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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
2505.22674 | Pawan Neupane | Pawan Neupane, Jian Liu and Jianlin Cheng | PSBench: a large-scale benchmark for estimating the accuracy of protein complex structural models | null | null | null | null | q-bio.BM cs.AI cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Predicting protein complex structures is essential for protein function analysis, protein design, and drug discovery. While AI methods like AlphaFold can predict accurate structural models for many protein complexes, reliably estimating the quality of these predicted models (estimation of model accuracy, or EMA) for model ranking and selection remains a major challenge. A key barrier to developing effective machine learning-based EMA methods is the lack of large, diverse, and well-annotated datasets for training and evaluation. To address this gap, we introduce PSBench, a benchmark suite comprising four large-scale, labeled datasets generated during the 15th and 16th community-wide Critical Assessment of Protein Structure Prediction (CASP15 and CASP16). PSBench includes over one million structural models covering a wide range of protein sequence lengths, complex stoichiometries, functional classes, and modeling difficulties. Each model is annotated with multiple complementary quality scores at the global, local, and interface levels. PSBench also provides multiple evaluation metrics and baseline EMA methods to facilitate rigorous comparisons. To demonstrate PSBench's utility, we trained and evaluated GATE, a graph transformer-based EMA method, on the CASP15 data. GATE was blindly tested in CASP16 (2024), where it ranked among the top-performing EMA methods. These results highlight PSBench as a valuable resource for advancing EMA research in protein complex modeling. PSBench is publicly available at: https://github.com/BioinfoMachineLearning/PSBench. | [
{
"version": "v1",
"created": "Tue, 13 May 2025 17:47:12 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Neupane",
"Pawan",
""
],
[
"Liu",
"Jian",
""
],
[
"Cheng",
"Jianlin",
""
]
] |
2505.22675 | Devanshu Saxena | Devanshu Saxena | Towards Real-Time Interpolation for Enhanced AUV Deep Sea Mapping | 8 pages | null | null | null | physics.ao-ph cs.NI cs.RO | http://creativecommons.org/licenses/by/4.0/ | Approximately seventy-one percent of the Earth is covered in water. Of that area, ninety-five percent of the ocean has never been explored or mapped. There are several engineering challenges that have prevented the exploration of the deep ocean through human or autonomous means. These challenges include but are not limited to high pressure, cold temperatures, little natural light, corrosion of materials, and communication. Ongoing research has been focused on trying to find optimal and low-cost solutions to effective communication between autonomous underwater vehicles (AUVs), and the surface or air. In this paper, an architecture is introduced that utilizes an edge computing approach to establish computation nearer to the source of data, allowing further exploration of the deep ocean. Taking the most common interpolation techniques used today in the field of bathymetry, the data are tested and analyzed to find the feasibility of switching from CPU to GPU computation. Specifically, the focus is on writing efficient interpolation algorithms that can be run on low-level GPUs, which can be carried onboard AUVs as payload. | [
{
"version": "v1",
"created": "Tue, 13 May 2025 18:46:40 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Saxena",
"Devanshu",
""
]
] |
2505.22676 | Igor Suslov | I. M. Suslov (P.L.Kapitza Institute for Physical Problems, Moscow, Russia) | Anderson transition in high dimension: comments to arXiv:2403.01974 | Latex, 4 pages | null | null | null | cond-mat.dis-nn | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In the recent submission arXiv:2403.01974, Altshuler et al suggested a new approach to the Anderson transition in high dimensions. The main idea consists in the use of the branching graphs instead of high-dimensional lattices: it does not look very convincing, but we do not want to stress this point. Since the authors welcome comments, we put forward a lot of objections to their exposition of the general situation. The arising hypothesis is given in the end. | [
{
"version": "v1",
"created": "Wed, 14 May 2025 01:18:37 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Suslov",
"I. M.",
"",
"P.L.Kapitza Institute for Physical Problems, Moscow, Russia"
]
] |
2505.22677 | Benjamin Riggan | Jisu Kim and Alex Mattingly and Eung-Joo Lee and Benjamin S. Riggan | Using Cross-Domain Detection Loss to Infer Multi-Scale Information for Improved Tiny Head Tracking | To appear at IEEE International Conference on Automatic Face and Gesture 2025 (FG2025) | null | null | null | cs.CV | http://creativecommons.org/licenses/by-sa/4.0/ | Head detection and tracking are essential for downstream tasks, but current methods often require large computational budgets, which increase latencies and ties up resources (e.g., processors, memory, and bandwidth). To address this, we propose a framework to enhance tiny head detection and tracking by optimizing the balance between performance and efficiency. Our framework integrates (1) a cross-domain detection loss, (2) a multi-scale module, and (3) a small receptive field detection mechanism. These innovations enhance detection by bridging the gap between large and small detectors, capturing high-frequency details at multiple scales during training, and using filters with small receptive fields to detect tiny heads. Evaluations on the CroHD and CrowdHuman datasets show improved Multiple Object Tracking Accuracy (MOTA) and mean Average Precision (mAP), demonstrating the effectiveness of our approach in crowded scenes. | [
{
"version": "v1",
"created": "Wed, 14 May 2025 02:29:54 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Kim",
"Jisu",
""
],
[
"Mattingly",
"Alex",
""
],
[
"Lee",
"Eung-Joo",
""
],
[
"Riggan",
"Benjamin S.",
""
]
] |
2505.22678 | Jiahao Yang | Jiahao Yang, Ran Fang, Ming Zhang and Jun Zhou | An Efficient deep learning model to Predict Stock Price Movement Based on Limit Order Book | null | null | null | null | q-fin.TR cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In high-frequency trading (HFT), leveraging limit order books (LOB) to model stock price movements is crucial for achieving profitable outcomes. However, this task is challenging due to the high-dimensional and volatile nature of the original data. Even recent deep learning models often struggle to capture price movement patterns effectively, particularly without well-designed features. We observed that raw LOB data exhibits inherent symmetry between the ask and bid sides, and the bid-ask differences demonstrate greater stability and lower complexity compared to the original data. Building on this insight, we propose a novel approach in which leverages the Siamese architecture to enhance the performance of existing deep learning models. The core idea involves processing the ask and bid sides separately using the same module with shared parameters. We applied our Siamese-based methods to several widely used strong baselines and validated their effectiveness using data from 14 military industry stocks in the Chinese A-share market. Furthermore, we integrated multi-head attention (MHA) mechanisms with the Long Short-Term Memory (LSTM) module to investigate its role in modeling stock price movements. Our experiments used raw data and widely used Order Flow Imbalance (OFI) features as input with some strong baseline models. The results show that our method improves the performance of strong baselines in over 75$% of cases, excluding the Multi-Layer Perception (MLP) baseline, which performed poorly and is not considered practical. Furthermore, we found that Multi-Head Attention can enhance model performance, particularly over shorter forecasting horizons. | [
{
"version": "v1",
"created": "Wed, 14 May 2025 12:46:21 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Yang",
"Jiahao",
""
],
[
"Fang",
"Ran",
""
],
[
"Zhang",
"Ming",
""
],
[
"Zhou",
"Jun",
""
]
] |
2505.22679 | Rolf Hoffmann | Rolf Hoffmann and Mariusz Bia{\l}ecki | Loop Patterns Formed by Cellular Automata | null | null | null | null | nlin.CG | http://creativecommons.org/licenses/by-nc-nd/4.0/ | A Cellular Automata (CA) rule is presented that can generate "loop patterns" in a 2D grid under fixed boundary conditions. A loop is a cyclically closed path represented by one-cells enclosed by zero-cells. A loop pattern can contain several loops that are not allowed to touch each other. The problem is solved by designing an appropriate set of tiles that can overlap and which are used in the CA rule. Templates are derived from the tiles which are used for local pattern matching. In order to drive the evolution to the desired patterns, noise is injected if the templates do not match or other constraints are not fulfilled. The general CA rule can be specialized by enabling certain conditions, and the characteristics of five rule variants are explained. Simulations illustrate that the CA rule can securely evolve stable loop patterns. The preliminary theoretical analysis of the obtained loop patterns raises many interesting research problems for the future -- several of them have been briefly discussed. | [
{
"version": "v1",
"created": "Wed, 14 May 2025 19:06:23 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Hoffmann",
"Rolf",
""
],
[
"Białecki",
"Mariusz",
""
]
] |
2505.22680 | Christian M. Kerskens | Christian Kerskens | Exploring Holography in Neuro-Vascular Dynamics | 10 pages | null | null | null | q-bio.NC | http://creativecommons.org/licenses/by/4.0/ | The holonomic brain theory, originally formulated to account for the need of non-local memory encoding in cognitive systems, could gain new theoretical traction when integrated with holographic principles from physics, most notably the AdS/CFT correspondence. Recent findings in neuroscience suggest that conformal field theories (CFTs), emerging at critical points across spatiotemporal scales in neural dynamics, are essential for brain function. Concurrently, black-brane geometries, long studied in gravitational physics, can find unexpected analogues in the interplay of active matter dynamics and the brain s neuroanatomical organization. Motivated by these parallels, we posit a generalized holographic framework and interrogate its validity through the fluid/gravity duality; a correspondence linking hydrodynamic equations to gravitational spacetime metrics. In this work, we explore the holographic principles at the Navier-Stokes regime, demonstrating that holography can model key neurophysiological mechanisms: cerebral autoregulation (the brain s hemodynamic self-stabilization) and neurovascular coupling (the dynamic neuron-bloodflow interplay). This work bridges holography, active matter physics, and neuroscience, proposing a unified framework to decode the brain s multiscale organization, its resilience to perturbations, and its computational capabilities. By grounding neurovascular physiology in gravitational duals, we open pathways to reinterpret brain function through the lens of emergent spacetime geometry. | [
{
"version": "v1",
"created": "Mon, 19 May 2025 09:43:25 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Kerskens",
"Christian",
""
]
] |
2505.22681 | Bekir Dan{\i}\c{s} | Bekir Dan{\i}\c{s} | General contractions in new type perturbed metric spaces | null | null | null | null | math.GM | http://creativecommons.org/licenses/by/4.0/ | We focus on the new type perturbed metric spaces and introduce a contraction mapping namely new type perturbed Kannan mappings. For these mappings, we show that Banach's fixed point theorem holds. Moreover, this new generalization of Banach's contraction principle does not depend on the continuity of the operator. | [
{
"version": "v1",
"created": "Thu, 22 May 2025 06:34:41 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Danış",
"Bekir",
""
]
] |
2505.22682 | Mengye Lyu | Xinxian Fan and Mengye Lyu | MRI Image Generation Based on Text Prompts | null | null | null | null | eess.IV cs.CV physics.med-ph | http://creativecommons.org/licenses/by/4.0/ | This study explores the use of text-prompted MRI image generation with the Stable Diffusion (SD) model to address challenges in acquiring real MRI datasets, such as high costs, limited rare case samples, and privacy concerns. The SD model, pre-trained on natural images, was fine-tuned using the 3T fastMRI dataset and the 0.3T M4Raw dataset, with the goal of generating brain T1, T2, and FLAIR images across different magnetic field strengths. The performance of the fine-tuned model was evaluated using quantitative metrics,including Fr\'echet Inception Distance (FID) and Multi-Scale Structural Similarity (MS-SSIM), showing improvements in image quality and semantic consistency with the text prompts. To further evaluate the model's potential, a simple classification task was carried out using a small 0.35T MRI dataset, demonstrating that the synthetic images generated by the fine-tuned SD model can effectively augment training datasets and improve the performance of MRI constrast classification tasks. Overall, our findings suggest that text-prompted MRI image generation is feasible and can serve as a useful tool for medical AI applications. | [
{
"version": "v1",
"created": "Fri, 23 May 2025 03:01:22 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Fan",
"Xinxian",
""
],
[
"Lyu",
"Mengye",
""
]
] |
2505.22683 | Shuqiang Wang | Xuhang Chen, Michael Kwok-Po Ng, Kim-Fung Tsang, Chi-Man Pun and Shuqiang Wang | ConnectomeDiffuser: Generative AI Enables Brain Network Construction from Diffusion Tensor Imaging | null | null | null | null | q-bio.NC cs.AI cs.CV | http://creativecommons.org/licenses/by/4.0/ | Brain network analysis plays a crucial role in diagnosing and monitoring neurodegenerative disorders such as Alzheimer's disease (AD). Existing approaches for constructing structural brain networks from diffusion tensor imaging (DTI) often rely on specialized toolkits that suffer from inherent limitations: operator subjectivity, labor-intensive workflows, and restricted capacity to capture complex topological features and disease-specific biomarkers. To overcome these challenges and advance computational neuroimaging instrumentation, ConnectomeDiffuser is proposed as a novel diffusion-based framework for automated end-to-end brain network construction from DTI. The proposed model combines three key components: (1) a Template Network that extracts topological features from 3D DTI scans using Riemannian geometric principles, (2) a diffusion model that generates comprehensive brain networks with enhanced topological fidelity, and (3) a Graph Convolutional Network classifier that incorporates disease-specific markers to improve diagnostic accuracy. ConnectomeDiffuser demonstrates superior performance by capturing a broader range of structural connectivity and pathology-related information, enabling more sensitive analysis of individual variations in brain networks. Experimental validation on datasets representing two distinct neurodegenerative conditions demonstrates significant performance improvements over other brain network methods. This work contributes to the advancement of instrumentation in the context of neurological disorders, providing clinicians and researchers with a robust, generalizable measurement framework that facilitates more accurate diagnosis, deeper mechanistic understanding, and improved therapeutic monitoring of neurodegenerative diseases such as AD. | [
{
"version": "v1",
"created": "Fri, 23 May 2025 15:03:58 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Chen",
"Xuhang",
""
],
[
"Ng",
"Michael Kwok-Po",
""
],
[
"Tsang",
"Kim-Fung",
""
],
[
"Pun",
"Chi-Man",
""
],
[
"Wang",
"Shuqiang",
""
]
] |
2505.22684 | Yufeng Wang | Yufeng Wang, Yiguang Bai, Tianqing Zhu, Ismail Ben Ayed, Jing Yuan | Recovering Fairness Directly from Modularity: a New Way for Fair Community Partitioning | 17pages, 5 figures | null | null | null | cs.SI cs.LG | http://creativecommons.org/licenses/by/4.0/ | Community partitioning is crucial in network analysis, with modularity optimization being the prevailing technique. However, traditional modularity-based methods often overlook fairness, a critical aspect in real-world applications. To address this, we introduce protected group networks and propose a novel fairness-modularity metric. This metric extends traditional modularity by explicitly incorporating fairness, and we prove that minimizing it yields naturally fair partitions for protected groups while maintaining theoretical soundness. We develop a general optimization framework for fairness partitioning and design the efficient Fair Fast Newman (FairFN) algorithm, enhancing the Fast Newman (FN) method to optimize both modularity and fairness. Experiments show FairFN achieves significantly improved fairness and high-quality partitions compared to state-of-the-art methods, especially on unbalanced datasets. | [
{
"version": "v1",
"created": "Tue, 27 May 2025 09:44:07 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Wang",
"Yufeng",
""
],
[
"Bai",
"Yiguang",
""
],
[
"Zhu",
"Tianqing",
""
],
[
"Ayed",
"Ismail Ben",
""
],
[
"Yuan",
"Jing",
""
]
] |
2505.22685 | Marcus Vroemen | Marcus J. Vroemen, Yuqian Chen, Yui Lo, Tengfei Xu, Weidong Cai, Fan Zhang, Josien P.W. Pluim, Lauren J. O'Donnell | DeepMultiConnectome: Deep Multi-Task Prediction of Structural Connectomes Directly from Diffusion MRI Tractography | 15 pages, 5 figures, 5 tables | null | null | null | eess.IV cs.AI cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Diffusion MRI (dMRI) tractography enables in vivo mapping of brain structural connections, but traditional connectome generation is time-consuming and requires gray matter parcellation, posing challenges for large-scale studies. We introduce DeepMultiConnectome, a deep-learning model that predicts structural connectomes directly from tractography, bypassing the need for gray matter parcellation while supporting multiple parcellation schemes. Using a point-cloud-based neural network with multi-task learning, the model classifies streamlines according to their connected regions across two parcellation schemes, sharing a learned representation. We train and validate DeepMultiConnectome on tractography from the Human Connectome Project Young Adult dataset ($n = 1000$), labeled with an 84 and 164 region gray matter parcellation scheme. DeepMultiConnectome predicts multiple structural connectomes from a whole-brain tractogram containing 3 million streamlines in approximately 40 seconds. DeepMultiConnectome is evaluated by comparing predicted connectomes with traditional connectomes generated using the conventional method of labeling streamlines using a gray matter parcellation. The predicted connectomes are highly correlated with traditionally generated connectomes ($r = 0.992$ for an 84-region scheme; $r = 0.986$ for a 164-region scheme) and largely preserve network properties. A test-retest analysis of DeepMultiConnectome demonstrates reproducibility comparable to traditionally generated connectomes. The predicted connectomes perform similarly to traditionally generated connectomes in predicting age and cognitive function. Overall, DeepMultiConnectome provides a scalable, fast model for generating subject-specific connectomes across multiple parcellation schemes. | [
{
"version": "v1",
"created": "Tue, 27 May 2025 10:56:37 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Vroemen",
"Marcus J.",
""
],
[
"Chen",
"Yuqian",
""
],
[
"Lo",
"Yui",
""
],
[
"Xu",
"Tengfei",
""
],
[
"Cai",
"Weidong",
""
],
[
"Zhang",
"Fan",
""
],
[
"Pluim",
"Josien P. W.",
""
],
[
"O'Donnell",
"Lauren J.",
""
]
] |
2505.22686 | Ange-Cl\'ement Akazan | Ange-Clement Akazan, Verlon Roel Mbingui, Gnankan Landry Regis N'guessan and Issa Karambal | Localized Weather Prediction Using Kolmogorov-Arnold Network-Based Models and Deep RNNs | null | null | null | null | cs.LG physics.ao-ph | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Weather forecasting is crucial for managing risks and economic planning, particularly in tropical Africa, where extreme events severely impact livelihoods. Yet, existing forecasting methods often struggle with the region's complex, non-linear weather patterns. This study benchmarks deep recurrent neural networks such as $\texttt{LSTM, GRU, BiLSTM, BiGRU}$, and Kolmogorov-Arnold-based models $(\texttt{KAN} and \texttt{TKAN})$ for daily forecasting of temperature, precipitation, and pressure in two tropical cities: Abidjan, Cote d'Ivoire (Ivory Coast) and Kigali (Rwanda). We further introduce two customized variants of $ \texttt{TKAN}$ that replace its original $\texttt{SiLU}$ activation function with $ \texttt{GeLU}$ and \texttt{MiSH}, respectively. Using station-level meteorological data spanning from 2010 to 2024, we evaluate all the models on standard regression metrics. $\texttt{KAN}$ achieves temperature prediction ($R^2=0.9986$ in Abidjan, $0.9998$ in Kigali, $\texttt{MSE} < 0.0014~^\circ C ^2$), while $\texttt{TKAN}$ variants minimize absolute errors for precipitation forecasting in low-rainfall regimes. The customized $\texttt{TKAN}$ models demonstrate improvements over the standard $\texttt{TKAN}$ across both datasets. Classical \texttt{RNNs} remain highly competitive for atmospheric pressure ($R^2 \approx 0.83{-}0.86$), outperforming $\texttt{KAN}$-based models in this task. These results highlight the potential of spline-based neural architectures for efficient and data-efficient forecasting. | [
{
"version": "v1",
"created": "Tue, 27 May 2025 18:01:57 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Akazan",
"Ange-Clement",
""
],
[
"Mbingui",
"Verlon Roel",
""
],
[
"N'guessan",
"Gnankan Landry Regis",
""
],
[
"Karambal",
"Issa",
""
]
] |
2505.22687 | Martino Andrea Scarpolini | Martino Andrea Scarpolini, Giovanni Vagnoli, Fabio Guglietta, Roberto Verzicco and Francesco Viola | Hemodynamic effects of intra- and supra- deployment locations for a bio-prosthetic aortic valve | null | null | null | null | physics.med-ph | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Aortic valve replacement is a key surgical procedure for treating aortic valve pathologies, such as stenosis and regurgitation. The precise placement of the prosthetic valve relative to the native aortic annulus plays a critical role in the post-operative hemodynamics. This study investigates how the positioning of a biological prosthetic valve-either intra-annular (within the native annulus) or supra-annular (slightly downstream, in the widened portion of the aortic root)-affects cardiac fluid dynamics. Using high-fidelity numerical simulations on a patient-specific left heart model derived from CT imaging, we simulate physiological flow conditions to isolate the impact of valve placement. Unlike previous clinical studies that compare different patients and valve models, our approach evaluates the same valve in both positions within a single virtual patient, ensuring a controlled comparison. Key hemodynamic parameters are assessed, including transvalvular pressure drop, effective orifice area, wall shear stress, and hemolysis. Results reveal that supra-annular implantation offers significant advantages: lower pressure gradients, larger orifice area, and reduced shear-induced stress. Furthermore, hemolysis analysis using advanced red blood cell stress models indicates a decreased risk of blood damage in the supra-annular configuration. These findings offer valuable insights to guide valve selection and implantation strategies, ultimately supporting improved patient outcomes. | [
{
"version": "v1",
"created": "Tue, 27 May 2025 20:08:26 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Scarpolini",
"Martino Andrea",
""
],
[
"Vagnoli",
"Giovanni",
""
],
[
"Guglietta",
"Fabio",
""
],
[
"Verzicco",
"Roberto",
""
],
[
"Viola",
"Francesco",
""
]
] |
2505.22688 | Venkata Raghuvamsi Palur | Palur Venkata Raghuvamsi, Siyuan Brandon Loh, Prasanta Bhattacharya, Joses Ho, Raphael Lee Tze Chuen, Alvin X. Han, Sebastian Maurer-Stroh | Investigating the effectiveness of multimodal data in forecasting SARS-COV-2 case surges | null | null | null | null | q-bio.QM cs.LG stat.ML | http://creativecommons.org/licenses/by/4.0/ | The COVID-19 pandemic response relied heavily on statistical and machine learning models to predict key outcomes such as case prevalence and fatality rates. These predictions were instrumental in enabling timely public health interventions that helped break transmission cycles. While most existing models are grounded in traditional epidemiological data, the potential of alternative datasets, such as those derived from genomic information and human behavior, remains underexplored. In the current study, we investigated the usefulness of diverse modalities of feature sets in predicting case surges. Our results highlight the relative effectiveness of biological (e.g., mutations), public health (e.g., case counts, policy interventions) and human behavioral features (e.g., mobility and social media conversations) in predicting country-level case surges. Importantly, we uncover considerable heterogeneity in predictive performance across countries and feature modalities, suggesting that surge prediction models may need to be tailored to specific national contexts and pandemic phases. Overall, our work highlights the value of integrating alternative data sources into existing disease surveillance frameworks to enhance the prediction of pandemic dynamics. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 01:00:02 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Raghuvamsi",
"Palur Venkata",
""
],
[
"Loh",
"Siyuan Brandon",
""
],
[
"Bhattacharya",
"Prasanta",
""
],
[
"Ho",
"Joses",
""
],
[
"Chuen",
"Raphael Lee Tze",
""
],
[
"Han",
"Alvin X.",
""
],
[
"Maurer-Stroh",
"Sebastian",
""
]
] |
2505.22689 | Jialong Guo | Jialong Guo, Xinghao Chen, Yehui Tang, Yunhe Wang | SlimLLM: Accurate Structured Pruning for Large Language Models | ICML 2025 | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Large language models(LLMs) have garnered significant attention and demonstrated impressive capabilities in a wide range of applications. However, due to their enormous computational costs, the deployment and application of LLMs are often severely limited. To address this issue, structured pruning is an effective solution to compress the parameters of LLMs. Determining the importance of each sub-module in LLMs and minimizing performance loss are critical issues that need to be carefully addressed in structured pruning. In this paper, we propose an effective and fast structured pruning method named SlimLLM for large language models. For channel and attention head pruning, we evaluate the importance based on the entire channel or head, rather than merely aggregating the importance of individual elements within a sub-module. This approach enables a more holistic consideration of the interdependence among elements within the sub-module. In addition, we design a simple linear regression strategy for the output matrix to quickly recover performance. We also propose layer-based importance ratio to determine the pruning ratio for each layer. Based on the LLaMA benchmark results, our SlimLLM outperforms other methods and achieves state-of-the-art performance. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 03:01:28 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Guo",
"Jialong",
""
],
[
"Chen",
"Xinghao",
""
],
[
"Tang",
"Yehui",
""
],
[
"Wang",
"Yunhe",
""
]
] |
2505.22690 | Sanasam Surendra Singh | Elangbam Chingkheinganba Meetei, S. Surendra Singh | Phase space analysis of cosmological models on $f(R,G,T)$ gravity | 16 pages, 12 figures | null | null | null | gr-qc hep-th | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In the present article, we developed a dynamical system in the context of modified $f(R,G,T)$ gravity, where $R$, $G$ and $T$ are Ricci scalar, Gauss-Bonnet term and energy-momentum tensor respectively. Development of the dynamical system is done by first defining 9 dimensionless variables and formulate a ordinary differential equations by taking derivative of the variables with respect to logarithmic time $N=\log a(t)$, where $a(t)$ is the scale factor. This formulation is applied to the model $f(R,G,T)=\alpha R^{l}+\beta G^{m} + \gamma T^{n} $, and its solutions and their corresponding stabilities are analysed in details. From the plot of density parameters vs $N=-log(1+z)$, we conclude that our Universe is currently dominated by dark energy, which is compatible with current observation. Taking $l=2$, cosmic parameters such as deceleration parameter $(q)$, equation of state $(\omega)$ and state finder parameters are also discussed by fixing one dimensionless variable, showing our Universe's expansion is accelerating with the present value of $q=-0.5$. Present value of $\omega=-0.66$ suggests that our Universe is in Quintessence phase. Lastly, the validity of the model with respect to the $\Lambda$CDM model is checked by using 77 Hubble, 15 BAO and 1024 pantheon datasets, which implies that our model is aligned with the $\Lambda$CDM model's bahaviour. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 03:51:11 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Meetei",
"Elangbam Chingkheinganba",
""
],
[
"Singh",
"S. Surendra",
""
]
] |
2505.22691 | An Do | Won Joon Sohn, Jeffrey Lim, Po T. Wang, Susan J. Shaw, Michelle Armacost, Hui Gong, Brian Lee, Darrin Lee, Payam Heydari, Richard A. Andersen, Charles Y. Liu, Zoran Nenadic, An H. Do | Early Assessment of Artificial Lower Extremity Sensory Response Times and Proprioceptive Acuity via Sensory Cortex Electrical Stimulation | 6 page conference proceeding pre-print, 3 figures, 4 tables | null | null | null | q-bio.NC q-bio.OT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Bi-directional brain computer interfaces (BD-BCIs) may restore brain-controlled walking and artificial leg sensation after spinal cord injury. Current BD-BCIs provide only simplistic "tingling" feedback, which lacks proprioceptive information to perceive critical gait events (leg swing, double support). This information must also be perceived adequately fast to facilitate timely motor responses. Here, we investigated utilizing primary sensory cortex (S1) direct cortical electrical stimulation (DCES) to deliver leg proprioceptive information and measured response times to artificial leg sensations. Subjects with subdural electrocorticogram electrodes over S1 leg areas participated in two tasks: (1) Proprioceptive acuity: subjects identified the difference between DCES-induced percepts emulating various leg swing speeds; (2) Sensory response: measuring subjects' reaction time to DCES-induced leg sensations, with DCES-hand, visual and auditory control conditions. Three subjects were recruited. Only one completed the proprioceptive assessment, achieving 80%, 70%, 60%, and 53% accuracy in discriminating between fast/slow, fast/medium, medium/slow, and same speeds, respectively (p-value=1.9x10$^{-5}$). Response times for leg/hand percepts were 1007$\pm$413/599$\pm$171 ms, visual leg/hand responses were 528$\pm$137/384$\pm$84 ms, and auditory leg/hand responses were 393$\pm$106/352$\pm$93 ms, respectively. These results suggest proprioceptive information can be delivered artificially, but perception may be significantly delayed. Future work should address improving acuity, reducing response times, and expanding sensory modalities. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 06:35:22 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Sohn",
"Won Joon",
""
],
[
"Lim",
"Jeffrey",
""
],
[
"Wang",
"Po T.",
""
],
[
"Shaw",
"Susan J.",
""
],
[
"Armacost",
"Michelle",
""
],
[
"Gong",
"Hui",
""
],
[
"Lee",
"Brian",
""
],
[
"Lee",
"Darrin",
""
],
[
"Heydari",
"Payam",
""
],
[
"Andersen",
"Richard A.",
""
],
[
"Liu",
"Charles Y.",
""
],
[
"Nenadic",
"Zoran",
""
],
[
"Do",
"An H.",
""
]
] |
2505.22692 | Jing Du | Jing Du, Haley Stone, Yang Yang, Ashna Desai, Hao Xue, Andreas Z\"ufle, Chandini Raina MacIntyre, Flora D. Salim | BLUE: Bi-layer Heterogeneous Graph Fusion Network for Avian Influenza Forecasting | 21 pages, 3 figures, 9 tables. The paper is under review | null | null | null | cs.SI | http://creativecommons.org/licenses/by/4.0/ | Accurate forecasting of avian influenza outbreaks within wild bird populations requires models that account for complex, multi-scale transmission patterns driven by various factors. Spatio-temporal GNN-based models have recently gained traction for infection forecasting due to their ability to capture relations and flow between spatial regions, but most existing frameworks rely solely on spatial connections and their connections. This overlooks valuable genetic information at the case level, such as cases in one region being genetically descended from strains in another, which is essential for understanding how infectious diseases spread through epidemiological linkages beyond geography. We address this gap with BLUE, a B}i-Layer heterogeneous graph fUsion nEtwork designed to integrate genetic, spatial, and ecological data for accurate outbreak forecasting. The framework 1) builds heterogeneous graphs from multiple information sources and multiple layers, 2) smooths across relation types, 3) performs fusion while retaining structural patterns, and 4) predicts future outbreaks via an autoregressive graph sequence model that captures transmission dynamics over time. To facilitate further research, we introduce \textbf{Avian-US} dataset, the dataset for avian influenza outbreak forecasting in the United States, incorporating genetic, spatial, and ecological data across locations. BLUE achieves superior performance over existing baselines, highlighting the value of incorporating multi-layer information into infectious disease forecasting. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 06:36:21 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Du",
"Jing",
""
],
[
"Stone",
"Haley",
""
],
[
"Yang",
"Yang",
""
],
[
"Desai",
"Ashna",
""
],
[
"Xue",
"Hao",
""
],
[
"Züfle",
"Andreas",
""
],
[
"MacIntyre",
"Chandini Raina",
""
],
[
"Salim",
"Flora D.",
""
]
] |
2505.22693 | Irfan Lone | Irfan Lone | Derivation of Fokker-Planck equation from Schrodinger dynamics | null | null | null | null | quant-ph | http://creativecommons.org/licenses/by/4.0/ | The Fokker_Planck equation can be derived in a consistent manner through a microscopic approach based on a unified scheme of classical and quantum mechanics. Here we shall derive it through a purely quantum mechanical approach based on the reversible Schrodinger dynamics. We also give a brief discussion of the path integral representation of the Fokker_Planck equation in light of our derivation. We conclude that, because of the use of the representation of eigenstates of the time-independent Hamiltonian in our derivation, the thermodynamical entropy in this case must correspond to a coarse-graining of the quantum entropy. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 11:51:30 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Lone",
"Irfan",
""
]
] |
2505.22694 | Kun Zhang | Dacao Zhang, Kun Zhang, Shimao Chu, Le Wu, Xin Li, Si Wei | MoRE: A Mixture of Low-Rank Experts for Adaptive Multi-Task Learning | This paper has been accepted to ACL 2025 Findings | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | With the rapid development of Large Language Models (LLMs), Parameter-Efficient Fine-Tuning (PEFT) methods have gained significant attention, which aims to achieve efficient fine-tuning of LLMs with fewer parameters. As a representative PEFT method, Low-Rank Adaptation (LoRA) introduces low-rank matrices to approximate the incremental tuning parameters and achieves impressive performance over multiple scenarios. After that, plenty of improvements have been proposed for further improvement. However, these methods either focus on single-task scenarios or separately train multiple LoRA modules for multi-task scenarios, limiting the efficiency and effectiveness of LoRA in multi-task scenarios. To better adapt to multi-task fine-tuning, in this paper, we propose a novel Mixture of Low-Rank Experts (MoRE) for multi-task PEFT. Specifically, instead of using an individual LoRA for each task, we align different ranks of LoRA module with different tasks, which we named low-rank experts. Moreover, we design a novel adaptive rank selector to select the appropriate expert for each task. By jointly training low-rank experts, MoRE can enhance the adaptability and efficiency of LoRA in multi-task scenarios. Finally, we conduct extensive experiments over multiple multi-task benchmarks along with different LLMs to verify model performance. Experimental results demonstrate that compared to traditional LoRA and its variants, MoRE significantly improves the performance of LLMs in multi-task scenarios and incurs no additional inference cost. We also release the model and code to facilitate the community. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 12:32:09 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Zhang",
"Dacao",
""
],
[
"Zhang",
"Kun",
""
],
[
"Chu",
"Shimao",
""
],
[
"Wu",
"Le",
""
],
[
"Li",
"Xin",
""
],
[
"Wei",
"Si",
""
]
] |
2505.22695 | Tengfei Lyu | Tengfei Lyu, Siyuan Feng, Hao Liu, and Hai Yang | LLM-ODDR: A Large Language Model Framework for Joint Order Dispatching and Driver Repositioning | null | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | Ride-hailing platforms face significant challenges in optimizing order dispatching and driver repositioning operations in dynamic urban environments. Traditional approaches based on combinatorial optimization, rule-based heuristics, and reinforcement learning often overlook driver income fairness, interpretability, and adaptability to real-world dynamics. To address these gaps, we propose LLM-ODDR, a novel framework leveraging Large Language Models (LLMs) for joint Order Dispatching and Driver Repositioning (ODDR) in ride-hailing services. LLM-ODDR framework comprises three key components: (1) Multi-objective-guided Order Value Refinement, which evaluates orders by considering multiple objectives to determine their overall value; (2) Fairness-aware Order Dispatching, which balances platform revenue with driver income fairness; and (3) Spatiotemporal Demand-Aware Driver Repositioning, which optimizes idle vehicle placement based on historical patterns and projected supply. We also develop JointDR-GPT, a fine-tuned model optimized for ODDR tasks with domain knowledge. Extensive experiments on real-world datasets from Manhattan taxi operations demonstrate that our framework significantly outperforms traditional methods in terms of effectiveness, adaptability to anomalous conditions, and decision interpretability. To our knowledge, this is the first exploration of LLMs as decision-making agents in ride-hailing ODDR tasks, establishing foundational insights for integrating advanced language models within intelligent transportation systems. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 13:14:55 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Lyu",
"Tengfei",
""
],
[
"Feng",
"Siyuan",
""
],
[
"Liu",
"Hao",
""
],
[
"Yang",
"Hai",
""
]
] |
2505.22696 | Eleni Nisioti | Eleni Nisioti, Joachim Winther Pedersen, Erwan Plantec, Milton L. Montero, Sebastian Risi | When Does Neuroevolution Outcompete Reinforcement Learning in Transfer Learning Tasks? | null | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | The ability to continuously and efficiently transfer skills across tasks is a hallmark of biological intelligence and a long-standing goal in artificial systems. Reinforcement learning (RL), a dominant paradigm for learning in high-dimensional control tasks, is known to suffer from brittleness to task variations and catastrophic forgetting. Neuroevolution (NE) has recently gained attention for its robustness, scalability, and capacity to escape local optima. In this paper, we investigate an understudied dimension of NE: its transfer learning capabilities. To this end, we introduce two benchmarks: a) in stepping gates, neural networks are tasked with emulating logic circuits, with designs that emphasize modular repetition and variation b) ecorobot extends the Brax physics engine with objects such as walls and obstacles and the ability to easily switch between different robotic morphologies. Crucial in both benchmarks is the presence of a curriculum that enables evaluating skill transfer across tasks of increasing complexity. Our empirical analysis shows that NE methods vary in their transfer abilities and frequently outperform RL baselines. Our findings support the potential of NE as a foundation for building more adaptable agents and highlight future challenges for scaling NE to complex, real-world problems. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 13:15:18 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Nisioti",
"Eleni",
""
],
[
"Pedersen",
"Joachim Winther",
""
],
[
"Plantec",
"Erwan",
""
],
[
"Montero",
"Milton L.",
""
],
[
"Risi",
"Sebastian",
""
]
] |
2505.22697 | Filippo Rinaldi | Filippo Rinaldi, Giacomo Capitani, Lorenzo Bonicelli, Donato Crisostomi, Federico Bolelli, Elisa Ficarra, Emanuele Rodol\`a, Simone Calderara, Angelo Porrello | Update Your Transformer to the Latest Release: Re-Basin of Task Vectors | Accepted at ICML 2025 | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Foundation models serve as the backbone for numerous specialized models developed through fine-tuning. However, when the underlying pretrained model is updated or retrained (e.g., on larger and more curated datasets), the fine-tuned model becomes obsolete, losing its utility and requiring retraining. This raises the question: is it possible to transfer fine-tuning to a new release of the model? In this work, we investigate how to transfer fine-tuning to a new checkpoint without having to re-train, in a data-free manner. To do so, we draw principles from model re-basin and provide a recipe based on weight permutations to re-base the modifications made to the original base model, often called task vector. In particular, our approach tailors model re-basin for Transformer models, taking into account the challenges of residual connections and multi-head attention layers. Specifically, we propose a two-level method rooted in spectral theory, initially permuting the attention heads and subsequently adjusting parameters within select pairs of heads. Through extensive experiments on visual and textual tasks, we achieve the seamless transfer of fine-tuned knowledge to new pre-trained backbones without relying on a single training step or datapoint. Code is available at https://github.com/aimagelab/TransFusion. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 13:55:12 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Rinaldi",
"Filippo",
""
],
[
"Capitani",
"Giacomo",
""
],
[
"Bonicelli",
"Lorenzo",
""
],
[
"Crisostomi",
"Donato",
""
],
[
"Bolelli",
"Federico",
""
],
[
"Ficarra",
"Elisa",
""
],
[
"Rodolà",
"Emanuele",
""
],
[
"Calderara",
"Simone",
""
],
[
"Porrello",
"Angelo",
""
]
] |
2505.22698 | Luca Fantin | Luca Fantin, Marco Antonelli, Margherita Cesetti, Daniele Irto, Bruno Zamengo, Francesco Silvestri | Design and testing of an agent chatbot supporting decision making with public transport data | null | null | null | null | cs.AI | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Assessing the quality of public transportation services requires the analysis of large quantities of data on the scheduled and actual trips and documents listing the quality constraints each service needs to meet. Interrogating such datasets with SQL queries, organizing and visualizing the data can be quite complex for most users. This paper presents a chatbot offering a user-friendly tool to interact with these datasets and support decision making. It is based on an agent architecture, which expands the capabilities of the core Large Language Model (LLM) by allowing it to interact with a series of tools that can execute several tasks, like performing SQL queries, plotting data and creating maps from the coordinates of a trip and its stops. This paper also tackles one of the main open problems of such Generative AI projects: collecting data to measure the system's performance. Our chatbot has been extensively tested with a workflow that asks several questions and stores the generated query, the retrieved data and the natural language response for each of them. Such questions are drawn from a set of base examples which are then completed with actual data from the database. This procedure yields a dataset for the evaluation of the chatbot's performance, especially the consistency of its answers and the correctness of the generated queries. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 14:31:14 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Fantin",
"Luca",
""
],
[
"Antonelli",
"Marco",
""
],
[
"Cesetti",
"Margherita",
""
],
[
"Irto",
"Daniele",
""
],
[
"Zamengo",
"Bruno",
""
],
[
"Silvestri",
"Francesco",
""
]
] |
2505.22699 | Alireza Amani | S. Ganjizadeh, and Alireza Amani, and M. A. Ramzanpour | Correspondence between particle creation and dark components interaction in the context of $f(\mathcal{G})$ gravity | 17 pages, 2 figures | International Journal of Geometric Methods in Modern Physics, Vol. 22, No. 04, 2450312 (2025) | 10.1142/S0219887824503122 | null | gr-qc hep-th | http://creativecommons.org/licenses/by-nc-sa/4.0/ | In this paper, we explore the particle creation scenario in the context of $f(\mathcal{G})$ gravity in flat-FLRW metric. For this purpose, from the perspective of thermodynamics and considering an adiabatic universe, we obtain the modified continuity equation in terms of the dynamic number of particles $N$. On the other hand, we obtain Friedmann's equations in $f(\mathcal{G})$ gravity and then write down the continuity equations of the components of matter and dark energy, taking into account the interaction between them. In what follows, by establishing a correspondence between the particle production scenario and $f(\mathcal{G})$ gravity, we obtain the cosmological parameters in terms of $N$. After that, we find cosmological solutions using power-law cosmology and compare them with Hubble parameter data. Finally, we fit the current model with Hubble data and plot the best fit in terms of the redshift parameter. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 15:02:47 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Ganjizadeh",
"S.",
""
],
[
"Amani",
"Alireza",
""
],
[
"Ramzanpour",
"M. A.",
""
]
] |
2505.22700 | Milo Viviani Mr. | Franco Flandoli, Matteo Palmieri, Milo Viviani | Aggregation of vortex structures in 2D: the blob-wave system and its role in zonal flows | 11 pages, 4 figures | null | null | null | math-ph math.MP physics.flu-dyn | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We give a rigorous mathematical result, supported by accurate numerical simulations, of the aggregation of a concentrated vortex blob with an underlying non constant vorticity field: the blob moves in the direction of the gradient of the field. It is a unique example of Lagrangian explanation of aggregation of vortex structures of the same sign in 2D inviscid fluids. A conceptual model of zonal flow maintenance and modulation based on this idea is then developed, which produces shapes in accordance with observations. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 16:20:19 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Flandoli",
"Franco",
""
],
[
"Palmieri",
"Matteo",
""
],
[
"Viviani",
"Milo",
""
]
] |
2505.22701 | Ziyue Kang | Ziyue Kang, Weichuan Zhang | Frequency-Adaptive Discrete Cosine-ViT-ResNet Architecture for Sparse-Data Vision | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | A major challenge in rare animal image classification is the scarcity of data, as many species usually have only a small number of labeled samples.
To address this challenge, we designed a hybrid deep-learning framework comprising a novel adaptive DCT preprocessing module, ViT-B16 and ResNet50 backbones, and a Bayesian linear classification head. To our knowledge, we are the first to introduce an adaptive frequency-domain selection mechanism that learns optimal low-, mid-, and high-frequency boundaries suited to the subsequent backbones.
Our network first captures image frequency-domain cues via this adaptive DCT partitioning. The adaptively filtered frequency features are then fed into ViT-B16 to model global contextual relationships, while ResNet50 concurrently extracts local, multi-scale spatial representations from the original image. A cross-level fusion strategy seamlessly integrates these frequency- and spatial-domain embeddings, and the fused features are passed through a Bayesian linear classifier to output the final category predictions. On our self-built 50-class wildlife dataset, this approach outperforms conventional CNN and fixed-band DCT pipelines, achieving state-of-the-art accuracy under extreme sample scarcity. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 17:39:58 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Kang",
"Ziyue",
""
],
[
"Zhang",
"Weichuan",
""
]
] |
2505.22702 | Wenceslao Arroyo-Machado | Wenceslao Arroyo-Machado, Nicolas Robinson-Garcia and Daniel Torres-Salinas | Are there stars in Bluesky after the return of Donald Trump to the White House? | arXiv admin note: substantial text overlap with arXiv:2412.05624 | null | null | null | cs.DL | http://creativecommons.org/licenses/by/4.0/ | This study examines the shift in the scientific community from X (formerly Twitter) to Bluesky, its impact on scientific communication, and consequently on social metrics (altmetrics). We analysed 14,497 publications from multidisciplinary and Library and Information Science (LIS) journals between January 2024 and March 2025. The results reveal a notable increase in Bluesky activity for multidisciplinary journals in November 2024, likely influenced by political and platform changes, with mentions multiplying for journals like Nature and Science. In LIS, the adoption of Bluesky is different and shows marked variation between European and United States journals. Although Bluesky remains a minority platform compared to X over the whole period, when focusing on user engagement after the United States elections, we see a much more even distribution between the two platforms. In two LIS journals, Bluesky even surpasses X, while in most others, the difference in user engagement was no longer as pronounced, marking a significant change from previous patterns in altmetrics. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 17:49:18 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Arroyo-Machado",
"Wenceslao",
""
],
[
"Robinson-Garcia",
"Nicolas",
""
],
[
"Torres-Salinas",
"Daniel",
""
]
] |
2505.22703 | Mohammad Yaghini | Mohammad Yaghini, Tudor Cebere, Michael Menart, Aur\'elien Bellet, Nicolas Papernot | Private Rate-Constrained Optimization with Applications to Fair Learning | null | null | null | null | cs.LG cs.CR stat.ML | http://creativecommons.org/licenses/by/4.0/ | Many problems in trustworthy ML can be formulated as minimization of the model error under constraints on the prediction rates of the model for suitably-chosen marginals, including most group fairness constraints (demographic parity, equality of odds, etc.). In this work, we study such constrained minimization problems under differential privacy (DP). Standard DP optimization techniques like DP-SGD rely on the loss function's decomposability into per-sample contributions. However, rate constraints introduce inter-sample dependencies, violating the decomposability requirement. To address this, we develop RaCO-DP, a DP variant of the Stochastic Gradient Descent-Ascent (SGDA) algorithm which solves the Lagrangian formulation of rate constraint problems. We demonstrate that the additional privacy cost of incorporating these constraints reduces to privately estimating a histogram over the mini-batch at each optimization step. We prove the convergence of our algorithm through a novel analysis of SGDA that leverages the linear structure of the dual parameter. Finally, empirical results on learning under group fairness constraints demonstrate that our method Pareto-dominates existing private learning approaches in fairness-utility trade-offs. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 17:55:01 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Yaghini",
"Mohammad",
""
],
[
"Cebere",
"Tudor",
""
],
[
"Menart",
"Michael",
""
],
[
"Bellet",
"Aurélien",
""
],
[
"Papernot",
"Nicolas",
""
]
] |
2505.22704 | Feng Yao | Feng Yao, Zilong Wang, Liyuan Liu, Junxia Cui, Li Zhong, Xiaohan Fu, Haohui Mai, Vish Krishnan, Jianfeng Gao, Jingbo Shang | Training Language Models to Generate Quality Code with Program Analysis Feedback | 10 pages, 3 figures | null | null | null | cs.CL cs.AI | http://creativecommons.org/licenses/by/4.0/ | Code generation with large language models (LLMs), often termed vibe coding, is increasingly adopted in production but fails to ensure code quality, particularly in security (e.g., SQL injection vulnerabilities) and maintainability (e.g., missing type annotations). Existing methods, such as supervised fine-tuning and rule-based post-processing, rely on labor-intensive annotations or brittle heuristics, limiting their scalability and effectiveness. We propose REAL, a reinforcement learning framework that incentivizes LLMs to generate production-quality code using program analysis-guided feedback. Specifically, REAL integrates two automated signals: (1) program analysis detecting security or maintainability defects and (2) unit tests ensuring functional correctness. Unlike prior work, our framework is prompt-agnostic and reference-free, enabling scalable supervision without manual intervention. Experiments across multiple datasets and model scales demonstrate that REAL outperforms state-of-the-art methods in simultaneous assessments of functionality and code quality. Our work bridges the gap between rapid prototyping and production-ready code, enabling LLMs to deliver both speed and quality. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 17:57:47 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Yao",
"Feng",
""
],
[
"Wang",
"Zilong",
""
],
[
"Liu",
"Liyuan",
""
],
[
"Cui",
"Junxia",
""
],
[
"Zhong",
"Li",
""
],
[
"Fu",
"Xiaohan",
""
],
[
"Mai",
"Haohui",
""
],
[
"Krishnan",
"Vish",
""
],
[
"Gao",
"Jianfeng",
""
],
[
"Shang",
"Jingbo",
""
]
] |
2505.22705 | Ting Yao | Qi Cai, Jingwen Chen, Yang Chen, Yehao Li, Fuchen Long, Yingwei Pan, Zhaofan Qiu, Yiheng Zhang, Fengbin Gao, Peihan Xu, Yimeng Wang, Kai Yu, Wenxuan Chen, Ziwei Feng, Zijian Gong, Jianzhuang Pan, Yi Peng, Rui Tian, Siyu Wang, Bo Zhao, Ting Yao, Tao Mei | HiDream-I1: A High-Efficient Image Generative Foundation Model with Sparse Diffusion Transformer | Source codes and models are available at https://github.com/HiDream-ai/HiDream-I1 and https://github.com/HiDream-ai/HiDream-E1 | null | null | null | cs.CV cs.MM | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recent advancements in image generative foundation models have prioritized quality improvements but often at the cost of increased computational complexity and inference latency. To address this critical trade-off, we introduce HiDream-I1, a new open-source image generative foundation model with 17B parameters that achieves state-of-the-art image generation quality within seconds. HiDream-I1 is constructed with a new sparse Diffusion Transformer (DiT) structure. Specifically, it starts with a dual-stream decoupled design of sparse DiT with dynamic Mixture-of-Experts (MoE) architecture, in which two separate encoders are first involved to independently process image and text tokens. Then, a single-stream sparse DiT structure with dynamic MoE architecture is adopted to trigger multi-model interaction for image generation in a cost-efficient manner. To support flexiable accessibility with varied model capabilities, we provide HiDream-I1 in three variants: HiDream-I1-Full, HiDream-I1-Dev, and HiDream-I1-Fast.
Furthermore, we go beyond the typical text-to-image generation and remould HiDream-I1 with additional image conditions to perform precise, instruction-based editing on given images, yielding a new instruction-based image editing model namely HiDream-E1. Ultimately, by integrating text-to-image generation and instruction-based image editing, HiDream-I1 evolves to form a comprehensive image agent (HiDream-A1) capable of fully interactive image creation and refinement. To accelerate multi-modal AIGC research, we have open-sourced all the codes and model weights of HiDream-I1-Full, HiDream-I1-Dev, HiDream-I1-Fast, HiDream-E1 through our project websites: https://github.com/HiDream-ai/HiDream-I1 and https://github.com/HiDream-ai/HiDream-E1. All features can be directly experienced via https://vivago.ai/studio. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 17:59:15 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Cai",
"Qi",
""
],
[
"Chen",
"Jingwen",
""
],
[
"Chen",
"Yang",
""
],
[
"Li",
"Yehao",
""
],
[
"Long",
"Fuchen",
""
],
[
"Pan",
"Yingwei",
""
],
[
"Qiu",
"Zhaofan",
""
],
[
"Zhang",
"Yiheng",
""
],
[
"Gao",
"Fengbin",
""
],
[
"Xu",
"Peihan",
""
],
[
"Wang",
"Yimeng",
""
],
[
"Yu",
"Kai",
""
],
[
"Chen",
"Wenxuan",
""
],
[
"Feng",
"Ziwei",
""
],
[
"Gong",
"Zijian",
""
],
[
"Pan",
"Jianzhuang",
""
],
[
"Peng",
"Yi",
""
],
[
"Tian",
"Rui",
""
],
[
"Wang",
"Siyu",
""
],
[
"Zhao",
"Bo",
""
],
[
"Yao",
"Ting",
""
],
[
"Mei",
"Tao",
""
]
] |
2505.22706 | Fabio Rigamonti | Fabio Rigamonti, Lorenzo Bertassi, Riccardo Buscicchio, Fabiola Cocchiararo, Stefano Covino, Massimo Dotti, Alberto Sesana, and Paola Severgnini | Variability in the massive black hole binary candidate SDSS J2320+0024: no evidence for periodic modulation | 10 pages, 5 figures, submitted to A&A | null | null | null | astro-ph.GA astro-ph.HE | http://creativecommons.org/publicdomain/zero/1.0/ | Massive black hole binaries (MBHBs) are a natural outcome of galaxy mergers, and they are expected to be among the loudest gravitational wave sources at low frequencies. SDSS J2320+0024 has been recently proposed as a promising MBHB candidate due to a possible periodicity in its light-curve and variability in the MgII emission line. In this work, we re-analyse the optical (g and r bands) light-curves of J2320+0024 within the framework of Bayesian model selection. When a periodicity is searched for together with red noise, the analysis of the g-band light-curve finds a peak in the posterior of the period at ~290 days. The posterior profile is too broad to result in a preference for the periodic models with respect to models including only red-noise. Furthermore, the same peak is not present in the analysis of the r-band light-curve. A periodic model without red-noise identifies a different (~1100 days) periodicity, and it is significantly statistically disfavoured with respect to the other tested models. In summary, no significant evidence in favour of a true periodic signal over red noise variability is found. Our analysis questions the robustness of the previously claimed periodicity and emphasizes the importance of rigorous statistical treatment. While our findings challenge the binary interpretation for J2320+0024, they do not rule it out. A statistically robust joint analysis of the photometric light-curves and of the evolving broad line profiles can shed further light on the real nature of this object. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:00:00 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Rigamonti",
"Fabio",
""
],
[
"Bertassi",
"Lorenzo",
""
],
[
"Buscicchio",
"Riccardo",
""
],
[
"Cocchiararo",
"Fabiola",
""
],
[
"Covino",
"Stefano",
""
],
[
"Dotti",
"Massimo",
""
],
[
"Sesana",
"Alberto",
""
],
[
"Severgnini",
"Paola",
""
]
] |
2505.22707 | Eric Lescano | Eric Lescano | The non-relastivistic limit of HSZ Theory | V1: 19 pages + references | null | null | null | hep-th | http://creativecommons.org/licenses/by/4.0/ | We study the non-relativistic (NR) limit of HSZ theory, a higher-derivative theory of gravity with exact and manifest T-duality invariance. Since the theory can be formulated using the generalized metric formalism, the HSZ Lagrangian remains convergent to all orders in derivatives when taking the NR limit. In this work, we analyze the three-derivative corrections to the symmetry transformations of the fields in the NR case, as well as the four-derivative action in the absence of a b-field. Interestingly, the corrections to the metric degrees of freedom cannot be fully trivialized, as in the relativistic case, in order to preserve the convergence of the theory. As HSZ theory interpolates order by order between heterotic and bosonic string theories, the results of this work can be interpreted as a truncation of the four-derivative structure of heterotic supergravity in the NR limit. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:00:00 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Lescano",
"Eric",
""
]
] |
2505.22708 | Steven B. Giddings | Steven B. Giddings | Quantum gravity observables: observation, algebras, and mathematical structure | 18 pages, 11 figures | null | null | null | hep-th | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The questions of describing observables and observation in quantum gravity appear to be centrally important to its physics. A relational approach holds significant promise, and a classification of different types of relational observables (gravitationally dressed, field relational, and more general) is outlined. Plausibly gravitationally dressed observables are particularly closely tied to the fundamental structure of the theory. These may be constructed in the quantum theory to leading order in Newton's constant, and raise important questions about localization of information. Approximate localization is given by a "standard dressing" construction of a "gravitational splitting." It is also argued that such gravitational dressings give a generalization of the crossed product construction, reducing to this and yielding type II von Neumann algebras in special cases. Gravity therefore introduces a significantly more general alteration of the algebraic structure of local quantum field theory, also with apparent connections to holography, but whose implications have not been fully understood. In particular, properties of the algebra of gravitationally dressed observables suggest a possible role for other non-algebraic structure on the Hilbert space for quantum gravity. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:00:00 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Giddings",
"Steven B.",
""
]
] |
2505.22709 | Kirk Barrow | Kirk S. S. Barrow, Thinh Huu Nguyen, Edward C. Skrabacz | Haskap Pie: A Halo finding Algorithm with efficient Sampling, K-means clustering, tree-Assembly, Particle tracking, Python modules, Inter-code applicability, and Energy solving | 27 pages, 14 figures, submitted to MNRAS | null | null | null | astro-ph.CO astro-ph.GA | http://creativecommons.org/licenses/by-nc-nd/4.0/ | We describe a new Python-based stand-alone halo finding algorithm, Haskap Pie, that combines several methods of halo finding and tracking into a single calculation. Our halo-finder flexibly solves halos for simulations produced by eight simulation codes (ART-I, ENZO, RAMSES, CHANGA, GADGET-3, GEAR, AREPO and GIZMO) and for both zoom-in or full-box N-body or hydrodynamical simulations without the need for additional tuning or user-specified modeling parameters. When compared to Rockstar and Consistent Trees, our halo-finder tracks subhalos much longer and more consistently, produces halos with better constrained physical parameters, and returns a much denser halo mass function for halos with more than 100 particles. Our results also compare favorably to recently described specialized particle-tracking extensions to Rockstar. Our algorithm is well-suited to a variety of studies of simulated galaxies and is particularly robust for a new generation of studies of merging and satellite galaxies. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:00:00 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Barrow",
"Kirk S. S.",
""
],
[
"Nguyen",
"Thinh Huu",
""
],
[
"Skrabacz",
"Edward C.",
""
]
] |
2505.22710 | Marianne Moore | Pouya Asadi, Marianne Moore, David E. Morrissey, Michael Shamma | Genesis of Baryon and Dark Matter Asymmetries through Ultraviolet Scattering Freeze-in | 42 pages + appendices, 8 figures | null | null | MIT-CTP/5873 | hep-ph | http://creativecommons.org/licenses/by/4.0/ | We introduce a new mechanism for the simultaneous generation of baryon and dark matter asymmetries through ultraviolet-dominated freeze-in scatterings. The mechanism relies on heavy Majorana neutrinos that connect the visible Standard Model sector to a dark sector through the neutrino portal. Following reheating of the visible sector to a temperature well below the heavy neutrino masses, we show that 2-to-2 scattering processes can populate the dark sector and generate both baryon and dark matter asymmetries. In some parameter regions, the dominant source of baryon asymmetry can be charge transfer from the dark sector, a process we call dark wash-in. We also demonstrate that annihilation of the dark matter to massless states within the dark sector can deplete the symmetric population without destroying the net baryon charge to leave only an asymmetric dark matter abundance today. Depending on the specific model parameters, the observed baryon and dark matter abundances can be attained with heavy neutrino masses $M_N \gtrsim 10^{10}$ GeV, and dark matter masses in the range 0.1 GeV $\lesssim m_\chi \lesssim 10^3$ GeV if the dark matter relic abundance is mainly asymmetric and even lower masses if it is symmetric. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:00:00 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Asadi",
"Pouya",
""
],
[
"Moore",
"Marianne",
""
],
[
"Morrissey",
"David E.",
""
],
[
"Shamma",
"Michael",
""
]
] |
2505.22711 | Matheus Hostert | Yasaman Farzan and Matheus Hostert | Astrophysical sources of dark particles as a solution to the KM3NeT and IceCube tension over KM3-230213A | 13 pages, 5 figures | null | null | null | hep-ph | http://creativecommons.org/licenses/by/4.0/ | We entertain the possibility that transient astrophysical sources can produce a flux of dark particles that induce ultra-high-energy signatures at neutrino telescopes such as IceCube and KM3NeT. We construct scenarios where such ``dark flux" can produce meta-stable dark particles inside the Earth that subsequently decay to muons, inducing through-going tracks in large-volume neutrino detectors. We consider such a scenario in light of the $\mathcal{O}(70)$~PeV ultra-high-energy muon observed by KM3NeT and argue that because of its location in the sky and the strong geometrical dependence of the signal, such events would not necessarily have been observed by IceCube. Our model relies on the upscattering of a new particle $X$ onto new metastable particles that decay to dimuons with decay lengths of $\mathcal{O}(100)$~km. This scenario can explain the observation by KM3NeT without being in conflict with the IceCube data. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:00:00 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Farzan",
"Yasaman",
""
],
[
"Hostert",
"Matheus",
""
]
] |
2505.22712 | Andrew Marszewski | Andrew Marszewski, Claude-Andr\'e Faucher-Gigu\`ere, Robert Feldmann, Guochao Sun | Explaining the Weak Evolution of the High-Redshift Mass-Metallicity Relation with Galaxy Burst Cycles | 17 pages, 8 figures | null | null | null | astro-ph.GA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Recent observations suggest a nearly constant gas-phase mass-metallicity relation (MZR) at $z \gtrsim 5$, in agreement with many theoretical predictions. This lack of evolution contrasts with observations at $z \lesssim 3$, which find an increasing normalization of the MZR with decreasing redshift. We analyze a high-redshift suite of FIRE-2 cosmological zoom-in simulations to identify the physical drivers of the MZR. Previous studies have explained the weak evolution of the high-redshift MZR in terms of weakly evolving or saturated gas fractions, but we find this alone does not explain the evolution in FIRE-2. Instead, stellar feedback following intense bursts of star formation drives enriched gas out of galaxies, resetting their interstellar medium and separating their histories into distinct ``burst cycles". We develop the ``Reduced Burst Model", a simplified gas-regulator model that successfully reproduces the simulated MZR and identifies the dominant drivers of its evolution. As redshift decreases, the metallicity of inflows within burst cycles increases at fixed stellar mass due to increased wind recycling of enriched gas. Meanwhile, the metal mass produced by stars per inflowing gas mass within these cycles decreases because of decreased star formation per gas mass inflowing into the galaxy. The effects of these two processes on the median metallicity largely cancel, holding the MZR constant for $z = 5 - 12$. At fixed stellar mass, the simulations predict lower gas metallicities at higher $\rm H\alpha$-derived star formation rates, in qualitative agreement with the fundamental metallicity relation (FMR), but this effect is reduced in rest UV-selected samples. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:00:00 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Marszewski",
"Andrew",
""
],
[
"Faucher-Giguère",
"Claude-André",
""
],
[
"Feldmann",
"Robert",
""
],
[
"Sun",
"Guochao",
""
]
] |
2505.22713 | Tom Rudelius | Matthew Reece, Tom Rudelius, Christopher Tudball | Co-Scaling and Alignment of Electric and Magnetic Towers | 73 pages, 12 figures | null | null | null | hep-th | http://creativecommons.org/licenses/by/4.0/ | Towers of electrically and magnetically charged states in quantum gravity often exhibit two important properties. First, the ratio of the mass (or tension) of electrically charged states to magnetically charged states is of order $e^2/(4\pi)$, which we refer to as "co-scaling." Second, in theories of multiple gauge fields, the towers of states that exhibit co-scaling have charges that point in approximately the same direction in charge space as measured by the gauge kinetic matrix, which we refer to as "alignment." After motivating these ideas with some heuristic arguments, we examine the spectrum of BPS states in the 5d supergravity landscape arising from M-theory on a Calabi-Yau threefold. In this setting, every tower of magnetically charged strings is paired with a corresponding tower of electrically charged particles that exhibits co-scaling and rapid alignment. In particular, this motivates a sharp mathematical characterization of the magnetic infinity cone in Calabi-Yau geometry. We conjecture that co-scaling and alignment are universal properties of towers of states which, in some limit in moduli space, have maximally divergent charge-to-mass ratios. Co-scaling is not a general feature of extremal black hole solutions in theories of gauge fields and scalars, suggesting that it is a principle of UV complete quantum gravity. We briefly remark on possible phenomenological applications, including to axion physics. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:00:00 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Reece",
"Matthew",
""
],
[
"Rudelius",
"Tom",
""
],
[
"Tudball",
"Christopher",
""
]
] |
2505.22714 | Rachit Kapoor | Rachit Kapoor, D. Yahne, V.O. Garlea, G. Hester | Localized Triplons and Site Stuffing in the Quantum Dimer Magnet BiYbGeO$_5$ | 6 pages, 5 figures | null | null | null | cond-mat.str-el cond-mat.mtrl-sci | http://creativecommons.org/licenses/by/4.0/ | Thermodynamic and muon spin-relaxation measurements have recently highlighted BiYbGeO$_5$ as a new example of a rare-earth-based quantum dimer magnet with isolated Yb$^{3+}$ spin-$\frac{1}{2}$ dimers. However, direct spectroscopic evidence of the triplet excitations and measurements of the structural disorder are lacking. In this work, polycrystalline BiYbGeO$_5$ was synthesized using conventional high-temperature solid-state methods and investigated via high-resolution neutron powder diffraction and inelastic neutron scattering. Diffraction measurements down to 58 mK reveal no signatures of magnetic order and indicate that nearly 20\% of Yb$^{3+}$ sites are replaced by non-magnetic Bi$^{3+}$, introducing significant structural disorder. Inelastic neutron scattering shows dispersionless triplon excitations, consistent with localized, non-interacting spin dimers. Fits to the triplet excitation spectrum identify an XXZ-type anisotropic exchange with $J_{XX}$ = 0.11(2) meV and $J_Z = 0.15(1)$ meV. These findings establish BiYbGeO$_5$ as a structurally disordered but magnetically well-isolated quantum dimer system, providing a model platform for studying the resilience of entangled spin states to site dilution. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:00:00 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Kapoor",
"Rachit",
""
],
[
"Yahne",
"D.",
""
],
[
"Garlea",
"V. O.",
""
],
[
"Hester",
"G.",
""
]
] |
2505.22715 | Yannick Stade | Yannick Stade, Wan-Hsuan Lin, Jason Cong, and Robert Wille | Routing-Aware Placement for Zoned Neutral Atom-based Quantum Computing | 9 pages, 10 figures | null | null | null | quant-ph cs.ET | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Quantum computing promises to solve previously intractable problems, with neutral atoms emerging as a promising technology. Zoned neutral atom architectures allow for immense parallelism and higher coherence times by shielding idling atoms from interference with laser beams. However, in addition to hardware, successful quantum computation requires sophisticated software support, particularly compilers that optimize quantum algorithms for hardware execution. In the compilation flow for zoned neutral atom architectures, the effective interplay of the placement and routing stages decides the overhead caused by rearranging the atoms during the quantum computation. Sub-optimal placements can lead to unnecessary serialization of the rearrangements in the subsequent routing stage. Despite this, all existing compilers treat placement and routing independently thus far - focusing solely on minimizing travel distances. This work introduces the first routing-aware placement method to address this shortcoming. It groups compatible movements into parallel rearrangement steps to minimize both rearrangement steps and travel distances. The implementation utilizing the A* algorithm reduces the rearrangement time by 17% on average and by 49% in the best case compared to the state-of-the-art. The complete code is publicly available in open-source as part of the Munich Quantum Toolkit (MQT) at https://github.com/munich-quantum-toolkit/qmap. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:00:01 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Stade",
"Yannick",
""
],
[
"Lin",
"Wan-Hsuan",
""
],
[
"Cong",
"Jason",
""
],
[
"Wille",
"Robert",
""
]
] |
2505.22716 | Sergio Ernesto Aguilar Gutierrez | Sergio E. Aguilar-Gutierrez | From chords to dynamical wormholes with matter: Towards a bulk double-scaled (SYK) algebra | 53pgs + Appendices. 8 figures | null | null | null | hep-th | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We formulate a bulk description of the double-scaled algebra of the DSSYK model. Based on the Hartle-Hawking (HH) state with matter chords, we derive several properties of the DSSYK model with heavy or light matter chords, without making assumptions about the specific dual theory, including its semiclassical thermodynamics, correlation functions, and {Krylov complexity}. These quantities are found from the saddle points of the DSSYK path integral preparing the HH state. We also construct a Lanczos algorithm that simultaneously evaluates Krylov state and operator complexity in the two-sided Hamiltonian system including finite temperature effects. In the semiclassical limit, both measures are encoded in the saddle points of the path integral. They have a bulk interpretation in terms of minimal geodesic lengths in AdS$_2$ space with backreaction due to a shockwave. Different saddle points correspond to geodesic distances with different asymptotic boundary time evolution, and they display different scrambling properties. This allows us to quantize the bulk dual theory of the DSSYK model, which we find is sine dilaton gravity, assuming the Schr\"odinger equation and momentum shift symmetry. At last, we deduce the boundary/bulk dictionary of the double-scaled algebra. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:00:01 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Aguilar-Gutierrez",
"Sergio E.",
""
]
] |
2505.22717 | Rapha\"el Errani | Rapha\"el Errani, Jorge Pe\~narrubia, Matthew G. Walker | Stellar Mass Segregation in Dark Matter Halos | submitted to ApJ, comments welcome | null | null | null | astro-ph.GA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We study the effect of stellar mass segregation driven by collisional relaxation within the potential well of a smooth dark matter halo. This effect is of particular relevance for old stellar systems with short crossing times, where small collisional perturbations accumulate over many dynamical time scales. We run collisional $N$-body simulations tailored to the ambiguous stellar systems Ursa Major 3/Unions 1, Delve 1 and Eridanus 3, modelling their stellar populations as two-component systems of high- and low-mass stars, respectively. For Ursa Major 3/Unions 1 (Delve 1), assuming a dynamical-to-stellar mass ratio of 10, we find that after 10 Gyr of evolution, the radial extent of its low-mass stars will be twice as large (40 per cent larger) than that of its high-mass stars. We show that weak tides do not alter this relative separation of half-light radii, whereas for the case of strong tidal fields, mass segregation facilitates the tidal stripping of low-mass stars. We further find that as the population of high-mass stars contracts and cools, the number of dynamically formed binaries within that population increases. Our results call for caution when using stellar mass segregation as a criterion to separate star clusters from dwarf galaxies, and suggest that mass segregation increases the abundance of massive binaries in the central regions of dark matter-dominated dwarf galaxies. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:00:01 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Errani",
"Raphaël",
""
],
[
"Peñarrubia",
"Jorge",
""
],
[
"Walker",
"Matthew G.",
""
]
] |
2505.22718 | Laura Reymond | Laura Reymond, Alexander Reeves, Pierre Zhang, Alexandre Refregier | $\texttt{SwiftC}_\ell$: fast differentiable angular power spectra beyond Limber | 18 pages, 7 figures, submitted to JCAP | null | null | null | astro-ph.CO | http://creativecommons.org/licenses/by/4.0/ | The upcoming stage IV wide-field surveys will provide high precision measurements of the large-scale structure (LSS) of the universe. Their interpretation requires fast and accurate theoretical predictions including large scales. For this purpose, we introduce $\texttt{SwiftC}_\ell$, a fast, accurate and differentiable $\texttt{JAX}$-based pipeline for the computation of the angular power spectrum beyond the Limber approximation. It uses a new FFTLog-based method which can reach arbitrary precision and includes interpolation along $k$, allowing for $k$-dependent growth factor and biases. $\texttt{SwiftC}_\ell$ includes a wide range of probes and effects such as galaxy clustering, including magnification bias, redshift-space distortions and primordial non-Gaussianity, weak lensing, including intrinsic alignment, cosmic microwave background (CMB) lensing and CMB integrated Sachs-Wolfe effect. We compare our pipeline to the other available beyond-Limber codes within the N5K challenge from the Rubin Observatory Legacy Survey of Space and Time (LSST) Dark Energy Science Collaboration. $\texttt{SwiftC}_\ell$ computes the 120 different angular power spectra over 103 $\ell$-multipoles in 5 ms on one GPU core. Using a pre-calculation, $\texttt{SwiftC}_\ell$ is thus about 40$\times$ faster than the winner of the N5K challenge with comparable accuracy. Furthermore, all outputs are auto-differentiable, facilitating gradient-based sampling and robust and accurate Fisher forecasts. We showcase a Markov Chain Monte Carlo on an LSST-like survey as well as a Fisher forecast, illustrating $\texttt{SwiftC}_\ell$'s differentiability, speed and reliability in measuring cosmological parameters. The code is publicly available at https://cosmo-gitlab.phys.ethz.ch/cosmo_public/swiftcl. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:00:01 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Reymond",
"Laura",
""
],
[
"Reeves",
"Alexander",
""
],
[
"Zhang",
"Pierre",
""
],
[
"Refregier",
"Alexandre",
""
]
] |
2505.22719 | Qiushi Wei | Jeff A. Dror, Qiushi Wei | On Pulsar Timing Detection of Ultralight Vector Dark Matter | 20 pages, 1 figure | null | null | null | hep-ph astro-ph.HE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Ultralight vector dark matter induces metric fluctuations that generate timing residuals in the arrival times of pulsar emissions through two distinct modes: a fast mode, sourced by coherent field oscillations, and a slow mode, arising from interference patterns. These modes enable the detection of vector dark matter with masses $m \sim 10^{-24} - 10^{-22}\ \mathrm{eV}$ and $m \sim 10^{-18} - 10^{-16}\ \mathrm{eV}$, respectively, using pulsar timing arrays. While previous studies have explored the fast mode, they neglect the full statistical treatment of the vector field and a precise treatment of its polarization structure. In this work, we investigate the timing residuals from both modes, fully accounting for the statistical properties of ultralight vector dark matter, assuming equipartition among its three polarization states. The two-point correlation functions of timing residuals that we derive serve as direct tools for identifying vector dark matter signatures as a stochastic background in pulsar timing data. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:00:01 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Dror",
"Jeff A.",
""
],
[
"Wei",
"Qiushi",
""
]
] |
2505.22720 | Guo-Yi Zhu | Malte P\"utz, Romain Vasseur, Andreas W.W. Ludwig, Simon Trebst, Guo-Yi Zhu | Flow to Nishimori universality in weakly monitored quantum circuits with qubit loss | 8 + 7 pages; 6 + 6 figures | null | null | null | quant-ph cond-mat.dis-nn cond-mat.stat-mech cond-mat.str-el | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In circuit-based quantum state preparation, qubit loss and coherent errors are circuit imperfections that imperil the formation of long-range entanglement beyond a certain threshold. The critical theory at the threshold is a continuous entanglement transition known to be described by a (2+0)-dimensional non-unitary conformal field theory which, for the two types of imperfections of certain circuits, is described by either percolation or Nishimori criticality, respectively. Here we study the threshold behavior when the two types of errors simultaneously occur and show that, when moving away from the Clifford-regime of projective stabilizer measurements, the percolation critical point becomes unstable and the critical theory flows to Nishimori universality. We track this critical renormalization group (RG) crossover flow by mapping out the entanglement phase diagrams, parametrized by the probability and strength of random weak measurements, of two dual protocols preparing surface code or GHZ-class cat states from a parent cluster state via constant-depth circuits. Extensive numerical simulations, using hybrid Gaussian fermion and tensor network / Monte Carlo sampling techniques on systems with more than a million qubits, demonstrate that an infinitesimal deviation from the Clifford regime leads to a sudden, strongly non-monotonic entanglement growth at the incipient non-unitary RG flow. We argue that spectra of scaling dimensions of both the percolation and Nishimori fixed points exhibit multifractality. For percolation, we provide the exact (non-quadratic) multifractal spectrum of exponents, while for the Nishimori fixed point we show high-precision numerical results for five leading exponents characterizing multifractality. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:00:01 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Pütz",
"Malte",
""
],
[
"Vasseur",
"Romain",
""
],
[
"Ludwig",
"Andreas W. W.",
""
],
[
"Trebst",
"Simon",
""
],
[
"Zhu",
"Guo-Yi",
""
]
] |
2505.22721 | Mostafa Ali | Naushad A. Kamar, Mostafa Ali, Mohammad Maghrebi | Markovian dissipation can stabilize a (localization) quantum phase transition | null | null | null | null | cond-mat.quant-gas quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Quantum phase transitions are a cornerstone of many-body physics at low temperatures but have remained elusive far from equilibrium. Driven open quantum systems -- a prominent non-equilibrium platform where coherent dynamics competes with Markovian dissipation from the environment -- often exhibit an effective classical behavior. In this work, we present a nontrivial quantum phase transition that is stabilized, rather than destroyed, by Markovian dissipation. We consider a variant of the paradigmatic spin-boson model where the spin is driven and bosons are subject to Markovian loss proportional to frequency (hence, vanishing at low frequencies). We show that the steady state exhibits a localization phase transition where the spin's dynamics is frozen, to be contrasted with the ground-state transition in the absence of dissipation. Furthermore, this transition occurs when the steady state becomes pure. The latter is not simply a dark state of dissipation but rather emerges from a nontrivial renormalization of the spin dynamics by low-frequency bosonic modes. Our work provides a nontrivial example where quantumness, typically reserved for ground states, also emerges in dynamical settings, with potential applications in quantum computation. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:00:01 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Kamar",
"Naushad A.",
""
],
[
"Ali",
"Mostafa",
""
],
[
"Maghrebi",
"Mohammad",
""
]
] |
2505.22722 | Michael Baker | Michael J. Baker, Joaquim Iguaz Juan, Aidan Symons, Andrea Thamm | Explaining the PeV Neutrino Fluxes at KM3NeT and IceCube with Quasi-Extremal Primordial Black Holes | 6 pages, 3 figures | null | null | null | hep-ph astro-ph.CO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The KM3NeT experiment has recently observed a neutrino with an energy around 100 PeV, and IceCube has detected five neutrinos with energies above 1 PeV. While there are no known astrophysical sources, exploding primordial black holes could have produced these high-energy neutrinos. For Schwarzschild black holes this interpretation results in tensions between the burst rates inferred from the KM3NeT and IceCube observations, and with indirect constraints from the extragalactic gamma ray background. In this letter we show that if there is a population of primordial black holes charged under a new dark $u(1)$ symmetry which spend most of their time in a quasi-extremal state, the neutrino emission at 1 PeV may be more suppressed than at 100 PeV. The burst rates implied by the KM3NeT and IceCube observations and the indirect constraints can then all be consistent at $1\sigma$. Furthermore, these black holes could constitute all of the observed dark matter in the universe. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:00:01 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Baker",
"Michael J.",
""
],
[
"Juan",
"Joaquim Iguaz",
""
],
[
"Symons",
"Aidan",
""
],
[
"Thamm",
"Andrea",
""
]
] |
2505.22723 | Max Uetrecht | Max Uetrecht, Igor F. Herbut, Michael M. Scherer, Emmanuel Stamou and Tom Steudtner | Quantum multicriticality and emergent symmetry in Dirac systems with two order parameters at three-loop order | 17 pages, 3 figures | null | null | null | cond-mat.str-el hep-th | http://creativecommons.org/licenses/by/4.0/ | Two-dimensional materials with interacting Dirac excitations can host quantum multicritical behavior near the phase boundaries of the semimetallic and two-ordered phases. We study such behavior in Gross--Neveu--Yukawa field theories where $N_f$ flavors of Dirac fermions are coupled to two order-parameter fields with $SO(N_A)$ and $SO(N_B)$ symmetry, respectively. To that end, we employ the perturbative renormalization group up to three-loop order in $4-\epsilon$ spacetime dimensions. We distinguish two key scenarios: (i) The two orders are compatible as characterized by anticommuting mass terms, and (ii) the orders are incompatible. For the first case, we explore the stability of a quantum multicritical point with emergent $SO(N_A\!+\!N_B)$ symmetry. We find that the stability is controlled by increasing the number of Dirac fermion flavors. Moreover, we extract the series expansion of the leading critical exponents for the chiral $SO(4)$ and $SO(5)$ models up to third order in $\epsilon$. Notably, we find a tendency towards rapidly growing expansion coefficients at higher orders, rendering an extrapolation to $\epsilon=1$ difficult. For the second scenario, we study a model with $SO(4) \simeq SO(3) \times SO(3)$ symmetry, which was recently suggested to describe criticality of antiferromagnetism and superconductivity in Dirac systems. However, it was also argued that a physically admissible renormalization-group fixed point only exists for $N_f$ above a critical number $N_{c}^>$. We determine the corresponding series expansion at three-loop order as $N_{c}^>\approx 16.83-7.14\epsilon-7.12\epsilon^2$. This suggests that the physical choice of $N_f=2$ may be a borderline case, where true criticality and pseudocriticality, as induced by fixed-point annihilation, are extremely challenging to distinguish. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:00:01 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Uetrecht",
"Max",
""
],
[
"Herbut",
"Igor F.",
""
],
[
"Scherer",
"Michael M.",
""
],
[
"Stamou",
"Emmanuel",
""
],
[
"Steudtner",
"Tom",
""
]
] |
2505.22724 | Ralph Pudritz | Ralph E. Pudritz, Alex J. Cridland, Julie Inglis, and Mathew Alessi | Connecting Planetary Composition with Formation: a New Paradigm Emerges | 65 pages of text, 14 pages of references, 26 figures; Review to be published as a Chapter in Handbook of Exoplanets, 2nd Edition, eds. Hans J. Deeg and Juan Antonio Belmonte (Springer). arXiv admin note: substantial text overlap with arXiv:1804.04070 | null | null | null | astro-ph.EP | http://creativecommons.org/licenses/by/4.0/ | Extensive ground and space based surveys have now characterized the properties of thousands of exoplanets; their radii, masses, orbits around their host stars, and the beginnings of accurate measurements of the chemical compositions of their atmospheres and cores. How are these properties linked to their formation in physically and chemically evolving protoplanetary disks wherein they accrete pebbles, planetesimals, and gas as they undergo migration? To address this challenge, our review assembles a large and varied body of exoplanet observations as well as recent Atacama Large Millimeter Array (ALMA) and James Webb Space Telescope (JWST) observations of disk structure, chemistry, kinematics, and winds. The latest advances in theory and MHD simulations that bear on these issues are also reviewed and compared with the observations. Taken together, this review argues that a new dynamic paradigm for planet formation is emerging wherein MHD disk winds and not disk turbulence play a central role in disk evolution and planet formation including: angular momentum transport, gap and ring formation. disk astrochemistry, and planet formation and migration. These processes leave their mark on the resulting atmospheric composition, radii, and orbital characteristics of exoplanet populations, offering the possibility of future observational tests. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:00:01 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Pudritz",
"Ralph E.",
""
],
[
"Cridland",
"Alex J.",
""
],
[
"Inglis",
"Julie",
""
],
[
"Alessi",
"Mathew",
""
]
] |
2505.22725 | Hendrik Hohmann | Jonas Beck, Jonathan Bodky, Matteo D\"urrnagel, Ronny Thomale, Julian Ingham, Lennart Klebl, Hendrik Hohmann | Kekul\'e order from diffuse nesting near higher-order Van Hove points | null | null | null | null | cond-mat.str-el cond-mat.mtrl-sci | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Translation symmetry-breaking order is assumed to be suppressed by the lack of Fermi surface nesting near certain higher-order Van Hove singularities (HOVHS). We show the anisotropic band-flattening inherent to such HOVHS, combined with broadening of the Fermi surface due to elevated critical temperatures, results in the Fermi surface becoming approximately nested at a wavevector unrelated to the precise shape of the Fermi surface - leading to a $\sqrt{3}\times\sqrt{3}$ Kekul\'e density wave formation. The effect is demonstrated using unbiased renormalization group calculations for a model of the breathing kagome lattice. Our mechanism - termed diffuse nesting - represents an entirely new notion in the study of Fermi surface instabilities. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:00:02 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Beck",
"Jonas",
""
],
[
"Bodky",
"Jonathan",
""
],
[
"Dürrnagel",
"Matteo",
""
],
[
"Thomale",
"Ronny",
""
],
[
"Ingham",
"Julian",
""
],
[
"Klebl",
"Lennart",
""
],
[
"Hohmann",
"Hendrik",
""
]
] |
2505.22726 | Kiril Hristov | Kiril Hristov, Peng-Ju Hu, and Yi Pang | Black hole thermodynamics at 4 derivatives, natural variables and BPS limits | 43 pages | null | null | USTC-ICTS/PCFT-25-21 | hep-th gr-qc | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We study Einstein-Maxwell theory in $D \geq 3$ spacetime dimensions including all Lorentz-invariant parity-even four-derivative couplings. Building on the results of arXiv:2312.11610, we consider static, charged, asymptotically flat black hole solutions to first order in the higher-derivative expansion. In $D=4$ and $D=5$, we compute the corrected black hole thermodynamics and compare with the Reall-Santos prescription based on the two-derivative background, highlighting a subtlety when both inner and outer horizons are involved. By introducing natural variables, as in arXiv:2304.07320, we recast the on-shell actions in terms of left- and right-moving chemical potentials, which significantly simplifies the analysis.
We also compute first-order thermodynamic corrections for the most general rotating black holes in $D=4$ and $D=5$, without modifying the background solutions. We identify a novel BPS-like limit in $D=4$, extending known supergravity results beyond their traditional domain of validity. Finally, in $D=5$, the analysis of BPS and almost BPS limits enables an independent verification of the five-dimensional BPS thermodynamics. We clarify the origin of a discrepancy in the literature concerning higher-derivative supergravity localization, sharpening the tension between direct computations and predictions based on the $D=4$/$D=5$ connection. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:00:02 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Hristov",
"Kiril",
""
],
[
"Hu",
"Peng-Ju",
""
],
[
"Pang",
"Yi",
""
]
] |
2505.22727 | Pavel E. Mancera Pi\~na | Pavel E. Mancera Pi\~na, Justin I. Read, Stacy Kim, Antonino Marasco, Jos\'e A. Benavides, Marcin Glowacki, Gabriele Pezzulli, and Claudia del P. Lagos | The galaxy-halo connection of disc galaxies over six orders of magnitude in stellar mass | Accepted for publication in A&A. 18 pages (10 figures) + appendices | null | null | null | astro-ph.GA astro-ph.CO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | (Abridged) The relations between stellar ($M_\ast$), gas ($M_{\rm gas}$), baryonic ($M_{\rm bar} = M_\ast + M_{\rm gas}$), and dark matter halo mass ($M_{200}$) provide unique constraints on galaxy formation and cosmology. The shape of the relations constrains how galaxies regulate their growth through gas accretion, star formation, and feedback; their scatter probes the stochasticity of galaxy assembly.
Here, we assemble a sample of 49 nearby gas-rich dwarf and massive disc galaxies with unmatched ancillary data. We obtain their gas kinematics and derive their dark matter properties through rotation curve decomposition. Our sample allows us to study the galaxy-halo connection across nearly six orders of magnitude in $M_\ast$. We find that the $M_{\rm gas}-M_{200}$ relation rises monotonically, with galaxies having around 4 per cent of the average cosmological baryon fraction in cold gas. Contrastingly, the $M_\ast-M_{200}$ relation shows a more complex behaviour. A particularly interesting finding is that of a population of baryon-deficient' dwarfs (BDDs) with stellar masses $\sim 1-1.5$ orders of magnitude lower than expected from current models. Yet, baryon-rich galaxies also exist, and we find a large spread in the baryon retention fraction across our galaxies. We compare our findings with semi-analytic and hydrodynamical galaxy formation simulations. While the simulations broadly reproduce most observed features, they struggle to match the BDDs and do not capture the diversity in baryon fractions. Understanding these differences will shed new light on how feedback regulates galaxy formation. Finally, we study the dark matter halo concentration-mass relation. We find that below $M_{200} \sim 10^{11}\,M_\odot$, the concentrations are systematically lower than expected. We discuss whether these results stem from the influence of baryonic physics or the environment. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:00:02 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Piña",
"Pavel E. Mancera",
""
],
[
"Read",
"Justin I.",
""
],
[
"Kim",
"Stacy",
""
],
[
"Marasco",
"Antonino",
""
],
[
"Benavides",
"José A.",
""
],
[
"Glowacki",
"Marcin",
""
],
[
"Pezzulli",
"Gabriele",
""
],
[
"Lagos",
"Claudia del P.",
""
]
] |
2505.22728 | Ted Johnson | Ted M. Johnson, Rebecca G. Martin, Stephen Lepp, Stephen H. Lubow | The fraction of polar aligned circumbinary disks | 15 pages, 9 figures. Accepted for publication in AJ | null | null | null | astro-ph.EP astro-ph.SR | http://creativecommons.org/licenses/by/4.0/ | Circumbinary gas disks that are misaligned to the binary orbital plane evolve toward either a coplanar or a polar-aligned configuration with respect to the binary host. The preferred alignment depends on the dynamics of the disk: whether it undergoes librating or circulating nodal precession, with librating disks evolving to polar inclinations and circulating disks evolving to coplanar. We quantify the fraction of binary star systems whose disks are expected to have polar orbits $f_\text{polar}$, extending previous work to include disks with non-zero mass. Our results suggest that, for low mass disks, the polar fraction is highly sensitive to the distribution of binary eccentricity with a higher fraction expected for higher binary eccentricities, $f_{\rm polar}\sim e_{\rm b}$. However, for massive discs, the fraction is independent of the binary eccentricity and $f_{\rm polar}\approx 0.37$. The value of $f_\text{polar}$ is always reduced in a population with a greater preference for low initial mutual inclination. We also explore the consequences of the finite lifetime and non-zero radial extent of a real disk, both of which affect a disk's ability to complete its evolution to a stationary configuration. Our findings can be used to make predictions given populations with well-understood distributions of binary eccentricity, initial mutual inclination, and disk angular momentum. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:00:03 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Johnson",
"Ted M.",
""
],
[
"Martin",
"Rebecca G.",
""
],
[
"Lepp",
"Stephen",
""
],
[
"Lubow",
"Stephen H.",
""
]
] |
2505.22729 | Visal So | Visal So, Midhuna Duraisamy Suganthi, Mingjian Zhu, Abhishek Menon, George Tomaras, Roman Zhuravel, Han Pu, Peter G. Wolynes, Jos\'e N. Onuchic, Guido Pagano | Quantum Simulations of Charge and Exciton Transfer in Multi-mode Models using Engineered Reservoirs | 10+7 pages, 4+6 figures, 1 table | null | null | null | quant-ph cond-mat.quant-gas physics.atom-ph physics.chem-ph | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Quantum simulation offers a route to study open-system molecular dynamics in non-perturbative regimes by programming the interactions among electronic, vibrational, and environmental degrees of freedom on similar energy scales. Trapped-ion systems possess this capability, with their native spins, phonons, and tunable dissipation integrated within a single platform. Here, we demonstrate an open-system quantum simulation of charge and exciton transfer in a multi-mode linear vibronic coupling model (LVCM). Employing tailored spin-phonon interactions alongside reservoir engineering techniques, we emulate a system with two dissipative vibrational modes coupled to donor and acceptor electronic sites and follow its excitation transfer in real time. We continuously drive the system from the charge transfer (CT) regime to the vibrationally assisted exciton transfer (VAET) regime by tuning the vibronic coupling strengths. We find that degenerate modes enhance CT and VAET rates at large energy gaps, while non-degenerate modes activate slow-mode pathways that reduce the energy-gap dependence, thus enlarging the window for efficient transfer. These results show that the presence of just one additional vibration reshapes non-perturbative quantum excitation transfer. Our work also establishes a scalable route to simulating chemically relevant, many-mode vibronic processes with engineered environments, guiding the design of next-generation organic photovoltaics and molecular electronics. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:00:04 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"So",
"Visal",
""
],
[
"Suganthi",
"Midhuna Duraisamy",
""
],
[
"Zhu",
"Mingjian",
""
],
[
"Menon",
"Abhishek",
""
],
[
"Tomaras",
"George",
""
],
[
"Zhuravel",
"Roman",
""
],
[
"Pu",
"Han",
""
],
[
"Wolynes",
"Peter G.",
""
],
[
"Onuchic",
"José N.",
""
],
[
"Pagano",
"Guido",
""
]
] |
2505.22730 | Lachlan Lancaster | Lachlan Lancaster, Jeong-Gyu Kim, Greg L. Bryan, Shyam H. Menon, Eve C. Ostriker, Chang-Goo Kim | The Co-Evolution of Stellar Wind-blown Bubbles and Photoionized Gas I: Physical Principles and a Semi-Analytic Model | 33 pages, 6 figures, Accepted for Publication in ApJ. Comments welcome | null | null | null | astro-ph.GA | http://creativecommons.org/licenses/by/4.0/ | We propose a new framework for the simultaneous feedback of stellar winds and photo-ionizing radiation from massive stars, distinguishing the locations where forces are applied, and consequences for internal spatio-temporal evolution of the whole feedback bubble (FB). We quantify the relative dynamical importance of wind-blown bubbles (WBB) versus the photoionized region (PIR) by the ratio of the radius at which the WBB is in pressure equilibrium with the PIR, $R_{\rm eq}$, to the Str\"{o}mgren radius, $R_{\rm St}$. $\zeta \equiv R_{\rm eq}/R_{\rm St}$ quantifies the dynamical dominance of WBBs ($\zeta > 1$) or the PIR ($\zeta < 1$). We calculate $\zeta$ and find that, for momentum-driven winds, $0.1 \lesssim \zeta \lesssim 1$ for the star-forming regions in (i) typical Milky Way-like giant molecular clouds (GMCs), (ii) the most massive of individual OB stars, and (iii) dense, low-metallicity environments, relevant in the early universe. In this regime, both WBBs and the PIR are dynamically important to the expansion of the FB. We develop a semi-analytic Co-Evolution Model (CEM) that takes into account the spatial distribution of forces and the back reactions of both the WBB and PIR. In the $\zeta <1$ regime where the CEM is most relevant, the model differs in the total FB momentum by up to 25% compared to naive predictions. In the weak-wind limit of $\zeta \ll 1$, applicable to individual OB stars or low-mass clusters, the CEM has factors $\gtrsim 2$ differences in WBB properties. In a companion paper we compare these models to three-dimensional, turbulent hydro-dynamical simulations. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:00:05 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Lancaster",
"Lachlan",
""
],
[
"Kim",
"Jeong-Gyu",
""
],
[
"Bryan",
"Greg L.",
""
],
[
"Menon",
"Shyam H.",
""
],
[
"Ostriker",
"Eve C.",
""
],
[
"Kim",
"Chang-Goo",
""
]
] |
2505.22731 | Fernando Iemini | Andrei Tsypilnikov and Matheus Fibger and Fernando Iemini | Exact analysis of AC sensors based on Floquet time crystals | 6+12 pages, 1+1 figure | null | null | null | quant-ph cond-mat.other cond-mat.stat-mech | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We discuss the behavior of general Floquet Time Crystals (FTCs), including prethermal ones, in closed systems acting as AC sensors. We provide an analytical treatment of their quantum Fisher information (QFI) dynamics, which characterizes the ultimate sensor accuracy. By tuning the direction and frequency of the AC field, we show how to resonantly induce transitions between macroscopic paired cat states in the FTC sensor, allowing a robust Heisenberg scaling precision (QFI $\sim N^2 t^2$) for exponentially long times in the system size. The QFI dynamics exhibit, moreover, a characteristic step-like structure in time due to the eventual dephasing along the cat subspaces. The behavior is discussed for different relevant initial preparations of the sensor, including ground states, low and high correlated states. Furthermore, we examine the performance of the sensor along the FTC phase transition, with the QFI capturing its critical exponents. We present our findings for both linear and nonlinear response regimes, and illustrate them for a specific FTC based on the long-range interacting LMG model. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:00:05 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Tsypilnikov",
"Andrei",
""
],
[
"Fibger",
"Matheus",
""
],
[
"Iemini",
"Fernando",
""
]
] |
2505.22732 | Joonas Hirvonen Dr | Joonas Hirvonen and Oliver Gould | Langer's nucleation rate reproduced on the lattice | 12 pages, 6 figures | null | null | null | hep-ph astro-ph.CO cond-mat.stat-mech hep-lat hep-th | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We show that Langer's rate of bubble nucleation is quantitatively correct up to small higher-loop corrections, in comparison to lattice simulations. These results are a significant advancement on decades of lattice studies showing only qualitative trends, and the first showing agreement for any conservative system. We confirm that the failure to fully thermalize the metastable phase explains discrepancies with recent lattice studies that found disagreement with Langer's rate. The key theoretical development is the translation of Langer's perturbative definition of a thermal metastable phase into a nonperturbative statement that can be implemented on the lattice. Our statistical and systematic errors are small enough to allow us to measure on the lattice the coefficient of the two-loop contribution, missing from the perturbative prediction. Our conclusions also exclude a possible systematic uncertainty in $^3$He experiments. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:00:06 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Hirvonen",
"Joonas",
""
],
[
"Gould",
"Oliver",
""
]
] |
2505.22733 | Lachlan Lancaster | Lachlan Lancaster, Chang-Goo Kim, Jeong-Gyu Kim, Eve C. Ostriker, Greg L. Bryan | The Co-Evolution of Stellar Wind-blown Bubbles and Photoionized Gas II: 3D RMHD Simulations and Tests of Semi-Analytic Models | 33 pages, 11 Figures. Submitted to ApJ. Comments welcome | null | null | null | astro-ph.GA | http://creativecommons.org/licenses/by/4.0/ | In a companion paper (Paper I) we presented a Co-Evolution Model (CEM) in which to consider the evolution of feedback bubbles driven by massive stars through both stellar winds and ionizing radiation, outlining when either of these effects is dominant and providing a model for how they evolve together. Here we present results from three-dimensional radiation magneto-hydrodynamical (RMHD) simulations of this scenario for parameters typical of massive star-forming clouds in the Milky Way: precisely the regime where we expect both feedback mechanisms to matter. While we find that the CEM agrees with the simulations to within 25% for key parameters and modestly outperforms previous idealized models, disagreements remain. We show that these deviations originate mainly from the CEM's lack of (i) background inhomogeneity caused by turbulence and (ii) time-variable momentum enhancements in the wind-blown bubble (WBB). Additionally, we find that photoionized gas acts similarly to magnetic fields ([as in Lancaster et al. 2024a) by decreasing the WBB's surface area. This causes a decrease in the amount of cooling at the WBB's interface, resulting in an enhanced WBB dynamical impact. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:00:08 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Lancaster",
"Lachlan",
""
],
[
"Kim",
"Chang-Goo",
""
],
[
"Kim",
"Jeong-Gyu",
""
],
[
"Ostriker",
"Eve C.",
""
],
[
"Bryan",
"Greg L.",
""
]
] |
2505.22734 | Brandon Barton | Brandon Barton, Juan Carrasquilla, Christopher Roth, Agnes Valenti | Interpretable Scaling Behavior in Sparse Subnetwork Representations of Quantum States | null | null | null | null | quant-ph cond-mat.dis-nn | http://creativecommons.org/licenses/by/4.0/ | The Lottery Ticket Hypothesis (LTH) posits that within overparametrized neural networks, there exist sparse subnetworks that are capable of matching the performance of the original model when trained in isolation from the original initialization. We extend this hypothesis to the unsupervised task of approximating the ground state of quantum many-body Hamiltonians, a problem equivalent to finding a neural-network compression of the lowest-lying eigenvector of an exponentially large matrix. Focusing on two representative quantum Hamiltonians, the transverse field Ising model (TFIM) and the toric code (TC), we demonstrate that sparse neural networks can reach accuracies comparable to their dense counterparts, even when pruned by more than an order of magnitude in parameter count. Crucially, and unlike the original LTH, we find that performance depends only on the structure of the sparse subnetwork, not on the specific initialization, when trained in isolation. Moreover, we identify universal scaling behavior that persists across network sizes and physical models, where the boundaries of scaling regions are determined by the underlying Hamiltonian. At the onset of high-error scaling, we observe signatures of a sparsity-induced quantum phase transition that is first-order in shallow networks. Finally, we demonstrate that pruning enhances interpretability by linking the structure of sparse subnetworks to the underlying physics of the Hamiltonian. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:00:08 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Barton",
"Brandon",
""
],
[
"Carrasquilla",
"Juan",
""
],
[
"Roth",
"Christopher",
""
],
[
"Valenti",
"Agnes",
""
]
] |
2505.22735 | Tong Sun | Tong Sun, Bowen Jiang, Hailong Lin, Borui Li, Yixiao Teng, Yi Gao, Wei Dong | TensorShield: Safeguarding On-Device Inference by Shielding Critical DNN Tensors with TEE | null | null | null | null | cs.CR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | To safeguard user data privacy, on-device inference has emerged as a prominent paradigm on mobile and Internet of Things (IoT) devices. This paradigm involves deploying a model provided by a third party on local devices to perform inference tasks. However, it exposes the private model to two primary security threats: model stealing (MS) and membership inference attacks (MIA). To mitigate these risks, existing wisdom deploys models within Trusted Execution Environments (TEEs), which is a secure isolated execution space. Nonetheless, the constrained secure memory capacity in TEEs makes it challenging to achieve full model security with low inference latency. This paper fills the gap with TensorShield, the first efficient on-device inference work that shields partial tensors of the model while still fully defending against MS and MIA. The key enabling techniques in TensorShield include: (i) a novel eXplainable AI (XAI) technique exploits the model's attention transition to assess critical tensors and shields them in TEE to achieve secure inference, and (ii) two meticulous designs with critical feature identification and latency-aware placement to accelerate inference while maintaining security. Extensive evaluations show that TensorShield delivers almost the same security protection as shielding the entire model inside TEE, while being up to 25.35$\times$ (avg. 5.85$\times$) faster than the state-of-the-art work, without accuracy loss. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:00:24 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Sun",
"Tong",
""
],
[
"Jiang",
"Bowen",
""
],
[
"Lin",
"Hailong",
""
],
[
"Li",
"Borui",
""
],
[
"Teng",
"Yixiao",
""
],
[
"Gao",
"Yi",
""
],
[
"Dong",
"Wei",
""
]
] |
2505.22736 | Alireza Allahyari | Alireza Allahyari, Mohammadreza Davari and David F. Mota | Lorentz Violation with Gravitational Waves: Constraints from NANOGrav and IPTA Data | null | null | null | null | astro-ph.CO astro-ph.HE gr-qc | http://creativecommons.org/licenses/by-sa/4.0/ | We explore a theoretical framework in which Lorentz symmetry is explicitly broken by incorporating derivative terms of the extrinsic curvature into the gravitational action. These modifications introduce a scale-dependent damping effect in the propagation of gravitational waves (GWs), governed by a characteristic energy scale denoted as $M_{{LV}}$ . We derive the modified spectral energy density of GWs within this model and confront it with recent observational data from the NANOGrav 15-year dataset and the second data release of the International Pulsar Timing Array (IPTA). Our analysis yields a lower bound on the Lorentz-violating energy scale, finding $M_{{LV}} > 10^{-19}$ GeV at 68\% confidence level. This result significantly improves upon previous constraints derived from LIGO/VIRGO binary merger observations. Our findings demonstrate the potential of pulsar timing arrays to probe fundamental symmetries of spacetime and offer new insights into possible extensions of general relativity. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:00:35 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Allahyari",
"Alireza",
""
],
[
"Davari",
"Mohammadreza",
""
],
[
"Mota",
"David F.",
""
]
] |
2505.22737 | Takumi Matsuzawa | Takumi Matsuzawa, Minhui Zhu, Nigel Goldenfeld, William T.M. Irvine | Nonlinear Diffusion and Decay of an Expanding Turbulent Blob | 70 pages, 28 figures | null | null | null | physics.flu-dyn | http://creativecommons.org/licenses/by/4.0/ | Turbulence, left unforced, decays and invades the surrounding quiescent fluid. Though ubiquitous, this simple phenomenon has proven hard to capture within a simple and general framework. Experiments in conventional turbulent flow chambers are inevitably complicated by proximity to boundaries and mean flow, obscuring the fundamental aspects of the relaxation to the quiescent fluid state. Here, we circumvent these issues by creating a spatially-localized blob of turbulent fluid using eight converging vortex generators focused towards the center of a tank of water, and observe its decay and spread over decades in time, using particle image velocimetry with a logarithmic sampling rate. The blob initially expands and decays until it reaches the walls of the tank and eventually transitions to a second regime of approximately spatially uniform decay. We interpret these dynamics within the framework of a nonlinear diffusion equation, which predicts that the ideal quiescent-turbulent fluid boundary is sharp and propagates non-diffusively, driven by turbulent eddies while decaying with characteristic scaling laws. We find direct evidence for this model within the expansion phase of our turbulent blob and use it to account for the detailed behavior we observe, in contrast to earlier studies. Our work provides a detailed spatially-resolved narrative for the behavior of turbulence once the forcing is removed, and demonstrates unexpectedly that the turbulent cascade leaves an indelible footprint far into the decay process. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:01:35 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Matsuzawa",
"Takumi",
""
],
[
"Zhu",
"Minhui",
""
],
[
"Goldenfeld",
"Nigel",
""
],
[
"Irvine",
"William T. M.",
""
]
] |
2505.22738 | Paul L. Schechter | Paul L. Schechter and Jeremy D. Schnittman | Basic Elements of Strong Gravitational Lensing | Comments: 28 pages, 15 figures. Contribution to ISSI conference on"Strong Gravitational Lensing." Accepted 2 May 2025 Space Sci. Rev | null | 10.1007/s11214-025-01171-9 | null | astro-ph.CO | http://creativecommons.org/licenses/by/4.0/ | Even when used to describe the same phenomenon, equations, graphics and words each give different perspectives and lead to complementary insights. The basic elements of strong gravitational lensing are introduced here favoring words and graphics over equations whenever possible. Fermat's principle is the fundamental driver of strong lensing. Three "D's'' encapsulate the essential effects of lensing: Delay, Deflection and Distortion. Gravity and geometry both contribute to the delay of photons from a lensed source. Their interplay determines how the images of a source are deflected and how they are stretched or compressed. Caustics and critical curves are explained. Images of doubly, triply, quadruply and quintuply lensed sources are displayed. A table of symbols, their definitions and distinctions provides a summary of the basic elements of strong lensing. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:01:47 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Schechter",
"Paul L.",
""
],
[
"Schnittman",
"Jeremy D.",
""
]
] |
2505.22739 | Samsuzzaman Afroz | Samsuzzaman Afroz, Suvodip Mukherjee | The Phase Space of Low-Mass Binary Compact Objects from LIGO-Virgo-KAGRA Catalog: Hints on the Chances of Different Formation Scenarios | 18 Pages, 6 figures, To be submitted to ApJ | null | null | null | astro-ph.HE astro-ph.CO gr-qc | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Gravitational wave (GW) observations have significantly advanced our understanding of binary compact object (BCO) formation, yet directly linking these observations to specific formation scenarios remains challenging. The BCO phase space provides a robust and data-driven approach to discover the likely formation scenarios of these binaries. In this study, we expand the previously introduced binary black hole phase-space technique to encompass low-mass compact objects (LMCOs), establishing a novel framework to investigate their diverse formation mechanisms. Applying this approach to selected low-mass events $(\lesssim 5 M_\odot)$ from the GWTC-3 catalog and the recently observed GW230529 event, we show for the first time the phase-space demonstration of the LMCOs and find the associated probabilities for different formation scenarios including neutron stars, astrophysical black holes, or primordial black holes. Our analysis includes the astrophysical modelling uncertainties in and how it causes degeneracy between different formation scenarios. In future, with improvements in GW detector sensitivity and with detection of more GW events, the LMCO phase-space framework will significantly strengthen our capacity to associate more likely formation scenarios over the other, thereby refining our understanding of compact object formation for both astrophysical and primordial scenarios, and its evolution across the cosmic redshift. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:01:53 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Afroz",
"Samsuzzaman",
""
],
[
"Mukherjee",
"Suvodip",
""
]
] |
2505.22740 | Matthias Steinhauser | Marvin Gerlach, Ulrich Nierste, Pascal Reeck, Vladyslav Shtabovenko, Matthias Steinhauser | Current-current operator contribution to the decay matrix in $B$-meson mixing at next-to-next-to-leading order of QCD | 59 pages | null | null | P3H-25-034, SI-HEP-2025-11, TTP25-016 | hep-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We compute next-to-next-to-leading order perturbative corrections to the decay width difference of mass eigenstates and the charge-parity asymmetry $a_{\rm fs}$ in flavour-specific decays of neutral $B$ mesons. In our calculation we take into account the full dependence on the charm and bottom quark masses for the current-current operator contributions up to three-loop order. Special emphasis is put on the proper construction of the so-called $|\Delta B|=2$ theory such that Fierz symmetry is preserved. We provide updated phenomenological predictions, for $\Delta\Gamma$, $\Delta\Gamma/\Delta M$ and $a_{\rm fs}$ for the $B_d$ and $B_s$ system, including a detailed analysis of the uncertainties of our predictions. The calculated NNLO correction reduce the perturbative uncertainty of the leading term of the $1/m_b$ expansion of the width difference $\dg_s$ in the $B_s$ system to the level of the current experimental error. The uncertainty of our prediction $\dg_s=({0.077}\pm 0.016)\,\mbox{ps}^{-1}$ is dominated by the sub-leading term of this expansion. We further illustrate how better future measurements of $\dg_d$ and $a_{\rm fs}^d$ will help to gain a better understanding of $B_d$-$\bar B_d$ mixing. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:02:28 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Gerlach",
"Marvin",
""
],
[
"Nierste",
"Ulrich",
""
],
[
"Reeck",
"Pascal",
""
],
[
"Shtabovenko",
"Vladyslav",
""
],
[
"Steinhauser",
"Matthias",
""
]
] |
2505.22741 | Evan Peters | Evan Peters | Sample importance for data-driven decoding | 15 + 15 pages, 5 figures | null | null | null | quant-ph | http://creativecommons.org/licenses/by-sa/4.0/ | Data-driven decoding (DDD) -- learning to decode syndromes of (quantum) error-correcting codes using training examples -- can be a difficult problem due to several atypical and poorly understood properties of the training data. We introduce a theory of example importance that clarifies these unusual aspects of DDD: For instance, we show that DDD of a simple error-correcting code is equivalent to a noisy, imbalanced binary classification problem. This suggests that an existing data augmentation technique -- turning the knob to increase error rates in training data -- actually introduces a tradeoff between class imbalance and label noise. We apply this technique in experiments showing robust improvements to decoder accuracy while explaining the failures of this technique in terms of example importance. We show similar improvements for decoding quantum codes involving multiple rounds of syndrome measurements and we characterize example importance in random stabilizer codes, suggesting broad applicability of both example importance and turning the knob for improving experimentally relevant data-driven decoders. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:02:54 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Peters",
"Evan",
""
]
] |
2505.22742 | Debosmita Pathak | Debosmita Pathak, Charlotte R. Christensen, Alyson M. Brooks, Ferah Munshi, Anna C. Wright, Courtney Carter | Survivors and Zombies: The Quenching and Disruption of Satellites around Milky Way Analogs | 20 pages, 7 figures, 2 tables, submitted to ApJ | null | null | null | astro-ph.GA | http://creativecommons.org/licenses/by/4.0/ | It is necessary to understand the full accretion history of the Milky Way in order to contextualize the properties of observed Milky Way satellite galaxies and the stellar halo. This paper compares the dynamical properties and star-formation histories of surviving and disrupted satellites around Milky Way-like galaxies using the DC Justice League suite of very high-resolution cosmological zoom-in simulations of Milky Way analogs and their halo environments. We analyze the full census of galaxies accreted within the past 12 Gyrs, which including both surviving satellites at $z=0$, and dwarf galaxies that disrupted and merged with the host prior to $z=0$. Our simulations successfully reproduce the trends in $M_*$-[Fe/H]-[$\alpha$/Fe] observed in surviving Milky Way satellites and disrupted stellar streams, indicating earlier star-formation for disrupted progenitors. We find the likelihood and timescales for quenching and disruption are strongly correlated with the mass and time of infall. In particular, none of the galaxies accreted more than 12 Gyrs ago survived, and only 20% of all accreted galaxies with $M_*>10^8M_\odot$ survive. Additionally, satellites with highly radial trajectories are more likely to quench and disrupt. Disruption proceeds quickly for $\geq10^6M_\odot$ satellites accreted $10{-}12$ Gyr ago, often on timescales similar to the $\sim300$ Myr snapshot spacing. For high-mass satellites, the disruption timescale is faster than the quenching timescale. As a result, 92% of disrupted galaxies remain star-forming up until disruption. In contrast, Ultra Faint Dwarfs (UFDs) tend to quench prior to accretion, and 94% of UFDs accreted up to 12 Gyr ago survive at $z=0$. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:03:23 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Pathak",
"Debosmita",
""
],
[
"Christensen",
"Charlotte R.",
""
],
[
"Brooks",
"Alyson M.",
""
],
[
"Munshi",
"Ferah",
""
],
[
"Wright",
"Anna C.",
""
],
[
"Carter",
"Courtney",
""
]
] |
2505.22743 | Weiyuan Gong | Sitan Chen, Weiyuan Gong, Jonas Haferkamp, Yihui Quek | Information-Computation Gaps in Quantum Learning via Low-Degree Likelihood | 88 pages, 2 figures | null | null | null | quant-ph cs.CC cs.DS cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In a variety of physically relevant settings for learning from quantum data, designing protocols that can computationally efficiently extract information remains largely an art, and there are important cases where we believe this to be impossible, that is, where there is an information-computation gap. While there is a large array of tools in the classical literature for giving evidence for average-case hardness of statistical inference problems, the corresponding tools in the quantum literature are far more limited. One such framework in the classical literature, the low-degree method, makes predictions about hardness of inference problems based on the failure of estimators given by low-degree polynomials. In this work, we extend this framework to the quantum setting.
We establish a general connection between state designs and low-degree hardness. We use this to obtain the first information-computation gaps for learning Gibbs states of random, sparse, non-local Hamiltonians. We also use it to prove hardness for learning random shallow quantum circuit states in a challenging model where states can be measured in adaptively chosen bases. To our knowledge, the ability to model adaptivity within the low-degree framework was open even in classical settings. In addition, we also obtain a low-degree hardness result for quantum error mitigation against strategies with single-qubit measurements.
We define a new quantum generalization of the planted biclique problem and identify the threshold at which this problem becomes computationally hard for protocols that perform local measurements. Interestingly, the complexity landscape for this problem shifts when going from local measurements to more entangled single-copy measurements.
We show average-case hardness for the "standard" variant of Learning Stabilizers with Noise and for agnostically learning product states. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:04:10 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Chen",
"Sitan",
""
],
[
"Gong",
"Weiyuan",
""
],
[
"Haferkamp",
"Jonas",
""
],
[
"Quek",
"Yihui",
""
]
] |
2505.22744 | Olga Smirnova | Aycke Roos, Pablo M. Maier, Andres F. Ordonez and Olga Smirnova | Geometry of chiral temporal structures II: The formalism | 18 pages, 1 figure | null | null | null | quant-ph | http://creativecommons.org/licenses/by-nc-nd/4.0/ | We develop a mathematical formalism underlying the emergence of enantio-sensitive molecular orientation due to photoionization or photoexitation of chiral molecules. We consider geometric quantities such as the Berry connection and Berry curvature in light-driven chiral electronic states in the space of complex light polarization vectors. The parametric dependence of the light-driven electronic wavefunction on such vectors emerges due to various possible mutual orientations between the laser field and a chiral molecule. Using the tools of differential geometry we show how the enantio-sensitive observables emerge from the geometry of the molecular response in such spaces. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:04:14 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Roos",
"Aycke",
""
],
[
"Maier",
"Pablo M.",
""
],
[
"Ordonez",
"Andres F.",
""
],
[
"Smirnova",
"Olga",
""
]
] |
2505.22745 | Mohit Raj Sah Mr. | Mohit Raj Sah, Akash Maurya, Suvodip Mukherjee, Prayush Kumar, Vida Saeedzadeh, Arif Babul, Chandra Kant Mishra, Kaushik Paul, Thomas R. Quinn, and Michael Tremmel | An Accurate Modeling of Nano-Hertz Gravitational Wave Signal from Eccentric Supermassive Binary Black Holes: Essential Step Towards a Robust Discovery | 21 pages, 9 figures | null | null | null | astro-ph.CO astro-ph.HE gr-qc | http://creativecommons.org/licenses/by-nc-nd/4.0/ | The stochastic gravitational wave background (SGWB) in the nanohertz (nHz) regime, detectable by pulsar timing arrays (PTAs), offers a promising avenue to probe the cosmic population of supermassive black hole binaries (SMBHBs). These SMBHBs are expected to retain substantial eccentricity throughout their evolution due to their formation history. In this study, we propose a new modeling scenario of the nHz SGWB by incorporating the eccentricity of SMBHBs into a multi-scale adaptive simulation-based framework. We employ a time-domain eccentric waveform model, \esigmahm{}, to generate realistic gravitational wave (GW) signals from an astrophysical population of SMBHB, including physical effects from sub-parsec scales to Gpc scales. The eccentric inspiraling binary, unlike circular binaries, emits GW signal in multiple frequencies. As a consequence, the SGWB energy density in each frequency bin is not independent; instead, the presence of eccentricity introduces a spectral correlation between different frequencies. We show that these spectral correlations are absent for circular binaries but become increasingly significant for populations with higher eccentricities. Our novel approach can capture this effect and opens up the window towards measuring this with a high signal-to-noise ratio with future observations. This work develops the frontier of nHz signal modeling using eccentricity at small scales and can model realistic nHz signal, which will be essential for robust inference from future observations to shed light on the astrophysical properties of SMBHBs. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:04:48 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Sah",
"Mohit Raj",
""
],
[
"Maurya",
"Akash",
""
],
[
"Mukherjee",
"Suvodip",
""
],
[
"Kumar",
"Prayush",
""
],
[
"Saeedzadeh",
"Vida",
""
],
[
"Babul",
"Arif",
""
],
[
"Mishra",
"Chandra Kant",
""
],
[
"Paul",
"Kaushik",
""
],
[
"Quinn",
"Thomas R.",
""
],
[
"Tremmel",
"Michael",
""
]
] |
2505.22746 | Jose Hernandez | Jose Guadalupe Hernandez, Attri Ghosh, Philip J. Freda, Yufei Meng, Nicholas Matsumoto, Jason H. Moore | StarBASE-GP: Biologically-Guided Automated Machine Learning for Genotype-to-Phenotype Association Analysis | null | null | null | null | cs.NE cs.LG | http://creativecommons.org/licenses/by/4.0/ | We present the Star-Based Automated Single-locus and Epistasis analysis tool - Genetic Programming (StarBASE-GP), an automated framework for discovering meaningful genetic variants associated with phenotypic variation in large-scale genomic datasets. StarBASE-GP uses a genetic programming-based multi-objective optimization strategy to evolve machine learning pipelines that simultaneously maximize explanatory power (r2) and minimize pipeline complexity. Biological domain knowledge is integrated at multiple stages, including the use of nine inheritance encoding strategies to model deviations from additivity, a custom linkage disequilibrium pruning node that minimizes redundancy among features, and a dynamic variant recommendation system that prioritizes informative candidates for pipeline inclusion. We evaluate StarBASE-GP on a cohort of Rattus norvegicus (brown rat) to identify variants associated with body mass index, benchmarking its performance against a random baseline and a biologically naive version of the tool. StarBASE-GP consistently evolves Pareto fronts with superior performance, yielding higher accuracy in identifying both ground truth and novel quantitative trait loci, highlighting relevant targets for future validation. By incorporating evolutionary search and relevant biological theory into a flexible automated machine learning framework, StarBASE-GP demonstrates robust potential for advancing variant discovery in complex traits. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:05:15 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Hernandez",
"Jose Guadalupe",
""
],
[
"Ghosh",
"Attri",
""
],
[
"Freda",
"Philip J.",
""
],
[
"Meng",
"Yufei",
""
],
[
"Matsumoto",
"Nicholas",
""
],
[
"Moore",
"Jason H.",
""
]
] |
2505.22747 | Shweta Sharma | Shweta Sharma, Abhishek Sharma, Abhinendra Singh | Frictional Contact Network in Dense Suspension Flow | 19 pages, 12 figures | null | null | null | cond-mat.soft | http://creativecommons.org/licenses/by/4.0/ | Dense particulate suspensions often exhibit a dramatic increase in viscosity in response to external deformation. This shear thickening behavior has been related to a transition from lubricated, unconstrained pairwise motion to a frictional contact network (FCN) at high stresses. Here, we study the characteristics of the FCN formed during shear thickening to investigate the role of constraints, emphasizing the impact of resistance to gear-like rolling. We contrast the FCN formed by sliding friction alone with that formed by particles with sliding and rolling constraints. Particles with sliding constraints only form a highly interconnected network with primary force chains in the compressive direction, which requires orthogonal support from other force chains. However, orthogonal support is not required for mechanical stability when particles have both sliding and rolling constraints. In addition, the force chains appear linear and longer, reducing the jamming volume fraction for rough/faceted particles. Finally, we propose a novel mechanical stability picture for rough/faceted particles with sliding and rolling constraints, which is crucial for understanding the flow behavior of real-life suspensions. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:05:41 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Sharma",
"Shweta",
""
],
[
"Sharma",
"Abhishek",
""
],
[
"Singh",
"Abhinendra",
""
]
] |
2505.22748 | Bingqing Hu | Bingqing Hu and Bin Nan | Nonparametric Estimation of Conditional Survival Function with Time-Varying Covariates Using DeepONet | null | null | null | null | stat.ME | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Traditional survival models often rely on restrictive assumptions such as proportional hazards or instantaneous effects of time-varying covariates on the hazard function, which limit their applicability in real-world settings. We consider the nonparametric estimation of the conditional survival function, which leverages the flexibility of neural networks to capture the complex, potentially long-term non-instantaneous effects of time-varying covariates. In this work, we use Deep Operator Networks (DeepONet), a deep learning architecture designed for operator learning, to model the arbitrary effects of both time-varying and time-invariant covariates. Specifically, our method relaxes commonly used assumptions in hazard regressions by modeling the conditional hazard function as an unknown nonlinear operator of entire histories of time-varying covariates. The estimation is based on a loss function constructed from the nonparametric full likelihood for censored survival data. Simulation studies demonstrate that our method performs well, whereas the Cox model yields biased results when the assumption of instantaneous time-varying covariate effects is violated. We further illustrate its utility with the ADNI data, for which it yields a lower integrated Brier score than the Cox model. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:05:49 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Hu",
"Bingqing",
""
],
[
"Nan",
"Bin",
""
]
] |
2505.22749 | Tamas Spisak | Tamas Spisak and Karl Friston | Self-orthogonalizing attractor neural networks emerging from the free energy principle | 22 pages main text, 5 pages appendix, 6 figures; interactive manuscript available at: https://pni-lab.github.io/fep-attractor-network Associated GitHub repository: https://github.com/pni-lab/fep-attractor-network | null | null | null | q-bio.NC cs.AI cs.LG cs.NE | http://creativecommons.org/licenses/by/4.0/ | Attractor dynamics are a hallmark of many complex systems, including the brain. Understanding how such self-organizing dynamics emerge from first principles is crucial for advancing our understanding of neuronal computations and the design of artificial intelligence systems. Here we formalize how attractor networks emerge from the free energy principle applied to a universal partitioning of random dynamical systems. Our approach obviates the need for explicitly imposed learning and inference rules and identifies emergent, but efficient and biologically plausible inference and learning dynamics for such self-organizing systems. These result in a collective, multi-level Bayesian active inference process. Attractors on the free energy landscape encode prior beliefs; inference integrates sensory data into posterior beliefs; and learning fine-tunes couplings to minimize long-term surprise. Analytically and via simulations, we establish that the proposed networks favor approximately orthogonalized attractor representations, a consequence of simultaneously optimizing predictive accuracy and model complexity. These attractors efficiently span the input subspace, enhancing generalization and the mutual information between hidden causes and observable effects. Furthermore, while random data presentation leads to symmetric and sparse couplings, sequential data fosters asymmetric couplings and non-equilibrium steady-state dynamics, offering a natural extension to conventional Boltzmann Machines. Our findings offer a unifying theory of self-organizing attractor networks, providing novel insights for AI and neuroscience. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:10:03 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Spisak",
"Tamas",
""
],
[
"Friston",
"Karl",
""
]
] |
2505.22750 | Mariano Mateos | Eduardo Casas and Mariano Mateos | Quadratic convergence of an SQP method for some optimization problems with applications to control theory | null | null | null | null | math.OC math.AP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We analyze a sequential quadratic programming algorithm for solving a class of abstract optimization problems. Assuming that the initial point is in an $L^2$ neighborhood of a local solution that satisfies no-gap second-order sufficient optimality conditions and a strict complementarity condition, we obtain stability and quadratic convergence in $L^q$ for all $q\in[p,\infty]$ where $p\geq 2$ depends on the problem. Many of the usual optimal control problems of partial differential equations fit into this abstract formulation. Some examples are given in the paper. Finally, a computational comparison with other versions of the SQP method is presented. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:10:11 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Casas",
"Eduardo",
""
],
[
"Mateos",
"Mariano",
""
]
] |
2505.22751 | Anderson Kohara | Anderson Kendi Kohara | A complex logistic equation for universal energy evolution in hadronic elastic scattering | 35 pages, 8 figures 1 table | null | null | null | hep-ph hep-th | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We introduce a universal evolution equation for elastic scattering of hadrons, derived from Regge field theory (RFT) and solved in closed analytical form. The equation emerges from a complex logistic structure and evolves initial amplitude profiles taken from existing models at fixed energy, reproducing both the differential cross sections and the integrated quantities in a broad energy range. We prove that it admits a unique solution for each initial condition and rigorously satisfies unitarity, the Froissart-Martin bound, and dispersion relations. The dynamics are governed by two physically meaningful parameters: the effective Pomeron mass $\epsilon_{\mathcal{P}}$ and the nonlinear coupling $\lambda$, both fitted at a single energy. Our approach offers a minimal, yet predictive framework for saturation and unitarization in elastic scattering and may provide a useful bridge toward small-$x$ QCD evolution. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:11:57 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Kohara",
"Anderson Kendi",
""
]
] |
2505.22752 | Peter Tankov | Rafik Mankour, Yassine Chafai, Hamada Saleh, Ghassen Ben Hassine, Thibaud Barreau and Peter Tankov | Climate Finance Bench | Dataset is available at https://github.com/Pladifes/climate_finance_bench | null | null | null | cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Climate Finance Bench introduces an open benchmark that targets question-answering over corporate climate disclosures using Large Language Models. We curate 33 recent sustainability reports in English drawn from companies across all 11 GICS sectors and annotate 330 expert-validated question-answer pairs that span pure extraction, numerical reasoning, and logical reasoning. Building on this dataset, we propose a comparison of RAG (retrieval-augmented generation) approaches. We show that the retriever's ability to locate passages that actually contain the answer is the chief performance bottleneck. We further argue for transparent carbon reporting in AI-for-climate applications, highlighting advantages of techniques such as Weight Quantization. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:12:29 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Mankour",
"Rafik",
""
],
[
"Chafai",
"Yassine",
""
],
[
"Saleh",
"Hamada",
""
],
[
"Hassine",
"Ghassen Ben",
""
],
[
"Barreau",
"Thibaud",
""
],
[
"Tankov",
"Peter",
""
]
] |
2505.22753 | Arseniy Pertzovsky | Arseniy Pertzovsky, Roni Stern, Ariel Felner, Roie Zivan | Enhancing Lifelong Multi-Agent Path-finding by Using Artificial Potential Fields | null | null | null | null | cs.AI cs.MA cs.RO | http://creativecommons.org/licenses/by/4.0/ | We explore the use of Artificial Potential Fields (APFs) to solve Multi-Agent Path Finding (MAPF) and Lifelong MAPF (LMAPF) problems. In MAPF, a team of agents must move to their goal locations without collisions, whereas in LMAPF, new goals are generated upon arrival. We propose methods for incorporating APFs in a range of MAPF algorithms, including Prioritized Planning, MAPF-LNS2, and Priority Inheritance with Backtracking (PIBT). Experimental results show that using APF is not beneficial for MAPF but yields up to a 7-fold increase in overall system throughput for LMAPF. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:13:10 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Pertzovsky",
"Arseniy",
""
],
[
"Stern",
"Roni",
""
],
[
"Felner",
"Ariel",
""
],
[
"Zivan",
"Roie",
""
]
] |
2505.22754 | Aparajitha Karthikeyan | P. S. Bhupal Dev, Bhaskar Dutta, Aparajitha Karthikeyan, Writasree Maitra, Louis E. Strigari, and Ankur Verma | `Dark' Matter Effect as a Novel Solution to the KM3-230213A Puzzle | 5 pages, 4 figures | null | null | MI-HET-859 | hep-ph | http://creativecommons.org/licenses/by/4.0/ | The recent KM3NeT observation of the ${\cal O}(100~{\rm PeV})$ event KM3-230213A is puzzling because IceCube with much larger effective area times exposure has not found any such events. We propose a novel solution to this conundrum in terms of dark matter (DM) scattering in the Earth's crust. We show that intermediate dark-sector particles that decay into muons are copiously produced when high-energy ($\sim100~\text{PeV}$) DM propagates through a sufficient amount of Earth overburden. The same interactions responsible for DM scattering in Earth also source the boosted DM flux from a high-luminosity blazar. We address the non-observation of similar events at IceCube via two examples of weakly coupled long-lived dark sector scenarios that satisfy all the lab-based constraints. We calculate the corresponding dark sector cross sections, lifetimes and blazar luminosities required to yield one event at KM3NeT, and also predict the number of IceCube events for these parameters that can be tested very soon. Our proposed DM explanation of the event can also be distinguished from a neutrino-induced event in future high-energy neutrino flavor analyses, large-scale DM direct detection experiments, as well as at future colliders. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:13:58 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Dev",
"P. S. Bhupal",
""
],
[
"Dutta",
"Bhaskar",
""
],
[
"Karthikeyan",
"Aparajitha",
""
],
[
"Maitra",
"Writasree",
""
],
[
"Strigari",
"Louis E.",
""
],
[
"Verma",
"Ankur",
""
]
] |
2505.22755 | Zlil Sela | Z. Sela | A canonical Makanin-Razborov diagram and a pseudo topology for sets of tuples in free groups, semigroups, associative algebras and Lie algebras I | null | null | null | null | math.GR math.LO math.RA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The JSJ decomposition and the Makanin-Razborov diagram were proved to be essential in studying varieties over free groups, semigroups and associative algebras. In this paper we suggest a unified conceptual approach to the applicability of these structures over all these algebraic categories. With a variety over each of these algebraic categories we naturally associate a set of tuples in a free group. Then we show how to associate a Makanin-Razborov diagram with any set of tuples over a free group. Furthermore, in case the MR diagram that is associated with a set of tuples is single ended, we prove that there is a canonical Makanin-Razborov diagram that can be associated with such a set. This canonical diagram is a main key in studying varieties over free semigroups, associative algebras and Lie algebras, and encodes the global structure of these varieties. It enables us to define a (pseudo) closure of a set of tuples over each of the algebraic objects, associate a rank with it (analogous to Shelah and Lascar ranks), and over free groups the closure provides a canonical envelope that is essential in studying the structure and the properties of definable sets. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:15:25 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Sela",
"Z.",
""
]
] |
2505.22756 | Core Francisco Park | Tian Qin, Core Francisco Park, Mujin Kwun, Aaron Walsman, Eran Malach, Nikhil Anand, Hidenori Tanaka, David Alvarez-Melis | Decomposing Elements of Problem Solving: What "Math" Does RL Teach? | null | null | null | null | cs.AI cs.CL cs.LG | http://creativecommons.org/publicdomain/zero/1.0/ | Mathematical reasoning tasks have become prominent benchmarks for assessing the reasoning capabilities of LLMs, especially with reinforcement learning (RL) methods such as GRPO showing significant performance gains. However, accuracy metrics alone do not support fine-grained assessment of capabilities and fail to reveal which problem-solving skills have been internalized. To better understand these capabilities, we propose to decompose problem solving into fundamental capabilities: Plan (mapping questions to sequences of steps), Execute (correctly performing solution steps), and Verify (identifying the correctness of a solution). Empirically, we find that GRPO mainly enhances the execution skill-improving execution robustness on problems the model already knows how to solve-a phenomenon we call temperature distillation. More importantly, we show that RL-trained models struggle with fundamentally new problems, hitting a 'coverage wall' due to insufficient planning skills. To explore RL's impact more deeply, we construct a minimal, synthetic solution-tree navigation task as an analogy for mathematical problem-solving. This controlled setup replicates our empirical findings, confirming RL primarily boosts execution robustness. Importantly, in this setting, we identify conditions under which RL can potentially overcome the coverage wall through improved exploration and generalization to new solution paths. Our findings provide insights into the role of RL in enhancing LLM reasoning, expose key limitations, and suggest a path toward overcoming these barriers. Code is available at https://github.com/cfpark00/RL-Wall. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:18:49 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Qin",
"Tian",
""
],
[
"Park",
"Core Francisco",
""
],
[
"Kwun",
"Mujin",
""
],
[
"Walsman",
"Aaron",
""
],
[
"Malach",
"Eran",
""
],
[
"Anand",
"Nikhil",
""
],
[
"Tanaka",
"Hidenori",
""
],
[
"Alvarez-Melis",
"David",
""
]
] |
2505.22757 | Ansar Aynetdinov | Ansar Aynetdinov, Alan Akbik | Pre-Training Curriculum for Multi-Token Prediction in Language Models | Accepted to ACL 2025 (Main) | null | null | null | cs.CL cs.AI | http://creativecommons.org/licenses/by/4.0/ | Multi-token prediction (MTP) is a recently proposed pre-training objective for language models. Rather than predicting only the next token (NTP), MTP predicts the next $k$ tokens at each prediction step, using multiple prediction heads. MTP has shown promise in improving downstream performance, inference speed, and training efficiency, particularly for large models. However, prior work has shown that smaller language models (SLMs) struggle with the MTP objective. To address this, we propose a curriculum learning strategy for MTP training, exploring two variants: a forward curriculum, which gradually increases the complexity of the pre-training objective from NTP to MTP, and a reverse curriculum, which does the opposite. Our experiments show that the forward curriculum enables SLMs to better leverage the MTP objective during pre-training, improving downstream NTP performance and generative output quality, while retaining the benefits of self-speculative decoding. The reverse curriculum achieves stronger NTP performance and output quality, but fails to provide any self-speculative decoding benefits. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:19:18 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Aynetdinov",
"Ansar",
""
],
[
"Akbik",
"Alan",
""
]
] |
2505.22758 | Aniruddha Nrusimha | Aniruddha Nrusimha, William Brandon, Mayank Mishra, Yikang Shen, Rameswar Panda, Jonathan Ragan-Kelley, and Yoon Kim | FlashFormer: Whole-Model Kernels for Efficient Low-Batch Inference | null | null | null | null | cs.LG cs.CL | http://creativecommons.org/licenses/by/4.0/ | The size and compute characteristics of modern large language models have led to an increased interest in developing specialized kernels tailored for training and inference. Existing kernels primarily optimize for compute utilization, targeting the large-batch training and inference settings. However, low-batch inference, where memory bandwidth and kernel launch overheads contribute are significant factors, remains important for many applications of interest such as in edge deployment and latency-sensitive applications. This paper describes FlashFormer, a proof-of-concept kernel for accelerating single-batch inference for transformer-based large language models. Across various model sizes and quantizations settings, we observe nontrivial speedups compared to existing state-of-the-art inference kernels. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:19:30 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Nrusimha",
"Aniruddha",
""
],
[
"Brandon",
"William",
""
],
[
"Mishra",
"Mayank",
""
],
[
"Shen",
"Yikang",
""
],
[
"Panda",
"Rameswar",
""
],
[
"Ragan-Kelley",
"Jonathan",
""
],
[
"Kim",
"Yoon",
""
]
] |
2505.22759 | Sara Papi | Sara Papi, Marco Gaido, Luisa Bentivogli, Alessio Brutti, Mauro Cettolo, Roberto Gretter, Marco Matassoni, Mohamed Nabih, Matteo Negri | FAMA: The First Large-Scale Open-Science Speech Foundation Model for English and Italian | null | null | null | null | cs.CL cs.AI cs.SD | http://creativecommons.org/licenses/by/4.0/ | The development of speech foundation models (SFMs) like Whisper and SeamlessM4T has significantly advanced the field of speech processing. However, their closed nature--with inaccessible training data and code--poses major reproducibility and fair evaluation challenges. While other domains have made substantial progress toward open science by developing fully transparent models trained on open-source (OS) code and data, similar efforts in speech remain limited. To fill this gap, we introduce FAMA, the first family of open science SFMs for English and Italian, trained on 150k+ hours of OS speech data. Moreover, we present a new dataset containing 16k hours of cleaned and pseudo-labeled speech for both languages. Results show that FAMA achieves competitive performance compared to existing SFMs while being up to 8 times faster. All artifacts, including code, datasets, and models, are released under OS-compliant licenses, promoting openness in speech technology research. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:19:34 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Papi",
"Sara",
""
],
[
"Gaido",
"Marco",
""
],
[
"Bentivogli",
"Luisa",
""
],
[
"Brutti",
"Alessio",
""
],
[
"Cettolo",
"Mauro",
""
],
[
"Gretter",
"Roberto",
""
],
[
"Matassoni",
"Marco",
""
],
[
"Nabih",
"Mohamed",
""
],
[
"Negri",
"Matteo",
""
]
] |
2505.22760 | Razvan-Andrei Lascu | Razvan-Andrei Lascu, Mateusz B. Majka | Non-convex entropic mean-field optimization via Best Response flow | 40 pages | null | null | null | math.OC cs.LG math.PR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We study the problem of minimizing non-convex functionals on the space of probability measures, regularized by the relative entropy (KL divergence) with respect to a fixed reference measure, as well as the corresponding problem of solving entropy-regularized non-convex-non-concave min-max problems. We utilize the Best Response flow (also known in the literature as the fictitious play flow) and study how its convergence is influenced by the relation between the degree of non-convexity of the functional under consideration, the regularization parameter and the tail behaviour of the reference measure. In particular, we demonstrate how to choose the regularizer, given the non-convex functional, so that the Best Response operator becomes a contraction with respect to the $L^1$-Wasserstein distance, which then ensures the existence of its unique fixed point, which is then shown to be the unique global minimizer for our optimization problem. This extends recent results where the Best Response flow was applied to solve convex optimization problems regularized by the relative entropy with respect to arbitrary reference measures, and with arbitrary values of the regularization parameter. Our results explain precisely how the assumption of convexity can be relaxed, at the expense of making a specific choice of the regularizer. Additionally, we demonstrate how these results can be applied in reinforcement learning in the context of policy optimization for Markov Decision Processes and Markov games with softmax parametrized policies in the mean-field regime. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:22:08 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Lascu",
"Razvan-Andrei",
""
],
[
"Majka",
"Mateusz B.",
""
]
] |
2505.22761 | Afila Ajithkumar Sophiya | Afila Ajithkumar Sophiya, Akarsh K Nair, Sepehr Maleki, Senthil K. Krishnababu | A comprehensive analysis of PINNs: Variants, Applications, and Challenges | null | null | null | null | cs.CE cs.AI | http://creativecommons.org/licenses/by/4.0/ | Physics Informed Neural Networks (PINNs) have been emerging as a powerful computational tool for solving differential equations. However, the applicability of these models is still in its initial stages and requires more standardization to gain wider popularity. Through this survey, we present a comprehensive overview of PINNs approaches exploring various aspects related to their architecture, variants, areas of application, real-world use cases, challenges, and so on. Even though existing surveys can be identified, they fail to provide a comprehensive view as they primarily focus on either different application scenarios or limit their study to a superficial level. This survey attempts to bridge the gap in the existing literature by presenting a detailed analysis of all these factors combined with recent advancements and state-of-the-art research in PINNs. Additionally, we discuss prevalent challenges in PINNs implementation and present some of the future research directions as well. The overall contributions of the survey can be summarised into three sections: A detailed overview of PINNs architecture and variants, a performance analysis of PINNs on different equations and application domains highlighting their features. Finally, we present a detailed discussion of current issues and future research directions. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:25:17 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Sophiya",
"Afila Ajithkumar",
""
],
[
"Nair",
"Akarsh K",
""
],
[
"Maleki",
"Sepehr",
""
],
[
"Krishnababu",
"Senthil K.",
""
]
] |
2505.22762 | Marco Colussi | Marco Colussi, Dragan Ahmetovic, Sergio Mascetti | MIAS-SAM: Medical Image Anomaly Segmentation without thresholding | null | null | null | null | cs.CV cs.AI cs.LG | http://creativecommons.org/licenses/by-nc-sa/4.0/ | This paper presents MIAS-SAM, a novel approach for the segmentation of anomalous regions in medical images. MIAS-SAM uses a patch-based memory bank to store relevant image features, which are extracted from normal data using the SAM encoder. At inference time, the embedding patches extracted from the SAM encoder are compared with those in the memory bank to obtain the anomaly map. Finally, MIAS-SAM computes the center of gravity of the anomaly map to prompt the SAM decoder, obtaining an accurate segmentation from the previously extracted features. Differently from prior works, MIAS-SAM does not require to define a threshold value to obtain the segmentation from the anomaly map. Experimental results conducted on three publicly available datasets, each with a different imaging modality (Brain MRI, Liver CT, and Retina OCT) show accurate anomaly segmentation capabilities measured using DICE score. The code is available at: https://github.com/warpcut/MIAS-SAM | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:25:37 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Colussi",
"Marco",
""
],
[
"Ahmetovic",
"Dragan",
""
],
[
"Mascetti",
"Sergio",
""
]
] |
2505.22763 | Daniel Jim\'enez-Aguilar | Mark P. Hertzberg, Daniel Jim\'enez-Aguilar | On the Probability of the Initial Conditions for Inflation and Slow Contraction | 30 pages, 5 figures | null | null | null | gr-qc astro-ph.CO hep-ph hep-th | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Some recent studies based on numerical relativity simulations claim that slow contraction/ekpyrosis is strongly preferred over inflation as the smoothing mechanism that brought the universe into the homogeneous, isotropic and flat state we observe today on large scales. In this paper, we evaluate the likelihood of the initial conditions employed in the aforementioned simulations by estimating the probability that a free scalar field dominating the universe at the beginning of inflation or ekpyrosis will be sufficiently homogeneous on scales comparable to the Hubble radius at that time. We explore the space of parameters that characterize the initial power spectrum of the scalar field, finding that either can be more likely than the other for a fixed choice of parameters. On the other hand, when we extremize over these parameters, we find that the maximal probability for inflation is much higher than that of ekpyrosis. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:27:39 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Hertzberg",
"Mark P.",
""
],
[
"Jiménez-Aguilar",
"Daniel",
""
]
] |
2505.22764 | Divya Shanmugam | Divya Shanmugam, Helen Lu, Swami Sankaranarayanan, John Guttag | Test-time augmentation improves efficiency in conformal prediction | null | null | null | null | cs.LG cs.CV | http://creativecommons.org/licenses/by/4.0/ | A conformal classifier produces a set of predicted classes and provides a probabilistic guarantee that the set includes the true class. Unfortunately, it is often the case that conformal classifiers produce uninformatively large sets. In this work, we show that test-time augmentation (TTA)--a technique that introduces inductive biases during inference--reduces the size of the sets produced by conformal classifiers. Our approach is flexible, computationally efficient, and effective. It can be combined with any conformal score, requires no model retraining, and reduces prediction set sizes by 10%-14% on average. We conduct an evaluation of the approach spanning three datasets, three models, two established conformal scoring methods, different guarantee strengths, and several distribution shifts to show when and why test-time augmentation is a useful addition to the conformal pipeline. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:29:31 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Shanmugam",
"Divya",
""
],
[
"Lu",
"Helen",
""
],
[
"Sankaranarayanan",
"Swami",
""
],
[
"Guttag",
"John",
""
]
] |
2505.22765 | Yossi Adi | Iddo Yosha, Gallil Maimon, Yossi Adi | StressTest: Can YOUR Speech LM Handle the Stress? | null | null | null | null | cs.CL cs.SD eess.AS | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Sentence stress refers to emphasis, placed on specific words within a spoken utterance to highlight or contrast an idea, or to introduce new information. It is often used to imply an underlying intention that is not explicitly stated. Recent advances in speech-aware language models (SLMs) have enabled direct processing of audio, allowing models to bypass transcription and access the full richness of the speech signal and perform audio reasoning tasks such as spoken question answering. Despite the crucial role of sentence stress in shaping meaning and speaker intent, it remains largely overlooked in evaluation and development of such models. In this work, we address this gap by introducing StressTest, a benchmark specifically designed to evaluate a model's ability to distinguish between interpretations of spoken sentences based on the stress pattern. We assess the performance of several leading SLMs and find that, despite their overall capabilities, they perform poorly on such tasks. To overcome this limitation, we propose a novel synthetic data generation pipeline, and create Stress17k, a training set that simulates change of meaning implied by stress variation. Then, we empirically show that optimizing models with this synthetic dataset aligns well with real-world recordings and enables effective finetuning of SLMs. Results suggest, that our finetuned model, StresSLM, significantly outperforms existing models on both sentence stress reasoning and detection tasks. Code, models, data, and audio samples - pages.cs.huji.ac.il/adiyoss-lab/stresstest. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:32:56 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Yosha",
"Iddo",
""
],
[
"Maimon",
"Gallil",
""
],
[
"Adi",
"Yossi",
""
]
] |
2505.22766 | Paula Fialho | Paula M. S. Fialho, Aldo Procacci | On the independent set polynomial of graphs and claw-free graphs | null | null | null | null | math.CO math-ph math.MP | http://creativecommons.org/licenses/by/4.0/ | We present two new contributions to the study of the independence polynomial $Z_\GI(z)$ of a finite simple graph $\GI = (\VU,\EE)$. First, we establish a novel combinatorial expression for $Z_\GI(z)$, inspired by the connection with the abstract polymer gas models in statistical mechanics, which offers a new structural interpretation of the polynomial and may be of independent interest. Second, we provide an improved lower bound for the zero-free region of $Z_\GI(z)$ in the complex plane for the important class of claw-free graphs. Our bound exceeds the classical Shearer radius and is derived through a refined application of the Fern\'andez-Procacci criterion using properties of the local neighborhood structure in claw-free graphs. These results strengthen the connection between statistical physics, combinatorics, and graph theory, and suggest new approaches for analytic exploration | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:34:34 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Fialho",
"Paula M. S.",
""
],
[
"Procacci",
"Aldo",
""
]
] |
2505.22767 | Eleni Vasilaki D.Phil. | Eleni Vasilaki | In Dialogue with Intelligence: Rethinking Large Language Models as Collective Knowledge | 6 pages, 1 table | null | null | null | cs.HC cs.AI | http://creativecommons.org/licenses/by/4.0/ | Large Language Models (LLMs) are typically analysed through architectural, behavioural, or training-data lenses. This article offers a theoretical and experiential re-framing: LLMs as dynamic instantiations of Collective human Knowledge (CK), where intelligence is evoked through dialogue rather than stored statically. Drawing on concepts from neuroscience and AI, and grounded in sustained interaction with ChatGPT-4, I examine emergent dialogue patterns, the implications of fine-tuning, and the notion of co-augmentation: mutual enhancement between human and machine cognition. This perspective offers a new lens for understanding interaction, representation, and agency in contemporary AI systems. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:36:00 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Vasilaki",
"Eleni",
""
]
] |
2505.22768 | Mert Cakiroglu | Mert Onur Cakiroglu, Idil Bilge Altun, Hasan Kurban, Elham Buxton and Mehmet Dalkilic | Multivariate de Bruijn Graphs: A Symbolic Graph Framework for Time Series Forecasting | null | null | null | null | cs.LG | http://creativecommons.org/licenses/by/4.0/ | Time series forecasting remains a challenging task for foundation models due to temporal heterogeneity, high dimensionality, and the lack of inherent symbolic structure. In this work, we propose DRAGON (Discrete Representation and Augmented Graph encoding Over deBruijN Graphs), a novel encoder that introduces Multivariate de Bruijn Graphs (MdBGs) to bridge the gap between symbolic representations and neural modeling. DRAGON discretizes continuous input sequences and maps them onto a fixed graph structure, enabling dynamic context recovery via graph-based attention. Integrated as an auxiliary module within a dual-branch architecture, DRAGON augments conventional CNN-based encoders with symbolic, structure-aware representations. All code developed for this study is available at: https://github.com/KurbanIntelligenceLab/MultdBG-Time-Series-Library | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:36:26 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Cakiroglu",
"Mert Onur",
""
],
[
"Altun",
"Idil Bilge",
""
],
[
"Kurban",
"Hasan",
""
],
[
"Buxton",
"Elham",
""
],
[
"Dalkilic",
"Mehmet",
""
]
] |
2505.22769 | Yaxiong Lei | Yaxiong Lei, Mingyue Zhao, Yuheng Wang, Shijing He, Yusuke Sugano, Yafei Wang, Kaixing Zhao, Mohamed Khamis, and Juan Ye | MAC-Gaze: Motion-Aware Continual Calibration for Mobile Gaze Tracking | 24 pages, 7 figures | null | null | null | cs.HC cs.CV | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Mobile gaze tracking faces a fundamental challenge: maintaining accuracy as users naturally change their postures and device orientations. Traditional calibration approaches, like one-off, fail to adapt to these dynamic conditions, leading to degraded performance over time. We present MAC-Gaze, a Motion-Aware continual Calibration approach that leverages smartphone Inertial measurement unit (IMU) sensors and continual learning techniques to automatically detect changes in user motion states and update the gaze tracking model accordingly. Our system integrates a pre-trained visual gaze estimator and an IMU-based activity recognition model with a clustering-based hybrid decision-making mechanism that triggers recalibration when motion patterns deviate significantly from previously encountered states. To enable accumulative learning of new motion conditions while mitigating catastrophic forgetting, we employ replay-based continual learning, allowing the model to maintain performance across previously encountered motion conditions. We evaluate our system through extensive experiments on the publicly available RGBDGaze dataset and our own 10-hour multimodal MotionGaze dataset (481K+ images, 800K+ IMU readings), encompassing a wide range of postures under various motion conditions including sitting, standing, lying, and walking. Results demonstrate that our method reduces gaze estimation error by 19.9% on RGBDGaze (from 1.73 cm to 1.41 cm) and by 31.7% on MotionGaze (from 2.81 cm to 1.92 cm) compared to traditional calibration approaches. Our framework provides a robust solution for maintaining gaze estimation accuracy in mobile scenarios. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:38:04 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Lei",
"Yaxiong",
""
],
[
"Zhao",
"Mingyue",
""
],
[
"Wang",
"Yuheng",
""
],
[
"He",
"Shijing",
""
],
[
"Sugano",
"Yusuke",
""
],
[
"Wang",
"Yafei",
""
],
[
"Zhao",
"Kaixing",
""
],
[
"Khamis",
"Mohamed",
""
],
[
"Ye",
"Juan",
""
]
] |
2505.22770 | Iacopo Nonis | Iacopo Nonis | Mutation of $\tau$-exceptional sequences for acyclic quivers over local algebras | 17 pages | null | null | null | math.RT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Let $k$ be an algebraically closed field. Let $R$ be a local commutative finite dimensional $k$-algebra and let $Q$ be a quiver with no loops or oriented cycles. We show that mutation of $\tau$-exceptional sequences over $\Lambda = R\otimes_k kQ \cong RQ$ in the sense of Buan, Hanson, and Marsh coincides with the classical mutation of exceptional sequences defined by Crawley-Boevey and Ringel. In particular, the braid group acts transitively on the set of complete $\tau$-exceptional sequences in $\text{mod }{\Lambda}$. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:38:49 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Nonis",
"Iacopo",
""
]
] |
2505.22771 | Christopher Ormerod | Christopher Ormerod | Automated Essay Scoring Incorporating Annotations from Automated Feedback Systems | 10 pages, AIME-Con Conference Submission | null | null | null | cs.CL cs.AI | http://creativecommons.org/licenses/by-sa/4.0/ | This study illustrates how incorporating feedback-oriented annotations into the scoring pipeline can enhance the accuracy of automated essay scoring (AES). This approach is demonstrated with the Persuasive Essays for Rating, Selecting, and Understanding Argumentative and Discourse Elements (PERSUADE) corpus. We integrate two types of feedback-driven annotations: those that identify spelling and grammatical errors, and those that highlight argumentative components. To illustrate how this method could be applied in real-world scenarios, we employ two LLMs to generate annotations -- a generative language model used for spell-correction and an encoder-based token classifier trained to identify and mark argumentative elements. By incorporating annotations into the scoring process, we demonstrate improvements in performance using encoder-based large language models fine-tuned as classifiers. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:39:56 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Ormerod",
"Christopher",
""
]
] |
2505.22772 | Claas Alexander Voelcker | Claas Voelcker, Anastasiia Pedan, Arash Ahmadian, Romina Abachi, Igor Gilitschenski, Amir-massoud Farahmand | Calibrated Value-Aware Model Learning with Stochastic Environment Models | null | null | null | null | cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The idea of value-aware model learning, that models should produce accurate value estimates, has gained prominence in model-based reinforcement learning. The MuZero loss, which penalizes a model's value function prediction compared to the ground-truth value function, has been utilized in several prominent empirical works in the literature. However, theoretical investigation into its strengths and weaknesses is limited. In this paper, we analyze the family of value-aware model learning losses, which includes the popular MuZero loss. We show that these losses, as normally used, are uncalibrated surrogate losses, which means that they do not always recover the correct model and value function. Building on this insight, we propose corrections to solve this issue. Furthermore, we investigate the interplay between the loss calibration, latent model architectures, and auxiliary losses that are commonly employed when training MuZero-style agents. We show that while deterministic models can be sufficient to predict accurate values, learning calibrated stochastic models is still advantageous. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:40:49 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Voelcker",
"Claas",
""
],
[
"Pedan",
"Anastasiia",
""
],
[
"Ahmadian",
"Arash",
""
],
[
"Abachi",
"Romina",
""
],
[
"Gilitschenski",
"Igor",
""
],
[
"Farahmand",
"Amir-massoud",
""
]
] |
2505.22773 | Dmytro Bozhko | Rostyslav O. Serha, Kaitlin H. McAllister, Fabian Majcen, Sebastian Knauer, Timmy Reimann, Carsten Dubs, Gennadii A. Melkov, Alexander A. Serga, Vasyl S. Tyberkevych, Andrii V. Chumak, Dmytro A. Bozhko | Ultra-long-living magnons in the quantum limit | null | null | null | null | cond-mat.mtrl-sci quant-ph | http://creativecommons.org/licenses/by/4.0/ | Coherence time is the property of a quantum system that determines how long a state can hold quantum information. This parameter is directly bound to their lifetime in solid-state systems, where quantum information could be stored in quasiparticles. For decades, quasiparticles associated with magnetization order disturbance - magnons, had reported lifetimes below one microsecond at gigahertz frequencies, restricting their use as a quantum information carrier. Here, we report on the observation of short-wavelength magnons with lifetimes exceeding 18{\mu}s at millikelvin temperatures. The experiment has been performed in an ultra-pure single-crystal Yttrium Iron Garnet sphere in a wide range of temperatures from ambient down to 30 mK. Our results open doors for using magnons as data carriers in modern solid-state quantum computing platforms. | [
{
"version": "v1",
"created": "Wed, 28 May 2025 18:43:02 GMT"
}
] | 2025-05-30T00:00:00 | [
[
"Serha",
"Rostyslav O.",
""
],
[
"McAllister",
"Kaitlin H.",
""
],
[
"Majcen",
"Fabian",
""
],
[
"Knauer",
"Sebastian",
""
],
[
"Reimann",
"Timmy",
""
],
[
"Dubs",
"Carsten",
""
],
[
"Melkov",
"Gennadii A.",
""
],
[
"Serga",
"Alexander A.",
""
],
[
"Tyberkevych",
"Vasyl S.",
""
],
[
"Chumak",
"Andrii V.",
""
],
[
"Bozhko",
"Dmytro A.",
""
]
] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.