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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
2503.07192 | Marco Faroni | Cesare Tonola, Marco Faroni, Saeed Abdolshah, Mazin Hamad, Sami Haddadin, Nicola Pedrocchi, Manuel Beschi | Reactive and Safety-Aware Path Replanning for Collaborative Applications | Submitted to IEEE | null | null | null | cs.RO | http://creativecommons.org/licenses/by/4.0/ | This paper addresses motion replanning in human-robot collaborative scenarios, emphasizing reactivity and safety-compliant efficiency. While existing human-aware motion planners are effective in structured environments, they often struggle with unpredictable human behavior, leading to safety measures that limit robot performance and throughput. In this study, we combine reactive path replanning and a safety-aware cost function, allowing the robot to adjust its path to changes in the human state. This solution reduces the execution time and the need for trajectory slowdowns without sacrificing safety. Simulations and real-world experiments show the method's effectiveness compared to standard human-robot cooperation approaches, with efficiency enhancements of up to 60\%. | [
{
"version": "v1",
"created": "Mon, 10 Mar 2025 11:22:33 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 10:09:30 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Tonola",
"Cesare",
""
],
[
"Faroni",
"Marco",
""
],
[
"Abdolshah",
"Saeed",
""
],
[
"Hamad",
"Mazin",
""
],
[
"Haddadin",
"Sami",
""
],
[
"Pedrocchi",
"Nicola",
""
],
[
"Beschi",
"Manuel",
""
]
] |
2503.07734 | Vladyslav Pushenko | Vladyslav Pushenko, Simone Scollo, Patrice Meunier, Emmanuel Villermaux and J\"org Schumacher | Diffuselet method for three-dimensional turbulent mixing of a cloudy air filament | null | null | null | null | physics.flu-dyn | http://creativecommons.org/licenses/by/4.0/ | The mixing properties of vapor content, temperature and particle fields are of paramount importance in cloud turbulence as they pertain to essential processes, such as cloud water droplet evaporation and entrainment. Our study examines the mixing of a single cloudy air (which implies droplet-laden) filament with its clear air environment, a characteristic process at the cloud edge, in two ways. The first consists of three-dimensional combined Euler-Lagrangian direct numerical simulations which describe the scalar supersaturation as an Eulerian field and the individual cloud water droplets as an ensemble of Lagrangian tracers. The second way builds on the recently developed diffuselet method, a kinematic Lagrangian framework that decomposes a scalar filament into a collection of small sections subject to deformation by local stirring and cross-sheet diffusion. The Schmidt number is Sc=0.7. The entrainment process causes a deformation of the supersaturated cloud filament in combination with diffusion until the system reaches a well-mixed equilibrium state, which implies for the present configuration that all droplets are evaporated. We compare the time dependence of the mean square and probability density function of the supersaturation field. For the initial period of the mixing process they agree very well; at later stages deviations caused by non-zero mean of the conserved scalar are observed. For the cases including cloud water droplets, we investigate also the impact of droplet number density and condensation growth response. Turbulence causes deviations from the d^2-law similar to recent experiments in sprays. A simulation at a Schmidt number that is by a factor of 100 larger than in clouds improves the agreement between simulation and diffuselet method significantly. The latter result promotes the diffuselet framework as an efficient parametrization for turbulent high-Sc... | [
{
"version": "v1",
"created": "Mon, 10 Mar 2025 18:00:12 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 07:16:20 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Pushenko",
"Vladyslav",
""
],
[
"Scollo",
"Simone",
""
],
[
"Meunier",
"Patrice",
""
],
[
"Villermaux",
"Emmanuel",
""
],
[
"Schumacher",
"Jörg",
""
]
] |
2503.08099 | Runxi Cheng | Runxi Cheng, Feng Xiong, Yongxian Wei, Wanyun Zhu, Chun Yuan | Whoever Started the Interference Should End It: Guiding Data-Free Model Merging via Task Vectors | 23 pages, 13 figures, 12 tables | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Model merging seeks to integrate task-specific expert models into a unified architecture while preserving multi-task generalization capabilities, yet parameter interference between constituent models frequently induces performance degradation. Although prior work has explored many merging strategies, resolving interference without additional data for retraining or test-time computation remains challenging. In this paper, we theoretically demonstrate that the task vectors of the linear layer constitute an approximate linear subspace for its corresponding input. Therefore, we can minimize interference under the guidance of task vectors. Based on this insight, we propose \textbf{WUDI-Merging} (\textbf{W}hoever started the interference sho\textbf{U}ld en\textbf{D} \textbf{I}t), a simple yet effective model merging method that eliminates interference without any additional data or rescaling coefficients. Comprehensive empirical evaluations across vision and language benchmarks demonstrate our method's superiority, achieving state-of-the-art performance in data-free model merging scenarios (average 10.9\% improvement versus baseline methods) while even outperforming mainstream test-time adaptation approaches by 3.3\%, and only very few computing resources are required. The code will be publicly available soon. | [
{
"version": "v1",
"created": "Tue, 11 Mar 2025 07:01:35 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 05:50:16 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Cheng",
"Runxi",
""
],
[
"Xiong",
"Feng",
""
],
[
"Wei",
"Yongxian",
""
],
[
"Zhu",
"Wanyun",
""
],
[
"Yuan",
"Chun",
""
]
] |
2503.08430 | Paul Haubenwallner | Paul Haubenwallner and Matthias Heller | Fully numerical Hartree-Fock calculations for atoms and small molecules with quantics tensor trains | 27 pages, 16 figures | Electron. Struct. 7 025006 (2025) | 10.1088/2516-1075/add9f8 | null | physics.chem-ph quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present a fully numerical framework for the optimization of molecule-specific quantum chemical basis functions within the quantics tensor train format using a finite-difference scheme. The optimization is driven by solving the Hartree-Fock equations (HF) with the density-matrix renormalization group (DMRG) algorithm on Cartesian grids that are iteratively refined. In contrast to the standard way of tackling the mean-field problem by expressing the molecular orbitals as linear combinations of atomic orbitals (LCAO) our method only requires as much basis functions as there are electrons within the system. Benchmark calculations for atoms and molecules with up to ten electrons show excellent agreement with LCAO calculations with large basis sets supporting the validity of the tensor network approach. Our work therefore offers a promising alternative to well-established HF-solvers and could pave the way to define highly accurate, fully numerical, molecule-adaptive basis sets, which, in the future, could lead to benefits for post-HF calculations. | [
{
"version": "v1",
"created": "Tue, 11 Mar 2025 13:44:23 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 13:25:03 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Haubenwallner",
"Paul",
""
],
[
"Heller",
"Matthias",
""
]
] |
2503.09756 | Elsa Prada | Carlos Pay\'a, Ram\'on Aguado, Pablo San-Jose, Elsa Prada | Josephson effect and critical currents in trivial and topological full-shell hybrid nanowires | 15 pages, 4 figures. Final version | Physical Review B 111, 235420 (2025) | 10.1103/8mzs-dx7h | null | cond-mat.mes-hall | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We perform microscopic numerical simulations of the Josephson effect through short junctions between two full-shell hybrid nanowires, comprised of a semiconductor core fully wrapped by a thin superconductor shell, both in the trivial and topological regimes. We explore the behavior of the current-phase relation and the critical current $I_c$ as a function of a threading flux for different models of the semiconductor core and different transparencies of the weak link. We find that $I_c$ is modulated with flux due to the Little-Parks (LP) effect and displays a characteristic skewness towards large fluxes within non-zero LP lobes, which is inherited from the skewness of a peculiar kind of subgap states known as Caroli-de Gennes-Matricon (CdGM) analogs. The appearance of Majorana zero modes at the junction in the topological phase is revealed in $I_c$ as fin-shaped peaks that stand out from the background at low junction transparencies. The competition between CdGMs of opposite electron- and hole-like character produces steps and dips in $I_c$. A rich phenomenology results, which includes 0-, $\pi$- and $\phi$-junction behaviors depending on the charge distribution across the wire core and the junction transparency. | [
{
"version": "v1",
"created": "Wed, 12 Mar 2025 18:56:37 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 10:24:30 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Payá",
"Carlos",
""
],
[
"Aguado",
"Ramón",
""
],
[
"San-Jose",
"Pablo",
""
],
[
"Prada",
"Elsa",
""
]
] |
2503.09760 | Bhabani Sankar Tripathy | Bhabani Sankar Tripathy, Nilmani Mathur, M. Padmanath | $bb\bar u\bar d$ and $bs\bar u\bar d$ tetraquarks from lattice QCD using two-meson and diquark-antidiquark variational basis | v3: 18 pages, 15 figures, 7 tables, Several text and tables corrections to improve clarity. Published in Phys.Rev.D. Results and conclusions unchanged | Phys.Rev.D 111 (2025) 11, 114504 | 10.1103/PhysRevD.111.114504 | TIFR/TH/25-6 | hep-lat hep-ex hep-ph | http://creativecommons.org/licenses/by/4.0/ | We present a lattice QCD investigation of isoscalar tetraquark systems involving bottom quarks with explicit flavor content $bb\bar{u}\bar{d}$ and $bs\bar{u}\bar{d}$. In the doubly bottom sector, the study focuses on axialvector $J^P=1^+$ quantum numbers, whereas in the $bs\bar{u}\bar{d}$ channel both axial vector $J^P=1^+$ and scalar $J^P=0^+$ quantum numbers are investigated in search of signatures for possible tetraquark bound states. The calculations are performed on four ensembles with dynamical quark fields up to the charm quark generated by the MILC Collaboration, with lattice spacings ranging from approximately 0.058 fm to 0.12 fm, and at different values of the valence light quark mass $m_{u/d}$, corresponding to pseudoscalar meson masses, $M_{ps}$=0.5, 0.6 and 0.7 GeV. The energy eigenvalues in the finite volume are determined by applying a variational procedure to correlation matrices constructed from two-meson interpolating operators and diquark-antidiquark operators. Continuum extrapolated elastic $S$-wave scattering amplitudes of $BB^*$, $KB^*$ and $KB$ are extracted from the ground state eigenenergies following a finite-volume analysis \'a la L\"uscher. The chiral and continuum extrapolated binding energy estimates for the isoscalar axialvector doubly bottom tetraquark $T_{bb}$ from the extracted elastic $BB^*$ $S$-wave scattering amplitudes is found to be $\Delta E_{T_{bb}}(1^+)=-116(^{+30}_{-36})$ MeV. In the $bs\bar{u}\bar{d}$, no statistically significant deviations were observed in the ground state energies from the respective elastic threshold energies, leading to no conclusive evidence for any bound states. | [
{
"version": "v1",
"created": "Wed, 12 Mar 2025 19:00:10 GMT"
},
{
"version": "v2",
"created": "Wed, 19 Mar 2025 13:03:21 GMT"
},
{
"version": "v3",
"created": "Wed, 11 Jun 2025 06:44:06 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Tripathy",
"Bhabani Sankar",
""
],
[
"Mathur",
"Nilmani",
""
],
[
"Padmanath",
"M.",
""
]
] |
2503.10873 | Pedro Pessoa | Pedro Pessoa, Paul Campitelli, Douglas P. Shepherd, S. Banu Ozkan, Steve Press\'e | Mamba time series forecasting with uncertainty quantification | null | null | null | null | stat.ML cs.LG nlin.CD | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | State space models, such as Mamba, have recently garnered attention in time series forecasting due to their ability to capture sequence patterns. However, in electricity consumption benchmarks, Mamba forecasts exhibit a mean error of approximately 8\%. Similarly, in traffic occupancy benchmarks, the mean error reaches 18\%. This discrepancy leaves us to wonder whether the prediction is simply inaccurate or falls within error given spread in historical data. To address this limitation, we propose a method to quantify the predictive uncertainty of Mamba forecasts. Here, we propose a dual-network framework based on the Mamba architecture for probabilistic forecasting, where one network generates point forecasts while the other estimates predictive uncertainty by modeling variance. We abbreviate our tool, Mamba with probabilistic time series forecasting, as Mamba-ProbTSF and the code for its implementation is available on GitHub (https://github.com/PessoaP/Mamba-ProbTSF). Evaluating this approach on synthetic and real-world benchmark datasets, we find Kullback-Leibler divergence between the learned distributions and the data--which, in the limit of infinite data, should converge to zero if the model correctly captures the underlying probability distribution--reduced to the order of $10^{-3}$ for synthetic data and $10^{-1}$ for real-world benchmark, demonstrating its effectiveness. We find that in both the electricity consumption and traffic occupancy benchmark, the true trajectory stays within the predicted uncertainty interval at the two-sigma level about 95\% of the time. We end with a consideration of potential limitations, adjustments to improve performance, and considerations for applying this framework to processes for purely or largely stochastic dynamics where the stochastic changes accumulate, as observed for example in pure Brownian motion or molecular dynamics trajectories. | [
{
"version": "v1",
"created": "Thu, 13 Mar 2025 20:39:38 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 16:50:28 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Pessoa",
"Pedro",
""
],
[
"Campitelli",
"Paul",
""
],
[
"Shepherd",
"Douglas P.",
""
],
[
"Ozkan",
"S. Banu",
""
],
[
"Pressé",
"Steve",
""
]
] |
2503.11052 | Jie Li | Zi-Xu Lu, Huai-Bing Zhu, Xuan Zuo, Jie Li | Preparing magnonic non-Gaussian states by adding a single magnon onto Gaussian states | to appear in PR Research | Phys. Rev. Research 7, 023242 (2025) | 10.1103/jr6g-dcnp | null | quant-ph cond-mat.mes-hall physics.optics | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Quantum magnonics based on YIG spheres provides a new arena for observing macroscopic quantum states. Here we propose to prepare two kinds of non-Gaussian magnonic states by adding a single magnon onto two Gaussian states, namely, coherent and thermal states. We adopt an optomagnonic system of a YIG sphere and use fast optical pulses to weakly activate the magnon-induced Stokes scattering. Conditioned on the detection of a polarized single photon, a single magnon can be added onto an initial Gaussian state. We use a microwave cavity to prepare an initial magnon coherent state and finally read out the generated single-magnon added coherent or thermal state. Both the non-Gaussian states of a magnon mode in a large-size YIG sphere are macroscopic quantum states, which exhibit many nonclassical properties, such as sub-Poissonian statistics, quadrature squeezing, and a negative Wigner function. Further, both states show a smooth transition from a quantum state to a classical state by changing the coherent or thermal magnon excitations and thus can be used for the fundamental study of the quantum-to-classical transition. | [
{
"version": "v1",
"created": "Fri, 14 Mar 2025 03:44:22 GMT"
},
{
"version": "v2",
"created": "Fri, 30 May 2025 01:29:46 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Lu",
"Zi-Xu",
""
],
[
"Zhu",
"Huai-Bing",
""
],
[
"Zuo",
"Xuan",
""
],
[
"Li",
"Jie",
""
]
] |
2503.11314 | Xinyu Tang | Xinyu Tang, Xiaolei Wang, Zhihao Lv, Yingqian Min, Wayne Xin Zhao, Binbin Hu, Ziqi Liu, Zhiqiang Zhang | Unlocking General Long Chain-of-Thought Reasoning Capabilities of Large Language Models via Representation Engineering | ACL 2025 | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recent advancements in long chain-of-thoughts(long CoTs) have significantly improved the reasoning capabilities of large language models(LLMs). Existing work finds that the capability of long CoT reasoning can be efficiently elicited by tuning on only a few examples and can easily transfer to other tasks. This motivates us to investigate whether long CoT reasoning is a general capability for LLMs. In this work, we conduct an empirical analysis for this question from the perspective of representation. We find that LLMs do encode long CoT reasoning as a general capability, with a clear distinction from vanilla CoTs. Furthermore, domain-specific representations are also required for the effective transfer of long CoT reasoning. Inspired by these findings, we propose GLoRE, a novel representation engineering method to unleash the general long CoT reasoning capabilities of LLMs. Extensive experiments demonstrate the effectiveness and efficiency of GLoRE in both in-domain and cross-domain scenarios. | [
{
"version": "v1",
"created": "Fri, 14 Mar 2025 11:30:37 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 03:55:09 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Tang",
"Xinyu",
""
],
[
"Wang",
"Xiaolei",
""
],
[
"Lv",
"Zhihao",
""
],
[
"Min",
"Yingqian",
""
],
[
"Zhao",
"Wayne Xin",
""
],
[
"Hu",
"Binbin",
""
],
[
"Liu",
"Ziqi",
""
],
[
"Zhang",
"Zhiqiang",
""
]
] |
2503.11544 | Parsa Rahimi Noshanagh | Parsa Rahimi, Damien Teney, Sebastien Marcel | AugGen: Synthetic Augmentation Can Improve Discriminative Models | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | The increasing reliance on large-scale datasets in machine learning poses significant privacy and ethical challenges, particularly in sensitive domains such as face recognition (FR). Synthetic data generation offers a promising alternative; however, most existing methods depend heavily on external datasets or pre-trained models, increasing complexity and resource demands. In this paper, we introduce AugGen, a self-contained synthetic augmentation technique. AugGen strategically samples from a class-conditional generative model trained exclusively on the target FR dataset, eliminating the need for external resources. Evaluated across 8 FR benchmarks, including IJB-C and IJB-B, our method achieves 1-12% performance improvements, outperforming models trained solely on real data and surpassing state-of-the-art synthetic data generation approaches, while using less real data. Notably, these gains often exceed those from architectural modifications, underscoring the value of synthetic augmentation in data-limited scenarios. Our findings demonstrate that carefully integrated synthetic data can both mitigate privacy constraints and substantially enhance discriminative performance in face recognition. Paper website: https://parsa-ra.github.io/auggen/. | [
{
"version": "v1",
"created": "Fri, 14 Mar 2025 16:10:21 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 12:32:30 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Rahimi",
"Parsa",
""
],
[
"Teney",
"Damien",
""
],
[
"Marcel",
"Sebastien",
""
]
] |
2503.11814 | Emanuele Roberto Nocera | Juan Cruz-Martinez, Toon Hasenack, Felix Hekhorn, Giacomo Magni, Emanuele R. Nocera, Tanjona R. Rabemananjara, Juan Rojo, Tanishq Sharma, Gijs van Seeventer | NNPDFpol2.0: a global determination of polarised PDFs and their uncertainties at next-to-next-to-leading order | 42 pages, PDF sets available from the NNPDF web page and from LHAPDF; version accepted for publication in JHEP | null | null | null | hep-ph hep-ex nucl-ex nucl-th | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present NNPDFpol2.0, a new set of collinear helicity parton distribution functions (PDFs) of the proton based on legacy measurements of structure functions in inclusive neutral-current longitudinally polarised deep-inelastic scattering (DIS), and of W -boson, single-inclusive, and di-jet production asymmetries in longitudinally polarised proton-proton collisions. The determination is accurate to next-to-next-to-leading order in the strong coupling, and includes heavy quark mass corrections in the analysis of DIS data. Uncertainties due to missing higher-order corrections are systematically incorporated by means of a covariance matrix determined by scale variations. NNPDFpol2.0 is based on a machine learning methodology, that makes use of Monte Carlo sampling for the representation of uncertainties into PDFs, of a neural network for the parametrisation of PDFs, of stochastic gradient descent for the optimisation of PDF parameters, and of hyperoptimisation for the selection of the best fitting model. We study the impact on PDFs of higher-order corrections, of the positivity constraint, and of the data. We demonstrate two phenomenological applications of NNPDFpol2.0, specifically the determination of the proton spin fraction carried by gluons and quarks, and of theoretical predictions for single-hadron production in longitudinally polarised DIS and proton-proton collisions. | [
{
"version": "v1",
"created": "Fri, 14 Mar 2025 19:00:01 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 09:13:45 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Cruz-Martinez",
"Juan",
""
],
[
"Hasenack",
"Toon",
""
],
[
"Hekhorn",
"Felix",
""
],
[
"Magni",
"Giacomo",
""
],
[
"Nocera",
"Emanuele R.",
""
],
[
"Rabemananjara",
"Tanjona R.",
""
],
[
"Rojo",
"Juan",
""
],
[
"Sharma",
"Tanishq",
""
],
[
"van Seeventer",
"Gijs",
""
]
] |
2503.12066 | Yuetong Yu | Yuetong Yu, Ruiyang Ge, Ilker Hacihaliloglu, Alexander Rauscher, Roger Tam, Sophia Frangou | Dataset Properties Shape the Success of Neuroimaging-Based Patient Stratification: A Benchmarking Analysis Across Clustering Algorithms | null | null | null | null | cs.LG q-bio.NC q-bio.QM | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Background: Data driven stratification of patients into biologically informed subtypes holds promise for precision neuropsychiatry, yet neuroimaging-based clustering methods often fail to generalize across cohorts. While algorithmic innovations have focused on model complexity, the role of underlying dataset characteristics remains underexplored. We hypothesized that cluster separation, size imbalance, noise, and the direction and magnitude of disease-related effects in the input data critically determine both within-algorithm accuracy and reproducibility. Methods: We evaluated 4 widely used stratification algorithms, HYDRA, SuStaIn, SmileGAN, and SurrealGAN, on a suite of synthetic brain-morphometry cohorts derived from the Human Connectome Project Young Adult dataset. Three global transformation patterns were applied to 600 pseudo-patients against 508 controls, followed by 4 within-dataset variations varying cluster count (k=2-6), overlap, and effect magnitude. Algorithm performance was quantified by accuracy in recovering the known ground-truth clusters. Results: Across 122 synthetic scenarios, data complexity consistently outweighed algorithm choice in predicting stratification success. Well-separated clusters yielded high accuracy for all methods, whereas overlapping, unequal-sized, or subtle effects reduced accuracy by up to 50%. SuStaIn could not scale beyond 17 features, HYDRA's accuracy varied unpredictably with data heterogeneity. SmileGAN and SurrealGAN maintained robust pattern detection but did not assign discrete cluster labels to individuals. Conclusions: The study results demonstrate the impact of statistical properties of input data across algorithms and highlight the need for using realistic dataset distributions when new algorithms are being developed and suggest greater focus on data-centric strategies that actively shape and standardize the input distributions. | [
{
"version": "v1",
"created": "Sat, 15 Mar 2025 09:44:00 GMT"
},
{
"version": "v2",
"created": "Tue, 10 Jun 2025 22:00:59 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Yu",
"Yuetong",
""
],
[
"Ge",
"Ruiyang",
""
],
[
"Hacihaliloglu",
"Ilker",
""
],
[
"Rauscher",
"Alexander",
""
],
[
"Tam",
"Roger",
""
],
[
"Frangou",
"Sophia",
""
]
] |
2503.12348 | Mo Zhou | Mo Zhou, Jianwei Wang, Xuanmeng Zhang, Dylan Campbell, Kai Wang, Long Yuan, Wenjie Zhang, Xuemin Lin | ProbDiffFlow: An Efficient Learning-Free Framework for Probabilistic Single-Image Optical Flow Estimation | 18 pages, 13 figures, accepted by Frontiers of Computer Science (FCS) | null | 10.1007/s11704-025-50259-6 | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper studies optical flow estimation, a critical task in motion analysis with applications in autonomous navigation, action recognition, and film production. Traditional optical flow methods require consecutive frames, which are often unavailable due to limitations in data acquisition or real-world scene disruptions. Thus, single-frame optical flow estimation is emerging in the literature. However, existing single-frame approaches suffer from two major limitations: (1) they rely on labeled training data, making them task-specific, and (2) they produce deterministic predictions, failing to capture motion uncertainty. To overcome these challenges, we propose ProbDiffFlow, a training-free framework that estimates optical flow distributions from a single image. Instead of directly predicting motion, ProbDiffFlow follows an estimation-by-synthesis paradigm: it first generates diverse plausible future frames using a diffusion-based model, then estimates motion from these synthesized samples using a pre-trained optical flow model, and finally aggregates the results into a probabilistic flow distribution. This design eliminates the need for task-specific training while capturing multiple plausible motions. Experiments on both synthetic and real-world datasets demonstrate that ProbDiffFlow achieves superior accuracy, diversity, and efficiency, outperforming existing single-image and two-frame baselines. | [
{
"version": "v1",
"created": "Sun, 16 Mar 2025 04:07:51 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 08:56:10 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Zhou",
"Mo",
""
],
[
"Wang",
"Jianwei",
""
],
[
"Zhang",
"Xuanmeng",
""
],
[
"Campbell",
"Dylan",
""
],
[
"Wang",
"Kai",
""
],
[
"Yuan",
"Long",
""
],
[
"Zhang",
"Wenjie",
""
],
[
"Lin",
"Xuemin",
""
]
] |
2503.13323 | Pedro H. C. Sant'Anna | Andrew Baker, Brantly Callaway, Scott Cunningham, Andrew Goodman-Bacon, Pedro H. C. Sant'Anna | Difference-in-Differences Designs: A Practitioner's Guide | null | null | null | null | econ.EM stat.ME | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Difference-in-differences (DiD) is arguably the most popular quasi-experimental research design. Its canonical form, with two groups and two periods, is well-understood. However, empirical practices can be ad hoc when researchers go beyond that simple case. This article provides an organizing framework for discussing different types of DiD designs and their associated DiD estimators. It discusses covariates, weights, handling multiple periods, and staggered treatments. The organizational framework, however, applies to other extensions of DiD methods as well. | [
{
"version": "v1",
"created": "Mon, 17 Mar 2025 16:01:22 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 01:30:29 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Baker",
"Andrew",
""
],
[
"Callaway",
"Brantly",
""
],
[
"Cunningham",
"Scott",
""
],
[
"Goodman-Bacon",
"Andrew",
""
],
[
"Sant'Anna",
"Pedro H. C.",
""
]
] |
2503.13386 | Nico C. X. Stuhlm\"uller | Nex C. X. Stuhlm\"uller, Ren\'e van Roij, Marjolein Dijkstra | Neuromorphic Computing with Microfluidic Memristors | 10 pages, 7 figures | null | null | null | cond-mat.soft | http://creativecommons.org/licenses/by-sa/4.0/ | Conical microfluidic channels filled with electrolytes exhibit volatile memristive behavior, offering a promising platform for energy-efficient, neuromorphic computing. Here, we integrate these iontronic channels as additional nonlinear elements in nonlinear Shinriki-inspired oscillators and demonstrate that they exhibit alternating chaotic and non-chaotic dynamics across a broad frequency range. Exploiting this behavior, we construct XOR and NAND gates by coupling three Memriki oscillators, and we further realize the full set of standard logic gates through combinations of NAND gates. Our results establish a new paradigm for iontronic computing and open avenues for scalable, low-power logical operations in microfluidic and bio-inspired systems. | [
{
"version": "v1",
"created": "Mon, 17 Mar 2025 17:16:47 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 12:56:38 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Stuhlmüller",
"Nex C. X.",
""
],
[
"van Roij",
"René",
""
],
[
"Dijkstra",
"Marjolein",
""
]
] |
2503.13437 | Dmitry Khudoteplov | Dmitry Khudoteplov, Alexei Morozov, Alexey Sleptsov | Can Yang-Baxter imply Lie algebra? | 12 pages | Phys. Lett. B 867 (2025) 139586 | 10.1016/j.physletb.2025.139586 | null | hep-th math-ph math.GT math.MP math.RT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Quantum knot invariants (like colored HOMFLY-PT or Kauffman polynomials) are a distinguished class of non-perturbative topological invariants. Any known way to construct them (via Chern-Simons theory or quantum R-matrix) starts with a finite simple Lie algebra. Another set of knot invariants - of finite type - is related to quantum invariants via a perturbative expansion. However can all finite type invariants be obtained in this way? Investigating this problem, P. Vogel discovered a way to polynomially parameterize the expansion coefficients with three parameters so that, at different specific values, this reproduces the answers for all simple Lie (super)algebras. Then it is easy to construct a polynomial $P_{alg}$ that vanishes for all simple Lie algebras, and the corresponding Vassiliev invariant would thus be absent from the perturbative expansion.
We review these Vogel claims pointing out at least two interesting implications of his construction. First, we discuss whether infinite-dimensional Lie algebras might enlarge Chern-Simons theory. Second, Vogel's construction implies an alternative axiomatization of simple Lie algebras - when we start from knot invariants and arrive at Lie algebras and their classification, which is opposite to conventional logic that we mentioned at the beginning. | [
{
"version": "v1",
"created": "Mon, 17 Mar 2025 17:58:31 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 07:33:49 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Khudoteplov",
"Dmitry",
""
],
[
"Morozov",
"Alexei",
""
],
[
"Sleptsov",
"Alexey",
""
]
] |
2503.13602 | Facundo Rodriguez | Facundo Rodriguez, Manuel Merch\'an, Daniela Gal\'arraga-Espinosa, Agustina del Valle Marsengo-Colazo, Antonio D. Montero-Dorta, Vicente Izzo, M. Celeste Artale | Central galaxy alignments. Dependence on the mass and the large-scale environment | 13 pages, 10 figures, Accepted to A&A | null | null | null | astro-ph.GA astro-ph.CO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Observations indicate that central galaxies show a significant alignment of their main shape axes with other galaxies in their group, as well as with the large-scale structure of the universe. Simulations have corroborated this finding, providing further insights into how the shape of the stellar component aligns with the surrounding dark matter halo. In this work, we aim at gaining a deeper understanding of galaxy alignments by quantifying how this property is related to the mass of the haloes hosting central galaxies and to the large-scale environment measured at different scales. By studying different angles, we describe how the alignments of central galaxies depend on the mass of the haloes they inhabit. We explore how the main axes of central galaxies align across different scales, both in three-dimensional and two-dimensional projections. We examine how halo mass influences these alignments and how they vary in the surrounding large-scale environment. To conduct this study, we employ the TNG300 hydrodynamical simulations and compare our results with the spectroscopic data from the Sloan Digital Sky Survey Data Release 18. Three types of alignment were analysed: between stellar and dark matter components, between satellite galaxies and the central galaxy, and between the central galaxy and its host halo. The results show that the alignment increases with halo mass and varies with the environment (clusters, filaments, cluster periphery, and others). However, after controlling for local density, we found that most of the observed trends disappear, except for a marginal influence of cosmic filaments on some of the considered alignment angles. The SDSS observations confirm a mass dependence similar to the simulations, although observational biases limit the detection of differences between the different environments. | [
{
"version": "v1",
"created": "Mon, 17 Mar 2025 18:00:05 GMT"
},
{
"version": "v2",
"created": "Tue, 10 Jun 2025 21:33:43 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Rodriguez",
"Facundo",
""
],
[
"Merchán",
"Manuel",
""
],
[
"Galárraga-Espinosa",
"Daniela",
""
],
[
"Marsengo-Colazo",
"Agustina del Valle",
""
],
[
"Montero-Dorta",
"Antonio D.",
""
],
[
"Izzo",
"Vicente",
""
],
[
"Artale",
"M. Celeste",
""
]
] |
2503.14756 | Hou In Ivan Tam | Hou In Ivan Tam, Hou In Derek Pun, Austin T. Wang, Angel X. Chang, Manolis Savva | SceneEval: Evaluating Semantic Coherence in Text-Conditioned 3D Indoor Scene Synthesis | Expanded dataset to 500 annotated scene descriptions with new scene types; added validation via extended manual evaluation and a new user study; clarified distinctions from prior metrics; included results using an open-source VLM; stated intent to release code and data; corrected terminology and typos. 24 pages with 8 figures and 6 tables | null | null | null | cs.GR cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Despite recent advances in text-conditioned 3D indoor scene generation, there remain gaps in the evaluation of these methods. Existing metrics primarily assess the realism of generated scenes by comparing them to a set of ground-truth scenes, often overlooking alignment with the input text - a critical factor in determining how effectively a method meets user requirements. We present SceneEval, an evaluation framework designed to address this limitation. SceneEval includes metrics for both explicit user requirements, such as the presence of specific objects and their attributes described in the input text, and implicit expectations, like the absence of object collisions, providing a comprehensive assessment of scene quality. To facilitate evaluation, we introduce SceneEval-500, a dataset of scene descriptions with annotated ground-truth scene properties. We evaluate recent scene generation methods using SceneEval and demonstrate its ability to provide detailed assessments of the generated scenes, highlighting strengths and areas for improvement across multiple dimensions. Our results show that current methods struggle at generating scenes that meet user requirements, underscoring the need for further research in this direction. | [
{
"version": "v1",
"created": "Tue, 18 Mar 2025 22:02:35 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 06:42:00 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Tam",
"Hou In Ivan",
""
],
[
"Pun",
"Hou In Derek",
""
],
[
"Wang",
"Austin T.",
""
],
[
"Chang",
"Angel X.",
""
],
[
"Savva",
"Manolis",
""
]
] |
2503.15121 | Carlos Lozano | Carlos Lozano and Jorge Ponsin | Analytic adjoint solution for incompressible potential flows | 20 pages. This article has been accepted by Physics of Fluids. The version of record can be found at https://doi.org/10.1063/5.0271828 | Physics of Fluids 37 (6), 1 June 2025 : 067127 | 10.1063/5.0271828 | null | physics.flu-dyn cs.NA math.NA physics.comp-ph | http://creativecommons.org/licenses/by-nc-nd/4.0/ | We obtain the analytic adjoint solution for two-dimensional (2D) incompressible potential flow for a cost function measuring aerodynamic force using the connection of the adjoint approach to Green's functions and also by establishing and exploiting its relation to the adjoint incompressible Euler equations. By comparison with the analytic solution, it is shown that the naive approach based on solving Laplace's equation for the adjoint variables can be ill-defined. The analysis of the boundary behavior of the analytic solution is used to discuss the proper formulation of the adjoint problem as well as the mechanism for incorporating the Kutta condition in the adjoint formulation | [
{
"version": "v1",
"created": "Wed, 19 Mar 2025 11:30:43 GMT"
},
{
"version": "v2",
"created": "Wed, 30 Apr 2025 07:36:11 GMT"
},
{
"version": "v3",
"created": "Wed, 11 Jun 2025 05:54:47 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Lozano",
"Carlos",
""
],
[
"Ponsin",
"Jorge",
""
]
] |
2503.15567 | Yanchen Luo | Yanchen Luo, Zhiyuan Liu, Yi Zhao, Sihang Li, Hengxing Cai, Kenji Kawaguchi, Tat-Seng Chua, Yang Zhang, Xiang Wang | Towards Unified and Lossless Latent Space for 3D Molecular Latent Diffusion Modeling | null | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | 3D molecule generation is crucial for drug discovery and material science, requiring models to process complex multi-modalities, including atom types, chemical bonds, and 3D coordinates. A key challenge is integrating these modalities of different shapes while maintaining SE(3) equivariance for 3D coordinates. To achieve this, existing approaches typically maintain separate latent spaces for invariant and equivariant modalities, reducing efficiency in both training and sampling. In this work, we propose \textbf{U}nified Variational \textbf{A}uto-\textbf{E}ncoder for \textbf{3D} Molecular Latent Diffusion Modeling (\textbf{UAE-3D}), a multi-modal VAE that compresses 3D molecules into latent sequences from a unified latent space, while maintaining near-zero reconstruction error. This unified latent space eliminates the complexities of handling multi-modality and equivariance when performing latent diffusion modeling. We demonstrate this by employing the Diffusion Transformer--a general-purpose diffusion model without any molecular inductive bias--for latent generation. Extensive experiments on GEOM-Drugs and QM9 datasets demonstrate that our method significantly establishes new benchmarks in both \textit{de novo} and conditional 3D molecule generation, achieving leading efficiency and quality. On GEOM-Drugs, it reduces FCD by 72.6\% over the previous best result, while achieving over 70\% relative average improvements in geometric fidelity. | [
{
"version": "v1",
"created": "Wed, 19 Mar 2025 08:56:13 GMT"
},
{
"version": "v2",
"created": "Thu, 03 Apr 2025 04:03:49 GMT"
},
{
"version": "v3",
"created": "Wed, 11 Jun 2025 07:48:53 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Luo",
"Yanchen",
""
],
[
"Liu",
"Zhiyuan",
""
],
[
"Zhao",
"Yi",
""
],
[
"Li",
"Sihang",
""
],
[
"Cai",
"Hengxing",
""
],
[
"Kawaguchi",
"Kenji",
""
],
[
"Chua",
"Tat-Seng",
""
],
[
"Zhang",
"Yang",
""
],
[
"Wang",
"Xiang",
""
]
] |
2503.15645 | Martin Gabelmann | Henning Bahl, Johannes Braathen, Martin Gabelmann, Sebastian Pa{\ss}ehr | Generic two-loop results for trilinear and quartic scalar self-interactions | 42 pages, 5 figures | null | null | null | hep-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Reconstructing the shape of the Higgs potential realised in Nature is a central part of the physics programme at the LHC and future colliders. In this context, accurate theoretical predictions for trilinear and quartic Higgs couplings are becoming increasingly important. In this paper, we present results that enable significant progress in the automation of these calculations at the two-loop level in a wide range of models. Specifically, we calculate the generic two-loop corrections for scalar n-point functions with n<=4 assuming that all external scalars are identical. Working in the zero-momentum approximation, we express the results in terms of generic couplings and masses. Additionally, by exploiting permutation invariances, we reduce the number of Feynman diagrams appearing to a substantially smaller set of basis diagrams. To ease the application of our setup, we also provide routines that allow to map our generic results to scalar two-loop amplitudes generated with the package FeynArts. We perform a series of calculations to cross-check our results with existing results in the literature. Moreover, we present new two-loop results for the trilinear Higgs coupling in the general singlet extension of the Standard Model. We also present the public Python package Tintegrals, which allows for fast and stable evaluations of all relevant two-loop integrals with vanishing external momenta. | [
{
"version": "v1",
"created": "Wed, 19 Mar 2025 19:00:11 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 17:56:09 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Bahl",
"Henning",
""
],
[
"Braathen",
"Johannes",
""
],
[
"Gabelmann",
"Martin",
""
],
[
"Paßehr",
"Sebastian",
""
]
] |
2503.15702 | Bryanne McDonough | Bryanne McDonough, Olivia Curtis, Tereasa Brainerd | Measuring Scaling Relationships: Fitting Technique Matters | 19 pages, 4 figures, 1 table; accepted to ApJ; interactive figure available in published article or via linked Zenodo repository in the computational notebook | null | 10.3847/2041-8213/ade13c | null | astro-ph.GA | http://creativecommons.org/licenses/by/4.0/ | Scaling relationships, both integrated and spatially resolved, arise due to the physical processes that govern galaxy evolution and are frequently measured in both observed and simulated data. However, the accuracy and comparability of these measurements are hindered by various differences between studies such as spatial resolution, sample selection criteria, and fitting technique. Here, we compare variations of standard least squares techniques to the ridge line method for identifying spatially resolved scaling relations ($\Sigma_*-\Sigma_{\rm SFR}$, $\Sigma_*-\Sigma_{\rm gas}$, and $\Sigma_{\rm gas}-\Sigma_{\rm SFR}$) for TNG100 galaxies. We find that using the ridge line technique to fit these scaling relations with a double linear function results in significantly better fits than fitting with ordinary least squares. We further illustrate the utility of the ridge line technique with an investigation into the dependence of rSFMS measurements on spatial resolution and smoothing scale. Specifically, we find that the slope of the rSFMS at low-$\Sigma_*$ is independent (within $2\sigma$) of spatial resolution and smoothing scale. Finally, we discuss the need for a consistent re-analysis of resolved scaling relations in the literature and physically motivate adoption of the ridge line technique over other fitting methods. | [
{
"version": "v1",
"created": "Wed, 19 Mar 2025 21:31:12 GMT"
},
{
"version": "v2",
"created": "Tue, 10 Jun 2025 21:32:20 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"McDonough",
"Bryanne",
""
],
[
"Curtis",
"Olivia",
""
],
[
"Brainerd",
"Tereasa",
""
]
] |
2503.16117 | Alexandre Verine | Alexandre Verine, Ahmed Mehdi Inane, Florian Le Bronnec, Benjamin Negrevergne, Yann Chevaleyre | Improving Discriminator Guidance in Diffusion Models | European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases - ECML PKDD 2025 | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | Discriminator Guidance has become a popular method for efficiently refining pre-trained Score-Matching Diffusion models. However, in this paper, we demonstrate that the standard implementation of this technique does not necessarily lead to a distribution closer to the real data distribution. Specifically, we show that training the discriminator using Cross-Entropy loss, as commonly done, can in fact increase the Kullback-Leibler divergence between the model and target distributions, particularly when the discriminator overfits. To address this, we propose a theoretically sound training objective for discriminator guidance that properly minimizes the KL divergence. We analyze its properties and demonstrate empirically across multiple datasets that our proposed method consistently improves over the conventional method by producing samples of higher quality. | [
{
"version": "v1",
"created": "Thu, 20 Mar 2025 13:04:43 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 12:24:13 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Verine",
"Alexandre",
""
],
[
"Inane",
"Ahmed Mehdi",
""
],
[
"Bronnec",
"Florian Le",
""
],
[
"Negrevergne",
"Benjamin",
""
],
[
"Chevaleyre",
"Yann",
""
]
] |
2503.16563 | Boulbaba Ben Amor Prof. | Aahan Singh and Engin Tekin and Maryam Nadeem and Nancy A. ElNaker and Mohammad Amaan Sayeed and Natalia Vassilieva and Boulbaba Ben Amor | Chem42: a Family of chemical Language Models for Target-aware Ligand Generation | null | null | null | null | cs.LG cs.AI cs.CL q-bio.BM | http://creativecommons.org/licenses/by/4.0/ | Revolutionizing drug discovery demands more than just understanding molecular interactions - it requires generative models that can design novel ligands tailored to specific biological targets. While chemical Language Models (cLMs) have made strides in learning molecular properties, most fail to incorporate target-specific insights, restricting their ability to drive de-novo ligand generation. Chem42, a cutting-edge family of generative chemical Language Models, is designed to bridge this gap. By integrating atomic-level interactions with multimodal inputs from Prot42, a complementary protein Language Model, Chem42 achieves a sophisticated cross-modal representation of molecular structures, interactions, and binding patterns. This innovative framework enables the creation of structurally valid, synthetically accessible ligands with enhanced target specificity. Evaluations across diverse protein targets confirm that Chem42 surpasses existing approaches in chemical validity, target-aware design, and predicted binding affinity. By reducing the search space of viable drug candidates, Chem42 could accelerate the drug discovery pipeline, offering a powerful generative AI tool for precision medicine. Our Chem42 models set a new benchmark in molecule property prediction, conditional molecule generation, and target-aware ligand design. The models are publicly available at huggingface.co/inceptionai. | [
{
"version": "v1",
"created": "Thu, 20 Mar 2025 07:07:30 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 05:18:10 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Singh",
"Aahan",
""
],
[
"Tekin",
"Engin",
""
],
[
"Nadeem",
"Maryam",
""
],
[
"ElNaker",
"Nancy A.",
""
],
[
"Sayeed",
"Mohammad Amaan",
""
],
[
"Vassilieva",
"Natalia",
""
],
[
"Amor",
"Boulbaba Ben",
""
]
] |
2503.16817 | Xiong Zeng | Xiong Zeng, Jing Yu, and Necmiye Ozay | System Identification Under Bounded Noise: Optimal Rates Beyond Least Squares | null | null | null | null | eess.SY cs.SY | http://creativecommons.org/licenses/by/4.0/ | System identification is a fundamental problem in control and learning, particularly in high-stakes applications where data efficiency is critical. Classical approaches, such as the ordinary least squares estimator (OLS), achieve an $O(1/\sqrt{T})$ convergence rate under Gaussian noise assumptions, where $T$ is the number of samples. This rate has been shown to match the lower bound. However, in many practical scenarios, noise is known to be bounded, opening the possibility of improving sample complexity. In this work, we establish the minimax lower bound for system identification under bounded noise, proving that the $O(1/T)$ convergence rate is indeed optimal. We further demonstrate that OLS remains limited to an $\Omega(1/\sqrt{T})$ convergence rate, making it fundamentally suboptimal in the presence of bounded noise. Finally, we instantiate two natural variations of OLS that obtain the optimal sample complexity. | [
{
"version": "v1",
"created": "Fri, 21 Mar 2025 03:13:32 GMT"
},
{
"version": "v2",
"created": "Sat, 17 May 2025 02:27:21 GMT"
},
{
"version": "v3",
"created": "Wed, 11 Jun 2025 02:38:28 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Zeng",
"Xiong",
""
],
[
"Yu",
"Jing",
""
],
[
"Ozay",
"Necmiye",
""
]
] |
2503.17132 | Shilin Lu | Siyuan Yang, Shilin Lu, Shizheng Wang, Meng Hwa Er, Zengwei Zheng, Alex C. Kot | Temporal-Guided Spiking Neural Networks for Event-Based Human Action Recognition | null | null | null | null | cs.CV cs.AI cs.CR cs.NE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper explores the promising interplay between spiking neural networks (SNNs) and event-based cameras for privacy-preserving human action recognition (HAR). The unique feature of event cameras in capturing only the outlines of motion, combined with SNNs' proficiency in processing spatiotemporal data through spikes, establishes a highly synergistic compatibility for event-based HAR. Previous studies, however, have been limited by SNNs' ability to process long-term temporal information, essential for precise HAR. In this paper, we introduce two novel frameworks to address this: temporal segment-based SNN (\textit{TS-SNN}) and 3D convolutional SNN (\textit{3D-SNN}). The \textit{TS-SNN} extracts long-term temporal information by dividing actions into shorter segments, while the \textit{3D-SNN} replaces 2D spatial elements with 3D components to facilitate the transmission of temporal information. To promote further research in event-based HAR, we create a dataset, \textit{FallingDetection-CeleX}, collected using the high-resolution CeleX-V event camera $(1280 \times 800)$, comprising 7 distinct actions. Extensive experimental results show that our proposed frameworks surpass state-of-the-art SNN methods on our newly collected dataset and three other neuromorphic datasets, showcasing their effectiveness in handling long-range temporal information for event-based HAR. | [
{
"version": "v1",
"created": "Fri, 21 Mar 2025 13:31:16 GMT"
},
{
"version": "v2",
"created": "Thu, 27 Mar 2025 11:35:37 GMT"
},
{
"version": "v3",
"created": "Wed, 11 Jun 2025 08:19:19 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Yang",
"Siyuan",
""
],
[
"Lu",
"Shilin",
""
],
[
"Wang",
"Shizheng",
""
],
[
"Er",
"Meng Hwa",
""
],
[
"Zheng",
"Zengwei",
""
],
[
"Kot",
"Alex C.",
""
]
] |
2503.18491 | Shuo Yang | Shuo Yang, Siwen Luo, Soyeon Caren Han, Eduard Hovy | MAGIC-VQA: Multimodal And Grounded Inference with Commonsense Knowledge for Visual Question Answering | Findings of ACL 2025 | null | null | null | cs.CL | http://creativecommons.org/licenses/by-sa/4.0/ | Visual Question Answering (VQA) requires reasoning across visual and textual modalities, yet Large Vision-Language Models (LVLMs) often lack integrated commonsense knowledge, limiting their robustness in real-world scenarios. To address this, we introduce MAGIC-VQA, a novel framework that enhances VQA by systematically integrating commonsense knowledge with LVLMs. MAGIC-VQA employs a three-stage process: (1) Explicit Knowledge Integration from external sources, (2) By-Type Post-Processing for contextual refinement, and (3) Implicit Knowledge Augmentation using a Graph Neural Network (GNN) for structured reasoning. While GNNs bring greater depth to structured inference, they enable superior relational inference beyond LVLMs. MAGIC-VQA bridges a key gap by unifying commonsensse knowledge with LVLM-driven reasoning, eliminating the need for extensive pre-training or complex prompt tuning. Our framework achieves state-of-the-art performance on benchmark datasets, significantly improving commonsense reasoning in VQA. | [
{
"version": "v1",
"created": "Mon, 24 Mar 2025 09:45:26 GMT"
},
{
"version": "v2",
"created": "Fri, 30 May 2025 16:34:32 GMT"
},
{
"version": "v3",
"created": "Wed, 11 Jun 2025 11:33:08 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Yang",
"Shuo",
""
],
[
"Luo",
"Siwen",
""
],
[
"Han",
"Soyeon Caren",
""
],
[
"Hovy",
"Eduard",
""
]
] |
2503.18927 | Shreyas Ramakrishna | S. Ramakrishna and S. Fritzsche | Interaction of a Poincar\'e beam with optically polarized atoms in the presence of constant magnetic field | 11 pages, 8 figures | null | null | null | physics.atom-ph | http://creativecommons.org/licenses/by/4.0/ | Recent studies have highlighted the frequent applications of structured light modes in optically pumped atomic magnetometers. In this work, we theoretically explore how a Poincar\'e beam probes an optically polarized atomic medium. Specifically, we consider atoms polarized by a plane wave with linear polarization, immersed in a constant external magnetic field. We analyze how the polarization of the pump and probe light fields, along with the external magnetic field, impact the absorption profile. To this end, we employ a density matrix approach based on the Liouville-von Neumann equation. Our results reveal that the absorption profile exhibits an asymmetric pattern that depends on the magnetic field strength and the mutual orientation of the pump and probe light propagation directions relative to the quantization axis. For illustration, we assume the incoming radiation drives an electric dipole transition, $5s \, ^2S_{1/2}$ ($F=1$) $\rightarrow$ $5p \, ^2P_{3/2}$ ($F=0$), in rubidium atoms subjected to a magnetic field. These findings may aid in designing future experiments on optically pumped atomic magnetometers utilizing structured light modes. | [
{
"version": "v1",
"created": "Mon, 24 Mar 2025 17:49:46 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 13:56:48 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Ramakrishna",
"S.",
""
],
[
"Fritzsche",
"S.",
""
]
] |
2503.19094 | Pavel Shvartsman | Pavel Shvartsman | Efficient Algorithms for Lipschitz Selections of Set-Valued Mappings in ${\bf R}^2$: long version | 87 pages, 30 figures. arXiv admin note: substantial text overlap with arXiv:2306.14042 | null | null | null | math.FA | http://creativecommons.org/publicdomain/zero/1.0/ | Let $F$ be a set-valued mapping from an $N$-element metric space $({\mathcal M},\rho)$ into the family of all closed half-planes in ${\bf R}^2$. In this paper, we provide an efficient algorithm for a Lipschitz selection of $F$, i.e., a Lipschitz mapping $f:{\mathcal M}\to{\bf R}^2$ such that $f(x)\in F(x)$ for all $x\in{\mathcal M}$.
Given a constant $\lambda>0$, this algorithm produces the following two outcomes: (1) The algorithm guarantees that there is no Lipschitz selection of $F$ with Lipschitz constant at most $\lambda$; (2) The algorithm returns a Lipschitz selection of $F$ with Lipschitz constant at most $3\lambda$.
The total work and storage required by this selection algorithm are at most $CN^2$ where $C$ is an absolute constant. | [
{
"version": "v1",
"created": "Mon, 24 Mar 2025 19:26:56 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 14:45:39 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Shvartsman",
"Pavel",
""
]
] |
2503.19323 | Victor Reiner | Trevor Karn, Victor Reiner | Invariant theory for wreath products acting on superpolynomials | Added references | null | null | null | math.CO math.AC | http://creativecommons.org/licenses/by/4.0/ | This paper considers a finite group $G$ acting linearly on the variables $V$ of a polynomial algebra, or an exterior algebra, or superpolynomial algebra with both commuting and anticommuting variables. In this setting, the Hilbert series for the $G$-invariant subalgebra turns out to determine the analogous Hilbert series for the wreath product $P[G]$ acting on $V^n$ for any permutation group $P$ inside the symmetric group $S_n$ on $n$ letters.
This leads to a structural result: one can collate the direct sum for all $n$ of the $S_n[G]$-invariant subalgebras to form a graded ring via an external shuffle product, whose structure turns out to be a superpolynomial algebra generated by the $G$-invariants. A parallel statement holds for the direct sum of all $S_n[G]$-antiinvariants, which forms a graded ring via an external signed shuffle product, isomorphic to the superexterior algebra generated by the $G$-invariants. | [
{
"version": "v1",
"created": "Tue, 25 Mar 2025 03:36:43 GMT"
},
{
"version": "v2",
"created": "Tue, 10 Jun 2025 19:02:47 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Karn",
"Trevor",
""
],
[
"Reiner",
"Victor",
""
]
] |
2503.20582 | Rupert Levene | Rupert H. Levene, Polona Oblak, Helena \v{S}migoc | Orthogonalisability of joins of graphs | null | null | null | null | math.CO math.SP | http://creativecommons.org/licenses/by/4.0/ | A graph is said to be orthogonalisable if the set of real symmetric matrices whose off-diagonal pattern is prescribed by its edges contains an orthogonal matrix. We determine some necessary and some sufficient conditions on the sizes of the connected components of two graphs for their join to be orthogonalisable. In some cases, those conditions coincide, and we present several families of joins of graphs that are orthogonalisable. | [
{
"version": "v1",
"created": "Wed, 26 Mar 2025 14:29:58 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 07:48:43 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Levene",
"Rupert H.",
""
],
[
"Oblak",
"Polona",
""
],
[
"Šmigoc",
"Helena",
""
]
] |
2503.21698 | Mohsen Yarmohammadi | Mohsen Yarmohammadi and Ulrich Z\"ulicke and Jamal Berakdar and Jacob Linder and James K. Freericks | Anisotropic light-tailored RKKY interaction in two-dimensional $d$-wave altermagnets | 16 pages, 8 figures | Phys. Rev. B 111, 224412 (2025) | 10.1103/k3xb-8pts | null | cond-mat.mes-hall | http://creativecommons.org/licenses/by/4.0/ | Altermagnets are known in spintronics for their intrinsic spin-splitting and unconventional magnetic responses, particularly to magnetic impurities. However, effectively controlling the magnetic exchange interactions in altermagnets is challenging for practical applications. Here, we propose using circularly polarized light to tune the Ruderman-Kittel-Kasuya-Yosida (RKKY) interaction in two-dimensional $d$-wave altermagnets. Using the real-space retarded Green's functions approach, our results show that while the Heisenberg and Ising exchanges dominate, a notable Dzyaloshinskii-Moriya (DM) interaction also plays a key role. Furthermore, the inherent strength of altermagnetism imprints chirp-like signatures into the magnetic responses, which can be dynamically tuned via light. We mainly demonstrate that gate-induced Rashba spin-orbit coupling is essential in response to light -- light selectively and anisotropically adjusts the DM interaction without affecting the other exchanges. Our findings further indicate that rotating the altermagnet by $45^\circ$ relative to the light's polarization direction generates a Dirac-like dispersion and different DM interactions. We finally extract critical thresholds where light reverses DM interactions along one axis or balances both in-plane components. The anisotropic light-driven control of RKKY interactions in 2D altermagnets not only highlights their unique properties but also opens new avenues for engineering tailored magnetic characteristics in spintronic applications. | [
{
"version": "v1",
"created": "Thu, 27 Mar 2025 17:03:27 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Yarmohammadi",
"Mohsen",
""
],
[
"Zülicke",
"Ulrich",
""
],
[
"Berakdar",
"Jamal",
""
],
[
"Linder",
"Jacob",
""
],
[
"Freericks",
"James K.",
""
]
] |
2503.21971 | Armin Abdollahi | Armin Abdollahi and Mehdi Kamal and Massoud Pedram | RocketPPA: Code-Level Power, Performance, and Area Prediction via LLM and Mixture of Experts | null | null | null | null | cs.LG cs.SE | http://creativecommons.org/licenses/by/4.0/ | This paper presents RocketPPA, a novel ultra-fast power, performance (delay), and area (PPA) estimator operating directly at the code-level abstraction using HDL code as input. The key technical innovation is its LLM-based regression model, which uniquely integrates a large language model (LLM) with a mixture-of-experts (MoE) architecture composed of multilayer perceptrons (MLPs). The LLM interprets the input HDL code and then utilizes its final hidden-layer representations to predict PPA metrics. Low-rank adaptation (LoRA) is used for parameter-efficient fine-tuning to enable efficient LLM training. Furthermore, the work includes the development of an LLM-based HDL code repair framework to generate a large and synthesizable training dataset. Experimental results on the VerilogEval benchmark demonstrate that RocketPPA achieves significant improvements in the accuracy of PPA estimation compared to previous state-of-the-art methods like Llama3-MetRex-8B. Specifically, at a 10% relative error threshold, RocketPPA enhances the pass rate for area prediction by 13.6%, delay by 9.4%, and power by 14.7%. At a 20% threshold, the improvements are 9.6% for area, 10.8% for delay, and 18.5% for power. Moreover, RocketPPA achieves a speedup of over 20x compared to MetRex and 30x over MasterRTL in processing the test set. The impact of RocketPPA is the potential to substantially accelerate the hardware design process by providing accurate PPA estimations early in the design cycle, thus avoiding the overhead of manual feature engineering and time-consuming synthesis flows. | [
{
"version": "v1",
"created": "Thu, 27 Mar 2025 20:35:09 GMT"
},
{
"version": "v2",
"created": "Tue, 29 Apr 2025 00:43:11 GMT"
},
{
"version": "v3",
"created": "Tue, 10 Jun 2025 22:20:25 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Abdollahi",
"Armin",
""
],
[
"Kamal",
"Mehdi",
""
],
[
"Pedram",
"Massoud",
""
]
] |
2503.22874 | Gabriel S\'anchez-Col\'on | Sarira Sahu, R. de J. Pacheco-Ak\'e, G. S\'anchez-Col\'on, D. I. P\'aez-S\'anchez, A. U. Puga Oliveros, Subhash Rajpoot | Constraining the redshift of PG 1553+113 using the photohadronic model | 11 pages, 4 tables, 3 figures, to be published in MNRAS | null | null | null | astro-ph.HE | http://creativecommons.org/licenses/by/4.0/ | Between 2005 and 2015, the BL Lacertae object PG 1553+113 exhibited multiple very high-energy (VHE; > 100 GeV) gamma-ray flares, which were detected by the Cherenkov telescopes, High Energy Stereoscopic System (HESS), Major Atmospheric Gamma Imaging Cherenkov (MAGIC), and the Very Energetic Radiation Imaging Telescope Array System (VERITAS). Despite the uncertainty surrounding its redshift (z), various studies have sought to estimate this value. In this study, seventeen independently observed VHE gamma-ray spectra of PG 1553+113 are analyzed using four distinct extragalactic background light (EBL) models alongside the photohadronic framework. A global $\chi^2$ fit is applied to all observational data to determine the best-fitting redshift for each EBL model. Additionally, confidence level (CL) intervals for the redshift are calculated across all EBL models. The findings demonstrate that the photohadronic framework effectively describes all observed spectra. Among the EBL models, the lowest total $\chi^2$ value of 58.82 (with 58 degrees of freedom) was obtained using the model from Saldana-Lopez et al. (2021), while the highest value of 75.67 was associated with the model from Dominguez et al. (2011). The 95 per cent CL intervals for the statistical error of the redshift of PG 1553+113 are 0.500 < z < 0.537 for the Saldana-Lopez et al. (2021) model and 0.491 < z < 0.527 for the Dominguez et al. (2011) model. These results highlight the consistency of the photohadronic approach in interpreting the observed VHE gamma-ray spectra. | [
{
"version": "v1",
"created": "Fri, 28 Mar 2025 21:01:02 GMT"
},
{
"version": "v2",
"created": "Tue, 10 Jun 2025 21:29:48 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Sahu",
"Sarira",
""
],
[
"Pacheco-Aké",
"R. de J.",
""
],
[
"Sánchez-Colón",
"G.",
""
],
[
"Páez-Sánchez",
"D. I.",
""
],
[
"Oliveros",
"A. U. Puga",
""
],
[
"Rajpoot",
"Subhash",
""
]
] |
2503.23004 | Kathleen MacWilliam | Stefano Damiano, Kathleen MacWilliam, Valerio Lorenzoni, Thomas Dietzen and Toon van Waterschoot | The trajectoRIR Database: Room Acoustic Recordings Along a Trajectory of Moving Microphones | 17 pages, 7 figures | null | null | null | eess.AS | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Data availability is essential to develop acoustic signal processing algorithms, especially when it comes to data-driven approaches that demand large and diverse training datasets. For this reason, an increasing number of databases have been published in recent years, including either room impulse responses (RIRs) or audio recordings during motion. In this paper we introduce the trajectoRIR database, an extensive, multi-array collection of both dynamic and stationary acoustic recordings along a controlled trajectory in a room. Specifically, the database features recordings using moving microphones and stationary RIRs spatially sampling the room acoustics along an L-shaped trajectory. This combination makes trajectoRIR unique and applicable in various tasks ranging from sound source localization and tracking to spatially dynamic sound field reconstruction, auralization and system identification. The recording room has a reverberation time of 0.5 seconds, and the three different microphone configurations employed include a dummy head, with additional reference microphones located next to the ears, 3 first-order Ambisonics microphones, two circular arrays of 16 and 4 channels, and a 12-channel linear array. The motion of the microphones was achieved using a robotic cart traversing a 4.62 meter-long rail at three speeds: [0.2, 0.4, 0.8] m/s. Audio signals were reproduced using two stationary loudspeakers. The collected database features 8648 stationary RIRs, as well as perfect sweeps, speech, music, and stationary noise recorded during motion. Python functions are included to access the recorded audio as well as to retrieve geometrical information. | [
{
"version": "v1",
"created": "Sat, 29 Mar 2025 07:59:16 GMT"
},
{
"version": "v2",
"created": "Tue, 10 Jun 2025 16:20:26 GMT"
},
{
"version": "v3",
"created": "Wed, 11 Jun 2025 09:53:03 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Damiano",
"Stefano",
""
],
[
"MacWilliam",
"Kathleen",
""
],
[
"Lorenzoni",
"Valerio",
""
],
[
"Dietzen",
"Thomas",
""
],
[
"van Waterschoot",
"Toon",
""
]
] |
2503.23940 | Debapratim Banerjee | Debapratim Banerjee | Operator limit of Wigner matrices I | First draft. Comments are welcome | null | null | null | math.PR math-ph math.FA math.MP math.ST stat.TH | http://creativecommons.org/licenses/by/4.0/ | We consider the Wigner matrix $W_{n}$ of dimension $n \times n$ as $n \to \infty$. The objective of this paper is two folds: first we construct an operator $\mathcal{W}$ on a suitable Hilbert space $\mathcal{H}$ and then define a suitable notion of convergence such that the matrices $W_{n}$ converge in that notion of convergence to $\mathcal{W}$. We further investigate some properties of $\mathcal{W}$ and $\mathcal{H}$. We show that $\mathcal{H}$ is a nontrivial extension of $L^{2}[0,1]$ with respect to the Lebesgue measure and the spectral measure of $\mathcal{W}$ at any function $f \in L^{2}[0,1]$ is almost surely the semicircular law. | [
{
"version": "v1",
"created": "Mon, 31 Mar 2025 10:45:01 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 14:58:15 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Banerjee",
"Debapratim",
""
]
] |
2504.00092 | Oscar Wistemar | Oscar Wistemar, Felix Ryde, Filip Alamaa | A generalized method to measure the Lorentz factor from gamma-ray burst photospheric emission | Published in ApJ, minor changes. 8 pages, 3 figures | The Astrophysical Journal, 986:118, 2025 | 10.3847/1538-4357/add52d | null | astro-ph.HE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The properties of gamma-ray bursts (GRBs) that are inferred from observations depend on the value of the bulk Lorentz factor, $\Gamma$. Consequently, accurately estimating it is an important aim. In this work, we present a method of measuring $\Gamma$ based on observed photospheric emission, which can also be used for highly dissipative flows that may lead to non-thermal spectral shapes. For the method to be applicable, two conditions need to be met: the photon number should be conserved in the later stages of the jet, and the original photon temperature must be inferred from the data. The case of dissipation via subphotospheric shocks is discussed in detail, and we show that the method is particularly efficient when a low-energy spectral break is identified. We demonstrate the capabilities of the method by applying it to two different GRB spectra. From one of the spectra, we obtain a value for $\Gamma$ with statistical uncertainties of only $\sim 15$\%, while for the other spectrum we only obtain an upper limit. | [
{
"version": "v1",
"created": "Mon, 31 Mar 2025 18:00:03 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 15:25:46 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Wistemar",
"Oscar",
""
],
[
"Ryde",
"Felix",
""
],
[
"Alamaa",
"Filip",
""
]
] |
2504.00115 | Shiyue Zhao | Shiyue Zhao, Junzhi Zhang, Neda Masoud, Heye Huang, Xiaohui Hou, and Chengkun He | SACA: A Scenario-Aware Collision Avoidance Framework for Autonomous Vehicles Integrating LLMs-Driven Reasoning | 11 pages,10 figures. This work has been submitted to the IEEE TVT for possible publication | null | null | null | cs.RO cs.SY eess.SY | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Reliable collision avoidance under extreme situations remains a critical challenge for autonomous vehicles. While large language models (LLMs) offer promising reasoning capabilities, their application in safety-critical evasive maneuvers is limited by latency and robustness issues. Even so, LLMs stand out for their ability to weigh emotional, legal, and ethical factors, enabling socially responsible and context-aware collision avoidance. This paper proposes a scenario-aware collision avoidance (SACA) framework for extreme situations by integrating predictive scenario evaluation, data-driven reasoning, and scenario-preview-based deployment to improve collision avoidance decision-making. SACA consists of three key components. First, a predictive scenario analysis module utilizes obstacle reachability analysis and motion intention prediction to construct a comprehensive situational prompt. Second, an online reasoning module refines decision-making by leveraging prior collision avoidance knowledge and fine-tuning with scenario data. Third, an offline evaluation module assesses performance and stores scenarios in a memory bank. Additionally, A precomputed policy method improves deployability by previewing scenarios and retrieving or reasoning policies based on similarity and confidence levels. Real-vehicle tests show that, compared with baseline methods, SACA effectively reduces collision losses in extreme high-risk scenarios and lowers false triggering under complex conditions. Project page: https://sean-shiyuez.github.io/SACA/. | [
{
"version": "v1",
"created": "Mon, 31 Mar 2025 18:04:15 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 03:34:15 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Zhao",
"Shiyue",
""
],
[
"Zhang",
"Junzhi",
""
],
[
"Masoud",
"Neda",
""
],
[
"Huang",
"Heye",
""
],
[
"Hou",
"Xiaohui",
""
],
[
"He",
"Chengkun",
""
]
] |
2504.00253 | Laura Hunter | Laura Congreve Hunter, Bur\c{c}in Mutlu-Pakdil, David J. Sand, Paul Bennet, Donghyeon J. Khim, Denija Crnojevi\'c, Amandine Doliva-Dolinsky, Emmanuel Durodola, Catherine Fielder, Rowan Goebel-Bain, Michael G. Jones, Ananthan Karunakaran, Kristine Spekkens, Dennis Zaritsky | Identifying Dwarfs of MC Analog GalaxiEs (ID-MAGE): The Search for Satellites Around Low-Mass Hosts | 16 pages, 9 figures, 3 page appendix | null | null | null | astro-ph.GA | http://creativecommons.org/licenses/by/4.0/ | We present results from ID-MAGE (Identifying Dwarfs of MC Analog GalaxiEs), a survey aimed at identifying and characterizing unresolved satellite galaxies around 36~nearby LMC- and SMC-mass hosts (D$=$4$-$10~Mpc). We use archival DESI Legacy Survey imaging data and perform an extensive search for dwarf satellites, extending out to a radius of 150~kpc ($\sim$$R_{vir}$). We identify 355 candidate satellite galaxies, including 24 new discoveries. Extensive tests with injected galaxies demonstrate that the survey is complete down to $M_V\sim-$9.0 (assuming the distance of the host) and $\mu_{0,V}\sim$26 mag arcsec$^{-2}$ (assuming a n$=$1 S\'ersic profile). We perform consistent photometry, via S\'ersic profile fitting, on all candidates and have initiated a comprehensive follow-up campaign to confirm and characterize candidates. Through a systematic visual inspection campaign, we classify the top candidates as high-likelihood satellites. On average, we find 4.0$\pm$1.4 high-likelihood candidate satellites per LMC-mass host and 2.1$\pm$0.6 per SMC-mass host which is within the range predicted by cosmological models. We use this sample to establish upper and lower estimates on the satellite luminosity function of LMC/SMC-mass galaxies. ID-MAGE nearly triples the number of low-mass galaxies surveyed for satellites with well-characterized completeness limits, providing a unique dataset to explore small-scale structure and dwarf galaxy evolution around low-mass hosts in diverse environments. | [
{
"version": "v1",
"created": "Mon, 31 Mar 2025 21:55:38 GMT"
},
{
"version": "v2",
"created": "Wed, 02 Apr 2025 18:43:25 GMT"
},
{
"version": "v3",
"created": "Tue, 10 Jun 2025 18:11:09 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Hunter",
"Laura Congreve",
""
],
[
"Mutlu-Pakdil",
"Burçin",
""
],
[
"Sand",
"David J.",
""
],
[
"Bennet",
"Paul",
""
],
[
"Khim",
"Donghyeon J.",
""
],
[
"Crnojević",
"Denija",
""
],
[
"Doliva-Dolinsky",
"Amandine",
""
],
[
"Durodola",
"Emmanuel",
""
],
[
"Fielder",
"Catherine",
""
],
[
"Goebel-Bain",
"Rowan",
""
],
[
"Jones",
"Michael G.",
""
],
[
"Karunakaran",
"Ananthan",
""
],
[
"Spekkens",
"Kristine",
""
],
[
"Zaritsky",
"Dennis",
""
]
] |
2504.00632 | Bing Li | Junjie Huang, Bing Li and Sanju Velani | Exponential mixing for Gibbs measures on self-conformal sets and applications | 90 pages | null | null | null | math.DS math.CA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we show that Gibbs measures on self-conformal sets generated by a $C^{1+\alpha}$ conformal IFS on $\mathbb{R}^d$ satisfying the OSC are exponentially mixing. We exploit this to obtain essentially sharp asymptotic counting statements for the recurrent and the shrinking target subsets associated with any such set. In particular, we provide explicit examples of dynamical systems for which the recurrent sets exhibit (unexpected) behaviour that is not present in the shrinking target setup. In the process of establishing our exponential mixing result we extend Mattila's rigidity theorem for self-similar sets to self-conformal sets without any separation condition and for arbitrary Gibbs measures. | [
{
"version": "v1",
"created": "Tue, 01 Apr 2025 10:30:33 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 14:46:54 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Huang",
"Junjie",
""
],
[
"Li",
"Bing",
""
],
[
"Velani",
"Sanju",
""
]
] |
2504.00649 | Rui-Hao Bi | Rui-Hao Bi, Wei Liu and Wenjie Dou | Universal Structure of Computing Moments for Exact Quantum Dynamics: Application to Arbitrary System-Bath Couplings | 22 pages, 6 figures | null | 10.1063/5.0273707 | null | quant-ph physics.chem-ph | http://creativecommons.org/licenses/by/4.0/ | We introduce a general procedure for computing higher-order moments of correlation functions in open quantum systems, extending the scope of our recent work on Memory Kernel Coupling Theory (MKCT) [W. Liu, Y. Su, Y. Wang, and W. Dou, arXiv:2407.01923 (2024)]. This approach is demonstrated for arbitrary system-bath coupling that can be expressed as polynomial, $H_{SB} = \hat{V} (\alpha_0 + \alpha_1 \hat{q} + \alpha_2 \hat{q}^2+ \dots)$, where we show that the recursive commutators of a system operator obey a universal hierarchy. Exploiting this structure, the higher-order moments are obtained by evaluating the expectation values of the system and bath operators separately, with bath expectation values derived from the derivatives of a generating function. We further apply MKCT to compute the dipole autocorrelation function for the spin-boson model with both linear and quadratic coupling, achieving agreement with the hierarchical equations of motion approach. Our findings suggest a promising path toward accurate dynamics for complex open quantum systems. | [
{
"version": "v1",
"created": "Tue, 01 Apr 2025 10:58:55 GMT"
},
{
"version": "v2",
"created": "Tue, 13 May 2025 06:35:30 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Bi",
"Rui-Hao",
""
],
[
"Liu",
"Wei",
""
],
[
"Dou",
"Wenjie",
""
]
] |
2504.00731 | Dhanika Mahipala | Dhanika Mahipala, Trym Tengesdal, B{\o}rge Rokseth, Tor Arne Johansen | Design and Validation of an Intention-Aware Probabilistic Framework for Trajectory Prediction: Integrating COLREGS, Grounding Hazards, and Planned Routes | IMPORTANT: This preprint is not the final version. The peer-reviewed and updated version is published in Ocean Engineering journal [https://doi.org/10.1016/j.oceaneng.2025.121564] | Ocean Engineering 335, 121564 (2025) | 10.1016/j.oceaneng.2025.121564 | null | cs.RO cs.SY eess.SY | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Collision avoidance capability is an essential component in an autonomous vessel navigation system. To this end, an accurate prediction of dynamic obstacle trajectories is vital. Traditional approaches to trajectory prediction face limitations in generalizability and often fail to account for the intentions of other vessels. While recent research has considered incorporating the intentions of dynamic obstacles, these efforts are typically based on the own-ship's interpretation of the situation. The current state-of-the-art in this area is a Dynamic Bayesian Network (DBN) model, which infers target vessel intentions by considering multiple underlying causes and allowing for different interpretations of the situation by different vessels. However, since its inception, there have not been any significant structural improvements to this model. In this paper, we propose enhancing the DBN model by incorporating considerations for grounding hazards and vessel waypoint information. The proposed model is validated using real vessel encounters extracted from historical Automatic Identification System (AIS) data. | [
{
"version": "v1",
"created": "Tue, 01 Apr 2025 12:43:33 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 08:20:13 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Mahipala",
"Dhanika",
""
],
[
"Tengesdal",
"Trym",
""
],
[
"Rokseth",
"Børge",
""
],
[
"Johansen",
"Tor Arne",
""
]
] |
2504.01325 | Tomoo Yokoyama | Tomoo Yokoyama | Coarse chain recurrence, Morse graphs with finite errors, and persistence of circulations | null | null | null | null | math.DS | http://creativecommons.org/licenses/by-nc-nd/4.0/ | This paper develops a topological framework to describe global dynamics under finite-energy perturbations, focusing on escapes from attracting basins, elimination of stagnation, and persistence of recurrent behaviors governed by finite total or stepwise energy. To this end, we introduce generalized concepts of chain recurrence and construct corresponding filtrations for dynamical systems on metric spaces. These filtrations refine Conley's classical theory and enable the quantitative modeling of dynamical behaviors under bounded energy inputs. In particular, we introduce concepts of "coarse chain recurrences" and Morse graphs with finite errors, which yield bifurcation diagrams depicting how chain recurrent components evolve as perturbation scales increase. Our framework also reveals singular limits, where the total injected energy vanishes. The framework applies to both maps and semiflows, providing tools to analyze robustness, sensitivity, and structural evolution of recurrent dynamics. Applications include toy models of flow control and circulation maintained under finite-energy constraints. | [
{
"version": "v1",
"created": "Wed, 02 Apr 2025 03:18:34 GMT"
},
{
"version": "v2",
"created": "Tue, 10 Jun 2025 23:33:18 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Yokoyama",
"Tomoo",
""
]
] |
2504.01604 | Panagiotis Petrantonakis | Vasileios Georgiadis and Panagiotis C. Petrantonakis | SpikeSift: A Computationally Efficient and Drift-Resilient Spike Sorting Algorithm | 24 pages, 6 figures, 4 tables | null | null | null | eess.SP q-bio.NC | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Objective: Spike sorting is a fundamental step in analysing extracellular recordings, enabling the isolation of single-neuron activity. However, it remains a challenging problem because extracellular traces mix overlapping spikes from neighbouring cells and are marred by recording instabilities such as electrode drift. Numerous algorithms have been proposed, yet many struggle to balance accuracy and computational efficiency, limiting their practicality for todays large-scale datasets. Approach: In response, we introduce SpikeSift, a spike-sorting algorithm expressly designed to mitigate drift while running on standard CPUs. SpikeSift (i) partitions long recordings into shorter, relatively stationary segments, (ii) carries out spike detection and clustering simultaneously through an iterative detect-and-subtract scheme within each segment, and (iii) preserves neuronal identity across segments via a fast template alignment stage that dispenses with continuous trajectory estimation. Main results: Extensive validation on paired intracellularly validated datasets and on biophysically realistic MEArec simulations covering elevated noise, diverse drift profiles, ultra short recordings and bursting activity, demonstrates that SpikeSift matches or exceeds the accuracy of state of the art methods while completing sorting an order of magnitude faster on a single desktop core. Significance: The combination of high fidelity, drift resilience, and modest computational demand renders SpikeSift broadly accessible while preserving data quality for downstream neurophysiological analysis | [
{
"version": "v1",
"created": "Wed, 02 Apr 2025 11:16:28 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 09:06:03 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Georgiadis",
"Vasileios",
""
],
[
"Petrantonakis",
"Panagiotis C.",
""
]
] |
2504.01738 | Philip Lippmann | Philip Lippmann and Jie Yang | Style over Substance: Distilled Language Models Reason Via Stylistic Replication | null | null | null | null | cs.CL cs.AI | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Specialized reasoning language models (RLMs) have demonstrated that scaling test-time computation through detailed reasoning traces significantly enhances performance. Although these traces effectively facilitate knowledge distillation into smaller, instruction-tuned models, the precise nature of transferred reasoning remains unclear. In this study, we investigate to what extent distilled models internalize replicated stylistic patterns during reasoning. To this end, we systematically analyze reasoning traces, identifying structural and lexical patterns that characterize successful reasoning. We then introduce two new datasets -- a dataset of emergent reasoning traces and a synthetic dataset explicitly constructed to replicate these stylistic patterns -- to precisely examine their influence on distilled models' reasoning capabilities. We find that models trained on the synthetic traces achieve comparable performance, indicating that distilled reasoning abilities rely significantly on surface-level patterns. Surprisingly, we observe an increase in performance even when the synthetic traces are altered to lead to the wrong answer. Our findings highlight how stylistic patterns can be leveraged to efficiently enhance LM reasoning across diverse model families. | [
{
"version": "v1",
"created": "Wed, 02 Apr 2025 13:50:20 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 11:31:47 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Lippmann",
"Philip",
""
],
[
"Yang",
"Jie",
""
]
] |
2504.01837 | Hao Wu | Hao Wu and Lei Yu | Entropic Isoperimetric and Cram\'er--Rao Inequalities for R\'enyi--Fisher Information | 27 pages | null | null | null | cs.IT math.IT math.PR math.ST stat.TH | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The de Bruijn identity states that Fisher information is equal to a half of the time-derivative of Shannon differential entropy along heat flow. In the same spirit, a generalized version of Fisher information, which we term the R\'enyi--Fisher information, is defined as a half of the time-derivative of R\'enyi differential entropy along heat flow. Based on this R\'enyi--Fisher information, we establish several sharp R\'enyi-entropic isoperimetric inequalities, which generalize the classic entropic isoperimetric inequality to the R\'enyi setting. Utilizing these isoperimetric inequalities, we extend the classical Cram\'er--Rao inequality from Fisher information to R\'enyi--Fisher information. We then use these generalized Cram\'er--Rao inequalities to determine the signs of derivatives of R\'enyi entropy along heat flow, strengthening existing results on the complete monotonicity of R\'enyi entropy. We lastly explore the implications of our R\'enyi-entropic isoperimetric inequalities for entropy power inequalities. We demonstrate that, unlike in the Shannon entropy case, the classic entropy power inequality does not admit a direct extension to R\'enyi entropy without introducing additional exponents or scaling factors. Furthermore, we establish a sharp R\'enyi entropy power inequality involving a scaling factor under the assumption that one of two independent random vectors is Gaussian. | [
{
"version": "v1",
"created": "Wed, 02 Apr 2025 15:44:02 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 02:40:02 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Wu",
"Hao",
""
],
[
"Yu",
"Lei",
""
]
] |
2504.02097 | Marta Chabowska | Marta A. Chabowska, Magdalena A. Za{\l}uska-Kotur | Engineering 2D Surface Patterns with the VicCa Model | 17 pages, 7 figures, submitted to J. Cryst. Growth | J. Cryst. Growth 667, 128245 (2025) | 10.1016/j.jcrysgro.2025.128245 | null | cond-mat.mtrl-sci | http://creativecommons.org/licenses/by/4.0/ | We employed the VicCA model to investigate the influence of step-edge potential on nucleation and pattern formation, aiming to gain deeper insights into island formation and growth. Our study explores fractal structures governed by general cellular automaton (CA) rules, as well as compact structures shaped by density-dependent attachment mechanisms. We demonstrate that modifications to the CA framework have a significant impact on surface patterning, emphasizing the critical role of adatom attachment rules and the substantial effect of potential well depth on the resulting surface morphology. | [
{
"version": "v1",
"created": "Wed, 02 Apr 2025 19:56:37 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Chabowska",
"Marta A.",
""
],
[
"Załuska-Kotur",
"Magdalena A.",
""
]
] |
2504.02132 | Ezzeldin Shereen | Ezzeldin Shereen, Dan Ristea, Shae McFadden, Burak Hasircioglu, Vasilios Mavroudis, Chris Hicks | One Pic is All it Takes: Poisoning Visual Document Retrieval Augmented Generation with a Single Image | 19 pages, 7 figures | null | null | null | cs.CL cs.CR cs.CV cs.IR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Multi-modal retrieval augmented generation (M-RAG) is instrumental for inhibiting hallucinations in large multi-modal models (LMMs) through the use of a factual knowledge base (KB). However, M-RAG introduces new attack vectors for adversaries that aim to disrupt the system by injecting malicious entries into the KB. In this paper, we present the first poisoning attack against M-RAG targeting visual document retrieval applications where the KB contains images of document pages. We propose two attacks, each of which require injecting only a single adversarial image into the KB. Firstly, we propose a universal attack that, for any potential user query, influences the response to cause a denial-of-service (DoS) in the M-RAG system. Secondly, we present a targeted attack against one or a group of user queries, with the goal of spreading targeted misinformation. For both attacks, we use a multi-objective gradient-based adversarial approach to craft the injected image while optimizing for both retrieval and generation. We evaluate our attacks against several visual document retrieval datasets, a diverse set of state-of-the-art retrievers (embedding models) and generators (LMMs), demonstrating the attack effectiveness in both the universal and targeted settings. We additionally present results including commonly used defenses, various attack hyper-parameter settings, ablations, and attack transferability. | [
{
"version": "v1",
"created": "Wed, 02 Apr 2025 21:08:33 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 15:07:55 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Shereen",
"Ezzeldin",
""
],
[
"Ristea",
"Dan",
""
],
[
"McFadden",
"Shae",
""
],
[
"Hasircioglu",
"Burak",
""
],
[
"Mavroudis",
"Vasilios",
""
],
[
"Hicks",
"Chris",
""
]
] |
2504.02613 | Asad Mahmood Mr. | Asad Mahmood, Thang X. Vu, Wali Ullah Khan, Symeon Chatzinotas, Bj\"orn Ottersten | UAV-Assisted 5G Networks: Mobility-Aware 3D Trajectory Optimization and Resource Allocation for Dynamic Environments | null | null | null | null | eess.SP | http://creativecommons.org/licenses/by/4.0/ | This work proposes a framework for the robust design of UAV-assisted wireless networks that combine 3D trajectory optimization with user mobility prediction to address dynamic resource allocation challenges. We proposed a sparse second-order prediction model for real-time user tracking coupled with heuristic user clustering to balance service quality and computational complexity. The joint optimization problem is formulated to maximize the minimum rate. It is then decomposed into user association, 3D trajectory design, and resource allocation subproblems, which are solved iteratively via successive convex approximation (SCA). Extensive simulations demonstrate: (1) near-optimal performance with $\epsilon \approx 0.67\%$ deviation from upper-bound solutions, (2) $16\%$ higher minimum rates for distant users compared to non-predictive 3D designs, and (3) $10-30\%$ faster outage mitigation than time-division benchmarks. The framework's adaptive speed control enables precise mobile user tracking while maintaining energy efficiency under constrained flight time. Results demonstrate superior robustness in edge-coverage scenarios, making it particularly suitable for $5G/6G$ networks. | [
{
"version": "v1",
"created": "Thu, 03 Apr 2025 14:12:58 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 10:35:07 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Mahmood",
"Asad",
""
],
[
"Vu",
"Thang X.",
""
],
[
"Khan",
"Wali Ullah",
""
],
[
"Chatzinotas",
"Symeon",
""
],
[
"Ottersten",
"Björn",
""
]
] |
2504.03908 | Zhaoting Chen | Zhaoting Chen, Steven Cunnington, Alkistis Pourtsidou, Laura Wolz, Marta Spinelli, Jos\'e Luis Bernal, Matilde Barberi-Squarotti, Stefano Camera, Isabella P. Carucci, Jos\'e Fonseca, Keith Grainge, Melis O. Irfan, Mario G. Santos, Jingying Wang (the MeerKLASS Collaboration) | Emission-line Stacking of 21cm Intensity Maps with MeerKLASS: Inference Pipeline and Application to the L-band Deep-field Data | 39 pages, 25 figures. Updated to match the version accepted by ApJS | null | 10.3847/1538-4365/add897 | null | astro-ph.CO astro-ph.GA | http://creativecommons.org/licenses/by/4.0/ | We present a novel analysis of observational systematics through the emission-line stacking of the MeerKLASS L-band deep-field intensity maps, following the detection in arXiv:2407.21626. A stacking signal is obtained by stacking the 21cm intensity map cubelets around the galaxy positions from the GAMA survey at $0.39\lesssim z \lesssim0.46$. An extensive simulation framework is built to study the viability of the stacking detection, the covariance estimation, and the model inference, which are then applied to the data. The statistical significance of the detection is $8.66\sigma$ when averaged into an angular map, and $7.45\sigma$ when averaged into a spectrum. The stacked spectrum exhibits an oscillating component of systematics, and we provide evidence that these systematics are a convolutional effect on the map data. The oscillation frequency matches the diffraction from the secondary reflector into the primary beam of the MeerKAT telescope. Bayesian inference can be used to constrain the systematics and the average HI emission of the galaxies. The fitting of the parameters gives a constraint on the systematics frequency $\nu_{\rm sys}\,[{\rm MHz}] = 17.90^{+6.53}_{-4.27}$. The posterior of the systematics amplitude reaches the wide prior and gives $A_{\rm sys}=0.50^{+0.33}_{-0.33}$. A tentative measurement of the average HI mass of the sources is achieved at $\log_{10}[\langle M_{HI}\rangle/M_\odot ]=9.84^{+0.48}_{-0.59}$, which is an underestimation limited by the narrow redshift bin, the strong degeneracy with the systematics, and the low-density galaxy sample. These shortfalls will be resolved for future MeerKLASS data to enable accurate measurements of the HI density through stacking of intensity maps. | [
{
"version": "v1",
"created": "Fri, 04 Apr 2025 20:07:06 GMT"
},
{
"version": "v2",
"created": "Tue, 10 Jun 2025 20:49:17 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Chen",
"Zhaoting",
"",
"the MeerKLASS Collaboration"
],
[
"Cunnington",
"Steven",
"",
"the MeerKLASS Collaboration"
],
[
"Pourtsidou",
"Alkistis",
"",
"the MeerKLASS Collaboration"
],
[
"Wolz",
"Laura",
"",
"the MeerKLASS Collaboration"
],
[
"Spinelli",
"Marta",
"",
"the MeerKLASS Collaboration"
],
[
"Bernal",
"José Luis",
"",
"the MeerKLASS Collaboration"
],
[
"Barberi-Squarotti",
"Matilde",
"",
"the MeerKLASS Collaboration"
],
[
"Camera",
"Stefano",
"",
"the MeerKLASS Collaboration"
],
[
"Carucci",
"Isabella P.",
"",
"the MeerKLASS Collaboration"
],
[
"Fonseca",
"José",
"",
"the MeerKLASS Collaboration"
],
[
"Grainge",
"Keith",
"",
"the MeerKLASS Collaboration"
],
[
"Irfan",
"Melis O.",
"",
"the MeerKLASS Collaboration"
],
[
"Santos",
"Mario G.",
"",
"the MeerKLASS Collaboration"
],
[
"Wang",
"Jingying",
"",
"the MeerKLASS Collaboration"
]
] |
2504.03943 | Julia Hsu | Imon Mia, Armi Tiihonen, Anna Ernst, Anusha Srivastava, Tonio Buonassisi, William Vandenberghe, and Julia W.P. Hsu | Multi-Variable Batch Bayesian Optimization in Materials Research: Synthetic Data Analysis of Noise Sensitivity and Problem Landscape Effects | null | null | null | null | stat.ML cond-mat.mtrl-sci cs.LG | http://creativecommons.org/licenses/by/4.0/ | Bayesian Optimization (BO) machine learning method is increasingly used to guide experimental optimization tasks in materials science. To emulate the large number of input variables and noise-containing results in experimental materials research, we perform batch BO simulation of six design variables with a range of noise levels. Two test cases relevant for materials science problems are examined: a needle-in-a-haystack case (Ackley function) that may be encountered in, e.g., molecule optimizations, and a smooth landscape with a local optimum in addition to the global optimum (Hartmann function) that may be encountered in, e.g., material composition optimization. We show learning curves, performance metrics, and visualization to effectively track the optimization progression and evaluate how the optimization outcomes are affected by noise, batch-picking method, choice of acquisition function, and exploration hyperparameter values. We find that the effects of noise depend on the problem landscape: noise degrades the optimization results of a needle-in-a-haystack search (Ackley) dramatically more. However, with increasing noise, we observe an increasing probability of landing on the local optimum in Hartmann. Therefore, prior knowledge of the problem domain structure and noise level is essential when designing BO for materials research experiments. Synthetic data studies -- with known ground truth and controlled noise levels -- enable us to isolate and evaluate the impact of different batch BO components, {\it e.g.}, acquisition policy, objective metrics, and hyperparameter values, before transitioning to the inherent uncertainties of real experimental systems. The results and methodology of this study will facilitate a greater utilization of BO in guiding experimental materials research, specifically in settings with a large number of design variables to optimize. | [
{
"version": "v1",
"created": "Fri, 04 Apr 2025 21:20:11 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 14:57:24 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Mia",
"Imon",
""
],
[
"Tiihonen",
"Armi",
""
],
[
"Ernst",
"Anna",
""
],
[
"Srivastava",
"Anusha",
""
],
[
"Buonassisi",
"Tonio",
""
],
[
"Vandenberghe",
"William",
""
],
[
"Hsu",
"Julia W. P.",
""
]
] |
2504.04118 | Darshan Kumar | Darshan Kumar (HNAS), Praveen Kumar Dhankar (SITN), Saibal Ray (GLA) and Fengge Zhang (HNAS) | Joint Analysis of Constraints on f(R) Parametrization from Recent Cosmological Observations | 15 pages, 6 figures, and 4 tables. Accepted for publication in Phys. Dark Univ | null | null | null | astro-ph.CO | http://creativecommons.org/licenses/by/4.0/ | In this study, we present constraints on the parameters of three well-known $f(R)$ gravity models, viz. (i) Hu-Sawicki, (ii) Starobinsky, and (iii) ArcTanh by using a joint analysis of recent cosmological observations. We perform analytical approximations for the Hubble parameter, $H(z)$, and cosmological distances in terms of the Hubble constant $(H_0)$, matter density $(\Omega_{m0})$, and a deviation parameter $b$ for each model. {Our analysis combines early and late-universe cosmological data from five cosmological observations:} (a) Hubble parameter measurements (Cosmic Chronometers), (b) Type Ia Supernovae (Union 3.0), (c) Baryon Acoustic Oscillations (DESI-2025), (d) Gamma-Ray Bursts (GRBs) and (e) Cosmic Microwave Background (CMB). We first optimize the models using each dataset independently, and subsequently, we perform a comprehensive joint analysis combining all four datasets. Our results show that the Hu-Sawicki and ArcTanh models do not deviate significantly from the $\Lambda$CDM model at 95% confidence level for individual datasets and remain consistent at 99% confidence level in the joint analysis. In contrast, the Starobinsky model shows a strong deviation and appears as a viable alternative to $\Lambda$CDM. We also constrain the transition redshift parameter ($z_t$), and check that the obtained value agrees with the values inferred from both early-time measurement (Planck) and late-time data from Type Ia Supernovae. These results support the potential support of $f(R)$ gravity to explain the late-time cosmic acceleration effectively. Finally, a statistical model comparison using $\chi^2_{\text{min}}$, AIC, and BIC indicates that all three $f(R)$ models are favored over $\Lambda$CDM, with the Starobinsky model receiving very strong support. | [
{
"version": "v1",
"created": "Sat, 05 Apr 2025 09:19:22 GMT"
},
{
"version": "v2",
"created": "Tue, 15 Apr 2025 12:01:11 GMT"
},
{
"version": "v3",
"created": "Wed, 11 Jun 2025 12:00:16 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Kumar",
"Darshan",
"",
"HNAS"
],
[
"Dhankar",
"Praveen Kumar",
"",
"SITN"
],
[
"Ray",
"Saibal",
"",
"GLA"
],
[
"Zhang",
"Fengge",
"",
"HNAS"
]
] |
2504.04211 | Junrong Lai | Junrong Lai, Changhong Li | Accelerated Bayesian Inference for Pulsar Timing Arrays: Normalizing Flows for Rapid Model Comparison Across Stochastic Gravitational-Wave Background Sources | null | null | null | null | astro-ph.CO | http://creativecommons.org/licenses/by-sa/4.0/ | The recent detection of nanohertz stochastic gravitational-wave backgrounds (SGWBs) by pulsar timing arrays (PTAs) promises unique insights into astrophysical and cosmological origins. However, traditional Markov Chain Monte Carlo (MCMC) approaches become prohibitively expensive for large datasets. We employ a normalizing flow (NF)-based machine learning framework to accelerate Bayesian inference in PTA analyses. For the first time, we perform Bayesian model comparison across SGWB source models in the framework of machine learning by training NF architectures on the PTA dataset (NANOGrav 15-year) and enabling direct evidence estimation via learned harmonic mean estimators. Our examples include 10 conventional SGWB source models such as supermassive black hole binaries, power-law spectrum, cosmic strings, domain walls, scalar-induced GWs, first-order phase transitions, and dual scenario/inflationary gravitational wave. Our approach jointly infers 20 red noise parameters and 2 SGWB parameters per model in $\sim 20$\,hours (including training), compared to $\sim 10$\,days with MCMC. Critically, the NF method preserves rigorous model selection accuracy, with small Hellinger distances ($\lesssim 0.3$) relative to MCMC posteriors, and reproduces MCMC-based Bayes factors across all tested scenarios. This scalable technique for SGWB source comparison will be essential for future PTA expansions and next-generation arrays such as the SKA, offering orders-of-magnitude efficiency gains without sacrificing physical interpretability. | [
{
"version": "v1",
"created": "Sat, 05 Apr 2025 15:44:57 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 13:51:54 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Lai",
"Junrong",
""
],
[
"Li",
"Changhong",
""
]
] |
2504.04745 | Ankush Raut | Ankush Raut, Xiaofeng Zhu, Maria Leonor Pacheco | Can LLMs Interpret and Leverage Structured Linguistic Representations? A Case Study with AMRs | 13 pages, 23 figures. Accepted at XLLM @ ACL 2025 | null | null | null | cs.CL | http://creativecommons.org/licenses/by/4.0/ | This paper evaluates the ability of Large Language Models (LLMs) to leverage contextual information in the form of structured linguistic representations. Specifically, we examine the impact of encoding both short and long contexts using Abstract Meaning Representation (AMR) structures across a diverse set of language tasks. We perform our analysis using 8-bit quantized and instruction-tuned versions of Llama 3.1 (8B), Phi-3, and Mistral 7B. Our results indicate that, for tasks involving short contexts, augmenting the prompt with the AMR of the original language context often degrades the performance of the underlying LLM. However, for tasks that involve long contexts, such as dialogue summarization in the SAMSum dataset, this enhancement improves LLM performance, for example, by increasing the zero-shot cosine similarity score of Llama 3.1 from 66% to 76%. This improvement is more evident in the newer and larger LLMs, but does not extend to the older or smaller ones. In addition, we observe that LLMs can effectively reconstruct the original text from a linearized AMR, achieving a cosine similarity of 81% in the best-case scenario. | [
{
"version": "v1",
"created": "Mon, 07 Apr 2025 05:38:40 GMT"
},
{
"version": "v2",
"created": "Sun, 08 Jun 2025 20:24:42 GMT"
},
{
"version": "v3",
"created": "Tue, 10 Jun 2025 19:47:54 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Raut",
"Ankush",
""
],
[
"Zhu",
"Xiaofeng",
""
],
[
"Pacheco",
"Maria Leonor",
""
]
] |
2504.04822 | Joaquin Garcia-Suarez | Joaquin Garcia-Suarez | Universal path decomposition of transfer and scattering matrices | 29 pages | null | null | null | physics.geo-ph cond-mat.other | http://creativecommons.org/licenses/by-sa/4.0/ | We report a universal identity: any entry of a one-dimensional transfer or scattering matrix comprising N layers equals a coherent sum of 2**(N-1) directed paths representing wave patterns with pre-defined amplitude and phase evolutions. Treating those paths as analytic building blocks, we derive closed-form results for arbitrary stratified media - optical, acoustic, elastic, or electronic - without resorting to matrix products or recursion. The combinatorial construction of paths turns layered system design into rule-based path engineering, illustrated with a design example that offers a reinterpretation of the quarter-wavelength principle. We also quantify the computational speed-up of the path method over classical transfer-matrix chaining and showcase two more cross-disciplinary applications (site-response seismology and quantum superlattices). This paradigm replaces numerical sweeps that employ the transfer matrix method with physically-transparent path-construction rules; its applicability spans across physical disciplines and scales: from nanometer optical coatings to kilometer-scale seismic strata. | [
{
"version": "v1",
"created": "Mon, 07 Apr 2025 08:22:38 GMT"
},
{
"version": "v2",
"created": "Wed, 09 Apr 2025 13:00:22 GMT"
},
{
"version": "v3",
"created": "Fri, 06 Jun 2025 16:15:09 GMT"
},
{
"version": "v4",
"created": "Tue, 10 Jun 2025 21:30:33 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Garcia-Suarez",
"Joaquin",
""
]
] |
2504.05369 | Adonis Bogris | A. Bogris, C. Simos, I. Simos, Y. Wang, A. Fichtner, S. Deligiannidis, N. S. Melis, C. Mesaritakis | Microwave Frequency Fiber Interferometry in Submarine Deployed Telecommunication Cables | null | Optical Fiber Communication Conference (OFC) 2025, Technical Digest Series (Optica Publishing Group, 2025), paper Th3F.1 | 10.1364/OFC.2025.Th3F.1 | null | physics.optics | http://creativecommons.org/licenses/by/4.0/ | We operated a microwave frequency fiber interferometer in a telecommunication cable in the Ionian Sea, Greece, for two months. The capability of detecting undersea micro earthquakes (magnitude~1.5), tides and ocean waves is reported | [
{
"version": "v1",
"created": "Mon, 07 Apr 2025 17:47:39 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Bogris",
"A.",
""
],
[
"Simos",
"C.",
""
],
[
"Simos",
"I.",
""
],
[
"Wang",
"Y.",
""
],
[
"Fichtner",
"A.",
""
],
[
"Deligiannidis",
"S.",
""
],
[
"Melis",
"N. S.",
""
],
[
"Mesaritakis",
"C.",
""
]
] |
2504.05659 | Ruijie Xu | Ruijie Xu | Integrability of Combinatorial Riemann Boundary Value Problem and Lattice Walks Avoiding a Quadrant | null | null | null | null | math.CO math-ph math.MP | http://creativecommons.org/licenses/by-sa/4.0/ | We introduce a general framework of matrix-form combinatorial Riemann boundary value problem (cRBVP) to characterize the integrability of functional equations arising in lattice walk enumerations. A matrix cRBVP is defined as integrable if it can be reduced to enough polynomial equations with one catalytic variable. Our central results establish that the integrability depends on the eigenspace of some matrix associated to the problem. For lattice walks in three quadrants, we demonstrate how the obstinate kernel method transforms a discrete difference equation into a $3\times 3$ matrix cRBVP. The special double-roots eigenvalue $1/4$ yields two independent polynomial equations in the problem. The other single-root eigenvalue yields a linear equation. We obtain three independent equations from a $3\times 3$ system. Crucially, our framework generalizes three-quadrant walks with Weyl symmetry to models satisfying only orbit-sum conditions. It explains many criteria about the orbit-sum proposed by various researchers and it also explains the counter-example of lattice walks starting outside the quadrant. | [
{
"version": "v1",
"created": "Tue, 08 Apr 2025 04:15:36 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 07:38:05 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Xu",
"Ruijie",
""
]
] |
2504.07129 | Christian Puntini | Christian Puntini | Near-Inertial Pollard Waves Modeling the Arctic Halocline | null | null | null | null | physics.ao-ph math.AP physics.flu-dyn | http://creativecommons.org/licenses/by/4.0/ | We present an explicit and exact solution to the governing equations describing the vertical structure of the Arctic Ocean region centered around the North Pole. The solution describes a stratified water column with three constant-density regions: a motionless bottom layer, a middle layer - the halocline - described by nonhydrostatic, near-inertial Pollard waves, and an upper layer presenting a mean current and a wave motion associated with the one in the halocline layer. | [
{
"version": "v1",
"created": "Thu, 03 Apr 2025 11:07:46 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 14:06:24 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Puntini",
"Christian",
""
]
] |
2504.07138 | Claudio Novelli | Claudio Novelli, Javier Argota S\'anchez-Vaquerizo, Dirk Helbing, Antonino Rotolo, Luciano Floridi | A Replica for our Democracies? On Using Digital Twins to Enhance Deliberative Democracy | null | null | null | null | cs.MA cs.CY cs.ET | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Deliberative democracy depends on carefully designed institutional frameworks, such as participant selection, facilitation methods, and decision-making mechanisms, that shape how deliberation performs. However, identifying optimal institutional designs for specific contexts remains challenging when relying solely on real-world observations or laboratory experiments: they can be expensive, ethically and methodologically tricky, or too limited in scale to give us clear answers. Computational experiments offer a complementary approach, enabling researchers to conduct large-scale investigations while systematically analyzing complex dynamics, emergent and unexpected collective behavior, and risks or opportunities associated with novel democratic designs. Therefore, this paper explores Digital Twin (DT) technology as a computational testing ground for deliberative systems (with potential applicability to broader institutional analysis). By constructing dynamic models that simulate real-world deliberation, DTs allow researchers and policymakers to rigorously test "what-if" scenarios across diverse institutional configurations in a controlled virtual environment. This approach facilitates evidence-based assessment of novel designs using synthetically generated data, bypassing the constraints of real-world or lab-based experimentation, and without societal disruption. The paper also discusses the limitations of this new methodological approach and suggests where future research should focus. | [
{
"version": "v1",
"created": "Mon, 07 Apr 2025 23:14:41 GMT"
},
{
"version": "v2",
"created": "Tue, 10 Jun 2025 23:11:07 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Novelli",
"Claudio",
""
],
[
"Sánchez-Vaquerizo",
"Javier Argota",
""
],
[
"Helbing",
"Dirk",
""
],
[
"Rotolo",
"Antonino",
""
],
[
"Floridi",
"Luciano",
""
]
] |
2504.07904 | Blake VanBerlo | Blake VanBerlo, Alexander Wong, Jesse Hoey, Robert Arntfield | The Efficacy of Semantics-Preserving Transformations in Self-Supervised Learning for Medical Ultrasound | 17 pages, 12 figures, 18 tables, Submitted to Medical Image Analysis | null | null | null | eess.IV cs.CV cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Data augmentation is a central component of joint embedding self-supervised learning (SSL). Approaches that work for natural images may not always be effective in medical imaging tasks. This study systematically investigated the impact of data augmentation and preprocessing strategies in SSL for lung ultrasound. Three data augmentation pipelines were assessed: (1) a baseline pipeline commonly used across imaging domains, (2) a novel semantic-preserving pipeline designed for ultrasound, and (3) a distilled set of the most effective transformations from both pipelines. Pretrained models were evaluated on multiple classification tasks: B-line detection, pleural effusion detection, and COVID-19 classification. Experiments revealed that semantics-preserving data augmentation resulted in the greatest performance for COVID-19 classification - a diagnostic task requiring global image context. Cropping-based methods yielded the greatest performance on the B-line and pleural effusion object classification tasks, which require strong local pattern recognition. Lastly, semantics-preserving ultrasound image preprocessing resulted in increased downstream performance for multiple tasks. Guidance regarding data augmentation and preprocessing strategies was synthesized for practitioners working with SSL in ultrasound. | [
{
"version": "v1",
"created": "Thu, 10 Apr 2025 16:26:47 GMT"
},
{
"version": "v2",
"created": "Tue, 10 Jun 2025 20:25:07 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"VanBerlo",
"Blake",
""
],
[
"Wong",
"Alexander",
""
],
[
"Hoey",
"Jesse",
""
],
[
"Arntfield",
"Robert",
""
]
] |
2504.08243 | Yulin An | Yulin An, Xueqi Zhao, Enrique del Castillo | Practical Implementation of an End-to-End Methodology for SPC of 3-D Part Geometry: A Case Study | 21 pages, 18 figures, Journal of Quality Technology (Accepted, to appear) | null | null | null | stat.AP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Del Castillo and Zhao (2020, 2021, 2022, 2024) have recently proposed a new methodology for the Statistical Process Control (SPC) of discrete parts whose 3-dimensional (3D) geometrical data are acquired with non-contact sensors. The approach is based on monitoring the spectrum of the Laplace-Beltrami (LB) operator of each scanned part estimated using finite element methods (FEM). The spectrum of the LB operator is an intrinsic summary of the geometry of a part, independent of the ambient space. Hence, registration of scanned parts is unnecessary when comparing them. The primary goal of this case study paper is to demonstrate the practical implementation of the spectral SPC methodology through multiple examples using real scanned parts acquired with an industrial-grade laser scanner, including 3D printed parts and commercial parts. We discuss the scanned mesh preprocessing needed in practice, including the type of remeshing found to be most beneficial for the FEM computations. For each part type, both the "phase I" and "phase II" stages of the spectral SPC methodology are showcased. In addition, we provide a new principled method to determine the number of eigenvalues of the LB operator to consider for efficient SPC of a given part geometry, and present an improved algorithm to automatically define a region of interest, particularly useful for large meshes. Computer codes that implement every method discussed in this paper, as well as all scanned part datasets used in the case studies, are made available and explained in the supplementary materials. | [
{
"version": "v1",
"created": "Fri, 11 Apr 2025 04:06:14 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 01:55:52 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"An",
"Yulin",
""
],
[
"Zhao",
"Xueqi",
""
],
[
"del Castillo",
"Enrique",
""
]
] |
2504.08380 | Peter Kr\"uger | Peter Kr\"uger | Circular dichroism in resonant photoelectron diffraction as a direct probe of sublattice magnetization in altermagnets | null | null | null | null | cond-mat.mtrl-sci | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Altermagnets are a new class of magnetic materials that are promising for spintronics technology. Here it is shown that in altermagnets, the circular dichroism (CD) in resonant photoelectron diffraction (RPED) contains a time-reversal odd signal, which provides a direct probe of the sublattice magnetization. RPED calculations are performed for MnTe at the Mn L2,3-edge resonance, using a combination of atomic multiplet and multiple scattering theory. A large magnetic CD is found for light helicity parallel to the N\'eel vector. This signal has the same angular distribution as the difference between the structural RPED of the two magnetic sublattices and its amplitude is approximately proportional to the X-ray magnetic CD in absorption of a single sublattice, thus providing a direct probe of the local magnetic moments. | [
{
"version": "v1",
"created": "Fri, 11 Apr 2025 09:32:09 GMT"
},
{
"version": "v2",
"created": "Tue, 10 Jun 2025 08:29:17 GMT"
},
{
"version": "v3",
"created": "Wed, 11 Jun 2025 09:40:51 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Krüger",
"Peter",
""
]
] |
2504.08534 | Alaa Eddine Mazouz | Alaa Mazouz, Duc Han Le, Van-Tam Nguyen | An FPGA Compiler for On-the-Fly Adaptive CNN Deployment and Reconfiguration | Submitted to IEEE Transactions on Computer-Aided Design of Integrated Circuits and Systems | null | null | null | cs.AR | http://creativecommons.org/licenses/by/4.0/ | We introduce ForgeMorph, a full-stack compiler for adaptive CNN deployment on FPGAs, combining design-time optimization with runtime reconfigurability. At compile time, the NeuroForge engine performs constraint-driven design space exploration, generating RTL mappings that are Pareto-optimal with respect to user-defined latency and resource budgets. Unlike existing FPGA compilers, which rely on static scheduling and manual tuning, NeuroForge leverages analytical performance models and multi-objective genetic algorithms to efficiently search large configuration spaces and propose highly optimized hardware implementations. At runtime, the NeuroMorph module enables dynamic reconfiguration of network width and depth without requiring redeployment. This is made possible by a novel training strategy, DistillCycle, which jointly trains the full model and its subnetworks using hierarchical knowledge distillation. As a result, each execution path maintains accuracy even under aggressive resource and power constraints. We demonstrate Forge-Morph on the Zynq-7100 using custom and benchmark models including MobileNetV2, ResNet-50, SqueezeNet, and YOLOv5. The system achieves up to 50x latency reduction and 32% lower power consumption at runtime, while matching or exceeding the efficiency of state-of-the-art compilers. ForgeMorph offers a unified solution for deployment scenarios that demand flexibility, performance, and hardware efficiency | [
{
"version": "v1",
"created": "Fri, 11 Apr 2025 13:43:02 GMT"
},
{
"version": "v2",
"created": "Tue, 10 Jun 2025 12:26:11 GMT"
},
{
"version": "v3",
"created": "Wed, 11 Jun 2025 09:53:35 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Mazouz",
"Alaa",
""
],
[
"Le",
"Duc Han",
""
],
[
"Nguyen",
"Van-Tam",
""
]
] |
2504.08760 | Yi-Cai Zhang | Guang-Xin Pang, Zhi Li, Shan-Zhong Li, Yan-Yang Zhang, Jun-Feng Liu, and Yi-Cai Zhang | Exact mobility line and mobility ring in the complex energy plane of a flat band lattice with a non-Hermitian quasiperiodic potential | 13 pages, 9 figures | Phys. Rev. B 111, 214205 (2025) | 10.1103/z96m-ckm6 | null | cond-mat.dis-nn cond-mat.quant-gas | http://creativecommons.org/publicdomain/zero/1.0/ | In this study, we investigate the problem of Anderson localization in a one-dimensional flat band lattice with a non-Hermitian quasiperiodic on-site potential. First of all, we discuss the influences of non-Hermitian potentials on the existence of critical states. Our findings show that, unlike in Hermitian cases, the non-Hermiticity of the potential leads to the disappearance of critical states and critical regions. Furthermore, we are able to accurately determine the Lyapunov exponents and the mobility edges. Our results reveal that the mobility edges form mobility lines and mobility rings in the complex energy plane. Within the mobility rings, the eigenstates are extended, while the localized states are located outside the mobility rings. For mobility line cases, only when the eigenenergies lie on the mobility lines, their corresponding eigenstates are extended states.Finally, as the energy approaches the mobility edges, we observe that, differently from Hermitian cases, here the critical index of the localization length is not a constant, but rather varies depending on the positions of the mobility edges. | [
{
"version": "v1",
"created": "Mon, 31 Mar 2025 01:55:12 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Pang",
"Guang-Xin",
""
],
[
"Li",
"Zhi",
""
],
[
"Li",
"Shan-Zhong",
""
],
[
"Zhang",
"Yan-Yang",
""
],
[
"Liu",
"Jun-Feng",
""
],
[
"Zhang",
"Yi-Cai",
""
]
] |
2504.10552 | Dmitry Ignatov PhD | Arash Torabi Goodarzi, Roman Kochnev, Waleed Khalid, Furui Qin, Tolgay Atinc Uzun, Yashkumar Sanjaybhai Dhameliya, Yash Kanubhai Kathiriya, Zofia Antonina Bentyn, Dmitry Ignatov, Radu Timofte | LEMUR Neural Network Dataset: Towards Seamless AutoML | null | null | null | null | cs.LG cs.AI cs.CV cs.DL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Neural networks are fundamental in artificial intelligence, driving progress in computer vision and natural language processing. High-quality datasets are crucial for their development, and there is growing interest in datasets composed of neural networks themselves to support benchmarking, automated machine learning (AutoML), and model analysis. We introduce LEMUR, an open source dataset of neural network models with well-structured code for diverse architectures across tasks such as object detection, image classification, segmentation, and natural language processing. LEMUR is primarily designed to provide a rich source of structured model representations and associated performance data, enabling the fine-tuning of large language models for AutoML applications. Leveraging Python and PyTorch, LEMUR enables seamless extension to new datasets and models while maintaining consistency. It integrates an Optuna-powered framework for evaluation, hyperparameter optimization, statistical analysis, and graphical insights. LEMUR VR extension enables the seamless deployment of models in virtual reality, optimizing their performance on resource-constrained devices. Providing tools for model evaluation, preprocessing, and database management, LEMUR supports researchers and practitioners in developing, testing, and analyzing neural networks. It offers an API that delivers comprehensive information about neural network models and their complete performance statistics with a single request, which can be used in experiments with code-generating large language models. The LEMUR and its plugins are accessible as open source projects under the MIT license at https://github.com/ABrain-One/nn-dataset, https://github.com/ABrain-One/nn-plots and https://github.com/ABrain-One/nn-vr. | [
{
"version": "v1",
"created": "Mon, 14 Apr 2025 09:08:00 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 08:16:41 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Goodarzi",
"Arash Torabi",
""
],
[
"Kochnev",
"Roman",
""
],
[
"Khalid",
"Waleed",
""
],
[
"Qin",
"Furui",
""
],
[
"Uzun",
"Tolgay Atinc",
""
],
[
"Dhameliya",
"Yashkumar Sanjaybhai",
""
],
[
"Kathiriya",
"Yash Kanubhai",
""
],
[
"Bentyn",
"Zofia Antonina",
""
],
[
"Ignatov",
"Dmitry",
""
],
[
"Timofte",
"Radu",
""
]
] |
2504.11171 | Johannes Jakubik | Johannes Jakubik, Felix Yang, Benedikt Blumenstiel, Erik Scheurer, Rocco Sedona, Stefano Maurogiovanni, Jente Bosmans, Nikolaos Dionelis, Valerio Marsocci, Niklas Kopp, Rahul Ramachandran, Paolo Fraccaro, Thomas Brunschwiler, Gabriele Cavallaro, Juan Bernabe-Moreno, Nicolas Long\'ep\'e | TerraMind: Large-Scale Generative Multimodality for Earth Observation | null | null | null | null | cs.CV cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present TerraMind, the first any-to-any generative, multimodal foundation model for Earth observation (EO). Unlike other multimodal models, TerraMind is pretrained on dual-scale representations combining both token-level and pixel-level data across modalities. On a token level, TerraMind encodes high-level contextual information to learn cross-modal relationships, while on a pixel level, TerraMind leverages fine-grained representations to capture critical spatial nuances. We pretrained TerraMind on nine geospatial modalities of a global, large-scale dataset. In this paper, we demonstrate that (i) TerraMind's dual-scale early fusion approach unlocks a range of zero-shot and few-shot applications for Earth observation, (ii) TerraMind introduces "Thinking-in-Modalities" (TiM) -- the capability of generating additional artificial data during finetuning and inference to improve the model output -- and (iii) TerraMind achieves beyond state-of-the-art performance in community-standard benchmarks for EO like PANGAEA. The pretraining dataset, the model weights, and our code are open-sourced under a permissive license. | [
{
"version": "v1",
"created": "Tue, 15 Apr 2025 13:17:39 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 17:44:41 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Jakubik",
"Johannes",
""
],
[
"Yang",
"Felix",
""
],
[
"Blumenstiel",
"Benedikt",
""
],
[
"Scheurer",
"Erik",
""
],
[
"Sedona",
"Rocco",
""
],
[
"Maurogiovanni",
"Stefano",
""
],
[
"Bosmans",
"Jente",
""
],
[
"Dionelis",
"Nikolaos",
""
],
[
"Marsocci",
"Valerio",
""
],
[
"Kopp",
"Niklas",
""
],
[
"Ramachandran",
"Rahul",
""
],
[
"Fraccaro",
"Paolo",
""
],
[
"Brunschwiler",
"Thomas",
""
],
[
"Cavallaro",
"Gabriele",
""
],
[
"Bernabe-Moreno",
"Juan",
""
],
[
"Longépé",
"Nicolas",
""
]
] |
2504.11562 | Borja Diez | Crist\'obal Corral, Borja Diez, Daniel Flores-Alfonso, Nelson Merino, Leonardo Sanhueza | Inhomogeneous metrics on complex bundles in Lovelock gravity | Matches published version | Phys. Rev. D 111 (2025) 124016 | 10.1103/PhysRevD.111.124016 | null | hep-th | http://creativecommons.org/licenses/by/4.0/ | We consider Lovelock gravity in arbitrary, even dimensions. We find a large class of new gravitational instantons by considering extended nontrivial circle bundles over K\"ahler manifolds. Concretely, we generalize the Page-Pope metric in the presence of higher-curvature corrections of the Lovelock class. A subset of these spaces admits analytic continuation into the Lorentzian sector, producing new stationary solutions in Lovelock gravity. The geometries are fully determined by a single algebraic equation. We also obtain necessary and sufficient conditions for Lovelock-constant K\"ahler manifolds to exist in Lovelock gravity. Finally, we find a wide class of Lovelock-Maxwell solutions beyond staticity, allowing us to obtain the electrovacuum extension of these instantons. | [
{
"version": "v1",
"created": "Tue, 15 Apr 2025 19:10:04 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 14:56:18 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Corral",
"Cristóbal",
""
],
[
"Diez",
"Borja",
""
],
[
"Flores-Alfonso",
"Daniel",
""
],
[
"Merino",
"Nelson",
""
],
[
"Sanhueza",
"Leonardo",
""
]
] |
2504.11639 | Ruy Exel | M. Dokuchaev, R. Exel, and H. Pinedo | Twisted Steinberg algebras, regular inclusions and induction | This update includes an alternative characterization of regular inclusions (Proposition 5.9) | null | null | null | math.OA math.RA math.RT | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Given a field $K$ and an ample (not necessarily Hausdorff) groupoid $G$, we define the concept of a line bundle over $G$ inspired by the well known concept from the theory of C*-algebras. If $E$ is such a line bundle, we construct the associated twisted Steinberg algebra in terms of sections of $E$, extending the original construction introduced independently by Steinberg in 2010, and by Clark, Farthing, Sims and Tomforde in a 2014 paper (originally announced in 2011). We also generalize (strictly, in the non-Hausdorff case) the 2023 construction of (cocycle) twisted Steinberg algebras of Armstrong, Clark, Courtney, Lin, Mccormick and Ramagge. We then extend Steinberg's theory of induction of modules, not only to the twisted case, but to the much more general case of regular inclusions of algebras. Among our main results, we show that, under appropriate conditions, every irreducible module is induced by an irreducible module over a certain abstractly defined isotropy algebra. We also describe a process of disintegration of modules and use it to prove a version of the Effros-Hahn conjecture, showing that every primitive ideal coincides with the annihilator of a module induced from isotropy. | [
{
"version": "v1",
"created": "Tue, 15 Apr 2025 22:02:44 GMT"
},
{
"version": "v2",
"created": "Thu, 17 Apr 2025 02:08:15 GMT"
},
{
"version": "v3",
"created": "Fri, 09 May 2025 18:08:59 GMT"
},
{
"version": "v4",
"created": "Tue, 10 Jun 2025 23:52:58 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Dokuchaev",
"M.",
""
],
[
"Exel",
"R.",
""
],
[
"Pinedo",
"H.",
""
]
] |
2504.12347 | Ville Heilala | Mika Set\"al\"a, Pieta Sikstr\"om, Ville Heilala, Tommi K\"arkk\"ainen | Assessment of Evolving Large Language Models in Upper Secondary Mathematics | null | null | null | null | cs.CL cs.AI cs.CY | http://creativecommons.org/licenses/by/4.0/ | Large language models (LLMs) have shown increasing promise in educational settings, yet their mathematical reasoning has been considered evolving. This study evaluates the mathematical capabilities of various LLMs using the Finnish matriculation examination, a high-stakes digital test for upper secondary education. Initial tests yielded moderate performance corresponding to mid-range grades, but later evaluations demonstrated substantial improvements as the language models evolved. Remarkably, some models achieved near-perfect or perfect scores, matching top student performance and qualifying for university admission. Our findings highlight the rapid advances in the mathematical proficiency of LLMs and illustrate their potential as underlying tools to support learning and teaching in a variety of ways. | [
{
"version": "v1",
"created": "Tue, 15 Apr 2025 18:31:54 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 10:49:35 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Setälä",
"Mika",
""
],
[
"Sikström",
"Pieta",
""
],
[
"Heilala",
"Ville",
""
],
[
"Kärkkäinen",
"Tommi",
""
]
] |
2504.12475 | Philip Hoggan Pr | Ali Bagci and Philip E Hoggan | Best practices in Quantum Monte Carlo for metal catalysis: CO hydrolysis on Pt(111) | 22 pages,2 figures. arXiv admin note: substantial text overlap with arXiv:2212.01823; text overlap with arXiv:2004.10565, arXiv:2202.00542 | null | null | null | cond-mat.mtrl-sci | http://creativecommons.org/licenses/by/4.0/ | Hydrogen production as a clean, sustainable replacement for fossil fuels is gathering pace. Research work and prototyping various aspects of hydrogen power is now a priority. Over 90 \% of all chemical manufacture uses a solid catalyst. This work describes catalytic selective hydrogen production optimising reactant structure on a model catalyst. Focus is on O-H bond dissociation in format radicals formed after carbon-monoxide is co-adsorbed with water at Pt(111). Finally, hydrogen gas is given off. Many chemical reactions involve bond-dissociation. This process is often the key to rate-limiting reaction steps at solid surfaces. %This is also true for reactions at solid surfaces, in which the dissociation step is often limiting but facilitated in comparison to gas phase reaction channels. Since bond-breaking is poorly described by Hartree-Fock and DFT methods, our embedded active site approach is used. This work demonstrates Quantum Monte Carlo (QMC) methodology using a very simple four primitive-cell layer model, oriented to expose Pt (111). QMC is a stochastic approach to solving the Schr{\"o}dinger equation recently came of age for heterogeneous systems involving solids. During hydrolysis of carbon monoxide, initial O-H bond stretch is rate-limiting. This dissociation energy is offset by Pt-H bond formation at the surface. The reactive formate (H-O-C=O) species formed, by initial hydrolysis of CO, also interact with a vicinal Pt. These are subsequently desorbed. They then produce carbon dioxide and hydrogen, with a H-atom dissociated from the formate species and another desorbed at the Pt(111) face.
Our approach allows a high-level configuration interaction (CI) wave-function to be used, expanded in plane-waves and embedded in the metal lattice exposing its close-packed face. The resulting periodic function is used to guide the QMC calculation. | [
{
"version": "v1",
"created": "Wed, 16 Apr 2025 20:25:46 GMT"
},
{
"version": "v2",
"created": "Wed, 23 Apr 2025 16:54:36 GMT"
},
{
"version": "v3",
"created": "Wed, 11 Jun 2025 16:47:09 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Bagci",
"Ali",
""
],
[
"Hoggan",
"Philip E",
""
]
] |
2504.12501 | Nathan Lambert | Nathan Lambert | Reinforcement Learning from Human Feedback | 131 pages. Web-native version at https://rlhfbook.com/ v2 adds more reasoning content | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Reinforcement learning from human feedback (RLHF) has become an important technical and storytelling tool to deploy the latest machine learning systems. In this book, we hope to give a gentle introduction to the core methods for people with some level of quantitative background. The book starts with the origins of RLHF -- both in recent literature and in a convergence of disparate fields of science in economics, philosophy, and optimal control. We then set the stage with definitions, problem formulation, data collection, and other common math used in the literature. The core of the book details every optimization stage in using RLHF, from starting with instruction tuning to training a reward model and finally all of rejection sampling, reinforcement learning, and direct alignment algorithms. The book concludes with advanced topics -- understudied research questions in synthetic data and evaluation -- and open questions for the field. | [
{
"version": "v1",
"created": "Wed, 16 Apr 2025 21:36:46 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 15:15:22 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Lambert",
"Nathan",
""
]
] |
2504.12663 | Xiaotian Zhang | Xiaotian Zhang, Ruizhe Chen, Yang Feng, Zuozhu Liu | Persona-judge: Personalized Alignment of Large Language Models via Token-level Self-judgment | ACL Finding | null | null | null | cs.CL cs.AI | http://creativecommons.org/licenses/by/4.0/ | Aligning language models with human preferences presents significant challenges, particularly in achieving personalization without incurring excessive computational costs. Existing methods rely on reward signals and additional annotated data, limiting their scalability and adaptability to diverse human values. To address these challenges, we introduce Persona-judge, a novel discriminative paradigm that enables training-free personalized alignment with unseen preferences. Instead of optimizing policy parameters through external reward feedback, Persona-judge leverages the intrinsic preference judgment capabilities of the model. Specifically, a draft model generates candidate tokens conditioned on a given preference, while a judge model, embodying another preference, cross-validates the predicted tokens whether to be accepted. Experimental results demonstrate that Persona-judge, using the inherent preference evaluation mechanisms of the model, offers a scalable and computationally efficient solution to personalized alignment, paving the way for more adaptive customized alignment. Our code is available here. | [
{
"version": "v1",
"created": "Thu, 17 Apr 2025 05:50:13 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 07:31:37 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Zhang",
"Xiaotian",
""
],
[
"Chen",
"Ruizhe",
""
],
[
"Feng",
"Yang",
""
],
[
"Liu",
"Zuozhu",
""
]
] |
2504.15374 | Lekhika Malhotra | Zafri A. Borboruah, Frank F. Deppisch, Anish Ghoshal, Lekhika Malhotra | Inflationary Gravitational Waves and Laboratory Searches as Complementary Probes of Right-handed Neutrinos | 18 pages including references, 16 figures including sub-figures, minor changes, more plots added | null | null | null | hep-ph astro-ph.CO | http://creativecommons.org/licenses/by/4.0/ | We analyze the damping of inflationary gravitational waves (GW) that re-enter the Hubble horizon before or during a post-inflationary era dominated by a meta-stable, right-handed neutrino (RHN), whose out-of-equilibrium decay releases entropy. Within a minimal type-I seesaw extension of the Standard Model (SM), we explore the conditions under which the population of thermally produced RHNs remain long-lived and cause a period of matter-domination. We find that the suppression of the GW spectrum occurs above a characteristic frequency determined by the RHN mass and active-sterile mixing. For RHN masses in the range $0.1$-$10$ GeV and mixing $10^{-12} \lesssim |V_{eN}|^2 \lesssim 10^{-5}$, we estimate such characteristic frequencies and the signal-to-noise ratio to assess the detection prospects in GW observatories such as THEIA, $\mu$-ARES, LISA, BBO and ET. Additionally we use LIGO data to put upper bounds on the reheating temperature after inflation, for a given blue-tilted GW spectrum. We find complementarity between GW signals and laboratory searches in SHiP, DUNE and LEGEND-1000. Notably, RHN masses of $0.2$-$2$ GeV and mixing $10^{-10} \lesssim |V_{eN}|^2 \lesssim 10^{-7}$ are testable in both laboratory experiments and GW observations. Additionally, GW experiments can probe the canonical seesaw regime of light neutrino mass generation, a region largely inaccessible to laboratory searches. | [
{
"version": "v1",
"created": "Mon, 21 Apr 2025 18:33:37 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 11:09:46 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Borboruah",
"Zafri A.",
""
],
[
"Deppisch",
"Frank F.",
""
],
[
"Ghoshal",
"Anish",
""
],
[
"Malhotra",
"Lekhika",
""
]
] |
2504.15629 | Harsh Maheshwari | Harsh Maheshwari, Srikanth Tenneti, Alwarappan Nakkiran | CiteFix: Enhancing RAG Accuracy Through Post-Processing Citation Correction | null | null | null | null | cs.IR cs.CL | http://creativecommons.org/licenses/by/4.0/ | Retrieval Augmented Generation (RAG) has emerged as a powerful application of Large Language Models (LLMs), revolutionizing information search and consumption. RAG systems combine traditional search capabilities with LLMs to generate comprehensive answers to user queries, ideally with accurate citations. However, in our experience of developing a RAG product, LLMs often struggle with source attribution, aligning with other industry studies reporting citation accuracy rates of only about 74% for popular generative search engines. To address this, we present efficient post-processing algorithms to improve citation accuracy in LLM-generated responses, with minimal impact on latency and cost. Our approaches cross-check generated citations against retrieved articles using methods including keyword + semantic matching, fine tuned model with BERTScore, and a lightweight LLM-based technique. Our experimental results demonstrate a relative improvement of 15.46% in the overall accuracy metrics of our RAG system. This significant enhancement potentially enables a shift from our current larger language model to a relatively smaller model that is approximately 12x more cost-effective and 3x faster in inference time, while maintaining comparable performance. This research contributes to enhancing the reliability and trustworthiness of AI-generated content in information retrieval and summarization tasks which is critical to gain customer trust especially in commercial products. | [
{
"version": "v1",
"created": "Tue, 22 Apr 2025 06:41:25 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 07:56:14 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Maheshwari",
"Harsh",
""
],
[
"Tenneti",
"Srikanth",
""
],
[
"Nakkiran",
"Alwarappan",
""
]
] |
2504.17656 | Ayush Jain | Ayush Jain, Rampi Ramprasad | polyGen: A Learning Framework for Atomic-level Polymer Structure Generation | null | null | null | null | cs.CE cond-mat.mtrl-sci cs.LG | http://creativecommons.org/licenses/by/4.0/ | Synthetic polymeric materials underpin fundamental technologies in the energy, electronics, consumer goods, and medical sectors, yet their development still suffers from prolonged design timelines. Although polymer informatics tools have supported speedup, polymer simulation protocols continue to face significant challenges in the on-demand generation of realistic 3D atomic structures that respect conformational diversity. Generative algorithms for 3D structures of inorganic crystals, bio-polymers, and small molecules exist, but have not addressed synthetic polymers because of challenges in representation and dataset constraints. In this work, we introduce polyGen, the first generative model designed specifically for polymer structures from minimal inputs such as the repeat unit chemistry alone. polyGen combines graph-based encodings with a latent diffusion transformer using positional biased attention for realistic conformation generation. Given the limited dataset of 3,855 DFT-optimized polymer structures, we incorporate joint training with small molecule data to enhance generation quality. We also establish structure matching criteria to benchmark our approach on this novel problem. polyGen overcomes the limitations of traditional crystal structure prediction methods for polymers, successfully generating realistic and diverse linear and branched conformations, with promising performance even on challenging large repeat units. As the first atomic-level proof-of-concept capturing intrinsic polymer flexibility, it marks a new capability in material structure generation. | [
{
"version": "v1",
"created": "Thu, 24 Apr 2025 15:26:00 GMT"
},
{
"version": "v2",
"created": "Tue, 10 Jun 2025 05:41:16 GMT"
},
{
"version": "v3",
"created": "Wed, 11 Jun 2025 01:33:57 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Jain",
"Ayush",
""
],
[
"Ramprasad",
"Rampi",
""
]
] |
2504.17834 | Haokai Zhang | Haokai Zhang, Shengtao Zhang, Zijian Cai, Heng Wang, Ruixuan Zhu, Zinan Zeng, Minnan Luo | Unveiling the Hidden: Movie Genre and User Bias in Spoiler Detection | ECML PKDD 2025 | null | null | null | cs.IR cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Spoilers in movie reviews are important on platforms like IMDb and Rotten Tomatoes, offering benefits and drawbacks. They can guide some viewers' choices but also affect those who prefer no plot details in advance, making effective spoiler detection essential. Existing spoiler detection methods mainly analyze review text, often overlooking the impact of movie genres and user bias, limiting their effectiveness. To address this, we analyze movie review data, finding genre-specific variations in spoiler rates and identifying that certain users are more likely to post spoilers. Based on these findings, we introduce a new spoiler detection framework called GUSD (The code is available at https://github.com/AI-explorer-123/GUSD) (Genre-aware and User-specific Spoiler Detection), which incorporates genre-specific data and user behavior bias. User bias is calculated through dynamic graph modeling of review history. Additionally, the R2GFormer module combines RetGAT (Retentive Graph Attention Network) for graph information and GenreFormer for genre-specific aggregation. The GMoE (Genre-Aware Mixture of Experts) model further assigns reviews to specialized experts based on genre. Extensive testing on benchmark datasets shows that GUSD achieves state-of-the-art results. This approach advances spoiler detection by addressing genre and user-specific patterns, enhancing user experience on movie review platforms. | [
{
"version": "v1",
"created": "Thu, 24 Apr 2025 15:34:35 GMT"
},
{
"version": "v2",
"created": "Mon, 28 Apr 2025 07:44:25 GMT"
},
{
"version": "v3",
"created": "Wed, 11 Jun 2025 14:58:32 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Zhang",
"Haokai",
""
],
[
"Zhang",
"Shengtao",
""
],
[
"Cai",
"Zijian",
""
],
[
"Wang",
"Heng",
""
],
[
"Zhu",
"Ruixuan",
""
],
[
"Zeng",
"Zinan",
""
],
[
"Luo",
"Minnan",
""
]
] |
2504.18574 | Aviv Bick | Aviv Bick, Eric Xing, Albert Gu | Understanding the Skill Gap in Recurrent Language Models: The Role of the Gather-and-Aggregate Mechanism | null | null | null | null | cs.LG cs.AI | http://creativecommons.org/licenses/by/4.0/ | State-space models (SSMs) offer efficient alternatives to Transformers for long sequences, but their fixed-size recurrent state limits capability on algorithmic tasks, such as retrieving past context. In this work, we examine how in-context retrieval operates in Transformer- and SSM-based language models and find that both rely on a similar Gather-and-Aggregate (G&A) mechanism: a Gather Head extracts relevant information pieces from context, which an Aggregate Head integrates into a single representation. In both architectures, G&A concentrates in a few heads, forming critical bottlenecks even for simple retrieval. For example, we show that disabling a single Gather or Aggregate Head in a pruned Llama-3.1-8B impairs retrieving the correct answer letter in MMLU, reducing its accuracy from 66% to 25% (random guessing). Moreover, this retrieval bottleneck can obscure limited knowledge demands of tasks as the pruned model succeeds on MMLU with functioning G&A heads yet fails on other knowledge benchmarks. The bottleneck similarly extends to tasks where SSMs typically underperform, such as GSM8K, BBH, and dialogue comprehension. We show that SSMs' retrieval challenges manifest in these heads, creating smoother attention patterns instead of the sharp token transitions effective G&A requires. Thus, the Transformer-SSM retrieval gap exists in just a few heads, rather than the entire language model. This suggests a unified explanation for Transformer vs. SSM performance gap while showing how to merge their strengths. We find that pretrained hybrid models, where SSMs are combined with a few attention layers, delegate the role of Aggregate Heads to attention. Similarly, replacing a single G&A head in a pretrained SSM with an attention variant boosts retrieval and benchmark scores. | [
{
"version": "v1",
"created": "Tue, 22 Apr 2025 16:15:19 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 11:06:08 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Bick",
"Aviv",
""
],
[
"Xing",
"Eric",
""
],
[
"Gu",
"Albert",
""
]
] |
2504.18756 | Rezowan Shuvo | Rezowan Shuvo, M S Mekala and Eyad Elyan | Multi-Stage Boundary-Aware Transformer Network for Action Segmentation in Untrimmed Surgical Videos | null | null | null | null | cs.CV | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Understanding actions within surgical workflows is critical for evaluating post-operative outcomes and enhancing surgical training and efficiency. Capturing and analyzing long sequences of actions in surgical settings is challenging due to the inherent variability in individual surgeon approaches, which are shaped by their expertise and preferences. This variability complicates the identification and segmentation of distinct actions with ambiguous boundary start and end points. The traditional models, such as MS-TCN, which rely on large receptive fields, that causes over-segmentation, or under-segmentation, where distinct actions are incorrectly aligned. To address these challenges, we propose the Multi-Stage Boundary-Aware Transformer Network (MSBATN) with hierarchical sliding window attention to improve action segmentation. Our approach effectively manages the complexity of varying action durations and subtle transitions by accurately identifying start and end action boundaries in untrimmed surgical videos. MSBATN introduces a novel unified loss function that optimises action classification and boundary detection as interconnected tasks. Unlike conventional binary boundary detection methods, our innovative boundary weighing mechanism leverages contextual information to precisely identify action boundaries. Extensive experiments on three challenging surgical datasets demonstrate that MSBATN achieves state-of-the-art performance, with superior F1 scores at 25% and 50%. thresholds and competitive results across other metrics. | [
{
"version": "v1",
"created": "Sat, 26 Apr 2025 01:07:56 GMT"
},
{
"version": "v2",
"created": "Tue, 10 Jun 2025 22:05:32 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Shuvo",
"Rezowan",
""
],
[
"Mekala",
"M S",
""
],
[
"Elyan",
"Eyad",
""
]
] |
2504.20223 | Ethan Partington | Ethan R. Partington, Edward M. Cackett, Rick Edelson, Keith Horne, Jake A. Miller, Aaron J. Barth, Jonathan Gelbord, and Juan V. Hern\'andez Santisteban | Testing X-ray Reprocessing and Mapping the Soft Excess of NGC 7469 with NICER | 18 pages, 8 figures, Accepted to ApJ | ApJ 986 81 (2025) | 10.3847/1538-4357/add3ef | null | astro-ph.HE astro-ph.GA | http://creativecommons.org/licenses/by/4.0/ | We present an X-ray/UV reverberation analysis of NGC 7469 across 210 days, using daily NICER observations with contemporaneous monitoring by Swift UVOT+XRT. We model the X-ray spectrum with a power law continuum and a soft excess during each NICER epoch. These emission sources demonstrate correlated flux variability with a lag consistent with zero days. We find that the power law emission is consistent with a compact X-ray corona, and that the soft excess can be explainedby reflected coronal emission from the inner accretion disk. We test the relationship between changes in the flux of the X-ray corona and the UVW2 continuum, finding strong correlation and a negative X-ray lag of less than one day. This is consistent with a scenario in which the X-ray corona drives the UVW2 light curve through thermal reprocessing | [
{
"version": "v1",
"created": "Mon, 28 Apr 2025 19:45:18 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Partington",
"Ethan R.",
""
],
[
"Cackett",
"Edward M.",
""
],
[
"Edelson",
"Rick",
""
],
[
"Horne",
"Keith",
""
],
[
"Miller",
"Jake A.",
""
],
[
"Barth",
"Aaron J.",
""
],
[
"Gelbord",
"Jonathan",
""
],
[
"Santisteban",
"Juan V. Hernández",
""
]
] |
2504.20462 | Qi Wang | Qi Wang, Xiao Zhang, Mingyi Li, Yuan Yuan, Mengbai Xiao, Fuzhen Zhuang and Dongxiao Yu | TAMO:Fine-Grained Root Cause Analysis via Tool-Assisted LLM Agent with Multi-Modality Observation Data in Cloud-Native Systems | null | null | null | null | cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | With the development of distributed systems, microservices and cloud native technologies have become central to modern enterprise software development. Despite bringing significant advantages, these technologies also increase system complexity and operational challenges. Traditional root cause analysis (RCA) struggles to achieve automated fault response, heavily relying on manual intervention. In recent years, large language models (LLMs) have made breakthroughs in contextual inference and domain knowledge integration, providing new solutions for Artificial Intelligence for Operations (AIOps). However, Existing LLM-based approaches face three key challenges: text input constraints, dynamic service dependency hallucinations, and context window limitations. To address these issues, we propose a tool-assisted LLM agent with multi-modality observation data, namely TAMO, for fine-grained RCA. It unifies multi-modal observational data into time-aligned representations to extract consistent features and employs specialized root cause localization and fault classification tools for perceiving the contextual environment. This approach overcomes the limitations of LLM in handling real-time changing service dependencies and raw observational data and guides LLM to generate repair strategies aligned with system contexts by structuring key information into a prompt. Experimental results show that TAMO performs well in root cause analysis when dealing with public datasets characterized by heterogeneity and common fault types, demonstrating its effectiveness. | [
{
"version": "v1",
"created": "Tue, 29 Apr 2025 06:50:48 GMT"
},
{
"version": "v2",
"created": "Wed, 30 Apr 2025 10:20:10 GMT"
},
{
"version": "v3",
"created": "Wed, 11 Jun 2025 02:57:02 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Wang",
"Qi",
""
],
[
"Zhang",
"Xiao",
""
],
[
"Li",
"Mingyi",
""
],
[
"Yuan",
"Yuan",
""
],
[
"Xiao",
"Mengbai",
""
],
[
"Zhuang",
"Fuzhen",
""
],
[
"Yu",
"Dongxiao",
""
]
] |
2504.20641 | Karl-Henning Rehren | Iustus C. Hemprich, Karl-Henning Rehren | Dressed fields for Quantum Chromodynamics | v2: 26 pages. More precise formulation of the Master Ward Identity. Other minor improvements | null | null | null | hep-th math-ph math.MP | http://creativecommons.org/licenses/by/4.0/ | String-localized QFT allows to explain Standard Model interactions in an autonomous way, committed to quantum principles rather than a "gauge principle", thus avoiding an indefinite state space and compensating ghosts. The resulting perturbative scattering matrix is known (at tree-level) to be insensitive to the non-locality of the auxiliary "string-localized fields" used in the construction. For the examples of Yang-Mills and QCD, we prove that it is actually equivalent to the perturbative S-matrix of gauge theory, restricted to physical particle states. The role of classical gauge invariance is revealed along the way. The main tool are "dressed fields", that are intermediate between free fields and interacting fields, and for which we give explicit formulas at all orders. The renormalization of loops, as well as non-perturbative issues are not adressed, but we hint at the possibility, enabled by our approach, that qualitative traces of confinement may be visible already at the level of the dressed fields. | [
{
"version": "v1",
"created": "Tue, 29 Apr 2025 11:11:54 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 14:15:25 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Hemprich",
"Iustus C.",
""
],
[
"Rehren",
"Karl-Henning",
""
]
] |
2504.20955 | Elias Mann | Elias L. Mann, Corin C. Wagen, Jonathon E. Vandezande, Arien M. Wagen, Spencer C. Schneider | Egret-1: Pretrained Neural Network Potentials for Efficient and Accurate Bioorganic Simulation | 33 pages, 8 figures | null | null | null | physics.chem-ph | http://creativecommons.org/licenses/by/4.0/ | Accurate simulation of atomic systems has the potential to revolutionize the design of molecules and materials. Unfortunately, exact solutions of the Schr\"odinger equation scale as O(N!) and remain inaccessible for systems with more than a handful of atoms, forcing scientists to accept steep tradeoffs between speed and accuracy and limiting the reliability and utility of the resultant simulations. Recent work in machine learning has demonstrated that neural network potentials (NNPs) can learn efficient approximations to quantum mechanics and resolve this tradeoff, but existing NNPs still suffer from limited accuracy relative to state-of-the-art quantum-chemical methods. Here, we present Egret-1, a family of large pretrained NNPs based on the MACE architecture with general applicability to main-group, organic, and biomolecular chemistry. We find that the Egret-1 models equal or exceed the accuracy of routinely employed quantum-chemical methods on a variety of standard tasks, including torsional scans, conformer ranking, and geometry optimization, while offering multiple-order-of-magnitude speedups relative to legacy methods. We also highlight important lacunae for future NNP research to investigate, and suggest strategies for building future high-quality models with increased scale and generality. | [
{
"version": "v1",
"created": "Tue, 29 Apr 2025 17:26:21 GMT"
},
{
"version": "v2",
"created": "Wed, 30 Apr 2025 15:29:40 GMT"
},
{
"version": "v3",
"created": "Thu, 01 May 2025 21:15:45 GMT"
},
{
"version": "v4",
"created": "Mon, 05 May 2025 15:18:40 GMT"
},
{
"version": "v5",
"created": "Wed, 11 Jun 2025 13:16:02 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Mann",
"Elias L.",
""
],
[
"Wagen",
"Corin C.",
""
],
[
"Vandezande",
"Jonathon E.",
""
],
[
"Wagen",
"Arien M.",
""
],
[
"Schneider",
"Spencer C.",
""
]
] |
2504.21025 | MD Thamed Bin Zaman Chowdhury | MD Thamed Bin Zaman Chowdhury, Moazzem Hossain, Md. Ridwanul Islam | Durghotona GPT: A Web Scraping and Large Language Model Based Framework to Generate Road Accident Dataset Automatically in Bangladesh | It has been accepted in IEEE 27th International Conference on Computer and Information Technology (ICCIT). Now, we are waiting for it to get published in IEEE Xplore | null | 10.1109/ICCIT64611.2024.11021969 | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Road accidents pose significant concerns globally. They lead to large financial losses, injuries, disabilities, and societal challenges. Accurate and timely accident data is essential for predicting and mitigating these events. This paper presents a novel framework named 'Durghotona GPT' that integrates web scraping and Large Language Models (LLMs) to automate the generation of comprehensive accident datasets from prominent national dailies in Bangladesh. The authors collected accident reports from three major newspapers: Prothom Alo, Dhaka Tribune, and The Daily Star. The collected news was then processed using the newest available LLMs: GPT-4, GPT-3.5, and Llama-3. The framework efficiently extracts relevant information, categorizes reports, and compiles detailed datasets. Thus, this framework overcomes limitations of manual data collection methods such as delays, errors, and communication gaps. The authors' evaluation demonstrates that Llama-3, an open-source model, performs comparably to GPT-4. It achieved 89% accuracy in the authors' evaluation. Therefore, it can be considered a cost-effective alternative for similar tasks. The results suggest that the framework developed by the authors can drastically enhance the quality and availability of accident data. As a result, it can support critical applications in traffic safety analysis, urban planning, and public health. The authors also developed an interface for 'Durghotona GPT' for ease of use as part of this paper. Future work will focus on expanding data collection methods and refining LLMs to further increase dataset accuracy and applicability. | [
{
"version": "v1",
"created": "Wed, 23 Apr 2025 04:30:42 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Chowdhury",
"MD Thamed Bin Zaman",
""
],
[
"Hossain",
"Moazzem",
""
],
[
"Islam",
"Md. Ridwanul",
""
]
] |
2504.21085 | Raphael Augusto Pereira De Oliveira | R. A. P. Oliveira, R. Poleski, P. Mr\'oz, A. Udalski, J. Skowron, M. Mr\'oz, M. K. Szyma\'nski, I. Soszy\'nski, K. Ulaczyk, P. Pietrukowicz, K. Rybicki, P. Iwanek, M. Wrona, M. Gromadzki | Automated detection and modeling of binary microlensing events in OGLE-IV data. I. Events with well-separated bumps | 26 pages, 7 figures, AcA in press. Long tables included as ancillary files | null | null | null | astro-ph.GA astro-ph.IM astro-ph.SR | http://creativecommons.org/licenses/by/4.0/ | Gravitational microlensing depends primarily on the lens mass and presents a larger occurrence rate in crowded regions, which makes it the best tool to uncover the initial mass function (IMF) of low-mass stars in the Galactic bulge. The bulge IMF can be obtained from the luminosity function measured with the Hubble Space Telescope if one knows the statistics of binary stellar systems in the bulge. We aim to analyse a statistically significant number of binary-lens/single-source and single-lens/binary-source events, in order to explore the lower-mass end of the bulge IMF even in unresolved binary systems. This paper deals with events with clearly separated bumps and no caustic crossing or approach, whereas other types will be analysed in following works. A fully-automated approach in the search and modeling of binary events was implemented. Event detection was carried out with a modified version of the algorithm used in previous studies. Model fitting was carried out with Markov chain Monte Carlo and nested sampling methods, in order to find the most probable solution among binary lens or binary source models. We retrieved 107 binary events in Optical Gravitational Lensing Experiment (OGLE) light curves spanning ten years in 9 high-cadence and 112 low-cadence fields towards the bulge. Several criteria were applied to reduce false positives, resulting in 59 most likely binary lenses and 48 binary sources. The tools were effective in detecting a bona-fide sample of binary events, with a distribution of Einstein timescales around 35-40 days and flat distributions for mass ratio and source flux ratio. After proper consideration of detection efficiency, the statistics for binary fraction and mass ratio will provide valuable constraints for the bulge IMF. | [
{
"version": "v1",
"created": "Tue, 29 Apr 2025 18:00:02 GMT"
},
{
"version": "v2",
"created": "Tue, 10 Jun 2025 18:00:25 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Oliveira",
"R. A. P.",
""
],
[
"Poleski",
"R.",
""
],
[
"Mróz",
"P.",
""
],
[
"Udalski",
"A.",
""
],
[
"Skowron",
"J.",
""
],
[
"Mróz",
"M.",
""
],
[
"Szymański",
"M. K.",
""
],
[
"Soszyński",
"I.",
""
],
[
"Ulaczyk",
"K.",
""
],
[
"Pietrukowicz",
"P.",
""
],
[
"Rybicki",
"K.",
""
],
[
"Iwanek",
"P.",
""
],
[
"Wrona",
"M.",
""
],
[
"Gromadzki",
"M.",
""
]
] |
2504.21199 | Terrance Liu | Terrance Liu, Eileen Xiao, Adam Smith, Pratiksha Thaker, Zhiwei Steven Wu | Generate-then-Verify: Reconstructing Data from Limited Published Statistics | First two authors contributed equally. Remaining authors are ordered alphabetically | null | null | null | stat.ML cs.CR cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We study the problem of reconstructing tabular data from aggregate statistics, in which the attacker aims to identify interesting claims about the sensitive data that can be verified with 100% certainty given the aggregates. Successful attempts in prior work have conducted studies in settings where the set of published statistics is rich enough that entire datasets can be reconstructed with certainty. In our work, we instead focus on the regime where many possible datasets match the published statistics, making it impossible to reconstruct the entire private dataset perfectly (i.e., when approaches in prior work fail). We propose the problem of partial data reconstruction, in which the goal of the adversary is to instead output a $\textit{subset}$ of rows and/or columns that are $\textit{guaranteed to be correct}$. We introduce a novel integer programming approach that first $\textbf{generates}$ a set of claims and then $\textbf{verifies}$ whether each claim holds for all possible datasets consistent with the published aggregates. We evaluate our approach on the housing-level microdata from the U.S. Decennial Census release, demonstrating that privacy violations can still persist even when information published about such data is relatively sparse. | [
{
"version": "v1",
"created": "Tue, 29 Apr 2025 22:06:04 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 16:54:47 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Liu",
"Terrance",
""
],
[
"Xiao",
"Eileen",
""
],
[
"Smith",
"Adam",
""
],
[
"Thaker",
"Pratiksha",
""
],
[
"Wu",
"Zhiwei Steven",
""
]
] |
2505.01015 | Jongwook Han | Jongwook Han, Dongmin Choi, Woojung Song, Eun-Ju Lee, Yohan Jo | Value Portrait: Assessing Language Models' Values through Psychometrically and Ecologically Valid Items | This paper has been accepted for publication at ACL 2025 | null | null | null | cs.CL cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The importance of benchmarks for assessing the values of language models has been pronounced due to the growing need of more authentic, human-aligned responses. However, existing benchmarks rely on human or machine annotations that are vulnerable to value-related biases. Furthermore, the tested scenarios often diverge from real-world contexts in which models are commonly used to generate text and express values. To address these issues, we propose the Value Portrait benchmark, a reliable framework for evaluating LLMs' value orientations with two key characteristics. First, the benchmark consists of items that capture real-life user-LLM interactions, enhancing the relevance of assessment results to real-world LLM usage. Second, each item is rated by human subjects based on its similarity to their own thoughts, and correlations between these ratings and the subjects' actual value scores are derived. This psychometrically validated approach ensures that items strongly correlated with specific values serve as reliable items for assessing those values. Through evaluating 44 LLMs with our benchmark, we find that these models prioritize Benevolence, Security, and Self-Direction values while placing less emphasis on Tradition, Power, and Achievement values. Also, our analysis reveals biases in how LLMs perceive various demographic groups, deviating from real human data. | [
{
"version": "v1",
"created": "Fri, 02 May 2025 05:26:50 GMT"
},
{
"version": "v2",
"created": "Tue, 10 Jun 2025 00:47:42 GMT"
},
{
"version": "v3",
"created": "Wed, 11 Jun 2025 05:38:26 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Han",
"Jongwook",
""
],
[
"Choi",
"Dongmin",
""
],
[
"Song",
"Woojung",
""
],
[
"Lee",
"Eun-Ju",
""
],
[
"Jo",
"Yohan",
""
]
] |
2505.01072 | Kota Yoshinaga | Kota Yoshinaga and Noritaka Shimizu and Takashi Nakatsukasa | Nuclear structure study using a hybrid approach of shell model and Gogny-type density functionals | null | null | null | null | nucl-th | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Nuclear density functional theory (DFT) is able to reproduce the saturation properties of nuclear matter, as well as properties of finite nuclei. Consequently, the DFT calculations are applicable to nuclei across a wide range of masses on nuclear chart. The Gogny-type density functional, which is equivalent to the mean-field calculations with finite-range density-dependent effective interactions, is a successful example. In contrast, the shell-model (configuration-interaction) calculation is a powerful tool to describe nuclear structure, especially spectroscopic properties. The shell model is able to take into account correlations beyond mean field in a truncated model space. In this work, we report investigation on $\textit{sd}$-shell nuclei and Ca isotopes using a hybrid approach of the shell model and Gogny-type DFT. | [
{
"version": "v1",
"created": "Fri, 02 May 2025 07:22:30 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 10:09:03 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Yoshinaga",
"Kota",
""
],
[
"Shimizu",
"Noritaka",
""
],
[
"Nakatsukasa",
"Takashi",
""
]
] |
2505.01267 | GaoZheng Pei | Gaozheng Pei, Ke Ma, Yingfei Sun, Qianqian Xu, Qingming Huang | Diffusion-based Adversarial Purification from the Perspective of the Frequency Domain | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The diffusion-based adversarial purification methods attempt to drown adversarial perturbations into a part of isotropic noise through the forward process, and then recover the clean images through the reverse process. Due to the lack of distribution information about adversarial perturbations in the pixel domain, it is often unavoidable to damage normal semantics. We turn to the frequency domain perspective, decomposing the image into amplitude spectrum and phase spectrum. We find that for both spectra, the damage caused by adversarial perturbations tends to increase monotonically with frequency. This means that we can extract the content and structural information of the original clean sample from the frequency components that are less damaged. Meanwhile, theoretical analysis indicates that existing purification methods indiscriminately damage all frequency components, leading to excessive damage to the image. Therefore, we propose a purification method that can eliminate adversarial perturbations while maximizing the preservation of the content and structure of the original image. Specifically, at each time step during the reverse process, for the amplitude spectrum, we replace the low-frequency components of the estimated image's amplitude spectrum with the corresponding parts of the adversarial image. For the phase spectrum, we project the phase of the estimated image into a designated range of the adversarial image's phase spectrum, focusing on the low frequencies. Empirical evidence from extensive experiments demonstrates that our method significantly outperforms most current defense methods. | [
{
"version": "v1",
"created": "Fri, 02 May 2025 13:41:14 GMT"
},
{
"version": "v2",
"created": "Mon, 09 Jun 2025 02:12:49 GMT"
},
{
"version": "v3",
"created": "Wed, 11 Jun 2025 07:31:27 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Pei",
"Gaozheng",
""
],
[
"Ma",
"Ke",
""
],
[
"Sun",
"Yingfei",
""
],
[
"Xu",
"Qianqian",
""
],
[
"Huang",
"Qingming",
""
]
] |
2505.01685 | Bin Hu | Bin Hu and Zhi-Hong Guan | A brain-inspired generative model for EEG-based cognitive state identification | null | null | null | null | eess.SP cs.NE | http://creativecommons.org/licenses/by/4.0/ | This article proposes a brain-inspired generative (BIG) model that merges an impulsive-attention neural network and a variational autoencoder (VAE) for identifying cognitive states based on electroencephalography (EEG) data. A hybrid learning method is presented for training the model by integrating gradient-based learning and heteroassociative memory. The BIG model is capable of achieving multi-task objectives: EEG classification, generating new EEG, and brain network interpretation, alleviating the limitations of excessive data training and high computational cost in conventional approaches. Experimental results on two public EEG datasets with different sampling rates demonstrate that the BIG model achieves a classification accuracy above 89\%, comparable with state-of-the-art methods, while reducing computational cost by nearly 11\% over the baseline EEGNet. Incorporating the generated EEG data for training, the BIG model exhibits comparative performance in a few-shot pattern. Ablation studies justify the poised brain-inspired characteristic regarding the impulsive-attention module and the hybrid learning method. Thanks to the performance advantages with interpretable outputs, this BIG model has application potential for building digital twins of the brain. | [
{
"version": "v1",
"created": "Sat, 03 May 2025 04:29:27 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 03:28:37 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Hu",
"Bin",
""
],
[
"Guan",
"Zhi-Hong",
""
]
] |
2505.01983 | Hang Zhou | Hang Zhou, Hans-Georg M\"uller | Association and Independence Test for Random Objects | null | null | null | null | stat.ME | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We develop a unified framework for testing independence and quantifying association between random objects that are located in general metric spaces. Special cases include functional and high-dimensional data as well as networks, covariance matrices and data on Riemannian manifolds, among other metric space-valued data. A key concept is the profile association, a measure based on distance profiles that intrinsically characterize the distributions of random objects in metric spaces. We rigorously establish a connection between the Hoeffding D statistic and the profile association and derive a permutation test with theoretical guarantees for consistency and power under alternatives to the null hypothesis of independence/no association. We extend this framework to the conditional setting, where the independence between random objects given a Euclidean predictor is of interest. In simulations across various metric spaces, the proposed profile independence test is found to outperform existing approaches. The practical utility of this framework is demonstrated with applications to brain connectivity networks derived from magnetic resonance imaging and age-at-death distributions for males and females obtained from human mortality data. | [
{
"version": "v1",
"created": "Sun, 04 May 2025 04:30:07 GMT"
},
{
"version": "v2",
"created": "Tue, 10 Jun 2025 18:48:27 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Zhou",
"Hang",
""
],
[
"Müller",
"Hans-Georg",
""
]
] |
2505.02604 | Yongding Tian | Yongding Tian, Zaid Al-Ars, Maksim Kitsak, Peter Hofstee | Low-Loss Space in Neural Networks is Continuous and Fully Connected | 17 pages, 10 figures | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | Visualizations of the loss landscape in neural networks suggest that minima are isolated points. However, both theoretical and empirical studies indicate that it is possible to connect two different minima with a path consisting of intermediate points that also have low loss. In this study, we propose a new algorithm which investigates low-loss paths in the full parameter space, not only between two minima. Our experiments on LeNet5, ResNet18, and Compact Convolutional Transformer architectures consistently demonstrate the existence of such continuous paths in the parameter space. These results suggest that the low-loss region is a fully connected and continuous space in the parameter space. Our findings provide theoretical insight into neural network over-parameterization, highlighting that parameters collectively define a high-dimensional low-loss space, implying parameter redundancy exists only within individual models and not throughout the entire low-loss space. Additionally, our work also provides new visualization methods and opportunities to improve model generalization by exploring the low-loss space that is closer to the origin. | [
{
"version": "v1",
"created": "Mon, 05 May 2025 12:16:55 GMT"
},
{
"version": "v2",
"created": "Tue, 10 Jun 2025 01:08:42 GMT"
},
{
"version": "v3",
"created": "Wed, 11 Jun 2025 01:28:36 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Tian",
"Yongding",
""
],
[
"Al-Ars",
"Zaid",
""
],
[
"Kitsak",
"Maksim",
""
],
[
"Hofstee",
"Peter",
""
]
] |
2505.03071 | Tom Denton | Vincent Dumoulin and Otilia Stretcu and Jenny Hamer and Lauren Harrell and Rob Laber and Hugo Larochelle and Bart van Merri\"enboer and Amanda Navine and Patrick Hart and Ben Williams and Timothy A.C. Lamont and Tries B. Razak and Mars Coral Restoration Team and Sheryn Brodie and Brendan Doohan and Phil Eichinski and Paul Roe and Lin Schwarzkopf and Tom Denton | The Search for Squawk: Agile Modeling in Bioacoustics | null | null | null | null | eess.AS | http://creativecommons.org/licenses/by/4.0/ | Passive acoustic monitoring (PAM) has shown great promise in helping ecologists understand the health of animal populations and ecosystems. However, extracting insights from millions of hours of audio recordings requires the development of specialized recognizers. This is typically a challenging task, necessitating large amounts of training data and machine learning expertise. In this work, we introduce a general, scalable and data-efficient system for developing recognizers for novel bioacoustic problems in under an hour. Our system consists of several key components that tackle problems in previous bioacoustic workflows: 1) highly generalizable acoustic embeddings pre-trained for birdsong classification minimize data hunger; 2) indexed audio search allows the efficient creation of classifier training datasets, and 3) precomputation of embeddings enables an efficient active learning loop, improving classifier quality iteratively with minimal wait time. Ecologists employed our system in three novel case studies: analyzing coral reef health through unidentified sounds; identifying juvenile Hawaiian bird calls to quantify breeding success and improve endangered species monitoring; and Christmas Island bird occupancy modeling. We augment the case studies with simulated experiments which explore the range of design decisions in a structured way and help establish best practices. Altogether these experiments showcase our system's scalability, efficiency, and generalizability, enabling scientists to quickly address new bioacoustic challenges. | [
{
"version": "v1",
"created": "Mon, 05 May 2025 23:34:15 GMT"
},
{
"version": "v2",
"created": "Wed, 07 May 2025 21:02:34 GMT"
},
{
"version": "v3",
"created": "Wed, 28 May 2025 12:44:00 GMT"
},
{
"version": "v4",
"created": "Tue, 10 Jun 2025 18:23:45 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Dumoulin",
"Vincent",
""
],
[
"Stretcu",
"Otilia",
""
],
[
"Hamer",
"Jenny",
""
],
[
"Harrell",
"Lauren",
""
],
[
"Laber",
"Rob",
""
],
[
"Larochelle",
"Hugo",
""
],
[
"van Merriënboer",
"Bart",
""
],
[
"Navine",
"Amanda",
""
],
[
"Hart",
"Patrick",
""
],
[
"Williams",
"Ben",
""
],
[
"Lamont",
"Timothy A. C.",
""
],
[
"Razak",
"Tries B.",
""
],
[
"Team",
"Mars Coral Restoration",
""
],
[
"Brodie",
"Sheryn",
""
],
[
"Doohan",
"Brendan",
""
],
[
"Eichinski",
"Phil",
""
],
[
"Roe",
"Paul",
""
],
[
"Schwarzkopf",
"Lin",
""
],
[
"Denton",
"Tom",
""
]
] |
2505.03086 | Sota Yoneda | Sota Yoneda, Hiromitsu Takeuchi | Proper Orthogonal Decomposition of a Superfluid Turbulent Wake | Movies showing the time evolution of the wake presented in the main text is available from https://www.youtube.com/playlist?list=PLtG3C55s-pk4Wn5XD7K-yohft-S6DI_Tm | Journal of the Physical Society of Japan 94, 073601 (2025) | 10.7566/JPSJ.94.073601 | null | cond-mat.quant-gas | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Superfluid turbulent wakes behind a square prism are studied theoretically and numerically by proper orthogonal decomposition (POD). POD is a data science approach that can efficiently extract the principal vibration modes of a physical system, and is widely used in hydrodynamics, including applications in wake structure analysis. It is not straightforward to apply the conventional POD method to superfluid wake systems, as the superfluid velocity field diverges at the center of a vortex whose circulation is quantized. We successfully established a POD method by applying appropriate blurring to the vorticity distribution in a two-dimensional superfluid wake. It is shown that a coherent structure corresponding to two parallel arrays of alternating quantum vortex bundles, called the "quasi-classical" K\'arm\'an vortex street, is latent as a distinctive major mode in the superfluid turbulent wakes that were naively thought to be "irregular". Since our method is also effective for fluid density, it can be applied to the experimental data analysis for ultra-cold atomic gases. | [
{
"version": "v1",
"created": "Tue, 06 May 2025 00:45:59 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 09:36:17 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Yoneda",
"Sota",
""
],
[
"Takeuchi",
"Hiromitsu",
""
]
] |
2505.03298 | Yanqi Qiu | Zhaofeng Lin, Yanqi Qiu, Mingjie Tan | Harmonic analysis of multiplicative chaos Part II: a unified approach to Fourier dimensions | 69 pages; further improvements in the setting of GMC are obtained in this version | null | null | null | math.PR math-ph math.DS math.FA math.MP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We introduce a unified approach for studying the polynomial Fourier decay of classical multiplicative chaos measures. As consequences, we obtain the precise Fourier dimensions for multiplicative chaos measures arising from the following key models: the sub-critical 1D and 2D GMC (which in particular resolves the Garban-Vargas conjecture); the sub-critical $d$-dimensional GMC with $d \ge 3$ when the parameter $\gamma$ is near the critical value; the canonical Mandelbrot random coverings; the canonical Mandelbrot cascades. For various other models, we establish the non-trivial lower bounds of the Fourier dimensions and in various cases we conjecture that they are all optimal and provide the exact values of Fourier dimensions. | [
{
"version": "v1",
"created": "Tue, 06 May 2025 08:28:11 GMT"
},
{
"version": "v2",
"created": "Mon, 12 May 2025 12:01:01 GMT"
},
{
"version": "v3",
"created": "Wed, 11 Jun 2025 01:35:55 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Lin",
"Zhaofeng",
""
],
[
"Qiu",
"Yanqi",
""
],
[
"Tan",
"Mingjie",
""
]
] |
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 | Comments 1: We corrected the authors' names. Revised methods from grounded theory to thematic analysis as it is more suitable. We also updated the reference of the systematic literature search. Comments 2: In Section II-A, a minor refinement of the definition of blockchain technology is made. In Section IV-A, the Gini coefficient metric was corrected | 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"
},
{
"version": "v3",
"created": "Wed, 11 Jun 2025 15:10:08 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Masama",
"Mansur Aliyu",
""
],
[
"Kannengießer",
"Niclas",
""
],
[
"Sunyaev",
"Ali",
""
]
] |
2505.04088 | Huanbin Zhang | Shang Zhang, Huanbin Zhang, Dali Feng, Yujie Cui, Ruoyan Xiong, and Cen He | SMMT: Siamese Motion Mamba with Self-attention for Thermal Infrared Target Tracking | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Thermal infrared (TIR) object tracking often suffers from challenges such as target occlusion, motion blur, and background clutter, which significantly degrade the performance of trackers. To address these issues, this paper pro-poses a novel Siamese Motion Mamba Tracker (SMMT), which integrates a bidirectional state-space model and a self-attention mechanism. Specifically, we introduce the Motion Mamba module into the Siamese architecture to ex-tract motion features and recover overlooked edge details using bidirectional modeling and self-attention. We propose a Siamese parameter-sharing strate-gy that allows certain convolutional layers to share weights. This approach reduces computational redundancy while preserving strong feature represen-tation. In addition, we design a motion edge-aware regression loss to improve tracking accuracy, especially for motion-blurred targets. Extensive experi-ments are conducted on four TIR tracking benchmarks, including LSOTB-TIR, PTB-TIR, VOT-TIR2015, and VOT-TIR 2017. The results show that SMMT achieves superior performance in TIR target tracking. | [
{
"version": "v1",
"created": "Wed, 07 May 2025 03:02:04 GMT"
},
{
"version": "v2",
"created": "Sun, 11 May 2025 01:58:20 GMT"
},
{
"version": "v3",
"created": "Wed, 11 Jun 2025 14:19:28 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Zhang",
"Shang",
""
],
[
"Zhang",
"Huanbin",
""
],
[
"Feng",
"Dali",
""
],
[
"Cui",
"Yujie",
""
],
[
"Xiong",
"Ruoyan",
""
],
[
"He",
"Cen",
""
]
] |
2505.04256 | Ashok K. Verma Prof | Ashok K. Verma, and P. Modak | Realizing high-temperature superconductivity in compressed molecular-hydrogen through Li doping | 30 pages,17 figures and 7 tables | null | null | null | cond-mat.supr-con | http://creativecommons.org/licenses/by/4.0/ | In this study, we explore lithium-doped stable molecular hydrogen structures by performing first-principles crystal structure searches across varying compositions in the Li-H system under high pressure. Our search reveals a cubic phase of LiH12, which shows promise as a high-temperature superconductor. Our Bader charge analysis suggests that electron transfer from Li to H atoms tunes the intra- and inter-molecular H-H distances, which are critical for the metallization of molecular hydrogen. This modulation alters the interaction between bonding and anti-bonding 1s states of hydrogen molecules. Furthermore, Li ions serve as stabilizers for the distorted H2 molecular network through ionic interactions. Numerical solutions to the fully anisotropic Migdal-Eliashberg equations reveals that this phase could exhibit superconductivity above 300 K at a pressure of 250 GPa, a pressure value that is typically achievable using a diamond anvil cell. Detailed analysis of species-specific phonons and the Eliashberg function shows that low- and intermediate-energy phonons are crucial in promoting strong electron-phonon coupling. Thus, our study establishes lithium doping as a promising approach to induce high-temperature superconductivity in compressed molecular hydrogen without causing molecular dissociation. | [
{
"version": "v1",
"created": "Wed, 07 May 2025 09:00:38 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 10:40:28 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Verma",
"Ashok K.",
""
],
[
"Modak",
"P.",
""
]
] |
2505.04317 | Ruize Zhang | Ruize Zhang, Sirui Xiang, Zelai Xu, Feng Gao, Shilong Ji, Wenhao Tang, Wenbo Ding, Chao Yu, Yu Wang | Mastering Multi-Drone Volleyball through Hierarchical Co-Self-Play Reinforcement Learning | null | null | null | null | cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we tackle the problem of learning to play 3v3 multi-drone volleyball, a new embodied competitive task that requires both high-level strategic coordination and low-level agile control. The task is turn-based, multi-agent, and physically grounded, posing significant challenges due to its long-horizon dependencies, tight inter-agent coupling, and the underactuated dynamics of quadrotors. To address this, we propose Hierarchical Co-Self-Play (HCSP), a hierarchical reinforcement learning framework that separates centralized high-level strategic decision-making from decentralized low-level motion control. We design a three-stage population-based training pipeline to enable both strategy and skill to emerge from scratch without expert demonstrations: (I) training diverse low-level skills, (II) learning high-level strategy via self-play with fixed low-level controllers, and (III) joint fine-tuning through co-self-play. Experiments show that HCSP achieves superior performance, outperforming non-hierarchical self-play and rule-based hierarchical baselines with an average 82.9% win rate and a 71.5% win rate against the two-stage variant. Moreover, co-self-play leads to emergent team behaviors such as role switching and coordinated formations, demonstrating the effectiveness of our hierarchical design and training scheme. The project page is at https://sites.google.com/view/hi-co-self-play. | [
{
"version": "v1",
"created": "Wed, 07 May 2025 11:04:36 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 12:47:48 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Zhang",
"Ruize",
""
],
[
"Xiang",
"Sirui",
""
],
[
"Xu",
"Zelai",
""
],
[
"Gao",
"Feng",
""
],
[
"Ji",
"Shilong",
""
],
[
"Tang",
"Wenhao",
""
],
[
"Ding",
"Wenbo",
""
],
[
"Yu",
"Chao",
""
],
[
"Wang",
"Yu",
""
]
] |
2505.04646 | Poria Azadi | Poria Azadi | Computational Irreducibility as the Foundation of Agency: A Formal Model Connecting Undecidability to Autonomous Behavior in Complex Systems | null | null | null | null | cs.AI cs.CC cs.IT math.IT | http://creativecommons.org/licenses/by-sa/4.0/ | This article presents a formal model demonstrating that genuine autonomy, the ability of a system to self-regulate and pursue objectives, fundamentally implies computational unpredictability from an external perspective. we establish precise mathematical connections, proving that for any truly autonomous system, questions about its future behavior are fundamentally undecidable. this formal undecidability, rather than mere complexity, grounds a principled distinction between autonomous and non-autonomous systems. our framework integrates insights from computational theory and biology, particularly regarding emergent agency and computational irreducibility, to explain how novel information and purpose can arise within a physical universe. the findings have significant implications for artificial intelligence, biological modeling, and philosophical concepts like free will. | [
{
"version": "v1",
"created": "Mon, 05 May 2025 21:24:50 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 13:38:17 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Azadi",
"Poria",
""
]
] |
2505.05173 | Andrei Zavarnitsine | Danila O. Revin, Andrei V. Zavarnitsine | Conjugate generation of sporadic almost simple groups | null | null | null | null | math.GR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | As defined by Guralnick and Saxl, given a nonabelian simple group $S$ and its nonidentity automorphism $x$, a natural number $\alpha_S(x)$ is the minimum number of conjugates of $x$ in $\langle x,S\rangle$ that generate a subgroup containing $S$. In this paper, for every sporadic group $S$ other than the Monster and an automorphism $x$ of $S$ of prime order, we complete the determination of the precise value of $\alpha_S(x)$. | [
{
"version": "v1",
"created": "Thu, 08 May 2025 12:20:54 GMT"
},
{
"version": "v2",
"created": "Wed, 11 Jun 2025 05:43:17 GMT"
}
] | 2025-06-12T00:00:00 | [
[
"Revin",
"Danila O.",
""
],
[
"Zavarnitsine",
"Andrei V.",
""
]
] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.