licenses/by/4.0/ | In this technical report, we empirically investigate the relationship between linguistic fluency and domain knowledge in the context of continual learning with large language models (LLMs). Specifically, we enhance the linguistic fluency of the Gemma2 LLM for the Lithuanian language by autoregressively pretraining its full parameter set on the first 10\% of the Lithuanian language component of the CulturaX dataset. To prevent catastrophic forgetting of the model's existing domain knowledge, we apply Elastic Weight Consolidation (EWC), leveraging Fisher information estimated using data from the Massive Multitask Language Understanding (MMLU) benchmark. In the post-training evaluations, we assess linguistic fluency through perplexity and evaluate domain knowledge using accuracy on a suite of language understanding benchmarks, including ARC-Easy, Belebele, GSM8K, HellaSwag, MMLU, TruthfulQA, and Winogrande, in both English and Lithuanian. The empirical results demonstrate that EWC not only mitigates catastrophic forgetting by preserving the model's performance in terms of both linguistic fluency and domain knowledge but also improves or maintains these capabilities for the newly added Lithuanian language. These findings highlight the potential for more efficient adaptation of general-purpose LLMs to under-represented languages without requiring access to the original training data. The accompanying codebase is openly accessible at https://github.com/Neurotechnology/LLM_EWC. | [
{
"version": "v1",
"created": "Fri, 09 May 2025 10:43:37 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 11:13:42 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Šliogeris",
"Vytenis",
""
],
[
"Daniušis",
"Povilas",
""
],
[
"Nakvosas",
"Artūras",
""
]
] |
2505.05968 | Dan Qiao | Dan Qiao, Wenhao Li, Shanchao Yang, Hongyuan Zha, Baoxiang Wang | Offline Multi-agent Reinforcement Learning via Score Decomposition | Working papers | null | null | null | cs.LG cs.MA | http://creativecommons.org/licenses/by/4.0/ | Offline cooperative multi-agent reinforcement learning (MARL) faces unique challenges due to distributional shifts, particularly stemming from the high dimensionality of joint action spaces and the presence of out-of-distribution joint action selections. In this work, we highlight that a fundamental challenge in offline MARL arises from the multi-equilibrium nature of cooperative tasks, which induces a highly multimodal joint behavior policy space coupled with heterogeneous-quality behavior data. This makes it difficult for individual policy regularization to align with a consistent coordination pattern, leading to the policy distribution shift problems. To tackle this challenge, we design a sequential score function decomposition method that distills per-agent regularization signals from the joint behavior policy, which induces coordinated modality selection under decentralized execution constraints. Then we leverage a flexible diffusion-based generative model to learn these score functions from multimodal offline data, and integrate them into joint-action critics to guide policy updates toward high-reward, in-distribution regions under a shared team reward. Our approach achieves state-of-the-art performance across multiple particle environments and Multi-agent MuJoCo benchmarks consistently. To the best of our knowledge, this is the first work to explicitly address the distributional gap between offline and online MARL, paving the way for more generalizable offline policy-based MARL methods. | [
{
"version": "v1",
"created": "Fri, 09 May 2025 11:42:31 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 09:41:09 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Qiao",
"Dan",
""
],
[
"Li",
"Wenhao",
""
],
[
"Yang",
"Shanchao",
""
],
[
"Zha",
"Hongyuan",
""
],
[
"Wang",
"Baoxiang",
""
]
] |
2505.07286 | Keyue Qiu | Keyue Qiu, Yuxuan Song, Zhehuan Fan, Peidong Liu, Zhe Zhang, Mingyue Zheng, Hao Zhou, Wei-Ying Ma | Piloting Structure-Based Drug Design via Modality-Specific Optimal Schedule | Accepted to ICML 2025 | null | null | null | q-bio.BM cs.AI cs.LG | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Structure-Based Drug Design (SBDD) is crucial for identifying bioactive molecules. Recent deep generative models are faced with challenges in geometric structure modeling. A major bottleneck lies in the twisted probability path of multi-modalities -- continuous 3D positions and discrete 2D topologies -- which jointly determine molecular geometries. By establishing the fact that noise schedules decide the Variational Lower Bound (VLB) for the twisted probability path, we propose VLB-Optimal Scheduling (VOS) strategy in this under-explored area, which optimizes VLB as a path integral for SBDD. Our model effectively enhances molecular geometries and interaction modeling, achieving state-of-the-art PoseBusters passing rate of 95.9% on CrossDock, more than 10% improvement upon strong baselines, while maintaining high affinities and robust intramolecular validity evaluated on held-out test set. Code is available at https://github.com/AlgoMole/MolCRAFT. | [
{
"version": "v1",
"created": "Mon, 12 May 2025 07:18:09 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 12:37:46 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Qiu",
"Keyue",
""
],
[
"Song",
"Yuxuan",
""
],
[
"Fan",
"Zhehuan",
""
],
[
"Liu",
"Peidong",
""
],
[
"Zhang",
"Zhe",
""
],
[
"Zheng",
"Mingyue",
""
],
[
"Zhou",
"Hao",
""
],
[
"Ma",
"Wei-Ying",
""
]
] |
2505.07608 | Chen Chen | LLM-Core Xiaomi: Bingquan Xia, Bowen Shen, Cici, Dawei Zhu, Di Zhang, Gang Wang, Hailin Zhang, Huaqiu Liu, Jiebao Xiao, Jinhao Dong, Liang Zhao, Peidian Li, Peng Wang, Shihua Yu, Shimao Chen, Weikun Wang, Wenhan Ma, Xiangwei Deng, Yi Huang, Yifan Song, Zihan Jiang, Bowen Ye, Can Cai, Chenhong He, Dong Zhang, Duo Zhang, Guoan Wang, Hao Tian, Haochen Zhao, Heng Qu, Hongshen Xu, Jun Shi, Kainan Bao, Kai Fang, Kang Zhou, Kangyang Zhou, Lei Li, Menghang Zhu, Nuo Chen, Qiantong Wang, Shaohui Liu, Shicheng Li, Shuhao Gu, Shuhuai Ren, Shuo Liu, Sirui Deng, Weiji Zhuang, Weiwei Lv, Wenyu Yang, Xin Zhang, Xing Yong, Xing Zhang, Xingchen Song, Xinzhe Xu, Xu Wang, Yihan Yan, Yu Tu, Yuanyuan Tian, Yudong Wang, Yue Yu, Zhenru Lin, Zhichao Song, Zihao Yue | MiMo: Unlocking the Reasoning Potential of Language Model -- From Pretraining to Posttraining | null | null | null | null | cs.CL cs.AI cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present MiMo-7B, a large language model born for reasoning tasks, with optimization across both pre-training and post-training stages. During pre-training, we enhance the data preprocessing pipeline and employ a three-stage data mixing strategy to strengthen the base model's reasoning potential. MiMo-7B-Base is pre-trained on 25 trillion tokens, with additional Multi-Token Prediction objective for enhanced performance and accelerated inference speed. During post-training, we curate a dataset of 130K verifiable mathematics and programming problems for reinforcement learning, integrating a test-difficulty-driven code-reward scheme to alleviate sparse-reward issues and employing strategic data resampling to stabilize training. Extensive evaluations show that MiMo-7B-Base possesses exceptional reasoning potential, outperforming even much larger 32B models. The final RL-tuned model, MiMo-7B-RL, achieves superior performance on mathematics, code and general reasoning tasks, surpassing the performance of OpenAI o1-mini. The model checkpoints are available at https://github.com/xiaomimimo/MiMo. | [
{
"version": "v1",
"created": "Mon, 12 May 2025 14:30:11 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 11:49:09 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Xiaomi",
"LLM-Core",
""
],
[
":",
"",
""
],
[
"Xia",
"Bingquan",
""
],
[
"Shen",
"Bowen",
""
],
[
"Cici",
"",
""
],
[
"Zhu",
"Dawei",
""
],
[
"Zhang",
"Di",
""
],
[
"Wang",
"Gang",
""
],
[
"Zhang",
"Hailin",
""
],
[
"Liu",
"Huaqiu",
""
],
[
"Xiao",
"Jiebao",
""
],
[
"Dong",
"Jinhao",
""
],
[
"Zhao",
"Liang",
""
],
[
"Li",
"Peidian",
""
],
[
"Wang",
"Peng",
""
],
[
"Yu",
"Shihua",
""
],
[
"Chen",
"Shimao",
""
],
[
"Wang",
"Weikun",
""
],
[
"Ma",
"Wenhan",
""
],
[
"Deng",
"Xiangwei",
""
],
[
"Huang",
"Yi",
""
],
[
"Song",
"Yifan",
""
],
[
"Jiang",
"Zihan",
""
],
[
"Ye",
"Bowen",
""
],
[
"Cai",
"Can",
""
],
[
"He",
"Chenhong",
""
],
[
"Zhang",
"Dong",
""
],
[
"Zhang",
"Duo",
""
],
[
"Wang",
"Guoan",
""
],
[
"Tian",
"Hao",
""
],
[
"Zhao",
"Haochen",
""
],
[
"Qu",
"Heng",
""
],
[
"Xu",
"Hongshen",
""
],
[
"Shi",
"Jun",
""
],
[
"Bao",
"Kainan",
""
],
[
"Fang",
"Kai",
""
],
[
"Zhou",
"Kang",
""
],
[
"Zhou",
"Kangyang",
""
],
[
"Li",
"Lei",
""
],
[
"Zhu",
"Menghang",
""
],
[
"Chen",
"Nuo",
""
],
[
"Wang",
"Qiantong",
""
],
[
"Liu",
"Shaohui",
""
],
[
"Li",
"Shicheng",
""
],
[
"Gu",
"Shuhao",
""
],
[
"Ren",
"Shuhuai",
""
],
[
"Liu",
"Shuo",
""
],
[
"Deng",
"Sirui",
""
],
[
"Zhuang",
"Weiji",
""
],
[
"Lv",
"Weiwei",
""
],
[
"Yang",
"Wenyu",
""
],
[
"Zhang",
"Xin",
""
],
[
"Yong",
"Xing",
""
],
[
"Zhang",
"Xing",
""
],
[
"Song",
"Xingchen",
""
],
[
"Xu",
"Xinzhe",
""
],
[
"Wang",
"Xu",
""
],
[
"Yan",
"Yihan",
""
],
[
"Tu",
"Yu",
""
],
[
"Tian",
"Yuanyuan",
""
],
[
"Wang",
"Yudong",
""
],
[
"Yu",
"Yue",
""
],
[
"Lin",
"Zhenru",
""
],
[
"Song",
"Zhichao",
""
],
[
"Yue",
"Zihao",
""
]
] |
2505.08151 | Joey Chan | Joey Chan, Zhen Chen, Ershun Pan | Foundation Models Knowledge Distillation For Battery Capacity Degradation Forecast | null | null | null | null | cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Accurate estimation of lithium-ion battery capacity degradation is critical for enhancing the reliability and safety of battery operations. Traditional expert models, tailored to specific scenarios, provide isolated estimations. With the rapid advancement of data-driven techniques, a series of general-purpose time-series foundation models have been developed. However, foundation models specifically designed for battery capacity degradation remain largely unexplored. To enable zero-shot generalization in battery degradation prediction using large model technology, this study proposes a degradation-aware fine-tuning strategy for time-series foundation models. We apply this strategy to fine-tune the Timer model on approximately 10 GB of open-source battery charge discharge data. Validation on our released CycleLife-SJTUIE dataset demonstrates that the fine-tuned Battery-Timer possesses strong zero-shot generalization capability in capacity degradation forecasting. To address the computational challenges of deploying large models, we further propose a knowledge distillation framework that transfers the knowledge of pre-trained foundation models into compact expert models. Distillation results across several state-of-the-art time-series expert models confirm that foundation model knowledge significantly improves the multi-condition generalization of expert models. | [
{
"version": "v1",
"created": "Tue, 13 May 2025 01:03:35 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 03:20:39 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Chan",
"Joey",
""
],
[
"Chen",
"Zhen",
""
],
[
"Pan",
"Ershun",
""
]
] |
2505.08302 | Oishee Bintey Hoque | Oishee Bintey Hoque, Nibir Chandra Mandal, Abhijin Adiga, Samarth Swarup, Sayjro Kossi Nouwakpo, Amanda Wilson, Madhav Marathe | Knowledge-Informed Deep Learning for Irrigation Type Mapping from Remote Sensing | Full version of the paper will be appearing at the Proceedings of the Thirty-Third International Joint Conference on Artificial Intelligence (IJCAI-25), Special Track on AI for Good | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Accurate mapping of irrigation methods is crucial for sustainable agricultural practices and food systems. However, existing models that rely solely on spectral features from satellite imagery are ineffective due to the complexity of agricultural landscapes and limited training data, making this a challenging problem. We present Knowledge-Informed Irrigation Mapping (KIIM), a novel Swin-Transformer based approach that uses (i) a specialized projection matrix to encode crop to irrigation probability, (ii) a spatial attention map to identify agricultural lands from non-agricultural lands, (iii) bi-directional cross-attention to focus complementary information from different modalities, and (iv) a weighted ensemble for combining predictions from images and crop information. Our experimentation on five states in the US shows up to 22.9\% (IoU) improvement over baseline with a 71.4% (IoU) improvement for hard-to-classify drip irrigation. In addition, we propose a two-phase transfer learning approach to enhance cross-state irrigation mapping, achieving a 51% IoU boost in a state with limited labeled data. The ability to achieve baseline performance with only 40% of the training data highlights its efficiency, reducing the dependency on extensive manual labeling efforts and making large-scale, automated irrigation mapping more feasible and cost-effective. | [
{
"version": "v1",
"created": "Tue, 13 May 2025 07:25:28 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 05:18:45 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Hoque",
"Oishee Bintey",
""
],
[
"Mandal",
"Nibir Chandra",
""
],
[
"Adiga",
"Abhijin",
""
],
[
"Swarup",
"Samarth",
""
],
[
"Nouwakpo",
"Sayjro Kossi",
""
],
[
"Wilson",
"Amanda",
""
],
[
"Marathe",
"Madhav",
""
]
] |
2505.08421 | Simone Machetti | Simone Machetti, Pasquale Davide Schiavone, Lara Orlandic, Darong Huang, Deniz Kasap, Giovanni Ansaloni, David Atienza | e-GPU: An Open-Source and Configurable RISC-V Graphic Processing Unit for TinyAI Applications | null | null | null | null | cs.AR | http://creativecommons.org/licenses/by/4.0/ | Graphics processing units (GPUs) excel at parallel processing, but remain largely unexplored in ultra-low-power edge devices (TinyAI) due to their power and area limitations, as well as the lack of suitable programming frameworks. To address these challenges, this work introduces embedded GPU (e-GPU), an open-source and configurable RISC-V GPU platform designed for TinyAI devices. Its extensive configurability enables area and power optimization, while a dedicated Tiny-OpenCL implementation provides a lightweight programming framework tailored to resource-constrained environments. To demonstrate its adaptability in real-world scenarios, we integrate the e-GPU with the eXtendible Heterogeneous Energy-Efficient Platform (X-HEEP) to realize an accelerated processing unit (APU) for TinyAI applications. Multiple instances of the proposed system, featuring varying e-GPU configurations, are implemented in TSMC's 16 nm SVT CMOS technology and are operated at 300 MHz and 0.8 V. Their area and leakage characteristics are analyzed to ensure alignment with TinyAI constraints. To assess both runtime overheads and application-level efficiency, we employ two benchmarks: General Matrix Multiply (GeMM) and bio-signal processing (TinyBio) workloads. The GeMM benchmark is used to quantify the scheduling overhead introduced by the Tiny-OpenCL framework. The results show that the delay becomes negligible for matrix sizes larger than 256x256 (or equivalent problem sizes). The TinyBio benchmark is then used to evaluate performance and energy improvements in the baseline host. The results demonstrate that the high-range e-GPU configuration with 16 threads achieves up to a 15.1x speed-up and reduces energy consumption by up to 3.1x, while incurring only a 2.5x area overhead and operating within a 28 mW power budget. | [
{
"version": "v1",
"created": "Tue, 13 May 2025 10:26:52 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 11:22:37 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Machetti",
"Simone",
""
],
[
"Schiavone",
"Pasquale Davide",
""
],
[
"Orlandic",
"Lara",
""
],
[
"Huang",
"Darong",
""
],
[
"Kasap",
"Deniz",
""
],
[
"Ansaloni",
"Giovanni",
""
],
[
"Atienza",
"David",
""
]
] |
2505.08465 | Phillip J Black | D. P. Zerafa, P. N. Wilkinson, C. J. Radcliffe, J. P. Leahy, I. W. A. Browne, P. J. Black | L-BASS: a project to produce an absolutely calibrated 1.4 GHz sky map, I -- Scientific rationale and system overview | null | RAS Techniques and Instruments, Volume 4, 2025, rzaf017 | 10.1093/rasti/rzaf017 | null | astro-ph.IM astro-ph.CO | http://creativecommons.org/licenses/by/4.0/ | L-BASS is an instrument designed to produce an absolutely calibrated map of the sky at a wavelength of 21 cm (L-band) with a radiometric accuracy of less than or equal to 0.1 K and with an angular resolution of 23 degrees. The prime motivations are to improve the temperature calibration of higher resolution maps and to investigate the steep spectrum radio background proposed by the ARCADE 2 team. The instrument consists of a pair of conical horn antennas which can scan independently in elevation; each antenna produces a circularly polarized output. The difference in signals from the antennas is measured with a continuous-comparison receiver connected to a digital spectrometer sampling the signal from 1400 MHz to 1425 MHz within the protected radio astronomy band. We describe the astrophysical motivation for the project, the design requirements and how these will be attained. | [
{
"version": "v1",
"created": "Tue, 13 May 2025 11:47:50 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Zerafa",
"D. P.",
""
],
[
"Wilkinson",
"P. N.",
""
],
[
"Radcliffe",
"C. J.",
""
],
[
"Leahy",
"J. P.",
""
],
[
"Browne",
"I. W. A.",
""
],
[
"Black",
"P. J.",
""
]
] |
2505.08467 | Leonardo Giani | Leonardo Giani, Rodrigo Von Marttens, Oliver Fabio Piattella | The matter with(in) CPL | v2, reference added, comments are welcome! | null | null | null | astro-ph.CO gr-qc | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We introduce a two-parameter phenomenological extension of the $\Lambda$CDM model in which the equation of state parameter of the ``dust'' fluid becomes different from zero for redshifts below a transition value $z_t$. Using data from DESI DR2 BAO, DESY5 Sn~Ia and CMB distance priors ($R,l_A,\omega_b$) data, we compare our model with the standard CPL parameterization $w_0-w_a$ for dynamical dark energy. Using the Deviance Information Criteria (DIC), we find that the two models are essentially indistinguishable ($\Delta$DIC $<$ 2) and preferred over $\Lambda$CDM with a significance $\geq 3 \sigma$. We discuss how this parameterization finds a natural interpretation in the context of cosmological backreaction and derive a prediction for the evolution of the growth factor, discussing its impact on low redshift $f\sigma_8$ measurements. | [
{
"version": "v1",
"created": "Tue, 13 May 2025 11:50:01 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 05:43:41 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Giani",
"Leonardo",
""
],
[
"Von Marttens",
"Rodrigo",
""
],
[
"Piattella",
"Oliver Fabio",
""
]
] |
2505.08502 | Peter Gaal | Daniel Schmidt, David. v. Stetten, Michael Agthe, Arwen R. Pearson, Goddfrey .S. Beddard, Briony A. Yorke, Friedjof Tellkamp and Peter Gaal | Demonstration of Advanced Timing Schemes in Time-Resolved X-ray Diffraction Measurements | 17 pages, 4 figures | null | null | null | cond-mat.mtrl-sci | http://creativecommons.org/licenses/by/4.0/ | We present time-resolved X-ray diffraction measurements using advanced timing schemes that provide high temporal resolution while also maintaining a high flux in the X-ray probe beam. The method employs patterned probe pulse sequences that are generated with the WaveGate solid-state pulse picker. We demonstrate the feasibility of our method at two different beamlines on millisecond and microsecond timescales. | [
{
"version": "v1",
"created": "Tue, 13 May 2025 12:32:29 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 14:05:06 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Schmidt",
"Daniel",
""
],
[
"Stetten",
"David. v.",
""
],
[
"Agthe",
"Michael",
""
],
[
"Pearson",
"Arwen R.",
""
],
[
"Beddard",
"Goddfrey . S.",
""
],
[
"Yorke",
"Briony A.",
""
],
[
"Tellkamp",
"Friedjof",
""
],
[
"Gaal",
"Peter",
""
]
] |
2505.08861 | Daniel Piacitelli | Daniel R. Piacitelli, Alyson M. Brooks, Charlotte Christensen, N. Nicole Sanchez, Yakov Faerman, Sijing Shen, Akaxia Cruz, Ben Keller, Thomas R. Quinn, James Wadsley | Marvelous Metals: Surveying the Circumgalactic Medium of Simulated Dwarf Galaxies | 31 pages, 16 figures | null | null | null | astro-ph.GA | http://creativecommons.org/licenses/by/4.0/ | Dwarf galaxies are uniquely sensitive to energetic feedback processes and are known to experience substantial mass and metal loss from their disk. Here, we investigate the circumgalactic medium (CGM) of 64 isolated dwarf galaxies ($6.0<$log(M$_*/M_{\odot}$)$<9.5$) at $z=0$ from the Marvel-ous Dwarfs and Marvelous Massive Dwarfs hydrodynamic simulations. Our galaxies produce column densities broadly consistent with current observations. We investigate these column densities in the context of mass and metal retention rates and the physical properties of the CGM. We find $48\pm11\%$ of all baryons within $R_{200c}$ reside in the CGM, with $\sim70\%$ of CGM mass existing in a warm gas phase, $10^{4.5}<T<10^{5.5}$ K that dominates beyond $r/R_{200c}\sim0.5$. Further, the warm and cool ($10^{4.0}<T<10^{4.5}$ K) gas phases each retain $5-10\%$ of metals formed by the dwarf galaxy. The significant fraction of mass and metals residing in the warm CGM phase provides an interpretation for the lack of observed low ion detections beyond $b/R_{200c}\sim0.5$ at $z\sim0$. We find a weak correlation between galaxy mass and total CGM metal retention despite the fraction of metals lost from the halo increasing from $\sim10\%$ to $>40\%$ towards lower masses. Our findings highlight the CGM (primarily its warm component) as a key reservoir of mass and metals for dwarf galaxies across stellar masses and underscore its importance in understanding the baryon cycle in the low-mass regime. Finally, we provide individual galaxy properties of our full sample and quantify the fraction of ultraviolet observable mass to support future observational programs, particularly those aimed at performing a metal budget around dwarf galaxies. | [
{
"version": "v1",
"created": "Tue, 13 May 2025 18:00:00 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 13:43:05 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Piacitelli",
"Daniel R.",
""
],
[
"Brooks",
"Alyson M.",
""
],
[
"Christensen",
"Charlotte",
""
],
[
"Sanchez",
"N. Nicole",
""
],
[
"Faerman",
"Yakov",
""
],
[
"Shen",
"Sijing",
""
],
[
"Cruz",
"Akaxia",
""
],
[
"Keller",
"Ben",
""
],
[
"Quinn",
"Thomas R.",
""
],
[
"Wadsley",
"James",
""
]
] |
2505.08929 | Valentin Agababaev | V. A. Agababaev, E. A. Prokhorchuk, D. A. Glazov, A. V. Malyshev, V. M. Shabaev, A. V. Volotka | QED effects in quadratic Zeeman splitting in highly charged hydrogen-like ions | 34 pages, 2 figures, 6 tables | null | null | null | physics.atom-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present ab initio calculations of one-electron quantum electrodynamical corrections to the second-order Zeeman splitting for the $1s_{1/2}$, $2s_{1/2}$, and $2p_{1/2}$ states in highly charged hydrogen-like ions. The self-energy correction is evaluated using the rigorous QED approach. The vacuum polarization correction is evaluated within the electric-loop approximation. Calculations are performed for the wide range of nuclear charge number: $Z = 14 - 92$. | [
{
"version": "v1",
"created": "Tue, 13 May 2025 19:53:26 GMT"
},
{
"version": "v2",
"created": "Wed, 04 Jun 2025 20:48:29 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Agababaev",
"V. A.",
""
],
[
"Prokhorchuk",
"E. A.",
""
],
[
"Glazov",
"D. A.",
""
],
[
"Malyshev",
"A. V.",
""
],
[
"Shabaev",
"V. M.",
""
],
[
"Volotka",
"A. V.",
""
]
] |
2505.09400 | Fernando Cordero | Fernando Cordero and Sophia-Marie Mellis and Emmanuel Schertzer | Structured coalescents, coagulation equations and multi-type branching processes | 26 pages, 5 figures | null | null | null | math.PR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Consider a structured population consisting of $d$ colonies, with migration rates that are proportional to a positive parameter $K$. We sample $N_K$ individuals distributed evenly across the $d$ colonies and trace their ancestral lineages back. Within a colony, we assume that pairs of ancestral lineages coalesce at a constant rate, as in a Kingman's coalescent. We identify each ancestral lineage with the set, or block, of its descendants in the sample and we encode the state of the system using a $d$-dimensional vector of empirical measures; the $i$-th component records the blocks present in colony $i$ and the initial location of the lineages composing each block. We are interested in the asymptotic behaviour of the process of empirical measures as $K\to\infty$. We consider two regimes: the critical sampling regime, where $N_K \sim K$ and the large sampling regime where $N_K \gg K$. After an appropriate time-space scaling, we show that the process of empirical measures converges to the solution of a $d$-dimensional coagulation equation. In the critical sampling regime, the solution can be represented in terms of a multi-type branching process. In the large sampling regime, the solution can be represented in terms of the entrance law of a multi-type continuous state branching process. | [
{
"version": "v1",
"created": "Wed, 14 May 2025 13:58:39 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 15:21:34 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Cordero",
"Fernando",
""
],
[
"Mellis",
"Sophia-Marie",
""
],
[
"Schertzer",
"Emmanuel",
""
]
] |
2505.09430 | Yutong Hu | Yutong Hu, Pinhao Song, Kehan Wen, Renaud Detry | Mini Diffuser: Fast Multi-task Diffusion Policy Training Using Two-level Mini-batches | null | null | null | null | cs.RO cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present a method that reduces, by an order of magnitude, the time and memory needed to train multi-task vision-language robotic diffusion policies. This improvement arises from a previously underexplored distinction between action diffusion and the image diffusion techniques that inspired it: In image generation, the target is high-dimensional. By contrast, in action generation, the dimensionality of the target is comparatively small, and only the image condition is high-dimensional. Our approach, \emph{Mini Diffuser}, exploits this asymmetry by introducing \emph{two-level minibatching}, which pairs multiple noised action samples with each vision-language condition, instead of the conventional one-to-one sampling strategy. To support this batching scheme, we introduce architectural adaptations to the diffusion transformer that prevent information leakage across samples while maintaining full conditioning access. In RLBench simulations, Mini-Diffuser achieves 95\% of the performance of state-of-the-art multi-task diffusion policies, while using only 5\% of the training time and 7\% of the memory. Real-world experiments further validate that Mini-Diffuser preserves the key strengths of diffusion-based policies, including the ability to model multimodal action distributions and produce behavior conditioned on diverse perceptual inputs. Code available at mini-diffuse-actor.github.io | [
{
"version": "v1",
"created": "Wed, 14 May 2025 14:34:40 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 14:01:16 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Hu",
"Yutong",
""
],
[
"Song",
"Pinhao",
""
],
[
"Wen",
"Kehan",
""
],
[
"Detry",
"Renaud",
""
]
] |
2505.09443 | Kiprian Berbatov | Kiprian Berbatov, Andrey P. Jivkov | Variational formulations of transport on combinatorial meshes | 40 pages, 7 figures, 40 references, submitted to Applied Mathematical Modelling | null | null | null | math-ph math.MP | http://creativecommons.org/licenses/by/4.0/ | We develop analogues of the primal and mixed variational formulations of the conservation laws describing transport of scalar physical properties for topological spaces referred to as cell complexes. The development is restricted to cell complexes with cells having connectivity of simple polytopes. Such spaces are suitable representations of physical systems composed of elements of different apparent topological dimensions, where all elements might have individual properties, and elements of a given dimension might interact via common elements of lower dimensions. This modelling foundation can be considered as intermediate between the foundation for particle-based modelling, which is the discrete topology, and the foundation for continuum modelling, which is the smooth topology. The new foundation offers advantages for the analysis of the behaviour of physical systems with complex internal structures. The development is based on a calculus with combinatorial differential forms which is a discrete analogue of the smooth exterior calculus with differential forms. We call the resulting calculus combinatorial mesh calculus (CMC) as it is based on combinatorial meshes, for which embedding is forgotten and only the connectivity between cells and measures of the cells are used in calculations. We discuss how the obtained formulation is specialised for several different problems, including mass diffusion, heat conduction, fluid flow in porous media, and charge diffusion, and provide details about the formulation of initial boundary value problems for these transport cases. Examples and results for selected boundary value problems are given to demonstrate the capabilities of the CMC formulations. | [
{
"version": "v1",
"created": "Wed, 14 May 2025 14:50:45 GMT"
},
{
"version": "v2",
"created": "Mon, 19 May 2025 18:09:57 GMT"
},
{
"version": "v3",
"created": "Thu, 05 Jun 2025 09:16:34 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Berbatov",
"Kiprian",
""
],
[
"Jivkov",
"Andrey P.",
""
]
] |
2505.09496 | Rui Miao | Rui Miao, Babak Shahbaba, Annie Qu | Reinforcement Learning for Individual Optimal Policy from Heterogeneous Data | null | null | null | null | stat.ML cs.LG | http://creativecommons.org/licenses/by/4.0/ | Offline reinforcement learning (RL) aims to find optimal policies in dynamic environments in order to maximize the expected total rewards by leveraging pre-collected data. Learning from heterogeneous data is one of the fundamental challenges in offline RL. Traditional methods focus on learning an optimal policy for all individuals with pre-collected data from a single episode or homogeneous batch episodes, and thus, may result in a suboptimal policy for a heterogeneous population. In this paper, we propose an individualized offline policy optimization framework for heterogeneous time-stationary Markov decision processes (MDPs). The proposed heterogeneous model with individual latent variables enables us to efficiently estimate the individual Q-functions, and our Penalized Pessimistic Personalized Policy Learning (P4L) algorithm guarantees a fast rate on the average regret under a weak partial coverage assumption on behavior policies. In addition, our simulation studies and a real data application demonstrate the superior numerical performance of the proposed method compared with existing methods. | [
{
"version": "v1",
"created": "Wed, 14 May 2025 15:44:10 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 16:23:46 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Miao",
"Rui",
""
],
[
"Shahbaba",
"Babak",
""
],
[
"Qu",
"Annie",
""
]
] |
2505.09833 | Emre Girgin | Tuba Girgin, Emre Girgin, Cagri Kilic | Learning Rock Pushability on Rough Planetary Terrain | Paper presented at the Workshop on Field Robotics, ICRA 2025, Atlanta, GA, United States | null | null | null | cs.RO cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In the context of mobile navigation in unstructured environments, the predominant approach entails the avoidance of obstacles. The prevailing path planning algorithms are contingent upon deviating from the intended path for an indefinite duration and returning to the closest point on the route after the obstacle is left behind spatially. However, avoiding an obstacle on a path that will be used repeatedly by multiple agents can hinder long-term efficiency and lead to a lasting reliance on an active path planning system. In this study, we propose an alternative approach to mobile navigation in unstructured environments by leveraging the manipulation capabilities of a robotic manipulator mounted on top of a mobile robot. Our proposed framework integrates exteroceptive and proprioceptive feedback to assess the push affordance of obstacles, facilitating their repositioning rather than avoidance. While our preliminary visual estimation takes into account the characteristics of both the obstacle and the surface it relies on, the push affordance estimation module exploits the force feedback obtained by interacting with the obstacle via a robotic manipulator as the guidance signal. The objective of our navigation approach is to enhance the efficiency of routes utilized by multiple agents over extended periods by reducing the overall time spent by a fleet in environments where autonomous infrastructure development is imperative, such as lunar or Martian surfaces. | [
{
"version": "v1",
"created": "Wed, 14 May 2025 22:23:30 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 15:00:47 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Girgin",
"Tuba",
""
],
[
"Girgin",
"Emre",
""
],
[
"Kilic",
"Cagri",
""
]
] |
2505.09999 | Yuxing Xiang | Yuxing Xiang, Xue Li, Kun Qian, Wenyuan Yu, Ennan Zhai, Xin Jin | ServeGen: Workload Characterization and Generation of Large Language Model Serving in Production | Released URL | null | null | null | cs.DC | http://creativecommons.org/licenses/by-sa/4.0/ | With the widespread adoption of Large Language Models (LLMs), serving LLM inference requests has become an increasingly important task, attracting active research advancements. Practical workloads play an essential role in this process: they are critical for motivating and benchmarking serving techniques and systems. However, the existing understanding of real-world LLM serving workloads is limited due to the lack of a comprehensive workload characterization. Prior analyses remain insufficient in scale and scope, thus failing to fully capture intricate workload characteristics.
In this paper, we fill the gap with an in-depth characterization of LLM serving workloads collected from our worldwide cloud inference serving service, covering not only language models but also emerging multimodal and reasoning models, and unveiling important new findings in each case. Moreover, based on our findings, we propose ServeGen, a principled framework for generating realistic LLM serving workloads by composing them on a per-client basis. A practical use case in production validates that ServeGen avoids 50% under-provisioning compared to naive workload generation, demonstrating ServeGen's advantage in performance benchmarking. ServeGen is available at https://github.com/alibaba/ServeGen. | [
{
"version": "v1",
"created": "Thu, 15 May 2025 06:24:08 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 05:46:07 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Xiang",
"Yuxing",
""
],
[
"Li",
"Xue",
""
],
[
"Qian",
"Kun",
""
],
[
"Yu",
"Wenyuan",
""
],
[
"Zhai",
"Ennan",
""
],
[
"Jin",
"Xin",
""
]
] |
2505.10033 | Luis F. W. Batista | Luis F. W. Batista, St\'ephanie Aravecchia, Seth Hutchinson, C\'edric Pradalier | Evaluating Robustness of Deep Reinforcement Learning for Autonomous Surface Vehicle Control in Field Tests | Presented at the 2025 IEEE ICRA Workshop on Field Robotics | null | null | null | cs.RO cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Despite significant advancements in Deep Reinforcement Learning (DRL) for Autonomous Surface Vehicles (ASVs), their robustness in real-world conditions, particularly under external disturbances, remains insufficiently explored. In this paper, we evaluate the resilience of a DRL-based agent designed to capture floating waste under various perturbations. We train the agent using domain randomization and evaluate its performance in real-world field tests, assessing its ability to handle unexpected disturbances such as asymmetric drag and an off-center payload. We assess the agent's performance under these perturbations in both simulation and real-world experiments, quantifying performance degradation and benchmarking it against an MPC baseline. Results indicate that the DRL agent performs reliably despite significant disturbances. Along with the open-source release of our implementation, we provide insights into effective training strategies, real-world challenges, and practical considerations for deploying DRLbased ASV controllers. | [
{
"version": "v1",
"created": "Thu, 15 May 2025 07:29:16 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 15:31:07 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Batista",
"Luis F. W.",
""
],
[
"Aravecchia",
"Stéphanie",
""
],
[
"Hutchinson",
"Seth",
""
],
[
"Pradalier",
"Cédric",
""
]
] |
2505.10424 | Marco Badran | Marco Badran and Jean Van Schaftingen | Stationary $p$-harmonic maps approaching planar singular harmonic maps to the circle | 24 pages | Journal of Mathematical Analysis and Applications Volume 551, Issue 2, Part 1, 15 November 2025, 129697 | 10.1016/j.jmaa.2025.129697 | null | math.AP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Given a bounded planar domain $\Omega \subset \mathbb{R}^2$, we show that any singular harmonic map into the circle $\mathbb{S}^1$ corresponding to a topologically nondegenerate critical point of the renormalised energy in the sense of Bethuel, Brezis and H\'elein is a limit of stationary $p$-harmonic maps for $p < 2$ as $p \to 2$ | [
{
"version": "v1",
"created": "Thu, 15 May 2025 15:39:29 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Badran",
"Marco",
""
],
[
"Van Schaftingen",
"Jean",
""
]
] |
2505.10661 | Domenique Zipperling | Domenique Zipperling, Luca Deck, Julia Lanzl, Niklas K\"uhl | It's only fair when I think it's fair: How Gender Bias Alignment Undermines Distributive Fairness in Human-AI Collaboration | null | ACM Conference on Fairness, Accountability, and Transparency 2025 (ACM FAccT 2025) | 10.1145/3715275.3732084 | null | cs.HC | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Human-AI collaboration is increasingly relevant in consequential areas where AI recommendations support human discretion. However, human-AI teams' effectiveness, capability, and fairness highly depend on human perceptions of AI. Positive fairness perceptions have been shown to foster trust and acceptance of AI recommendations. Yet, work on confirmation bias highlights that humans selectively adhere to AI recommendations that align with their expectations and beliefs -- despite not being necessarily correct or fair. This raises the question whether confirmation bias also transfers to the alignment of gender bias between human and AI decisions. In our study, we examine how gender bias alignment influences fairness perceptions and reliance. The results of a 2x2 between-subject study highlight the connection between gender bias alignment, fairness perceptions, and reliance, demonstrating that merely constructing a ``formally fair'' AI system is insufficient for optimal human-AI collaboration; ultimately, AI recommendations will likely be overridden if biases do not align. | [
{
"version": "v1",
"created": "Thu, 15 May 2025 19:10:12 GMT"
},
{
"version": "v2",
"created": "Wed, 04 Jun 2025 18:13:56 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Zipperling",
"Domenique",
""
],
[
"Deck",
"Luca",
""
],
[
"Lanzl",
"Julia",
""
],
[
"Kühl",
"Niklas",
""
]
] |
2505.10796 | Wei-Wei Zhang | Wei-Wei Zhang, Xiaopeng Huang, Shenglin Shan, Wei Zhao, Beiya Yang, Wei Pan, Haobin Shi | Quantum data generation in a denoising model with multiscale entanglement renormalization network | 10 pages, 12 figures | Phys. Scr. 100 065120 (2025) | 10.1088/1402-4896/add8c8 | null | quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Quantum technology has entered the era of noisy intermediate-scale quantum (NISQ) information processing. The technological revolution of machine learning represented by generative models heralds a great prospect of artificial intelligence, and the huge amount of data processes poses a big challenge to existing computers. The generation of large quantities of quantum data will be a challenge for quantum artificial intelligence. In this work, we present an efficient noise-resistant quantum data generation method that can be applied to various types of NISQ quantum processors, where the target quantum data belongs to a certain class and our proposal enables the generation of various quantum data belonging to the target class. Specifically, we propose a quantum denoising probability model (QDM) based on a multiscale entanglement renormalization network (MERA) for the generation of quantum data. To show the feasibility and practicality of our scheme, we demonstrate the generations of the classes of GHZ-like states and W-like states with a success rate above 99%. Our MREA QDM can also be used to denoise multiple types of quantum data simultaneously. We show the success rate of denoising both GHZ-like and W-like states with a single qubit noise environment of noise level within 1/4 can approximate to be 100%, and with two other types of noise environment with noise level within 1/4 can be above 90%. Our quantum data generation scheme provides new ideas and prospects for quantum generative models in the NISQ era. | [
{
"version": "v1",
"created": "Fri, 16 May 2025 02:18:01 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Zhang",
"Wei-Wei",
""
],
[
"Huang",
"Xiaopeng",
""
],
[
"Shan",
"Shenglin",
""
],
[
"Zhao",
"Wei",
""
],
[
"Yang",
"Beiya",
""
],
[
"Pan",
"Wei",
""
],
[
"Shi",
"Haobin",
""
]
] |
2505.11224 | Eva-Maria Ahrer | Eva-Maria Ahrer, Siddharth Gandhi, Lili Alderson, James Kirk, Johanna Teske, Richard A. Booth, Catriona H. McDonald, Duncan A. Christie, Alastair B. Claringbold, Rebecca Nealon, Vatsal Panwar, Dimitri Veras, Hannah R. Wakeford, Peter J. Wheatley, Maria Zamyatina | Tracing the formation and migration history: molecular signatures in the atmosphere of misaligned hot Jupiter WASP-94Ab using JWST NIRSpec/G395H | 19 pages, 15 figures, accepted for publication in MNRAS | null | 10.1093/mnras/staf819 | null | astro-ph.EP | http://creativecommons.org/licenses/by/4.0/ | The discovery of hot Jupiters that orbit very close to their host stars has long challenged traditional models of planetary formation and migration. Characterising their atmospheric composition - mainly in the form of the carbon-to-oxygen (C/O) ratio and metallicity - can provide insights into their formation locations and evolution pathways. With JWST we can characterise the atmospheres of these types of planets more precisely than previously possible, primarily because it allows us to determine both their atmospheric oxygen and carbon composition. Here, we present a JWST NIRSpec/G395H transmission spectrum from 2.8-5.1$\mu m$ of WASP-94Ab, an inflated hot Jupiter with a retrograde misaligned orbit around its F-type host star. We find a relatively cloud-free atmosphere, with absorption features of H$_2$O and CO$_2$ at detection significances of $\sim 4\sigma$ and $\sim 11\sigma$, respectively. In addition, we detect tentative evidence of CO absorption at $\sim3\sigma$, as well as hints of sulphur with the detection of H$_2$S at a $\sim 2.5\sigma$ confidence level. Our favoured equilibrium chemistry model determines a C/O ratio of $0.49^{+0.08}_{-0.13}$ for WASP-94Ab's atmosphere, which is substellar compared to the star's C/O ratio of $0.68 \pm 0.10$. The retrieved atmospheric metallicity is similar to the star's metallicity as both are $\sim 2\times$ solar. We find that this sub-stellar C/O ratio and stellar metallicity can be best explained by pebble accretion or planetesimal accretion in combination with large-distance migration of the planet. | [
{
"version": "v1",
"created": "Fri, 16 May 2025 13:19:16 GMT"
},
{
"version": "v2",
"created": "Wed, 04 Jun 2025 21:27:38 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Ahrer",
"Eva-Maria",
""
],
[
"Gandhi",
"Siddharth",
""
],
[
"Alderson",
"Lili",
""
],
[
"Kirk",
"James",
""
],
[
"Teske",
"Johanna",
""
],
[
"Booth",
"Richard A.",
""
],
[
"McDonald",
"Catriona H.",
""
],
[
"Christie",
"Duncan A.",
""
],
[
"Claringbold",
"Alastair B.",
""
],
[
"Nealon",
"Rebecca",
""
],
[
"Panwar",
"Vatsal",
""
],
[
"Veras",
"Dimitri",
""
],
[
"Wakeford",
"Hannah R.",
""
],
[
"Wheatley",
"Peter J.",
""
],
[
"Zamyatina",
"Maria",
""
]
] |
2505.11292 | Tatsuya Nakada | Y. Abe, S. Arai, S. Araki, H. Araki, Y. Arimoto, A. Aryshev, S. Asai, R. Bajpai, T. Behnke, S. Belomestnykh, I. Bozovic, J.E. Brau, K. Buesser, P.N. Burrows, N. Catalan-Lasheras, E. Cenni, S. Chen, J. Clark, D. Delikaris, M. Demarteau, D. Denisov, S. Doebert, T. Dohmae, R. Dowd, G. Dugan, M. Egi, Y. Enomoto, A. Faus Golfe, F. Forti, K. Fujii, M. Fukuda, F. Gaede, Y. Gao, L. Garcia-Tabares, R.L. Geng, T. Goto, J. Grames, K. Hanagaki, T. Hara, K. Hara, C. Hensel, M. Hiraki, S. Hirose, T. Honma, Y. Horii, K. Inami, M. Ishino, H. Ito, M. Iwasaki, D. Jeans, O. Jinnouchi, R. Katayama, K. Kawagoe, S. Kessoku, T. Kitahara, T. Koseki, K. Krueger, T. Kubo, K. Kubo, A. Kumar, M. Kurata, M. Kuriki, A. J. Lankford, A. Latina, R. Laxdal, M. Liepe, J. List, B. List, C. Madec, J. Maeda, T.W. Markiewicz, T. Masubuchi, T. Matsumoto, K. Mazumdar, P. McBride, P.A. McIntosh, S. Michizono, T. Miura, L. Monaco, G. Moortgat-Pick, T. Mori, Y. Morikawa, H. Murayama, T. Nakada, H. Nakai, H. Nakajima, E. Nakamura, K. Nakamura, K. Nakanishi, T. Nakaya, O. Napoly, S. Narita, M. Nojiri, H. Oide, Y. Okada, T. Okugi, C. Oliver Amoros, M. Omet, W. Ootani, J. Osborne, B. Parker, J.R. Patterson, R. P\"oschl, M.E. Peskin, K. Popov, S. Posen, S. Riemann, R.A. Rimmer, A. Robson, M. C. Ross, R. Ruber, D. L. Rubin, T. Saeki, N. Saito, H. Sakai, T. Sanuki, M. Sato, S. Shanab, H. Shimizu, P. Sievers, F. Simon, N. Solyak, S. Stapnes, J. Strube, T. Suehara, R. Takahashi, T. Takahashi, G. Taylor, N. Terunuma, J. Tian, M. Titov, M. Togawa, K. Tsumura, R. Ueki, K. Umemori, P. Urquijo, Y. Ushiroda, C. Vernieri, E. Viklund, M. Vos, N. Walker, Y. Watanabe, H. Weise, A.P. White, T. Yamada, A. Yamamoto, Y. Yamamoto, M. Yamauchi, K. Yokoya, A.F. Zarnecki, J. Zhang, M. Zobov | Status of the International Linear Collider | null | null | null | null | hep-ex | http://creativecommons.org/licenses/by/4.0/ | This paper is not a proposal for a CERN future project but provides information on the International Linear Collider (ILC) considered for Japan in order to facilitate the European Strategy discussion in a global context. It describes progress to date, ongoing engineering studies, updated cost estimate for the machine at $\sqrt{s}=250~\rm GeV$ and the situation in Japan. The physics of the ILC is not presented here, but jointly for all Linear Collider projects in a separate document ``A Linear Collider Vision for the Future of Particle Physics'' submitted for the forthcoming European Strategy deliberations. | [
{
"version": "v1",
"created": "Fri, 16 May 2025 14:25:38 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 15:51:03 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Abe",
"Y.",
""
],
[
"Arai",
"S.",
""
],
[
"Araki",
"S.",
""
],
[
"Araki",
"H.",
""
],
[
"Arimoto",
"Y.",
""
],
[
"Aryshev",
"A.",
""
],
[
"Asai",
"S.",
""
],
[
"Bajpai",
"R.",
""
],
[
"Behnke",
"T.",
""
],
[
"Belomestnykh",
"S.",
""
],
[
"Bozovic",
"I.",
""
],
[
"Brau",
"J. E.",
""
],
[
"Buesser",
"K.",
""
],
[
"Burrows",
"P. N.",
""
],
[
"Catalan-Lasheras",
"N.",
""
],
[
"Cenni",
"E.",
""
],
[
"Chen",
"S.",
""
],
[
"Clark",
"J.",
""
],
[
"Delikaris",
"D.",
""
],
[
"Demarteau",
"M.",
""
],
[
"Denisov",
"D.",
""
],
[
"Doebert",
"S.",
""
],
[
"Dohmae",
"T.",
""
],
[
"Dowd",
"R.",
""
],
[
"Dugan",
"G.",
""
],
[
"Egi",
"M.",
""
],
[
"Enomoto",
"Y.",
""
],
[
"Golfe",
"A. Faus",
""
],
[
"Forti",
"F.",
""
],
[
"Fujii",
"K.",
""
],
[
"Fukuda",
"M.",
""
],
[
"Gaede",
"F.",
""
],
[
"Gao",
"Y.",
""
],
[
"Garcia-Tabares",
"L.",
""
],
[
"Geng",
"R. L.",
""
],
[
"Goto",
"T.",
""
],
[
"Grames",
"J.",
""
],
[
"Hanagaki",
"K.",
""
],
[
"Hara",
"T.",
""
],
[
"Hara",
"K.",
""
],
[
"Hensel",
"C.",
""
],
[
"Hiraki",
"M.",
""
],
[
"Hirose",
"S.",
""
],
[
"Honma",
"T.",
""
],
[
"Horii",
"Y.",
""
],
[
"Inami",
"K.",
""
],
[
"Ishino",
"M.",
""
],
[
"Ito",
"H.",
""
],
[
"Iwasaki",
"M.",
""
],
[
"Jeans",
"D.",
""
],
[
"Jinnouchi",
"O.",
""
],
[
"Katayama",
"R.",
""
],
[
"Kawagoe",
"K.",
""
],
[
"Kessoku",
"S.",
""
],
[
"Kitahara",
"T.",
""
],
[
"Koseki",
"T.",
""
],
[
"Krueger",
"K.",
""
],
[
"Kubo",
"T.",
""
],
[
"Kubo",
"K.",
""
],
[
"Kumar",
"A.",
""
],
[
"Kurata",
"M.",
""
],
[
"Kuriki",
"M.",
""
],
[
"Lankford",
"A. J.",
""
],
[
"Latina",
"A.",
""
],
[
"Laxdal",
"R.",
""
],
[
"Liepe",
"M.",
""
],
[
"List",
"J.",
""
],
[
"List",
"B.",
""
],
[
"Madec",
"C.",
""
],
[
"Maeda",
"J.",
""
],
[
"Markiewicz",
"T. W.",
""
],
[
"Masubuchi",
"T.",
""
],
[
"Matsumoto",
"T.",
""
],
[
"Mazumdar",
"K.",
""
],
[
"McBride",
"P.",
""
],
[
"McIntosh",
"P. A.",
""
],
[
"Michizono",
"S.",
""
],
[
"Miura",
"T.",
""
],
[
"Monaco",
"L.",
""
],
[
"Moortgat-Pick",
"G.",
""
],
[
"Mori",
"T.",
""
],
[
"Morikawa",
"Y.",
""
],
[
"Murayama",
"H.",
""
],
[
"Nakada",
"T.",
""
],
[
"Nakai",
"H.",
""
],
[
"Nakajima",
"H.",
""
],
[
"Nakamura",
"E.",
""
],
[
"Nakamura",
"K.",
""
],
[
"Nakanishi",
"K.",
""
],
[
"Nakaya",
"T.",
""
],
[
"Napoly",
"O.",
""
],
[
"Narita",
"S.",
""
],
[
"Nojiri",
"M.",
""
],
[
"Oide",
"H.",
""
],
[
"Okada",
"Y.",
""
],
[
"Okugi",
"T.",
""
],
[
"Amoros",
"C. Oliver",
""
],
[
"Omet",
"M.",
""
],
[
"Ootani",
"W.",
""
],
[
"Osborne",
"J.",
""
],
[
"Parker",
"B.",
""
],
[
"Patterson",
"J. R.",
""
],
[
"Pöschl",
"R.",
""
],
[
"Peskin",
"M. E.",
""
],
[
"Popov",
"K.",
""
],
[
"Posen",
"S.",
""
],
[
"Riemann",
"S.",
""
],
[
"Rimmer",
"R. A.",
""
],
[
"Robson",
"A.",
""
],
[
"Ross",
"M. C.",
""
],
[
"Ruber",
"R.",
""
],
[
"Rubin",
"D. L.",
""
],
[
"Saeki",
"T.",
""
],
[
"Saito",
"N.",
""
],
[
"Sakai",
"H.",
""
],
[
"Sanuki",
"T.",
""
],
[
"Sato",
"M.",
""
],
[
"Shanab",
"S.",
""
],
[
"Shimizu",
"H.",
""
],
[
"Sievers",
"P.",
""
],
[
"Simon",
"F.",
""
],
[
"Solyak",
"N.",
""
],
[
"Stapnes",
"S.",
""
],
[
"Strube",
"J.",
""
],
[
"Suehara",
"T.",
""
],
[
"Takahashi",
"R.",
""
],
[
"Takahashi",
"T.",
""
],
[
"Taylor",
"G.",
""
],
[
"Terunuma",
"N.",
""
],
[
"Tian",
"J.",
""
],
[
"Titov",
"M.",
""
],
[
"Togawa",
"M.",
""
],
[
"Tsumura",
"K.",
""
],
[
"Ueki",
"R.",
""
],
[
"Umemori",
"K.",
""
],
[
"Urquijo",
"P.",
""
],
[
"Ushiroda",
"Y.",
""
],
[
"Vernieri",
"C.",
""
],
[
"Viklund",
"E.",
""
],
[
"Vos",
"M.",
""
],
[
"Walker",
"N.",
""
],
[
"Watanabe",
"Y.",
""
],
[
"Weise",
"H.",
""
],
[
"White",
"A. P.",
""
],
[
"Yamada",
"T.",
""
],
[
"Yamamoto",
"A.",
""
],
[
"Yamamoto",
"Y.",
""
],
[
"Yamauchi",
"M.",
""
],
[
"Yokoya",
"K.",
""
],
[
"Zarnecki",
"A. F.",
""
],
[
"Zhang",
"J.",
""
],
[
"Zobov",
"M.",
""
]
] |
2505.12139 | Daniel Bragg | Daniel Bragg | On the logarithmic Hodge-de Rham spectral sequence for curves on K3 surfaces | Some errors corrected. Comments still welcome | null | null | null | math.AG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We show that if $X$ is a supersingular K3 surface then there exists a curve $D$ on $X$ such that the logarithmic Hodge-de Rham spectral sequence for $(X,D)$ is nondegenerate. | [
{
"version": "v1",
"created": "Sat, 17 May 2025 20:48:31 GMT"
},
{
"version": "v2",
"created": "Wed, 04 Jun 2025 19:39:33 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Bragg",
"Daniel",
""
]
] |
2505.12195 | Zihao Zhang | Chunpeng Wang, Zihao Zhang | Structural stability of supersonic spiral flows with large angular velocity for the Euler-Poisson system | arXiv admin note: substantial text overlap with arXiv:2503.15839 | null | null | null | math.AP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper concerns the structural stability of smooth cylindrically symmetric supersonic spiral flows with large angular velocity for the steady Euler-Poisson system in a concentric cylinder. We establish the existence and uniqueness of some smooth supersonic Euler-Poisson flows with nonzero angular velocity and vorticity including both cylindrical spiral flows and axisymmetric spiral flows. The deformation-curl-Poisson decomposition for the steady Euler-Poisson system is utilized to deal with the hyperbolic-elliptic mixed structure in the supersonic region. For smooth cylindrical supersonic spiral flows, the key point lies on the well-posedness of a boundary value problem for a linear second order hyperbolic-elliptic coupled system, which is achieved by finding an appropriate multiplier to obtain the important basic energy estimates. The nonlinear structural stability is established by designing a two-layer iteration and combining the estimates for the hyperbolic-elliptic system and the transport equations. For smooth axisymmetric supersonic spiral flows, we use the special structure of the steady Euler-Poisson system to derive a priori estimates of the linearized second order elliptic system, which enable us to establish the structural stability of the background supersonic flow within the class of axisymmetric flows. | [
{
"version": "v1",
"created": "Sun, 18 May 2025 02:11:29 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 02:02:51 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Wang",
"Chunpeng",
""
],
[
"Zhang",
"Zihao",
""
]
] |
2505.12240 | Junhong Fan | Daomin Cao, Junhong Fan, Guolin Qin, Jie Wan | Dynamics and leapfrogging phenomena of multiple helical vortices for 3D incompressible Euler equations | null | null | null | null | math.AP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we investigate the time evolution of multiple interacted helical vortices without swirl for the incompressible Euler equations in $\mathbb R^3$. Assuming that the initial helical symmetric vorticity is concentrated within an $\ep$ neighborhood of $N$ distinct helices with vanishing mutual distance of order $O(\frac{1}{|\ln \ep|})$, and each vortex core possesses a vorticity mass of order $O(\frac{1}{|\ln \ep|^2})$, we show that as $\ep\to 0$, the motion of the helical vortices converges to a dynamical system for positive times. Notably, in the case of two interacting helical vortices with initial mutual separation $ \frac{\rho_0}{|\ln \ep|}$, by selecting sufficiently small $\rho_0$, our analysis extends to longer timescales encompassing several periods. This result provides the first mathematical justification for the phenomena in numerical observation termed "leapfrogging of Kelvin waves" reported in e.g. [N. Hietala et al., Phys. Rev. Fluids, 2016]. | [
{
"version": "v1",
"created": "Sun, 18 May 2025 05:33:36 GMT"
},
{
"version": "v2",
"created": "Wed, 04 Jun 2025 06:01:19 GMT"
},
{
"version": "v3",
"created": "Thu, 05 Jun 2025 04:59:53 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Cao",
"Daomin",
""
],
[
"Fan",
"Junhong",
""
],
[
"Qin",
"Guolin",
""
],
[
"Wan",
"Jie",
""
]
] |
2505.12268 | Pratim Chowdhary | Pratim Chowdhary, Peter Chin, Deepernab Chakrabarty | $K$-MSHC: Unmasking Minimally Sufficient Head Circuits in Large Language Models with Experiments on Syntactic Classification Tasks | null | null | null | null | cs.CL cs.LG | http://creativecommons.org/licenses/by/4.0/ | Understanding which neural components drive specific capabilities in mid-sized language models ($\leq$10B parameters) remains a key challenge. We introduce the $(\bm{K}, \epsilon)$-Minimum Sufficient Head Circuit ($K$-MSHC), a methodology to identify minimal sets of attention heads crucial for classification tasks as well as Search-K-MSHC, an efficient algorithm for discovering these circuits. Applying our Search-K-MSHC algorithm to Gemma-9B, we analyze three syntactic task families: grammar acceptability, arithmetic verification, and arithmetic word problems. Our findings reveal distinct task-specific head circuits, with grammar tasks predominantly utilizing early layers, word problems showing pronounced activity in both shallow and deep regions, and arithmetic verification demonstrating a more distributed pattern across the network. We discover non-linear circuit overlap patterns, where different task pairs share computational components at varying levels of importance. While grammar and arithmetic share many "weak" heads, arithmetic and word problems share more consistently critical "strong" heads. Importantly, we find that each task maintains dedicated "super-heads" with minimal cross-task overlap, suggesting that syntactic and numerical competencies emerge from specialized yet partially reusable head circuits. | [
{
"version": "v1",
"created": "Sun, 18 May 2025 07:15:01 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 01:45:59 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Chowdhary",
"Pratim",
""
],
[
"Chin",
"Peter",
""
],
[
"Chakrabarty",
"Deepernab",
""
]
] |
2505.12612 | Xuhao Ren | Chuan Zhang, Xuhao Ren, Zhangcheng Huang, Jinwen Liang, Jianzong Wang, Liehuang Zhu | EPSpatial: Achieving Efficient and Private Statistical Analytics of Geospatial Data | There are some errors that need to be corrected | null | null | null | cs.CR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Geospatial data statistics involve the aggregation and analysis of location data to derive the distribution of clients within geospatial. The need for privacy protection in geospatial data analysis has become paramount due to concerns over the misuse or unauthorized access of client location information. However, existing private geospatial data statistics mainly rely on privacy computing techniques such as cryptographic tools and differential privacy, which leads to significant overhead and inaccurate results. In practical applications, geospatial data is frequently generated by mobile devices such as smartphones and IoT sensors. The continuous mobility of clients and the need for real-time updates introduce additional complexity. To address these issues, we first design \textit{spatially distributed point functions (SDPF)}, which combines a quad-tree structure with distributed point functions, allowing clients to succinctly secret-share values on the nodes of an exponentially large quad-tree. Then, we use Gray code to partition the region and combine SDPF with it to propose $\mathtt{EPSpatial}$, a scheme for accurate, efficient, and private statistical analytics of geospatial data. Moreover, considering clients' frequent movement requires continuous location updates, we leverage the region encoding property to present an efficient update algorithm.Security analysis shows that $\mathtt{EPSpatial}$ effectively protects client location privacy. Theoretical analysis and experimental results on real datasets demonstrate that $\mathtt{EPSpatial}$ reduces computational and communication overhead by at least $50\%$ compared to existing statistical schemes. | [
{
"version": "v1",
"created": "Mon, 19 May 2025 01:51:30 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 02:45:07 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Zhang",
"Chuan",
""
],
[
"Ren",
"Xuhao",
""
],
[
"Huang",
"Zhangcheng",
""
],
[
"Liang",
"Jinwen",
""
],
[
"Wang",
"Jianzong",
""
],
[
"Zhu",
"Liehuang",
""
]
] |
2505.13182 | Jianfeng Xu | Jianfeng Xu | Information Science Principles of Machine Learning: A Causal Chain Meta-Framework Based on Formalized Information Mapping | null | null | null | null | cs.LO cs.AI | http://creativecommons.org/licenses/by/4.0/ | [Objective] This study focuses on addressing the current lack of a unified formal theoretical framework in machine learning, as well as the deficiencies in interpretability and ethical safety assurance. [Methods] A formal information model is first constructed, utilizing sets of well-formed formulas to explicitly define the ontological states and carrier mappings of typical components in machine learning. Learnable and processable predicates, along with learning and processing functions, are introduced to analyze the logical deduction and constraint rules of the causal chains within models. [Results] A meta-framework for machine learning theory (MLT-MF) is established. Based on this framework, universal definitions for model interpretability and ethical safety are proposed. Furthermore, three key theorems are proved: the equivalence of model interpretability and information recoverability, the assurance of ethical safety, and the estimation of generalization error. [Limitations] The current framework assumes ideal conditions with noiseless information-enabling mappings and primarily targets model learning and processing logic in static scenarios. It does not yet address information fusion and conflict resolution across ontological spaces in multimodal or multi-agent systems. [Conclusions] This work overcomes the limitations of fragmented research and provides a unified theoretical foundation for systematically addressing the critical challenges currently faced in machine learning. | [
{
"version": "v1",
"created": "Mon, 19 May 2025 14:39:41 GMT"
},
{
"version": "v2",
"created": "Wed, 28 May 2025 06:30:45 GMT"
},
{
"version": "v3",
"created": "Fri, 30 May 2025 13:42:46 GMT"
},
{
"version": "v4",
"created": "Tue, 03 Jun 2025 11:32:24 GMT"
},
{
"version": "v5",
"created": "Thu, 05 Jun 2025 03:15:07 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Xu",
"Jianfeng",
""
]
] |
2505.13427 | Lingxiao Du | Lingxiao Du, Fanqing Meng, Zongkai Liu, Zhixiang Zhou, Ping Luo, Qiaosheng Zhang, Wenqi Shao | MM-PRM: Enhancing Multimodal Mathematical Reasoning with Scalable Step-Level Supervision | null | null | null | null | cs.AI cs.CV | http://creativecommons.org/licenses/by/4.0/ | While Multimodal Large Language Models (MLLMs) have achieved impressive progress in vision-language understanding, they still struggle with complex multi-step reasoning, often producing logically inconsistent or partially correct solutions. A key limitation lies in the lack of fine-grained supervision over intermediate reasoning steps. To address this, we propose MM-PRM, a process reward model trained within a fully automated, scalable framework. We first build MM-Policy, a strong multimodal model trained on diverse mathematical reasoning data. Then, we construct MM-K12, a curated dataset of 10,000 multimodal math problems with verifiable answers, which serves as seed data. Leveraging a Monte Carlo Tree Search (MCTS)-based pipeline, we generate over 700k step-level annotations without human labeling. The resulting PRM is used to score candidate reasoning paths in the Best-of-N inference setup and achieves significant improvements across both in-domain (MM-K12 test set) and out-of-domain (OlympiadBench, MathVista, etc.) benchmarks. Further analysis confirms the effectiveness of soft labels, smaller learning rates, and path diversity in optimizing PRM performance. MM-PRM demonstrates that process supervision is a powerful tool for enhancing the logical robustness of multimodal reasoning systems. We release all our codes and data at https://github.com/ModalMinds/MM-PRM. | [
{
"version": "v1",
"created": "Mon, 19 May 2025 17:55:08 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 05:29:41 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Du",
"Lingxiao",
""
],
[
"Meng",
"Fanqing",
""
],
[
"Liu",
"Zongkai",
""
],
[
"Zhou",
"Zhixiang",
""
],
[
"Luo",
"Ping",
""
],
[
"Zhang",
"Qiaosheng",
""
],
[
"Shao",
"Wenqi",
""
]
] |
2505.13438 | Penghui Qi | Penghui Qi, Zichen Liu, Tianyu Pang, Chao Du, Wee Sun Lee, Min Lin | Optimizing Anytime Reasoning via Budget Relative Policy Optimization | null | null | null | null | cs.LG cs.AI cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Scaling test-time compute is crucial for enhancing the reasoning capabilities of large language models (LLMs). Existing approaches typically employ reinforcement learning (RL) to maximize a verifiable reward obtained at the end of reasoning traces. However, such methods optimize only the final performance under a large and fixed token budget, which hinders efficiency in both training and deployment. In this work, we present a novel framework, AnytimeReasoner, to optimize anytime reasoning performance, which aims to improve token efficiency and the flexibility of reasoning under varying token budget constraints. To achieve this, we truncate the complete thinking process to fit within sampled token budgets from a prior distribution, compelling the model to summarize the optimal answer for each truncated thinking for verification. This introduces verifiable dense rewards into the reasoning process, facilitating more effective credit assignment in RL optimization. We then optimize the thinking and summary policies in a decoupled manner to maximize the cumulative reward. Additionally, we introduce a novel variance reduction technique, Budget Relative Policy Optimization (BRPO), to enhance the robustness and efficiency of the learning process when reinforcing the thinking policy. Empirical results in mathematical reasoning tasks demonstrate that our method consistently outperforms GRPO across all thinking budgets under various prior distributions, enhancing both training and token efficiency. | [
{
"version": "v1",
"created": "Mon, 19 May 2025 17:58:44 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 13:55:06 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Qi",
"Penghui",
""
],
[
"Liu",
"Zichen",
""
],
[
"Pang",
"Tianyu",
""
],
[
"Du",
"Chao",
""
],
[
"Lee",
"Wee Sun",
""
],
[
"Lin",
"Min",
""
]
] |
2505.13590 | Paula Denise L\'opez | Paula D. L\'opez, Francesca Fragkoudi, Sof\'ia A. Cora, Cecilia Scannapieco, R\"udiger Pakmor, Robert J. J. Grand, Facundo G\'omez and Federico Marinacci | Formation and evolution of boxy/peanut bulges in the Auriga cosmological simulations | 19 pages, 18 figures, 2 tables - Accepted for publication on MNRAS | MNRAS, Volume 540, Issue 3, July 2025, Pages from 2031 to 2048 | 10.1093/mnras/staf818 | null | astro-ph.GA | http://creativecommons.org/licenses/by/4.0/ | Boxy/peanut (b/p) or X-shaped bulges have been extensively explored with theory and numerical simulations of isolated galaxies. However, it is only recently that advances in hydrodynamical cosmological simulations have made it possible to explore b/p bulges in a cosmological setting, with much remaining to be understood about their formation and evolution. By using the Auriga magneto-hydrodynamical cosmological zoom-in simulations, we characterise the structural parameters of b/p bulges and how they form and evolve throughout cosmic history. We develop a method for estimating the b/p strength that allows us to identify the formation time and size of these structures. We find that b/p bulges in Auriga form between $\sim 1.1-1.6\, \mathrm Gyr$ after bar formation, following a `buckling' episode; some galaxies undergo multiple bucklings and events of b/p growth, with some b/p structures `dissolving' between buckling events. We find that at $z=0$, the b/p bulges have an extent of almost half the bar length. Finally, we analyse the evolution of the b/p fraction over redshift, finding that at $z=0$, two thirds of galaxies host a bar, and of these, $45$ per cent have a b/p. This b/p fraction is within the observed range at $z=0$, although on the low end as compared to some observational studies. The b/p fraction decreases to $20$ per cent at $z=0.5$, and falls to zero at $z \sim 1$; this is in line with the observed trend of declining b/p fraction with redshift. We discuss possible culprits for the apparent mismatch in b/p occurrence between observations and cosmological simulations, what causes them to form (or not) in these simulations, and what this might reveal about models of galaxy formation and evolution. | [
{
"version": "v1",
"created": "Mon, 19 May 2025 18:00:00 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"López",
"Paula D.",
""
],
[
"Fragkoudi",
"Francesca",
""
],
[
"Cora",
"Sofía A.",
""
],
[
"Scannapieco",
"Cecilia",
""
],
[
"Pakmor",
"Rüdiger",
""
],
[
"Grand",
"Robert J. J.",
""
],
[
"Gómez",
"Facundo",
""
],
[
"Marinacci",
"Federico",
""
]
] |
2505.13816 | James Glimm | James Glimm and Jarret Petrillo | Smooth Solutions of the Navier-Stokes Equation | Corrected version number in one reference. arXiv admin note: text overlap with arXiv:2410.09261 | null | null | null | math.AP math-ph math.MP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Smooth solutions of the Navier-Stokes equation with smooth but otherwise unconstrained initial conditions are constructed, to solve the Millennium fluids problem in the positive. The smooth solutions are the mean values of general weak solutions and are alternately characterized as the entropy production minimizing solutions. The construction occurs in a finite periodic cube. | [
{
"version": "v1",
"created": "Tue, 20 May 2025 01:54:56 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 17:03:34 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Glimm",
"James",
""
],
[
"Petrillo",
"Jarret",
""
]
] |
2505.13856 | Ruqin Zhou | Ruqin Zhou, Chenguang Dai, Wanshou Jiang, Yongsheng Zhang, Hanyun Wang, San Jiang | SuperMapNet for Long-Range and High-Accuracy Vectorized HD Map Construction | 13 pages, 9 figures | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Vectorized HD map is essential for autonomous driving. Remarkable work has been achieved in recent years, but there are still major issues: (1) in the generation of the BEV features, single modality-based methods are of limited perception capability, while direct concatenation-based multi-modal methods fail to capture synergies and disparities between different modalities, resulting in limited ranges with feature holes; (2) in the classification and localization of map elements, only point information is used without the consideration of element infor-mation and neglects the interaction between point information and element information, leading to erroneous shapes and element entanglement with low accuracy. To address above issues, we introduce SuperMapNet for long-range and high-accuracy vectorized HD map construction. It uses both camera images and LiDAR point clouds as input, and first tightly couple semantic information from camera images and geometric information from LiDAR point clouds by a cross-attention based synergy enhancement module and a flow-based disparity alignment module for long-range BEV feature generation. And then, local features from point queries and global features from element queries are tightly coupled by three-level interactions for high-accuracy classification and localization, where Point2Point interaction learns local geometric information between points of the same element and of each point, Element2Element interaction learns relation constraints between different elements and semantic information of each elements, and Point2Element interaction learns complement element information for its constituent points. Experiments on the nuScenes and Argoverse2 datasets demonstrate superior performances, surpassing SOTAs over 14.9/8.8 mAP and 18.5/3.1 mAP under hard/easy settings, respectively. The code is made publicly available1. | [
{
"version": "v1",
"created": "Tue, 20 May 2025 03:06:22 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 01:19:22 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Zhou",
"Ruqin",
""
],
[
"Dai",
"Chenguang",
""
],
[
"Jiang",
"Wanshou",
""
],
[
"Zhang",
"Yongsheng",
""
],
[
"Wang",
"Hanyun",
""
],
[
"Jiang",
"San",
""
]
] |
2505.14068 | Zhenyu Li | Zhenyu Li, Tianyi Shang, Pengjie Xu, Zhaojun Deng | Place Recognition Meet Multiple Modalitie: A Comprehensive Review, Current Challenges and Future Directions | 67 pages | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Place recognition is a cornerstone of vehicle navigation and mapping, which is pivotal in enabling systems to determine whether a location has been previously visited. This capability is critical for tasks such as loop closure in Simultaneous Localization and Mapping (SLAM) and long-term navigation under varying environmental conditions. In this survey, we comprehensively review recent advancements in place recognition, emphasizing three representative methodological paradigms: Convolutional Neural Network (CNN)-based approaches, Transformer-based frameworks, and cross-modal strategies. We begin by elucidating the significance of place recognition within the broader context of autonomous systems. Subsequently, we trace the evolution of CNN-based methods, highlighting their contributions to robust visual descriptor learning and scalability in large-scale environments. We then examine the emerging class of Transformer-based models, which leverage self-attention mechanisms to capture global dependencies and offer improved generalization across diverse scenes. Furthermore, we discuss cross-modal approaches that integrate heterogeneous data sources such as Lidar, vision, and text description, thereby enhancing resilience to viewpoint, illumination, and seasonal variations. We also summarize standard datasets and evaluation metrics widely adopted in the literature. Finally, we identify current research challenges and outline prospective directions, including domain adaptation, real-time performance, and lifelong learning, to inspire future advancements in this domain. The unified framework of leading-edge place recognition methods, i.e., code library, and the results of their experimental evaluations are available at https://github.com/CV4RA/SOTA-Place-Recognitioner. | [
{
"version": "v1",
"created": "Tue, 20 May 2025 08:16:37 GMT"
},
{
"version": "v2",
"created": "Thu, 22 May 2025 08:37:43 GMT"
},
{
"version": "v3",
"created": "Thu, 05 Jun 2025 12:31:11 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Li",
"Zhenyu",
""
],
[
"Shang",
"Tianyi",
""
],
[
"Xu",
"Pengjie",
""
],
[
"Deng",
"Zhaojun",
""
]
] |
2505.14767 | Seth Musser | Meng Cheng, Seth Musser, Amir Raz, Nathan Seiberg, T. Senthil | Ordering the topological order in the fractional quantum Hall effect | 83 pages, 4 figures, 2 tables; v2 contains updated references and minor edits | null | null | null | cond-mat.str-el cond-mat.mes-hall hep-th | http://creativecommons.org/licenses/by/4.0/ | We discuss the possible topological order/topological quantum field theory of different quantum Hall systems. Given the value of the Hall conductivity, we constrain the global symmetry of the low-energy theory and its anomaly. Specifically, the one-form global symmetry and its anomaly are presented as the organizing principle of these systems. This information is powerful enough to lead to a unique minimal topological order (or a small number of minimal topological orders). Almost all of the known experimentally discovered topological orders are these minimal theories. Since this work is interdisciplinary, we made a special effort to relate to people with different backgrounds by providing translations between different perspectives. | [
{
"version": "v1",
"created": "Tue, 20 May 2025 18:00:00 GMT"
},
{
"version": "v2",
"created": "Wed, 04 Jun 2025 18:23:16 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Cheng",
"Meng",
""
],
[
"Musser",
"Seth",
""
],
[
"Raz",
"Amir",
""
],
[
"Seiberg",
"Nathan",
""
],
[
"Senthil",
"T.",
""
]
] |
2505.14971 | Prashanth Vijayaraghavan | Prashanth Vijayaraghavan, Soroush Vosoughi, Lamogha Chiazor, Raya Horesh, Rogerio Abreu de Paula, Ehsan Degan, Vandana Mukherjee | DECASTE: Unveiling Caste Stereotypes in Large Language Models through Multi-Dimensional Bias Analysis | 7 (content pages) + 2 (reference pages) + 5 (Appendix pages), 5 figures, 6 Tables, IJCAI 2025 | null | null | null | cs.CL cs.CY | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Recent advancements in large language models (LLMs) have revolutionized natural language processing (NLP) and expanded their applications across diverse domains. However, despite their impressive capabilities, LLMs have been shown to reflect and perpetuate harmful societal biases, including those based on ethnicity, gender, and religion. A critical and underexplored issue is the reinforcement of caste-based biases, particularly towards India's marginalized caste groups such as Dalits and Shudras. In this paper, we address this gap by proposing DECASTE, a novel, multi-dimensional framework designed to detect and assess both implicit and explicit caste biases in LLMs. Our approach evaluates caste fairness across four dimensions: socio-cultural, economic, educational, and political, using a range of customized prompting strategies. By benchmarking several state-of-the-art LLMs, we reveal that these models systematically reinforce caste biases, with significant disparities observed in the treatment of oppressed versus dominant caste groups. For example, bias scores are notably elevated when comparing Dalits and Shudras with dominant caste groups, reflecting societal prejudices that persist in model outputs. These results expose the subtle yet pervasive caste biases in LLMs and emphasize the need for more comprehensive and inclusive bias evaluation methodologies that assess the potential risks of deploying such models in real-world contexts. | [
{
"version": "v1",
"created": "Tue, 20 May 2025 23:19:13 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 00:42:04 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Vijayaraghavan",
"Prashanth",
""
],
[
"Vosoughi",
"Soroush",
""
],
[
"Chiazor",
"Lamogha",
""
],
[
"Horesh",
"Raya",
""
],
[
"de Paula",
"Rogerio Abreu",
""
],
[
"Degan",
"Ehsan",
""
],
[
"Mukherjee",
"Vandana",
""
]
] |
2505.15080 | Sergey Pankov | Sergey Pankov, Georges Harik | SUS backprop: linear backpropagation algorithm for long inputs in transformers | 21 pages, 9 figures; main results unchanged, Fig.5 updated, some text rearranged | null | null | null | cs.LG cs.AI cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | It is straightforward to design an unbiased gradient estimator that stochastically cuts the backpropagation flow through any part of a computational graph. By cutting the parts that have little effect on the computation, one can potentially save a significant amount of backpropagation computation in exchange for a minimal increase in the stochastic gradient variance, in some situations. Such a situation occurs in the attention mechanism of the transformer architecture. For long sequences, attention becomes the limiting factor, as its compute requirements increase quadratically with sequence length $n$. At the same time, most attention weights become very small, as most attention heads tend to connect a given token with only a small fraction of other tokens in the sequence. These weights become promising targets for cutting backpropagation. We propose a simple probabilistic rule controlled by a single parameter $c$ that cuts back-propagation through most attention weights, leaving at most $c$ interactions per token per attention head. This brings a factor of $c/n$ reduction in the compute required for the attention backpropagation, turning it from quadratic $O(n^2)$ to linear complexity $O(nc)$. We have empirically verified that, for a typical transformer model, cutting about $99\%$ of the attention gradient flow (i.e. choosing $c \sim 25-30$) results in relative gradient variance increase of only about $1\%$ for $n \sim 2000$, and it decreases with $n$. This approach is amenable to efficient sparse matrix implementation, thus being promising for making the cost of a backward pass negligible relative to the cost of a forward pass when training a transformer model on long sequences. | [
{
"version": "v1",
"created": "Wed, 21 May 2025 04:00:38 GMT"
},
{
"version": "v2",
"created": "Wed, 04 Jun 2025 19:53:25 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Pankov",
"Sergey",
""
],
[
"Harik",
"Georges",
""
]
] |
2505.15724 | Johannes Schlenk | Lara Bohnenblust, Harald Ita, Manfred Kraus, Johannes Schlenk | Gravitational Bremsstrahlung in Black-Hole Scattering at $\mathcal{O}(G^3)$: Quadratic-in-Spin Effects | 45 pages, 8 figures, ancillary files; v2: added references, minor text edits, comments on SUSY property | null | null | null | hep-th gr-qc | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We are employing a supersymmetric variant of the worldline quantum field theory (WQFT) formalism to compute the far-field momentum-space gravitational waveform emitted during the scattering of two spinning black holes at next-to-leading order (NLO) in the post-Minkowskian expansion. Our results are accurate up to quadratic-in-spin contributions, which means we report for the very first time the waveform observable at the order $\mathcal{O}(G^3\mathcal{S}^2)$. Our computation is based on mapping $n$-body tree-level amplitudes in such a way that we can obtain the $(n-2)$-loop two-body waveform integrand. We discuss in detail this procedure and highlight the similarity of the resulting structures with those obtained in the scattering-amplitude approach. As a by product of our computational approach, we also obtain, for the first time, the leading-order waveform for three-body scattering of spinning black holes. We validated our results in various ways but most notably, we find exact agreement for the NLO waveform integrand obtained from the WQFT and the classical limit of scattering amplitudes in QFT. | [
{
"version": "v1",
"created": "Wed, 21 May 2025 16:31:25 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 15:09:08 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Bohnenblust",
"Lara",
""
],
[
"Ita",
"Harald",
""
],
[
"Kraus",
"Manfred",
""
],
[
"Schlenk",
"Johannes",
""
]
] |
2505.15862 | Long Wang | Long Wang, Jiongzhi Zheng, Zhengda Xiong, ChuMin Li, Kun He | Bandit based Dynamic Candidate Edge Selection in Solving Traveling Salesman Problems | null | null | null | null | cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Algorithms designed for routing problems typically rely on high-quality candidate edges to guide their search, aiming to reduce the search space and enhance the search efficiency. However, many existing algorithms, like the classical Lin-Kernighan-Helsgaun (LKH) algorithm for the Traveling Salesman Problem (TSP), often use predetermined candidate edges that remain static throughout local searches. This rigidity could cause the algorithm to get trapped in local optima, limiting its potential to find better solutions. To address this issue, we propose expanding the candidate sets to include other promising edges, providing them an opportunity for selection. Specifically, we incorporate multi-armed bandit models to dynamically select the most suitable candidate edges in each iteration, enabling LKH to make smarter choices and lead to improved solutions. Extensive experiments on multiple TSP benchmarks show the excellent performance of our method. Moreover, we employ this bandit-based method to LKH-3, an extension of LKH tailored for solving various TSP variant problems, and our method also significantly enhances LKH-3's performance across typical TSP variants. | [
{
"version": "v1",
"created": "Wed, 21 May 2025 06:11:00 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 06:16:06 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Wang",
"Long",
""
],
[
"Zheng",
"Jiongzhi",
""
],
[
"Xiong",
"Zhengda",
""
],
[
"Li",
"ChuMin",
""
],
[
"He",
"Kun",
""
]
] |
2505.16142 | Shicheng Xu | Shicheng Xu, Liang Pang, Yunchang Zhu, Jia Gu, Zihao Wei, Jingcheng Deng, Feiyang Pan, Huawei Shen, Xueqi Cheng | Distilling the Implicit Multi-Branch Structure in LLMs' Reasoning via Reinforcement Learning | 15 pages | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Distilling reasoning paths from teacher to student models via supervised fine-tuning (SFT) provides a shortcut for improving the reasoning ability of smaller Large Language Models (LLMs). However, the reasoning paths generated by teacher models often reflect only surface-level traces of their underlying authentic reasoning. Insights from cognitive neuroscience suggest that authentic reasoning involves a complex interweaving between meta-reasoning (which selects appropriate sub-problems from multiple candidates) and solving (which addresses the sub-problem). This implies authentic reasoning has an implicit multi-branch structure. Supervised fine-tuning collapses this rich structure into a flat sequence of token prediction in the teacher's reasoning path, preventing effective distillation of this structure to students. To address this limitation, we propose RLKD, a reinforcement learning (RL)-based distillation framework guided by a novel Generative Structure Reward Model (GSRM). Our GSRM converts reasoning paths into multiple meta-reasoning-solving steps and computes rewards to measure structural alignment between student and teacher reasoning. RLKD combines this reward with RL, enabling student LLMs to internalize the teacher's implicit multi-branch reasoning structure rather than merely mimicking fixed output paths. Experiments show RLKD surpasses standard SFT-RL pipelines even when trained on 0.1% of data under an RL-only regime, unlocking greater student reasoning potential than SFT-based distillation. | [
{
"version": "v1",
"created": "Thu, 22 May 2025 02:36:36 GMT"
},
{
"version": "v2",
"created": "Thu, 05 Jun 2025 08:57:42 GMT"
}
] | 2025-06-06T00:00:00 | [
[
"Xu",
"Shicheng",
""
],
[
"Pang",
"Liang",
""
],
[
"Zhu",
"Yunchang",
""
],
[
"Gu",
"Jia",
""
],
[
"Wei",
"Zihao",
""
],
[
"Deng",
"Jingcheng",
""
],
[
"Pan",
"Feiyang",
""
],
[
"Shen",
"Huawei",
""
],
[
"Cheng",
"Xueqi",
""
]
] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.