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 |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
2506.03203 | Victor Luder | Victor Luder, Michele Magno | Self-Sustaining Multi-Sensor LoRa-Based Activity Monitoring for Community Workout Parks | null | null | null | null | cs.NI | http://creativecommons.org/publicdomain/zero/1.0/ | With the rise of the Internet of Things (IoT), more sensors are deployed around us, covering a wide range of applications from industry and agriculture to urban environments such as smart cities. Throughout these applications the sensors collect data of various characteristics and support city planners and decision-makers in their work processes, ultimately maximizing the impact of public funds. This paper introduces the design and implementation of a self-sustaining wireless sensor node designed to continuously monitor the utilization of community street workout parks. The proposed sensor node monitors activity by leveraging acceleration data capturing micro-vibrations that propagate through the steel structures of the workout equipment. This allows us to detect activity duration with an average measured error of only 2.8 seconds. The sensor is optimized with an energy-aware, adaptive sampling and transmission algorithm which, in combination with the Long Range Wide Area Network (LoRaWAN), reduces power consumption to just 1.147 mW in normal operation and as low as 0.712 mW in low-power, standby mode allowing 46 days of battery runtime. In addition, the integrated energy-harvesting circuit was tested in the field. By monitoring the battery voltage for multiple days, it was shown that the sensor is capable of operating sustainably year-round without external power sources. To evaluate the sensor effectiveness, we conducted a week-long field test in Zurich, placing sensors at various street workout parks throughout the city. Analysis of the collected data revealed clear patterns in park usage depending on day and location. This dataset is made publicly available through our online dashboard. Finally, we showcase the potential of IoT for city applications in combination with an accessible data interface for decision-makers. | [
{
"version": "v1",
"created": "Mon, 02 Jun 2025 18:52:08 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Luder",
"Victor",
""
],
[
"Magno",
"Michele",
""
]
] |
2506.03204 | Mihai Prunescu | Mihai Prunescu | On the first-order theory of the remainder | 2 pages | null | null | null | math.LO | http://creativecommons.org/licenses/by/4.0/ | It is proved that the first-order theory of the structure (N,mod) is undecidable. Here mod denotes the operation of computing the remainder for any division between positive integers; i.e. x mod y is the remainder obtained by the division x : y. | [
{
"version": "v1",
"created": "Mon, 02 Jun 2025 19:19:21 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Prunescu",
"Mihai",
""
]
] |
2506.03205 | Umberto Gon\c{c}alves De Sousa | Umberto Gon\c{c}alves de Sousa | Q-ARDNS-Multi: A Multi-Agent Quantum Reinforcement Learning Framework with Meta-Cognitive Adaptation for Complex 3D Environments | 17 pages, 5 figures | null | null | null | cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper presents Q-ARDNS-Multi, an advanced multi-agent quantum reinforcement learning (QRL) framework that extends the ARDNS-FN-Quantum model, where Q-ARDNS-Multi stands for "Quantum Adaptive Reward-Driven Neural Simulator - Multi-Agent". It integrates quantum circuits with RY gates, meta-cognitive adaptation, and multi-agent coordination mechanisms for complex 3D environments. Q-ARDNS-Multi leverages a 2-qubit quantum circuit for action selection, a dual-memory system inspired by human cognition, a shared memory module for agent cooperation, and adaptive exploration strategies modulated by reward variance and intrinsic motivation. Evaluated in a $10 \times 10 \times 3$ GridWorld environment with two agents over 5000 episodes, Q-ARDNS-Multi achieves success rates of 99.6\% and 99.5\% for Agents 0 and 1, respectively, outperforming Multi-Agent Deep Deterministic Policy Gradient (MADDPG) and Soft Actor-Critic (SAC) in terms of success rate, stability, navigation efficiency, and collision avoidance. The framework records mean rewards of $-304.2891 \pm 756.4636$ and $-295.7622 \pm 752.7103$, averaging 210 steps to goal, demonstrating its robustness in dynamic settings. Comprehensive analyses, including learning curves, reward distributions, statistical tests, and computational efficiency evaluations, highlight the contributions of quantum circuits and meta-cognitive adaptation. By bridging quantum computing, cognitive science, and multi-agent RL, Q-ARDNS-Multi offers a scalable, human-like approach for applications in robotics, autonomous navigation, and decision-making under uncertainty. | [
{
"version": "v1",
"created": "Mon, 02 Jun 2025 20:43:33 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"de Sousa",
"Umberto Gonçalves",
""
]
] |
2506.03206 | Nadav Timor | Nadav Timor, Jonathan Mamou, Oren Pereg, Hongyang Zhang, David Harel | Out-of-Vocabulary Sampling Boosts Speculative Decoding | null | null | null | null | cs.LG cs.CL | http://creativecommons.org/licenses/by/4.0/ | Speculative decoding relies on fast and accurate drafters. Recent state-of-the-art language models employ larger and larger vocabularies, which significantly slows down drafters. One promising approach to boost the efficiency of speculative decoding is to use drafters with smaller vocabularies. However, existing sampling methods cannot draw out-of-vocabulary tokens, creating a tradeoff between drafters' vocabulary size and acceptance rates. This paper introduces Redistributing Drafter Kernels (RDK), the first out-of-vocabulary sampler that effectively recovers acceptance rates by virtually restoring pruned target tokens. RDK leverages token-affinity priors to reallocate drafter mass towards high-overlap regions. We prove mathematically that RDK can achieve higher acceptance rates than vanilla and state-of-the-art samplers. We provide an efficient first-order approximation of RDK and prove that it reduces redistribution times from $O(N^2)$ to $O(N)$, enabling lightweight implementations for large vocabularies. Our experiments demonstrate that this linear-time RDK significantly boosts acceptance rates even after extreme pruning (removing more than 75% of the drafter's vocabulary), where existing samplers fail. RDK opens the door to extremely pruned drafters, which were previously impractical. | [
{
"version": "v1",
"created": "Mon, 02 Jun 2025 21:08:02 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Timor",
"Nadav",
""
],
[
"Mamou",
"Jonathan",
""
],
[
"Pereg",
"Oren",
""
],
[
"Zhang",
"Hongyang",
""
],
[
"Harel",
"David",
""
]
] |
2506.03207 | Md Nahid Hasan Shuvo | Md Nahid Hasan Shuvo and Moinul Hossain | Fingerprinting Deep Learning Models via Network Traffic Patterns in Federated Learning | 7 pages, 4 Figures, Accepted to publish in Proceedings of the 2025 ACM Workshop on Wireless Security and Machine Learning (WiseML 2025), July 3, 2025, Arlington, VA, USA | null | null | null | cs.LG cs.AI cs.CR | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Federated Learning (FL) is increasingly adopted as a decentralized machine learning paradigm due to its capability to preserve data privacy by training models without centralizing user data. However, FL is susceptible to indirect privacy breaches via network traffic analysis-an area not explored in existing research. The primary objective of this research is to study the feasibility of fingerprinting deep learning models deployed within FL environments by analyzing their network-layer traffic information. In this paper, we conduct an experimental evaluation using various deep learning architectures (i.e., CNN, RNN) within a federated learning testbed. We utilize machine learning algorithms, including Support Vector Machines (SVM), Random Forest, and Gradient-Boosting, to fingerprint unique patterns within the traffic data. Our experiments show high fingerprinting accuracy, achieving 100% accuracy using Random Forest and around 95.7% accuracy using SVM and Gradient Boosting classifiers. This analysis suggests that we can identify specific architectures running within the subsection of the network traffic. Hence, if an adversary knows about the underlying DL architecture, they can exploit that information and conduct targeted attacks. These findings suggest a notable security vulnerability in FL systems and the necessity of strengthening it at the network level. | [
{
"version": "v1",
"created": "Mon, 02 Jun 2025 21:37:20 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Shuvo",
"Md Nahid Hasan",
""
],
[
"Hossain",
"Moinul",
""
]
] |
2506.03208 | Oleg Smirnov | O.Smirnov | Intrinsic Resolution of Organic Scintillators | 31 pages, 8 figures, 3 tables | null | null | null | physics.ins-det | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Fluctuations in photon production within scintillators contribute to the overall energy resolution of scintillation detectors. While typically negligible, this contribution - known as intrinsic resolution (IR) - presents significant challenges for both experimental measurement and Monte Carlo simulation. Using the phenomenological model proposed in~[1], we interpreted multiple datasets of IR measurements in organic scintillators. Although our results generally agree with the IR dataset examined in~[1], a comprehensive evaluation of systematic uncertainties highlights the necessity for more detailed investigation. Based on our findings, we offer specific recommendations for future measurements. | [
{
"version": "v1",
"created": "Mon, 02 Jun 2025 22:09:55 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Smirnov",
"O.",
""
]
] |
2506.03209 | Shuheng Chen | Tinghuan Li, Shuheng Chen, Junyi Fan, Elham Pishgar, Kamiar Alaei, Greg Placencia, Maryam Pishgar | Predicting Postoperative Stroke in Elderly SICU Patients: An Interpretable Machine Learning Model Using MIMIC Data | null | null | null | null | q-bio.QM cs.AI cs.LG | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Postoperative stroke remains a critical complication in elderly surgical intensive care unit (SICU) patients, contributing to prolonged hospitalization, elevated healthcare costs, and increased mortality. Accurate early risk stratification is essential to enable timely intervention and improve clinical outcomes. We constructed a combined cohort of 19,085 elderly SICU admissions from the MIMIC-III and MIMIC-IV databases and developed an interpretable machine learning (ML) framework to predict in-hospital stroke using clinical data from the first 24 hours of Intensive Care Unit (ICU) stay. The preprocessing pipeline included removal of high-missingness features, iterative Singular Value Decomposition (SVD) imputation, z-score normalization, one-hot encoding, and class imbalance correction via the Adaptive Synthetic Sampling (ADASYN) algorithm. A two-stage feature selection process-combining Recursive Feature Elimination with Cross-Validation (RFECV) and SHapley Additive exPlanations (SHAP)-reduced the initial 80 variables to 20 clinically informative predictors. Among eight ML models evaluated, CatBoost achieved the best performance with an AUROC of 0.8868 (95% CI: 0.8802--0.8937). SHAP analysis and ablation studies identified prior cerebrovascular disease, serum creatinine, and systolic blood pressure as the most influential risk factors. Our results highlight the potential of interpretable ML approaches to support early detection of postoperative stroke and inform decision-making in perioperative critical care. | [
{
"version": "v1",
"created": "Mon, 02 Jun 2025 22:53:12 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Li",
"Tinghuan",
""
],
[
"Chen",
"Shuheng",
""
],
[
"Fan",
"Junyi",
""
],
[
"Pishgar",
"Elham",
""
],
[
"Alaei",
"Kamiar",
""
],
[
"Placencia",
"Greg",
""
],
[
"Pishgar",
"Maryam",
""
]
] |
2506.03210 | Qiusheng Huang | Qiusheng Huang, Yuan Niu, Xiaohui Zhong, Anboyu Guo, Lei Chen, Dianjun Zhang, Xuefeng Zhang, Hao Li | FuXi-Ocean: A Global Ocean Forecasting System with Sub-Daily Resolution | null | null | null | null | cs.LG cs.AI physics.ao-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Accurate, high-resolution ocean forecasting is crucial for maritime operations and environmental monitoring. While traditional numerical models are capable of producing sub-daily, eddy-resolving forecasts, they are computationally intensive and face challenges in maintaining accuracy at fine spatial and temporal scales. In contrast, recent data-driven approaches offer improved computational efficiency and emerging potential, yet typically operate at daily resolution and struggle with sub-daily predictions due to error accumulation over time. We introduce FuXi-Ocean, the first data-driven global ocean forecasting model achieving six-hourly predictions at eddy-resolving 1/12{\deg} spatial resolution, reaching depths of up to 1500 meters. The model architecture integrates a context-aware feature extraction module with a predictive network employing stacked attention blocks. The core innovation is the Mixture-of-Time (MoT) module, which adaptively integrates predictions from multiple temporal contexts by learning variable-specific reliability , mitigating cumulative errors in sequential forecasting. Through comprehensive experimental evaluation, FuXi-Ocean demonstrates superior skill in predicting key variables, including temperature, salinity, and currents, across multiple depths. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 00:52:31 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Huang",
"Qiusheng",
""
],
[
"Niu",
"Yuan",
""
],
[
"Zhong",
"Xiaohui",
""
],
[
"Guo",
"Anboyu",
""
],
[
"Chen",
"Lei",
""
],
[
"Zhang",
"Dianjun",
""
],
[
"Zhang",
"Xuefeng",
""
],
[
"Li",
"Hao",
""
]
] |
2506.03211 | Wanting Yang | Wanting Yang, Zehui Xiong, Qianqian Yang, Ping Zhang, Merouane Debbah, Rahim Tafazolli | Channel-adaptive Cross-modal Generative Semantic Communication for Point Cloud Transmission | null | null | null | null | cs.CV cs.NI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | With the rapid development of autonomous driving and extended reality, efficient transmission of point clouds (PCs) has become increasingly important. In this context, we propose a novel channel-adaptive cross-modal generative semantic communication (SemCom) for PC transmission, called GenSeC-PC. GenSeC-PC employs a semantic encoder that fuses images and point clouds, where images serve as non-transmitted side information. Meanwhile, the decoder is built upon the backbone of PointDif. Such a cross-modal design not only ensures high compression efficiency but also delivers superior reconstruction performance compared to PointDif. Moreover, to ensure robust transmission and reduce system complexity, we design a streamlined and asymmetric channel-adaptive joint semantic-channel coding architecture, where only the encoder needs the feedback of average signal-to-noise ratio (SNR) and available bandwidth. In addition, rectified denoising diffusion implicit models is employed to accelerate the decoding process to the millisecond level, enabling real-time PC communication. Unlike existing methods, GenSeC-PC leverages generative priors to ensure reliable reconstruction even from noisy or incomplete source PCs. More importantly, it supports fully analog transmission, improving compression efficiency by eliminating the need for error-free side information transmission common in prior SemCom approaches. Simulation results confirm the effectiveness of cross-modal semantic extraction and dual-metric guided fine-tuning, highlighting the framework's robustness across diverse conditions, including low SNR, bandwidth limitations, varying numbers of 2D images, and previously unseen objects. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 01:14:58 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Yang",
"Wanting",
""
],
[
"Xiong",
"Zehui",
""
],
[
"Yang",
"Qianqian",
""
],
[
"Zhang",
"Ping",
""
],
[
"Debbah",
"Merouane",
""
],
[
"Tafazolli",
"Rahim",
""
]
] |
2506.03212 | Svetlana Kushpil | Svetlana Kushpil (on behalf of the ALICE Collaboration) | Synchronous and asynchronous Data Quality Control of the ALICE Inner Tracking System in the LHC Run 3 | 4 pages, 6 figures, PIXEL 2024 | null | null | null | physics.ins-det hep-ex | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The Inner Tracking System (ITS) of the ALICE experiment at the CERN Large Hadron Collider (LHC) is the largest Monolithic Active Pixel Sensor technology application in high-energy physics. The upgraded version of the tracking system, called ITS2, consists of seven concentric layers of ALPIDE monolithic active pixel sensors produced in the 180 nm CMOS process, covering a total sensitive area of about 10 m${}^2$. The ALPIDE sensor features a pixel pitch of 27 $\mu$m $\times$ 29 $\mu$m and a position resolution of about 5 $\mu$m. The very low material budget, 0.36\% $X_{0}$/layer for the three innermost layers and 1.10\% $X_{0}$/layer for the outer layers, in combination with the small radial distance of only 23 mm from the beam, leads to an excellent impact parameter resolution at low transverse momentum. This makes the detector well suited for experimentally challeging physics measurements such as the reconstruction of low transverse momentum heavy-flavor particles in the heavy-ion collision environment. This contribution provides an overview of the ITS2 data Quality Control system (QC), a framework designed to synchronously monitor the detector operating parameters and provide asynchronous reconstruction of the collected data, with the goal of guaranteeing a stable and efficient data taking. The monitoring for fake-hit rate, front-end electronics status, data integrity, cluster and track distributions, are presented, together with an overview of the ITS2 performance during the recent Run 3 pp and Pb--Pb data taking campaigns, as extracted from the QC asynchronous reconstruction. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 02:39:37 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Kushpil",
"Svetlana",
"",
"on behalf of the ALICE Collaboration"
]
] |
2506.03213 | Abdullah Al Mamun | Abdullah Al Mamun, Miaohua Zhang, David Ahmedt-Aristizabal, Zeeshan Hayder, Mohammad Awrangjeb | ConMamba: Contrastive Vision Mamba for Plant Disease Detection | null | null | null | null | cs.CV | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Plant Disease Detection (PDD) is a key aspect of precision agriculture. However, existing deep learning methods often rely on extensively annotated datasets, which are time-consuming and costly to generate. Self-supervised Learning (SSL) offers a promising alternative by exploiting the abundance of unlabeled data. However, most existing SSL approaches suffer from high computational costs due to convolutional neural networks or transformer-based architectures. Additionally, they struggle to capture long-range dependencies in visual representation and rely on static loss functions that fail to align local and global features effectively. To address these challenges, we propose ConMamba, a novel SSL framework specially designed for PDD. ConMamba integrates the Vision Mamba Encoder (VME), which employs a bidirectional State Space Model (SSM) to capture long-range dependencies efficiently. Furthermore, we introduce a dual-level contrastive loss with dynamic weight adjustment to optimize local-global feature alignment. Experimental results on three benchmark datasets demonstrate that ConMamba significantly outperforms state-of-the-art methods across multiple evaluation metrics. This provides an efficient and robust solution for PDD. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 03:01:38 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Mamun",
"Abdullah Al",
""
],
[
"Zhang",
"Miaohua",
""
],
[
"Ahmedt-Aristizabal",
"David",
""
],
[
"Hayder",
"Zeeshan",
""
],
[
"Awrangjeb",
"Mohammad",
""
]
] |
2506.03214 | Shuqiang Wang | Yi Guo, Yihang Dong, Michael Kwok-Po Ng, Shuqiang Wang | A Pre-trained Framework for Multilingual Brain Decoding Using Non-invasive Recordings | null | null | null | null | q-bio.NC cs.AI cs.CL | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Brain-computer interfaces (BCIs) with speech decoding from brain recordings have broad application potential in fields such as clinical rehabilitation and cognitive neuroscience. However, current decoding methods remain limited to single-language, single-subject, and single neuroimaging modality settings, restricting their clinical applicability and generalizability. Here we propose a joint multilingual, multi-subject and multimodal decoding framework. It maps diverse brain recordings into a unified semantic space defined by a pre-trained multilingual model (PMM), enabling decoding across multiple languages, multiple subjects and multiple neuroimaging modalities. The proposed framework is validated using non-invasive brain recordings from 159 participants across four languages. Experimental results show that it exhibits strong generalization across multilingual, multi-subject, and multimodal settings. More importantly, the proposed framework can promote linguistic fairness, which is vital for underrepresented languages in BCI applications. The unified semantic space enables cross-lingual mapping enhancement, allowing the framework to boost the decoding performance of underrepresented languages, thereby promoting linguistic fairness. Overall, the proposed framework establishes a new potential paradigm for brain decoding, opening new paths for broader applications of BCI. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 04:34:22 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Guo",
"Yi",
""
],
[
"Dong",
"Yihang",
""
],
[
"Ng",
"Michael Kwok-Po",
""
],
[
"Wang",
"Shuqiang",
""
]
] |
2506.03215 | Sourabhashis Das | Sourabhashis Das, Wentang Kuo, and Yu-Ru Liu | A subset generalization of the Erd\H{o}s-Kac theorem over number fields with applications | Submitted for publication, 33 pages. arXiv admin note: substantial text overlap with arXiv:2506.02432 | null | null | null | math.NT | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Let $\omega(n)$ denote the number of distinct prime factors of a natural number $n$. In 1940, Erd\H{o}s and Kac established that $\omega(n)$ obeys the Gaussian distribution over natural numbers. In 2004, the third author generalized their theorem to all abelian monoids. In this work, we extend the work of the third author to any subset of the set of ideals of a number field satisfying some additional conditions. Finally, we apply this theorem to prove the Erd\H{o}s-Kac theorem over $h$-free and over $h$-full ideals of the number field. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 04:57:43 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Das",
"Sourabhashis",
""
],
[
"Kuo",
"Wentang",
""
],
[
"Liu",
"Yu-Ru",
""
]
] |
2506.03216 | Arbind Agrahari Baniya | Arbind Agrahari Baniya, Tsz-Kwan Lee, Peter Eklund and Sunil Aryal | A Survey of Deep Learning Video Super-Resolution | This paper has been published in IEEE Transactions on Emerging Topics in Computational Intelligence, vol. 8, no. 4, pp. 2655-2676, Aug. 2024, doi: 10.1109/TETCI.2024.3398015 | in IEEE Transactions on Emerging Topics in Computational Intelligence, vol. 8, no. 4, pp. 2655-2676, Aug. 2024 | 10.1109/TETCI.2024.3398015 | null | eess.IV cs.CV cs.LG | http://creativecommons.org/licenses/by/4.0/ | Video super-resolution (VSR) is a prominent research topic in low-level computer vision, where deep learning technologies have played a significant role. The rapid progress in deep learning and its applications in VSR has led to a proliferation of tools and techniques in the literature. However, the usage of these methods is often not adequately explained, and decisions are primarily driven by quantitative improvements. Given the significance of VSR's potential influence across multiple domains, it is imperative to conduct a comprehensive analysis of the elements and deep learning methodologies employed in VSR research. This methodical analysis will facilitate the informed development of models tailored to specific application needs. In this paper, we present an overarching overview of deep learning-based video super-resolution models, investigating each component and discussing its implications. Furthermore, we provide a synopsis of key components and technologies employed by state-of-the-art and earlier VSR models. By elucidating the underlying methodologies and categorising them systematically, we identified trends, requirements, and challenges in the domain. As a first-of-its-kind survey of deep learning-based VSR models, this work also establishes a multi-level taxonomy to guide current and future VSR research, enhancing the maturation and interpretation of VSR practices for various practical applications. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 05:42:19 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Baniya",
"Arbind Agrahari",
""
],
[
"Lee",
"Tsz-Kwan",
""
],
[
"Eklund",
"Peter",
""
],
[
"Aryal",
"Sunil",
""
]
] |
2506.03217 | Pierrick Coupe | Pierrick Coup\'e, Boris Mansencal, Flor\'eal Morandat, Sergio Morell-Ortega, Nicolas Villain, Jose V. Manj\'on, Vincent Planche | petBrain: A New Pipeline for Amyloid, Tau Tangles and Neurodegeneration Quantification Using PET and MRI | null | null | null | null | eess.IV cs.CV | http://creativecommons.org/licenses/by-nc-nd/4.0/ | INTRODUCTION: Quantification of amyloid plaques (A), neurofibrillary tangles (T2), and neurodegeneration (N) using PET and MRI is critical for Alzheimer's disease (AD) diagnosis and prognosis. Existing pipelines face limitations regarding processing time, variability in tracer types, and challenges in multimodal integration.
METHODS: We developed petBrain, a novel end-to-end processing pipeline for amyloid-PET, tau-PET, and structural MRI. It leverages deep learning-based segmentation, standardized biomarker quantification (Centiloid, CenTauR, HAVAs), and simultaneous estimation of A, T2, and N biomarkers. The pipeline is implemented as a web-based platform, requiring no local computational infrastructure or specialized software knowledge.
RESULTS: petBrain provides reliable and rapid biomarker quantification, with results comparable to existing pipelines for A and T2. It shows strong concordance with data processed in ADNI databases. The staging and quantification of A/T2/N by petBrain demonstrated good agreement with CSF/plasma biomarkers, clinical status, and cognitive performance.
DISCUSSION: petBrain represents a powerful and openly accessible platform for standardized AD biomarker analysis, facilitating applications in clinical research. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 07:44:04 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Coupé",
"Pierrick",
""
],
[
"Mansencal",
"Boris",
""
],
[
"Morandat",
"Floréal",
""
],
[
"Morell-Ortega",
"Sergio",
""
],
[
"Villain",
"Nicolas",
""
],
[
"Manjón",
"Jose V.",
""
],
[
"Planche",
"Vincent",
""
]
] |
2506.03218 | Alina Wernick | Alina Wernick, Kristof Meding | Beware! The AI Act Can Also Apply to Your AI Research Practices | null | null | null | null | cs.CY cs.AI cs.LG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The EU has become one of the vanguards in regulating the digital age. A particularly important regulation in the Artificial Intelligence (AI) domain is the EU AI Act, which entered into force in 2024. The AI Act specifies -- due to a risk-based approach -- various obligations for providers of AI systems. These obligations, for example, include a cascade of documentation and compliance measures, which represent a potential obstacle to science. But do these obligations also apply to AI researchers? This position paper argues that, indeed, the AI Act's obligations could apply in many more cases than the AI community is aware of. In our analysis of the AI Act and its applicability, we contribute the following: 1.) We give a high-level introduction to the AI Act aimed at non-legal AI research scientists. 2.) We explain with everyday research examples why the AI Act applies to research. 3.) We analyse the exceptions of the AI Act's applicability and state that especially scientific research exceptions fail to account for current AI research practices. 4.) We propose changes to the AI Act to provide more legal certainty for AI researchers and give two recommendations for AI researchers to reduce the risk of not complying with the AI Act. We see our paper as a starting point for a discussion between policymakers, legal scholars, and AI researchers to avoid unintended side effects of the AI Act on research. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:01:36 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Wernick",
"Alina",
""
],
[
"Meding",
"Kristof",
""
]
] |
2506.03219 | Philipp Lepold | Philipp Lepold, Tobias R\"oddiger, Michael Beigl | HARNode: A Time-Synchronised, Open-Source, Multi-Device, Wearable System for Ad Hoc Field Studies | 6 pages, 6 figures | null | null | null | cs.NI | http://creativecommons.org/licenses/by/4.0/ | Research into human activity recognition (HAR) suffers from a lack of comprehensive, freely available field data. Commercial systems are rarely open source, offer little expandability, and have shortcomings in node synchronisation, data throughput, placement clarity, setup complexity, and cost. As a result, only a few intuitively placed sensors are often used, and field trials are generally reduced. HARNode addresses these obstacles as a fully open-source hardware and software platform for rapid field applications. Each sensor node combines an ESP32-S3 module (AtomS3) with a display, a 9-axis IMU (Bosch BMX160), pressure/temperature sensors (Bosch BMP388) and an I2C port for extensions; the operating time is up to 8 hours. Data is streamed via Wi-Fi, while NTP-based time synchronisation provides an average clock accuracy of $\approx$ 1 ms. Manufacturing is carried out exclusively with commercially available components, an online PCB service - requiring little hardware knowledge - and a compact, 3D-printed housing with Velcro straps, allowing almost any number of highly synchronised nodes to be flexibly attached to the body. Its performance was demonstrated in a proof-of-concept study with ten test subjects, each wearing eleven HARNodes; the entire setup took less than five minutes per person. An example application goal was to detect the transition from level walking to climbing stairs. A random forest classifier evaluated the benefits of sensor overprovisioning: the best combination of seven nodes achieved $\approx$ 98\% accuracy (binary: level walking vs. approaching stairs), matching the result of all eleven positions. A single sensor on the foot achieved $\approx$ 90\% accuracy. These results demonstrate the suitability of HARNode as an ultra-fast ad hoc field system and support evidence-based sensor placement. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 08:42:18 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Lepold",
"Philipp",
""
],
[
"Röddiger",
"Tobias",
""
],
[
"Beigl",
"Michael",
""
]
] |
2506.03220 | Muhammad Sharif | Malick Sallah and M. Sharif | Investigating Strange Stars in Rastall Theory | 27 pages, 11 figures | Chinese J. Physics 95(2025)1141 | null | null | gr-qc | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This study explores the structural formation of various spherically symmetric anisotropic stars within the framework of Rastall theory. To achieve this, we derive modified field equations which are then resolved using the Finch-Skea ansatz, which involve unknown parameters $(A_1,A_2,A_3)$. These parameters are found by using appropriate constraints given by the junction conditions, in addition to observational data from some selected stars. The $\mathbb{EOS}$ given by the $\mathbb{MIT}$ bag model is employed to examine the interior structure and various physical properties of these compact objects. For calculated values of the bag constant $\mathcal{B}$ and two values of the Rastall parameter, $\xi = 0.3, 0.5$, we investigate the regularity and viability of the state variables. Additionally, we analyze stability of the developed model by employing three distinct criteria. We find that the obtained model is stable and provides an accurate approximation for the mass and radius of strange stars when the Rastall parameter $\xi = 0.3$ is considered. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 09:12:08 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Sallah",
"Malick",
""
],
[
"Sharif",
"M.",
""
]
] |
2506.03221 | Lena John | Lena John, Tim Wittenborg, S\"oren Auer, Oliver Karras | Human-In-The-Loop Workflow for Neuro- Symbolic Scholarly Knowledge Organization | null | null | null | null | cs.DL | http://creativecommons.org/licenses/by/4.0/ | As the volume of scientific literature continues to grow, efficient knowledge organization is an increasingly challenging task. Traditional structuring of scientific content is time-consuming and requires significant domain expertise, increasing the need for tool support. Our goal is to create a Human-in-the-Loop (HITL) workflow that supports researchers in creating and structuring scientific knowledge, leveraging neural models and knowledge graphs, exemplified using the Open Research Knowledge Graph (ORKG). The workflow aims to automate key steps, including data extraction and knowledge structuring, while keeping user oversight through human validation. We developed a modular framework implementing the workflow and evaluated it along the Quality Improvement Paradigm (QIP) with participants from the ORKG user community. The evaluation indicated that the framework is highly usable and provides practical support. It significantly reduces the time and effort required to transition from a research interest to literature-based answers by streamlining the import of information into a knowledge graph. Participants evaluated the framework with an average System Usability Scale (SUS) score of 84.17, an A+ -- the highest achievable rating. They also reported that it improved their time spent, previously between 4 hours and two weeks, down to an average of 24:40 minutes. The tool streamlines the creation of scientific corpora and extraction of structured knowledge for KG integration by leveraging LLMs and user-defined models, significantly accelerating the review process. However, human validation remains essential throughout the extraction process, and future work is needed to improve extraction accuracy and entity linking to existing knowledge resources. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 09:12:52 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"John",
"Lena",
""
],
[
"Wittenborg",
"Tim",
""
],
[
"Auer",
"Sören",
""
],
[
"Karras",
"Oliver",
""
]
] |
2506.03222 | Gabriele Gatto | Gabriele Gatto | High-energy dynamics of QCD: Theoretical and phenomenological results | 144 pages, 240 references, 29 figures, Ph.D. Thesis | null | null | null | hep-ph hep-th | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This work investigates the behavior of hadronic matter in the high-energy Regge-Gribov (semi-hard) regime of Quantum Chromodynamics (QCD), accessible through current and future colliders such as the LHC, EIC, and FCC. Central to the analysis is the Balitsky-Fadin-Kuraev-Lipatov (BFKL) formalism, with detailed treatment of the BFKL equation in both leading and next-to-leading logarithmic approximations. A major focus is placed on the computation of real next-to-leading order (NLO) corrections to the Higgs boson impact factor, incorporating finite top-quark mass effects. These corrections are essential for improving precision in Higgs production processes at large rapidity separations. The study also explores the semi-inclusive production of exotic tetraquark states, employing a hybrid framework that combines collinear and BFKL dynamics within a variable-flavor number scheme. Updated fragmentation functions for bottomonium-like states are provided, offering improved predictions at 14 TeV and 100 TeV. Additionally, the work addresses diffractive di-hadron production in the small-x saturation regime using the Color Glass Condensate (CGC) formalism, presenting preliminary leading-order analytical results. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 09:55:28 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Gatto",
"Gabriele",
""
]
] |
2506.03223 | Levente Bodnar | Levente Bodn\'ar, Jared Le\'on, Xizhi Liu, Oleg Pikhurko | The Tur\'{a}n density of short tight cycles | 28 pages, ancilliary files. arXiv admin note: substantial text overlap with arXiv:2412.21011 | null | null | null | math.CO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The $3$-uniform tight $\ell$-cycle $C_\ell^{3}$ is the $3$-graph on $\{1,\dots,\ell\}$ consisting of all $\ell$ consecutive triples in the cyclic order. Let $\mathcal{C}$ be either the pair $\{C_{4}^{3}, C_{5}^{3}\}$ or the single tight $\ell$-cycle $C_{\ell}^{3}$ for some $\ell\ge 7$ not divisible by $3$.
We show that the Tur\'an density of $\mathcal{C}$, that is, the asymptotically maximal edge density of a large $\mathcal{C}$-free $3$-graph, is equal to $2\sqrt{3} - 3$. We also establish the corresponding Erd\H{o}s-Simonovits-type stability result, informally stating that all almost maximum $\mathcal{C}$-free graphs are close in the edit distance to a 2-part recursive construction. This extends the earlier analogous results of Kam\v{c}ev-Letzter-Pokrovskiy ["The Tur\'an density of tight cycles in three-uniform hypergraphs", Int. Math. Res. Not. 6 (2024), 4804-4841] that apply for sufficiently large $\ell$ only.
Additionally, we prove a finer structural result that allows us to determine the maximum number of edges in a $\{C_{4}^{3}, C_{5}^{3}\}$-free $3$-graph with a given number of vertices up to an additive $O(1)$ error term. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 09:56:20 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Bodnár",
"Levente",
""
],
[
"León",
"Jared",
""
],
[
"Liu",
"Xizhi",
""
],
[
"Pikhurko",
"Oleg",
""
]
] |
2506.03224 | Jinwei Zeng | Jinwei Zeng, Yu Liu, Guozhen Zhang, Jingtao Ding, Yuming Lin, Jian Yuan, Yong Li | OpenCarbon: A Contrastive Learning-based Cross-Modality Neural Approach for High-Resolution Carbon Emission Prediction Using Open Data | Accepted by IJCAI 2025 | null | null | null | cs.CV cs.AI physics.soc-ph | http://creativecommons.org/licenses/by/4.0/ | Accurately estimating high-resolution carbon emissions is crucial for effective emission governance and mitigation planning. While conventional methods for precise carbon accounting are hindered by substantial data collection efforts, the rise of open data and advanced learning techniques offers a promising solution. Once an open data-based prediction model is developed and trained, it can easily infer emissions for new areas based on available open data. To address this, we incorporate two modalities of open data, satellite images and point-of-interest (POI) data, to predict high-resolution urban carbon emissions, with satellite images providing macroscopic and static and POI data offering fine-grained and relatively dynamic functionality information. However, estimating high-resolution carbon emissions presents two significant challenges: the intertwined and implicit effects of various functionalities on carbon emissions, and the complex spatial contiguity correlations that give rise to the agglomeration effect. Our model, OpenCarbon, features two major designs that target the challenges: a cross-modality information extraction and fusion module to extract complementary functionality information from two modules and model their interactions, and a neighborhood-informed aggregation module to capture the spatial contiguity correlations. Extensive experiments demonstrate our model's superiority, with a significant performance gain of 26.6\% on R2. Further generalizability tests and case studies also show OpenCarbon's capacity to capture the intrinsic relation between urban functionalities and carbon emissions, validating its potential to empower efficient carbon governance and targeted carbon mitigation planning. Codes and data are available: https://github.com/JinweiZzz/OpenCarbon. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 10:12:10 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Zeng",
"Jinwei",
""
],
[
"Liu",
"Yu",
""
],
[
"Zhang",
"Guozhen",
""
],
[
"Ding",
"Jingtao",
""
],
[
"Lin",
"Yuming",
""
],
[
"Yuan",
"Jian",
""
],
[
"Li",
"Yong",
""
]
] |
2506.03225 | Wa\"el Doulazmi | Wa\"el Doulazmi, Auguste Lehuger, Marin Toromanoff, Valentin Charraut, Thibault Buhet, Fabien Moutarde | Multiple-Frequencies Population-Based Training | Accepted at RLC25 | null | null | null | cs.LG cs.AI cs.NE | http://creativecommons.org/licenses/by/4.0/ | Reinforcement Learning's high sensitivity to hyperparameters is a source of instability and inefficiency, creating significant challenges for practitioners. Hyperparameter Optimization (HPO) algorithms have been developed to address this issue, among them Population-Based Training (PBT) stands out for its ability to generate hyperparameters schedules instead of fixed configurations. PBT trains a population of agents, each with its own hyperparameters, frequently ranking them and replacing the worst performers with mutations of the best agents. These intermediate selection steps can cause PBT to focus on short-term improvements, leading it to get stuck in local optima and eventually fall behind vanilla Random Search over longer timescales. This paper studies how this greediness issue is connected to the choice of evolution frequency, the rate at which the selection is done. We propose Multiple-Frequencies Population-Based Training (MF-PBT), a novel HPO algorithm that addresses greediness by employing sub-populations, each evolving at distinct frequencies. MF-PBT introduces a migration process to transfer information between sub-populations, with an asymmetric design to balance short and long-term optimization. Extensive experiments on the Brax suite demonstrate that MF-PBT improves sample efficiency and long-term performance, even without actually tuning hyperparameters. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 11:19:21 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Doulazmi",
"Waël",
""
],
[
"Lehuger",
"Auguste",
""
],
[
"Toromanoff",
"Marin",
""
],
[
"Charraut",
"Valentin",
""
],
[
"Buhet",
"Thibault",
""
],
[
"Moutarde",
"Fabien",
""
]
] |
2506.03226 | Elena V. Mikheeva | V. N. Lukash (Astro Space Center, P. N. Lebedev Physical Institute, Russian Academy of Sciences, Russia), E. V. Mikheeva (Astro Space Center, P. N. Lebedev Physical Institute, Russian Academy of Sciences) | Cascade Relaxation of the Gravitating Vacuum as a Generator of the Evolving Universe | 10 pages, 2 figures | JETP Letters, 2025, Vol. 121, No.6, pp. 399-410 | 10.1134/S0021364024604718 | null | gr-qc | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The cascade relaxation of a polarized vacuum in the expanding Universe is a chain of evolutionary epochs of decreasing its density with the exit of dominant fields (each in its own time) from the initial zero states to nonzero values, from the dominant scalar field in the early Universe to subsequent ones, including the $\Lambda$-term in the modern Universe. The cascade vacuum relaxation creates the entire observable cosmology from the Friedmann model with small perturbations of the metric with non-power-law power spectra from which primordial black holes could have arisen, and gravitational waves over a wide wavenumber range to the formation of dark matter and energy, early galaxies, supermassive black holes, and the large-scale structure of the Universe. An observational model of cascade vacuum relaxation in the early Universe has been build, which contains two constants determined by observational data and does not require information on the potential of the fields. A solution has been obtained for the general relativity vacuum attractor, including, in addition to the two previously mentioned constants, the third constant (not yet limited by observations), which leads to an additional power of density perturbations on a small scale ($k>10\,$Mpc$^{-1}$) in the form of a ``bump'', a two-power-law spectrum, etc. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 11:49:21 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Lukash",
"V. N.",
"",
"Astro Space Center, P. N. Lebedev Physical Institute, Russian Academy of Sciences, Russia"
],
[
"Mikheeva",
"E. V.",
"",
"Astro Space Center, P. N. Lebedev Physical Institute, Russian Academy of Sciences"
]
] |
2506.03227 | Abdelrahman Sayed Sayed | Abdelrahman Sayed Sayed, Pierre-Jean Meyer, Mohamed Ghazel | Bridging Neural ODE and ResNet: A Formal Error Bound for Safety Verification | 17 pages, 5 figures, Accepted for publication in the proceedings of the 8th International Symposium on AI Verification SAIV 2025 | null | null | null | cs.LG cs.AI | http://creativecommons.org/licenses/by-nc-sa/4.0/ | A neural ordinary differential equation (neural ODE) is a machine learning model that is commonly described as a continuous depth generalization of a residual network (ResNet) with a single residual block, or conversely, the ResNet can be seen as the Euler discretization of the neural ODE. These two models are therefore strongly related in a way that the behaviors of either model are considered to be an approximation of the behaviors of the other. In this work, we establish a more formal relationship between these two models by bounding the approximation error between two such related models. The obtained error bound then allows us to use one of the models as a verification proxy for the other, without running the verification tools twice: if the reachable output set expanded by the error bound satisfies a safety property on one of the models, this safety property is then guaranteed to be also satisfied on the other model. This feature is fully reversible, and the initial safety verification can be run indifferently on either of the two models. This novel approach is illustrated on a numerical example of a fixed-point attractor system modeled as a neural ODE. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 11:56:22 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Sayed",
"Abdelrahman Sayed",
""
],
[
"Meyer",
"Pierre-Jean",
""
],
[
"Ghazel",
"Mohamed",
""
]
] |
2506.03228 | Jos\'e Garre-Rubio | Jose Garre Rubio | A simple mechanism for unstable degeneracies in local Hamiltonians | Was this known or written somewhere? Comments are welcome | null | null | null | quant-ph | http://creativecommons.org/licenses/by/4.0/ | If a local Hamiltonian eigenstate is mapped to another state by local operators commuting with the Hamiltonian terms, the latter is also an eigenstate. This basic observation implies a no-go result for being a unique ground state, short-range entangled and protected against local perturbations. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 12:26:06 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Rubio",
"Jose Garre",
""
]
] |
2506.03229 | Qian-Wei Wang | Qian-Wei Wang, Yuqiu Xie, Letian Zhang, Zimo Liu and Shu-Tao Xia | Pre-trained Vision-Language Models Assisted Noisy Partial Label Learning | null | null | null | null | cs.CV cs.AI | http://creativecommons.org/licenses/by-nc-sa/4.0/ | In the context of noisy partial label learning (NPLL), each training sample is associated with a set of candidate labels annotated by multiple noisy annotators. With the emergence of high-performance pre-trained vision-language models (VLMs) such as CLIP, LLaVa and GPT-4V, the direction of using these models to replace time-consuming manual annotation workflows and achieve "manual-annotation-free" training for downstream tasks has become a highly promising research avenue. This paper focuses on learning from noisy partial labels annotated by pre-trained VLMs and proposes an innovative collaborative consistency regularization (Co-Reg) method. Unlike the symmetric noise primarily addressed in traditional noisy label learning, the noise generated by pre-trained models is instance-dependent, embodying the underlying patterns of the pre-trained models themselves, which significantly increases the learning difficulty for the model. To address this, we simultaneously train two neural networks that implement collaborative purification of training labels through a "Co-Pseudo-Labeling" mechanism, while enforcing consistency regularization constraints in both the label space and feature representation space. Our method can also leverage few-shot manually annotated valid labels to further enhance its performances. Comparative experiments with different denoising and disambiguation algorithms, annotation manners, and pre-trained model application schemes fully validate the effectiveness of the proposed method, while revealing the broad prospects of integrating weakly-supervised learning techniques into the knowledge distillation process of pre-trained models. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 12:48:54 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Wang",
"Qian-Wei",
""
],
[
"Xie",
"Yuqiu",
""
],
[
"Zhang",
"Letian",
""
],
[
"Liu",
"Zimo",
""
],
[
"Xia",
"Shu-Tao",
""
]
] |
2506.03230 | Zi Yang | Selcuk Gurses, Aozhong Zhang, Yanxia Deng, Xun Dong, Xin Li, Naigang Wang, Penghang Yin, Zi Yang | DiaBlo: Diagonal Blocks Are Sufficient For Finetuning | null | null | null | null | cs.LG cs.AI cs.CL math.OC | http://creativecommons.org/licenses/by/4.0/ | Finetuning is a critical step for adapting large language models (LLMs) to domain-specific downstream tasks. To mitigate the substantial computational and memory costs of full-model fine-tuning, Parameter-Efficient Finetuning (PEFT) methods have been proposed to update only a small subset of model parameters. However, performance gaps between PEFT approaches and full-model fine-tuning still exist. In this work, we present DiaBlo, a simple yet effective PEFT approach that updates only the diagonal blocks of selected model weight matrices. Unlike Low Rank Adaptation (LoRA) and its variants, DiaBlo eliminates the need for low rank matrix products, thereby avoiding the reliance on auxiliary initialization schemes or customized optimization strategies to improve convergence. This design leads to stable and robust convergence while maintaining comparable memory efficiency and training speed to LoRA. We conduct extensive experiments across a range of tasks, including commonsense reasoning, arithmetic reasoning, code generation, and safety alignment, to evaluate the effectiveness and efficiency of DiaBlo. Across these benchmarks, DiaBlo demonstrates strong and consistent performance while maintaining high memory efficiency and fast finetuning speed. Codes are available at https://github.com/ziyangjoy/DiaBlo. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 13:47:59 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Gurses",
"Selcuk",
""
],
[
"Zhang",
"Aozhong",
""
],
[
"Deng",
"Yanxia",
""
],
[
"Dong",
"Xun",
""
],
[
"Li",
"Xin",
""
],
[
"Wang",
"Naigang",
""
],
[
"Yin",
"Penghang",
""
],
[
"Yang",
"Zi",
""
]
] |
2506.03231 | Yajie Zhou | Yajie Zhou, Jiajun Ruan, Eric S. Wang, Sadjad Fouladi, Francis Y. Yan, Kevin Hsieh, Zaoxing Liu | NetPress: Dynamically Generated LLM Benchmarks for Network Applications | null | null | null | null | cs.NI cs.AI cs.LG | http://creativecommons.org/licenses/by/4.0/ | Despite growing interest in domain-specific benchmarking of large language models (LLMs) and agents, current evaluations remain limited to static, small-scale datasets, especially in high-stakes tasks like network operations that demand reliability for deployments. We present NetPress, an automated benchmark generation framework for evaluating LLM agents in network applications. NetPress introduces a unified abstraction with state and action, enabling dynamic generation of diverse query sets along with corresponding ground truths. At runtime, users can specify benchmark configurations to generate millions of queries on the fly. In addition to dynamic benchmark construction, NetPress integrates with network emulators to provide realistic environment feedback, supporting comprehensive evaluation across correctness, safety, and latency. We instantiate NetPress on three representative applications, revealing interesting fine-grained differences in agent behavior that static, correctness-only benchmarks often miss. NetPress moves LLM evaluation toward realistic, scalable testing in infrastructure-centric domains, helping close the gap between benchmark performance and real-world deployment readiness. Code is available at https://github.com/Froot-NetSys/NetPress. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 14:04:22 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Zhou",
"Yajie",
""
],
[
"Ruan",
"Jiajun",
""
],
[
"Wang",
"Eric S.",
""
],
[
"Fouladi",
"Sadjad",
""
],
[
"Yan",
"Francis Y.",
""
],
[
"Hsieh",
"Kevin",
""
],
[
"Liu",
"Zaoxing",
""
]
] |
2506.03232 | Nicholas Horton | Oren Tirschwell and Nicholas Jon Horton | Pivoting the paradigm: the role of spreadsheets in K-12 data science | null | null | null | null | stat.OT cs.CY | http://creativecommons.org/licenses/by/4.0/ | Spreadsheet tools are widely accessible to and commonly used by K-12 students and teachers. They have an important role in data collection and organization. Beyond data organization, spreadsheets also make data visible and easy to interact with, facilitating student engagement in data exploration and analysis. Though not suitable for all circumstances, spreadsheets can and do help foster data and computing skills for K-12 students. This paper 1) reviews prior frameworks on K-12 data tools; 2) proposes data-driven learning outcomes that can be accomplished by incorporating spreadsheets into the curriculum; and 3) discusses how spreadsheets can help develop data acumen and computational fluency. We provide example class activities, identify challenges and barriers to adoption, suggest pedagogical approaches to ease the learning curve for instructors and students, and discuss the need for professional development to facilitate deeper use of spreadsheets for data science and STEM disciplines. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 14:22:59 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Tirschwell",
"Oren",
""
],
[
"Horton",
"Nicholas Jon",
""
]
] |
2506.03233 | Andrea Ferrario | Andrea Ferrario | A Trustworthiness-based Metaphysics of Artificial Intelligence Systems | To appear in the proceedings of 2025 ACM Conference on Fairness, Accountability, and Transparency (FAccT '25) | null | null | null | cs.AI cs.CY cs.LG | http://creativecommons.org/licenses/by/4.0/ | Modern AI systems are man-made objects that leverage machine learning to support our lives across a myriad of contexts and applications. Despite extensive epistemological and ethical debates, their metaphysical foundations remain relatively under explored. The orthodox view simply suggests that AI systems, as artifacts, lack well-posed identity and persistence conditions -- their metaphysical kinds are no real kinds. In this work, we challenge this perspective by introducing a theory of metaphysical identity of AI systems. We do so by characterizing their kinds and introducing identity criteria -- formal rules that answer the questions "When are two AI systems the same?" and "When does an AI system persist, despite change?" Building on Carrara and Vermaas' account of fine-grained artifact kinds, we argue that AI trustworthiness provides a lens to understand AI system kinds and formalize the identity of these artifacts by relating their functional requirements to their physical make-ups. The identity criteria of AI systems are determined by their trustworthiness profiles -- the collection of capabilities that the systems must uphold over time throughout their artifact histories, and their effectiveness in maintaining these capabilities. Our approach suggests that the identity and persistence of AI systems is sensitive to the socio-technical context of their design and utilization via their trustworthiness, providing a solid metaphysical foundation to the epistemological, ethical, and legal discussions about these artifacts. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 15:45:46 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Ferrario",
"Andrea",
""
]
] |
2506.03234 | Hongwei Yao | Kaiwen Duan, Hongwei Yao, Yufei Chen, Ziyun Li, Tong Qiao, Zhan Qin, Cong Wang | BadReward: Clean-Label Poisoning of Reward Models in Text-to-Image RLHF | null | null | null | null | cs.LG cs.AI cs.CR | http://creativecommons.org/licenses/by/4.0/ | Reinforcement Learning from Human Feedback (RLHF) is crucial for aligning text-to-image (T2I) models with human preferences. However, RLHF's feedback mechanism also opens new pathways for adversaries. This paper demonstrates the feasibility of hijacking T2I models by poisoning a small fraction of preference data with natural-appearing examples. Specifically, we propose BadReward, a stealthy clean-label poisoning attack targeting the reward model in multi-modal RLHF. BadReward operates by inducing feature collisions between visually contradicted preference data instances, thereby corrupting the reward model and indirectly compromising the T2I model's integrity. Unlike existing alignment poisoning techniques focused on single (text) modality, BadReward is independent of the preference annotation process, enhancing its stealth and practical threat. Extensive experiments on popular T2I models show that BadReward can consistently guide the generation towards improper outputs, such as biased or violent imagery, for targeted concepts. Our findings underscore the amplified threat landscape for RLHF in multi-modal systems, highlighting the urgent need for robust defenses. Disclaimer. This paper contains uncensored toxic content that might be offensive or disturbing to the readers. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:01:04 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Duan",
"Kaiwen",
""
],
[
"Yao",
"Hongwei",
""
],
[
"Chen",
"Yufei",
""
],
[
"Li",
"Ziyun",
""
],
[
"Qiao",
"Tong",
""
],
[
"Qin",
"Zhan",
""
],
[
"Wang",
"Cong",
""
]
] |
2506.03235 | Redouane El Ouardi | Redouane El Ouardi, Amine Bouali, Imad El Bojaddaini, Ahmed Errahmani, and Taoufik Ouali | Model-Independent Reconstruction of $f(T)$ Gravity Using Genetic Algorithms | null | null | null | null | astro-ph.CO | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this paper, we use genetic algorithms, a specific machine learning technique, to achieve a model-independent reconstruction of $f(T)$ gravity. By using $H(z)$ data derived from cosmic chronometers and radial Baryon Acoustic Oscillation method, including the latest Dark Energy Spectroscopic Instrument (DESI) data, we reconstruct the Hubble rate which is the basis parameter for reconstructing $f(T)$ gravity without any assumptions. In this reconstruction process, we use the current value of the Hubble rate, $H_0$, derived by genetic algorithms. The reconstructed $f(T)$ function is consistent with the standard $\Lambda$CDM cosmology within the 1$\sigma$ confidence level across a broad temporal range. The mean $f(T)$ curve, adopting a quadratic form, prompts us to parametrize it using a second degree polynomial. This quadratic deviation from the $\Lambda$CDM scenario is mildly favored by the data. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 16:33:21 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Ouardi",
"Redouane El",
""
],
[
"Bouali",
"Amine",
""
],
[
"Bojaddaini",
"Imad El",
""
],
[
"Errahmani",
"Ahmed",
""
],
[
"Ouali",
"Taoufik",
""
]
] |
2506.03236 | Nico Orce | C. Ngwetsheni, J. N. Orce, P. Navr\'atil, P. E. Garrett, T. Faestermann, A. Bergmaier, M. Frosini, V. Bildstein, B. A. Brown, C. Burbadge, T. Duguet, K. Hady\'nska-Kl\c{e}k, M. Mahgoub, C. V. Mehl, A. Pastore, A. Radich, S. Triambak | Extremely large oblate deformation of the first excited state in $^{12}$C: a new challenge to modern nuclear theory | (11 pages, 6 figures) | null | null | null | nucl-ex nucl-th physics.data-an | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | A Coulomb-excitation study of the high-lying first excited state at 4.439 MeV in the nucleus $^{12}$C has been carried out using the $^{208}$Pb($^{12}$C,$^{12}$C$^*$)$^{208}$Pb$^*$ reaction at 56 MeV and the {\sc Q3D} magnetic spectrograph at the Maier-Leibnitz Laboratorium in Munich. High-statistics achieved with an average beam intensity of approximately 10$^{11}$ ions/s together with state-of-the-art {\it ab initio} calculations of the nuclear dipole polarizability permitted the accurate determination of the spectroscopic quadrupole moment, $Q_{_S}(2_{_1}^+) = +0.076(30)$~eb, in agreement with previous measurements. Combined with previous work, a weighted average of $Q_{_S}(2_{_1}^+) = +0.090(14)$ eb is determined, which includes the re-analysis of a similar experiment by Vermeer and collaborators, $Q_{_S}(2_{_1}^+) = +0.103(20)$~eb. Such a large oblate deformation challenges modern nuclear theory and emphasizes the need of $\alpha$ clustering and associated triaxiality effects for full convergence of $E2$ collective properties. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 17:48:29 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Ngwetsheni",
"C.",
""
],
[
"Orce",
"J. N.",
""
],
[
"Navrátil",
"P.",
""
],
[
"Garrett",
"P. E.",
""
],
[
"Faestermann",
"T.",
""
],
[
"Bergmaier",
"A.",
""
],
[
"Frosini",
"M.",
""
],
[
"Bildstein",
"V.",
""
],
[
"Brown",
"B. A.",
""
],
[
"Burbadge",
"C.",
""
],
[
"Duguet",
"T.",
""
],
[
"Hadyńska-Klȩk",
"K.",
""
],
[
"Mahgoub",
"M.",
""
],
[
"Mehl",
"C. V.",
""
],
[
"Pastore",
"A.",
""
],
[
"Radich",
"A.",
""
],
[
"Triambak",
"S.",
""
]
] |
2506.03237 | Jigang Fan | Jigang Fan, Quanlin Wu, Shengjie Luo, Liwei Wang | UniSite: The First Cross-Structure Dataset and Learning Framework for End-to-End Ligand Binding Site Detection | null | null | null | null | q-bio.QM cs.AI cs.LG q-bio.BM | http://creativecommons.org/licenses/by-nc-sa/4.0/ | The detection of ligand binding sites for proteins is a fundamental step in Structure-Based Drug Design. Despite notable advances in recent years, existing methods, datasets, and evaluation metrics are confronted with several key challenges: (1) current datasets and methods are centered on individual protein-ligand complexes and neglect that diverse binding sites may exist across multiple complexes of the same protein, introducing significant statistical bias; (2) ligand binding site detection is typically modeled as a discontinuous workflow, employing binary segmentation and subsequent clustering algorithms; (3) traditional evaluation metrics do not adequately reflect the actual performance of different binding site prediction methods. To address these issues, we first introduce UniSite-DS, the first UniProt (Unique Protein)-centric ligand binding site dataset, which contains 4.81 times more multi-site data and 2.08 times more overall data compared to the previously most widely used datasets. We then propose UniSite, the first end-to-end ligand binding site detection framework supervised by set prediction loss with bijective matching. In addition, we introduce Average Precision based on Intersection over Union (IoU) as a more accurate evaluation metric for ligand binding site prediction. Extensive experiments on UniSite-DS and several representative benchmark datasets demonstrate that IoU-based Average Precision provides a more accurate reflection of prediction quality, and that UniSite outperforms current state-of-the-art methods in ligand binding site detection. The dataset and codes will be made publicly available at https://github.com/quanlin-wu/unisite. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 17:49:41 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Fan",
"Jigang",
""
],
[
"Wu",
"Quanlin",
""
],
[
"Luo",
"Shengjie",
""
],
[
"Wang",
"Liwei",
""
]
] |
2506.03238 | Ziheng Zhao | Ziheng Zhao, Lisong Dai, Ya Zhang, Yanfeng Wang, Weidi Xie | Rethinking Whole-Body CT Image Interpretation: An Abnormality-Centric Approach | null | null | null | null | eess.IV cs.AI cs.CV | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Automated interpretation of CT images-particularly localizing and describing abnormal findings across multi-plane and whole-body scans-remains a significant challenge in clinical radiology. This work aims to address this challenge through four key contributions: (i) On taxonomy, we collaborate with senior radiologists to propose a comprehensive hierarchical classification system, with 404 representative abnormal findings across all body regions; (ii) On data, we contribute a dataset containing over 14.5K CT images from multiple planes and all human body regions, and meticulously provide grounding annotations for over 19K abnormalities, each linked to the detailed description and cast into the taxonomy; (iii) On model development, we propose OminiAbnorm-CT, which can automatically ground and describe abnormal findings on multi-plane and whole-body CT images based on text queries, while also allowing flexible interaction through visual prompts; (iv) On benchmarks, we establish three representative evaluation tasks based on real clinical scenarios. Through extensive experiments, we show that OminiAbnorm-CT can significantly outperform existing methods on all the tasks and metrics. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 17:57:34 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Zhao",
"Ziheng",
""
],
[
"Dai",
"Lisong",
""
],
[
"Zhang",
"Ya",
""
],
[
"Wang",
"Yanfeng",
""
],
[
"Xie",
"Weidi",
""
]
] |
2506.03239 | Alexey Gorshkov | Alexey V. Gorshkov, Daniel Cohen, Arbel Haim, Amit Rotem, Or Golan, Gihwan Kim, Andreas Butler, Connor T. Hann, Oskar Painter, Fernando G.S.L. Brand\~ao, Alex Retzker | Cavity-mediated cross-cross-resonance gate | 40 pages, 22 figures | null | null | null | quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We propose a cavity-mediated gate between two transmon qubits or other nonlinear superconducting elements. The gate is realized by driving both qubits at a frequency that is near-resonant with the frequency of the cavity. Since both qubits are subject to a cross-resonant drive, we call this gate a cross-cross-resonance gate. In close analogy with gates between trapped-ion qubits, in phase space, the state of the cavity makes a circle whose area depends on the state of the two qubits, realizing a controlled-phase gate. We propose two schemes for canceling the dominant error, which is the dispersive coupling. We also show that this cross-cross-resonance gate allows one to realize simultaneous gates between multiple pairs of qubits coupled via the same metamaterial composed of an array of coupled cavities or other linear mediators. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:00:00 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Gorshkov",
"Alexey V.",
""
],
[
"Cohen",
"Daniel",
""
],
[
"Haim",
"Arbel",
""
],
[
"Rotem",
"Amit",
""
],
[
"Golan",
"Or",
""
],
[
"Kim",
"Gihwan",
""
],
[
"Butler",
"Andreas",
""
],
[
"Hann",
"Connor T.",
""
],
[
"Painter",
"Oskar",
""
],
[
"Brandão",
"Fernando G. S. L.",
""
],
[
"Retzker",
"Alex",
""
]
] |
2506.03240 | Sten Delos | M. Sten Delos | The cusp-halo relation | 25 pages, 24 figures. Associated code at https://github.com/delos/cusp-halo-relation | null | null | null | astro-ph.CO astro-ph.GA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Every halo of collisionless dark matter contains a $\rho = A r^{-1.5}$ density cusp at its center. This prompt cusp is a relic of the halo's earliest moments and has a mass comparable to the cutoff scale in the spectrum of initial density perturbations. In this work, we provide a framework to predict for each halo the coefficient $A$ of its central cusp. We also present a "cusp-NFW" functional form that accurately describes the density profile of a halo with a prompt cusp at its center. Accurate characterization of each halo's central cusp is of particular importance in the study of warm dark matter models, for which the spectral cutoff is on an astrophysically relevant mass scale. To facilitate easy incorporation of prompt cusps into any halo modeling approach, we provide a code package that implements the cusp-halo relation and the cusp-NFW density profile. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:00:00 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Delos",
"M. Sten",
""
]
] |
2506.03241 | Fernando G\'omez-Ruiz | Pablo D\'iez-Valle, Fernando J. G\'omez-Ruiz, Diego Porras, Juan Jos\'e Garc\'ia-Ripoll | Universal Resources for QAOA and Quantum Annealing | 11 pages, 8 Figures | null | null | null | quant-ph cond-mat.stat-mech | http://creativecommons.org/licenses/by/4.0/ | The Quantum Approximate Optimization Algorithm (QAOA) is a variational ansatz that resembles the Trotterized dynamics of a Quantum Annealing (QA) protocol. This work formalizes this connection formally and empirically, showing the angles of a multilayer QAOA circuit converge to universal QA trajectories. Furthermore, the errors in both QAOA circuits and QA paths act as thermal excitations in pseudo-Boltzmann probability distributions whose temperature decreases with the invested resource -- i.e. integrated angles or total time -- and which in QAOA also contain a higher temperature arising from the Trotterization. This also means QAOA and QA are cooling protocols and simulators of partition functions whose target temperature can be tuned by rescaling the universal trajectory. The average cooling power of both methods exhibits favorable algebraic scalings with respect to the target temperature and problem size, whereby in QAOA the coldest temperature is inversely proportional to the number of layers, $T\sim 1/p$, and to the integrated angles -- or integrated interactions in QA. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:00:00 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Díez-Valle",
"Pablo",
""
],
[
"Gómez-Ruiz",
"Fernando J.",
""
],
[
"Porras",
"Diego",
""
],
[
"García-Ripoll",
"Juan José",
""
]
] |
2506.03242 | Shunta Shimizu | Shunta Shimizu, Nobunari Kashikawa, Satoshi Kikuta, Yoshihiro Takeda, Junya Arita, Ryo Emori, Kentaro Koretomo | Age dependence of Ly\alpha\ escape fraction of Ly\alpha\ emitters and their significant role in cosmic reionization | 19 pages, 14 figures, submitted to MNRAS | null | null | null | astro-ph.GA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We study the Ly$\alpha$ escape fraction, $f_{\mathrm{esc}}^{\mathrm{Ly\alpha}}$ of Ly$\alpha$ emitters (LAEs) identified by Subaru/HSC narrowband imaging at $z = 2.2-6.6$, using publicly available deep imaging data from HST and JWST. We perform SED fitting for 127 LAEs at $0.4 - 5.0\, \mathrm{\mu m}$ to estimate their physical properties robustly, and confirm that two distinct LAE populations exist: young LAEs (< 100 Myr) and old LAEs (> 100 Myr). Young LAEs are characterized by burst-like star formation activity and low dust content, significantly differing from Lyman-break galaxies (LBGs) at the same stellar mass, while old LAEs show similar star formation activity to LBGs, yet with lower dust content and more compact morphology in rest-UV/optical than LBGs. The $f_{\mathrm{esc}}^{\mathrm{Ly\alpha}}$ of LAEs is anticorrelated with stellar mass, and this correlation is found to depend on the age of LAEs, such that old LAEs show a weaker anticorrelation than young LAEs, and tend to exhibit higher $f_{\mathrm{esc}}^{\mathrm{Ly\alpha}}$ than young LAEs at a given stellar mass. This implies that Ly$\alpha$ photons escape more efficiently from old LAEs, possibly through the low-density channels of HI and dust created by outflows. The average $f_{\mathrm{esc}}^{\mathrm{Ly\alpha}}$ of young LAEs remains nearly constant at $\sim40$% at $z = 2.2-6.6$, suggesting that the previously observed evolution of global $f_{\mathrm{esc}}^{\mathrm{Ly\alpha}}$ of star-forming galaxies (SFGs) is due to the changes in the LAE fraction among the SFGs. Converting $f_{\mathrm{esc}}^{\mathrm{Ly\alpha}}$ to Lyman continuum escape fraction using empirical relations, we demonstrate that LAEs alone can supply the ionizing photons necessary for reionization at $z\sim6$, causing rapid and late reionization. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:00:00 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Shimizu",
"Shunta",
""
],
[
"Kashikawa",
"Nobunari",
""
],
[
"Kikuta",
"Satoshi",
""
],
[
"Takeda",
"Yoshihiro",
""
],
[
"Arita",
"Junya",
""
],
[
"Emori",
"Ryo",
""
],
[
"Koretomo",
"Kentaro",
""
]
] |
2506.03243 | Marko Shuntov | Marko Shuntov, Hollis B. Akins, Louise Paquereau, Caitlin M. Casey, Olivier Ilbert, Rafael C. Arango-Toro, Henry Joy McCracken, Maximilien Franco, Santosh Harish, Jeyhan S. Kartaltepe, Anton M. Koekemoer, Lilan Yang, Marc Huertas-Company, Edward M. Berman, Jacqueline E. McCleary, Sune Toft, Rapha\"el Gavazzi, Mark J. Achenbach, Emmanuel Bertin, Malte Brinch, Jackie Champagne, Nima Chartab, Nicole E. Drakos, Eiichi Egami, Ryan Endsley, Andreas L. Faisst, Xiaohui Fan, Carter Flayhart, William G. Hartley, Hossein Hatamnia, Ghassem Gozaliasl, Fabrizio Gentile, Iris Jermann, Shuowen Jin, Koki Kakiichi, Ali Ahmad Khostovan, Martin K\"ummel, Clotilde Laigle, Ronaldo Laishram, Erini Lambrides, Daizhong Liu, Jianwei Lyu, Georgios Magdis, Bahram Mobasher, Thibaud Moutard, Alvio Renzini, Brant E. Robertson, Marc Schefer, Diana Scognamiglio, Nick Scoville, Zahra Sattari, David B. Sanders, Sina Taamoli, Benny Trakhtenbrot, Francesco Valentino, Feige Wang, John R. Weaver, Jinyl Yang | COSMOS2025: The COSMOS-Web galaxy catalog of photometry, morphology, redshifts, and physical parameters from JWST, HST, and ground-based imaging | null | null | null | null | astro-ph.GA | http://creativecommons.org/licenses/by-nc-nd/4.0/ | We present COSMOS2025, the COSMOS-Web catalog of photometry, morphology, photometric redshifts and physical parameters for more than 700,000 galaxies in the Cosmic Evolution Survey (COSMOS) field. This catalog is based on our \textit{James Webb Space Telescope} 255\,h COSMOS-Web program, which provides deep near-infrared imaging in four NIRCam (F115W, F150W, F277W, F444W) and one MIRI (F770W) filter over the central $\sim 0.54 {\, \rm deg}^2$ ($\sim 0.2 {\, \rm deg}^2$ for MIRI) in COSMOS. These data are combined with ground- and space-based data to derive photometric measurements of NIRCam-detected sources using both fixed-aperture photometry (on the space-based bands) and a profile-fitting technique on all 37 bands spanning 0.3-8 micron. We provide morphology for all sources from complementary techniques including profile fitting and machine-learning classification. We derive photometric redshifts, physical parameters and non-parametric star formation histories from spectral energy distribution (SED) fitting. The catalog has been extensively validated against previous COSMOS catalogs and other surveys. Photometric redshift accuracy measured using spectroscopically confirmed galaxies out to $z\sim9$ reaches $\sigma_{\rm MAD} = 0.012$ at $m_{\rm F444W}<28$ and remains at $\sigma_{\rm MAD} \lesssim 0.03$ as a function of magnitude, color, and galaxy type. This represents a factor of $\sim 2$ improvement at 26 AB mag compared to COSMOS2020. The catalog is approximately 80\% complete at $\log(M_{\star}/{\rm M}_{\odot}) \sim 9$ at $z \sim 10$ and at $\log(M_{\star}/{\rm M}_{\odot}) \sim 7$ at $z \sim 0.2$, representing a gain of 1\,dex compared to COSMOS2020. COSMOS2025 represents the definitive COSMOS-Web catalog. It is provided with complete documentation, together with redshift probability distributions, and it is ready for scientific exploitation today. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:00:00 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Shuntov",
"Marko",
""
],
[
"Akins",
"Hollis B.",
""
],
[
"Paquereau",
"Louise",
""
],
[
"Casey",
"Caitlin M.",
""
],
[
"Ilbert",
"Olivier",
""
],
[
"Arango-Toro",
"Rafael C.",
""
],
[
"McCracken",
"Henry Joy",
""
],
[
"Franco",
"Maximilien",
""
],
[
"Harish",
"Santosh",
""
],
[
"Kartaltepe",
"Jeyhan S.",
""
],
[
"Koekemoer",
"Anton M.",
""
],
[
"Yang",
"Lilan",
""
],
[
"Huertas-Company",
"Marc",
""
],
[
"Berman",
"Edward M.",
""
],
[
"McCleary",
"Jacqueline E.",
""
],
[
"Toft",
"Sune",
""
],
[
"Gavazzi",
"Raphaël",
""
],
[
"Achenbach",
"Mark J.",
""
],
[
"Bertin",
"Emmanuel",
""
],
[
"Brinch",
"Malte",
""
],
[
"Champagne",
"Jackie",
""
],
[
"Chartab",
"Nima",
""
],
[
"Drakos",
"Nicole E.",
""
],
[
"Egami",
"Eiichi",
""
],
[
"Endsley",
"Ryan",
""
],
[
"Faisst",
"Andreas L.",
""
],
[
"Fan",
"Xiaohui",
""
],
[
"Flayhart",
"Carter",
""
],
[
"Hartley",
"William G.",
""
],
[
"Hatamnia",
"Hossein",
""
],
[
"Gozaliasl",
"Ghassem",
""
],
[
"Gentile",
"Fabrizio",
""
],
[
"Jermann",
"Iris",
""
],
[
"Jin",
"Shuowen",
""
],
[
"Kakiichi",
"Koki",
""
],
[
"Khostovan",
"Ali Ahmad",
""
],
[
"Kümmel",
"Martin",
""
],
[
"Laigle",
"Clotilde",
""
],
[
"Laishram",
"Ronaldo",
""
],
[
"Lambrides",
"Erini",
""
],
[
"Liu",
"Daizhong",
""
],
[
"Lyu",
"Jianwei",
""
],
[
"Magdis",
"Georgios",
""
],
[
"Mobasher",
"Bahram",
""
],
[
"Moutard",
"Thibaud",
""
],
[
"Renzini",
"Alvio",
""
],
[
"Robertson",
"Brant E.",
""
],
[
"Schefer",
"Marc",
""
],
[
"Scognamiglio",
"Diana",
""
],
[
"Scoville",
"Nick",
""
],
[
"Sattari",
"Zahra",
""
],
[
"Sanders",
"David B.",
""
],
[
"Taamoli",
"Sina",
""
],
[
"Trakhtenbrot",
"Benny",
""
],
[
"Valentino",
"Francesco",
""
],
[
"Wang",
"Feige",
""
],
[
"Weaver",
"John R.",
""
],
[
"Yang",
"Jinyl",
""
]
] |
2506.03244 | Fabio Pacucci | Fabio Pacucci, Abraham Loeb | Cosmic Outliers: Low-Spin Halos Explain the Abundance, Compactness, and Redshift Evolution of the Little Red Dots | Submitted to The Astrophysical Journal Letters. 11 pages, 5 figures | null | null | null | astro-ph.GA astro-ph.CO astro-ph.HE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The Little Red Dots (LRDs) are high-redshift galaxies uncovered by JWST, characterized by small effective radii ($R_{\rm eff} \sim 80-300$ pc), number densities intermediate between typical galaxies and quasars, and a redshift distribution peaked at $z \sim 5$. We present a theoretical model in which the LRDs descend from dark matter halos in the extreme low-spin tail of the angular momentum distribution. Within this framework, we explain their three key observational signatures: (i) abundance, (ii) compactness, and (iii) redshift distribution. Our model focuses on observed, not modeled, properties; it is thus independent of whether they are powered primarily by a black hole or stars. We find that the assumption that the prototypical LRD at $z \sim 5$ originates from halos in the lowest $\sim 1\%$ of the spin distribution is sufficient to reproduce both their observed number densities and physical sizes. The redshift evolution of their observability is driven by the interplay between the evolving compact disk fraction and cosmological surface brightness dimming. This effect leads to a well-defined "LRDs Era" at $4<z<8$, during which the LRDs are common and detectable; at $z<4$, they are bright but rare, while at $z>8$, they are common but faint. Finally, we test the predicted redshift trend against observational data, finding excellent agreement. Additional observational support comes from their excess small-scale clustering and spectral signatures of extreme core densities, both of which are expected outcomes of galaxy formation in low-spin halos. These findings suggest that the LRDs are not a fundamentally distinct population but the natural manifestation of galaxies forming in the rarest, lowest angular momentum environments. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:00:00 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Pacucci",
"Fabio",
""
],
[
"Loeb",
"Abraham",
""
]
] |
2506.03245 | Sergio Martin-Alvarez | Sergio Martin-Alvarez, Debora Sijacki, Martin G. Haehnelt, Alice Concas, Yuxuan Yuan, Roberto Maiolino, Risa H. Wechsler, Francisco Rodr\'iguez Montero, Marion Farcy, Mahsa Sanati, Yohan Dubois, Joki Rosdahl, Enrique Lopez-Rodriguez and Susan E. Clark | The Pandora project. II: how non-thermal physics drives bursty star formation and temperate mass-loaded outflows in dwarf galaxies | Submitted to MNRAS. 15 pages of main text, 9 figures, 3 appendices | null | null | null | astro-ph.GA astro-ph.CO | http://creativecommons.org/licenses/by/4.0/ | Dwarf galaxies provide powerful laboratories for studying galaxy formation physics. Their early assembly, shallow gravitational potentials, and bursty, clustered star formation histories make them especially sensitive to the processes that regulate baryons through multi-phase outflows. Using high-resolution, cosmological zoom-in simulations of a dwarf galaxy from \textit{the Pandora suite}, we explore the impact of stellar radiation, magnetic fields, and cosmic ray feedback on star formation, outflows, and metal retention. We find that our purely hydrodynamical model without non-thermal physics - in which supernova feedback is boosted to reproduce realistic stellar mass assembly - drives violent, overly enriched outflows that suppress the metal content of the host galaxy. Including radiation reduces the clustering of star formation and weakens feedback. However, the additional incorporation of cosmic rays produces fast, mass-loaded, multi-phase outflows consisting of both ionized and neutral gas components, in better agreement with observations. These outflows, which entrain a denser, more temperate ISM, exhibit broad metallicity distributions while preserving metals within the galaxy. Furthermore, the star formation history becomes more bursty, in agreement with recent JWST findings. These results highlight the essential role of non-thermal physics in galaxy evolution and the need to incorporate it in future galaxy formation models. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:00:01 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Martin-Alvarez",
"Sergio",
""
],
[
"Sijacki",
"Debora",
""
],
[
"Haehnelt",
"Martin G.",
""
],
[
"Concas",
"Alice",
""
],
[
"Yuan",
"Yuxuan",
""
],
[
"Maiolino",
"Roberto",
""
],
[
"Wechsler",
"Risa H.",
""
],
[
"Montero",
"Francisco Rodríguez",
""
],
[
"Farcy",
"Marion",
""
],
[
"Sanati",
"Mahsa",
""
],
[
"Dubois",
"Yohan",
""
],
[
"Rosdahl",
"Joki",
""
],
[
"Lopez-Rodriguez",
"Enrique",
""
],
[
"Clark",
"Susan E.",
""
]
] |
2506.03246 | Davide Gaiotto | Davide Gaiotto, Suriyah Rajalingam Kannagi and Sergio Sanjurjo | D-branes and the planar limit of Chern-Simons theory I: Link invariants | 106 pages, 33 figures | null | null | null | hep-th math.QA math.SG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We revisit the Holographic duality between $SU(N)_\kappa$ Chern-Simons theory and the A-model Topological String Theory. We develop a strategy to systematically compute the large $N$ saddles for correlation functions of Wilson lines in antisymmetric powers $\Lambda^\bullet \mathbb{C}^N$ of the fundamental representation. The mathematical structures which appear in the calculation match in detail the data of dual A-model D-branes. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:00:01 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Gaiotto",
"Davide",
""
],
[
"Kannagi",
"Suriyah Rajalingam",
""
],
[
"Sanjurjo",
"Sergio",
""
]
] |
2506.03247 | Nikolay Ebel | Nikolay Ebel, Tom Kennedy, Slava Rychkov | Tensor Renormalization Group Meets Computer Assistance | 42+4 pages, 11 figures, 1 table, open source code | null | null | null | cond-mat.stat-mech cond-mat.str-el hep-th math-ph math.MP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Tensor renormalization group, originally devised as a numerical technique, is emerging as a rigorous analytical framework for studying lattice models in statistical physics. Here we introduce a new renormalization map - the 2x1 map - which coarse-grains the lattice anisotropically by a factor of two in one direction followed by a 90-degree rotation. We develop a novel graphical language that translates the action of the 2x1 map into a system of inequalities on tensor components, with rigorous estimates in the Hilbert-Schmidt norm. We define a finite-dimensional "bounding box" called the hat-tensor, and a master function governing its RG flow. Iterating this function numerically, we establish convergence to the high-temperature fixed point for tensors lying within a quantifiable neighborhood. Our main theorem shows that tensors with deviations bounded by 0.02 in 63 orthogonal sectors flow to the fixed point. We also apply the method to specific models - the 2D Ising and XY models - obtaining explicit bounds on their high-temperature phase. This work brings the Tensor RG program closer towards a rigorous, computer-assisted construction of critical fixed points. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:00:01 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Ebel",
"Nikolay",
""
],
[
"Kennedy",
"Tom",
""
],
[
"Rychkov",
"Slava",
""
]
] |
2506.03248 | Zachary Claytor | Zachary R. Claytor and Jamie Tayar | New Rotation Periods from the Kepler Bonus Background Light Curves | Accepted to ApJ. 18 pages, 12 figures, 3 tables. Machine readable versions of Tables 1 and 2 are available as ancillary files | null | null | null | astro-ph.SR astro-ph.EP astro-ph.IM | http://creativecommons.org/licenses/by/4.0/ | The Kepler field hosts the best studied sample of field star rotation periods. However, due to Kepler's large 4" pixels, many of its light curves are at high risk of contamination from background sources. The new Kepler Bonus Background light curves are de-blended using a PSF algorithm, providing light curves of over 400,000 new background sources in addition to over 200,000 re-analyzed Kepler prime targets. These light curves provide the opportunity to search for new rotation periods. Here we apply a convolutional neural network trained on synthetic spot-modulated light curves to regress rotation periods from the Kepler Bonus light curves. We obtained periods for 32,159 total sources, 19,650 of which had previously been measured and 9,811 of which are new periods for both Kepler prime and background sources. Our method also detected 608 pulsation frequencies from asteroseismic oscillations in red giants. We validate our Kepler prime periods against literature values and present the full period sample. We find excellent agreement with previously-known literature periods, validating deep learning as a viable class of period determination methods. Comparing the periods and light curves of foreground-background pairs, we find that as many as 63% of periodic background light curves are still blended with the foreground, highlighting limitations of the de-blending technique. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:00:01 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Claytor",
"Zachary R.",
""
],
[
"Tayar",
"Jamie",
""
]
] |
2506.03249 | Riccardo Gonzo | Francesco Alessio, Riccardo Gonzo, Canxin Shi | Dirac brackets for classical radiative observables | 10 pages + ancillary file with 2PM spinning observables | null | null | null | hep-th gr-qc | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We introduce a new coherent state expansion of the exponential representation of the S-matrix for the classical gravitational two-body problem. By combining the Kosower-Maybee-O'Connell (KMOC) formalism with the Dirac bracket structure emerging in the classical limit, we derive compact and gauge-invariant expressions for scattering observables in the presence of radiation. This causal formulation bypasses the calculation of KMOC cuts and provides a direct link between observables and a minimal set of classical matrix elements extracted from amplitudes. We illustrate our method with several examples, including the impulse, spin kick, angular momentum, waveform and the related radiative fluxes. Finally, using our formalism we evaluate for the first time the spin kick and the change in angular momentum of each particle up to $\mathcal{O}(G^2 s_1^{j_1} s_2^{j_2})$ with $j_1+ j_2 \leq 11$. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:00:02 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Alessio",
"Francesco",
""
],
[
"Gonzo",
"Riccardo",
""
],
[
"Shi",
"Canxin",
""
]
] |
2506.03250 | Kristian M{\ae}land | Kristian M{\ae}land, Masoud Bahari, Bj\"orn Trauzettel | Phonon-Mediated Intrinsic Topological Superconductivity in Fermi Arcs | 6+10 pages, 3+4 figures | null | null | null | cond-mat.supr-con cond-mat.mtrl-sci | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We propose that phonons can mediate topological superconductivity on the surface of Weyl semimetals. Weyl semimetals are gapless topological materials with nondegenerate zero energy surface states known as Fermi arcs. We derive the phonon spectrum and electron-phonon coupling in an effective model of a Weyl semimetal and apply weak-coupling Bardeen-Cooper-Schrieffer theory of superconductivity. In a slab geometry, we demonstrate that surface superconductivity dominates over bulk superconductivity in a wide range of chemical potentials. The superconducting gap function realizes spinless chiral $p$-wave superconductivity in the Fermi arcs, leading to Majorana bound states in the core of vortices. Furthermore, we show a suppression of the absolute value of the gap in the center of the arc, which is not captured by a local Hubbard attraction. The suppression is due to the nonlocal origin of electron-phonon coupling, leading to a layer dependence which has important consequences for topological surface states. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:00:02 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Mæland",
"Kristian",
""
],
[
"Bahari",
"Masoud",
""
],
[
"Trauzettel",
"Björn",
""
]
] |
2506.03251 | Francesco Goto Goto | A. Jabed, F. Goto, B. Frimpong, D. Armanno, A. Longa, M. Michiardi, A. Damascelli, P. Hofmann, G. Jargot, H. Ibrahim, F. L\'egar\'e, N. Gauthier, S. Beaulieu, F. Boschini | Control of intervalley scattering in Bi$_2$Te$_3$ via temperature-dependent band renormalization | null | null | null | null | cond-mat.str-el cond-mat.mtrl-sci | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The control of out-of-equilibrium electron dynamics in topological insulators is essential to unlock their potential in next-generation quantum technologies. However, the role of temperature on the renormalization of the electronic band structure and, consequently, on electron scattering processes is still elusive. Here, using high-resolution time- and angle-resolved photoemission spectroscopy (TR-ARPES), we show that even a modest ($\sim$15 meV) renormalization of the conduction band of Bi$_2$Te$_3$ can critically affect bulk and surface electron scattering processes. Supported by a kinetic Monte Carlo toy-model, we show that temperature-induced changes in the bulk band structure modulate the intervalley electron-phonon scattering rate, reshaping the out-of-equilibrium response. This work establishes temperature as an effective control knob for engineering scattering pathways in topological insulators. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:00:02 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Jabed",
"A.",
""
],
[
"Goto",
"F.",
""
],
[
"Frimpong",
"B.",
""
],
[
"Armanno",
"D.",
""
],
[
"Longa",
"A.",
""
],
[
"Michiardi",
"M.",
""
],
[
"Damascelli",
"A.",
""
],
[
"Hofmann",
"P.",
""
],
[
"Jargot",
"G.",
""
],
[
"Ibrahim",
"H.",
""
],
[
"Légaré",
"F.",
""
],
[
"Gauthier",
"N.",
""
],
[
"Beaulieu",
"S.",
""
],
[
"Boschini",
"F.",
""
]
] |
2506.03252 | Anthony Massidda | Roberta Angius and Sergio Luigi Cacciatori and Anthony Massidda | Wall crossing structure from quantum phenomena to Feynman Integrals | 41 pages + Appendix | null | null | null | hep-th | http://creativecommons.org/licenses/by/4.0/ | A growing body of evidence suggests that the complexity of Feynman integrals is best understood through geometry. Recent mathematical developments [Kontsevich and Soibelman, arXiv:2402.07343] have illuminated the role of exponential integrals as periods of twisted de Rham cocycles over Betti cycles, providing a structured approach to tackle this problem in many situations. In this paper, we apply these concepts to show how families of physically relevant integrals, ranging from exponentials to logarithmic multivalued functions, can be recast as twisted periods of differential forms over homology cycles. In the case of holomorphic exponents, we provide explicit decompositions as thimble expansions and reveal a geometric wall-crossing structure behind the analytic continuation in parameters. We then show that the generalization to multivalued functions provides the right framework to describe Feynman integrals in the Baikov representation, where the multivaluedness is governed by the logarithm of the Baikov polynomial. In this context, the thimble decomposition aligns with the decomposition into Master Integrals. We highlight how the wall-crossing structure allows for a sharp count of independent Master Integrals (or periods), circumventing complications arising from Stokes phenomena. Additionally, we study the large-parameter expansions of these integrals, whose coefficients correspond to periods of standard (co-)homology associated with families of algebraic varieties, and which reveal the dominant basis elements in different sectors of the wall crossing structure. This unifies perturbative expansions and geometric representation theory under a single cohomological framework. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:00:02 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Angius",
"Roberta",
""
],
[
"Cacciatori",
"Sergio Luigi",
""
],
[
"Massidda",
"Anthony",
""
]
] |
2506.03253 | Alberto Castellano | Christian Aoufia, Alberto Castellano, Luis Ib\'a\~nez | Laplacians in Various Dimensions and the Swampland | 50 pages, 5 figures, 2 tables | null | null | IFT-UAM/CSIC-25-61 | hep-th | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | The species cutoff is a moduli-dependent quantity signaling the onset of quantum gravitational phenomena, whose form can be oftentimes determined from higher-derivative and higher-curvature corrections within low-energy gravitational EFTs. In this work, we point out that these Wilson coefficients are eigenfunctions of an appropriate second-order elliptic operator defined over moduli space in theories with more than four supercharges. This was already known to be the case for the leading $\mathcal{R}^4$-correction to the two-derivative (bosonic) action of maximal supergravity in $d\leq 10$. Here, we reconsider this fact from the Swampland point of view and show how, in $d=10,9,8$, solving a Laplace equation imposes non-trivial restrictions on the species hull vectors. We further argue that this property is also satisfied in settings with less supersymmetry. In particular, we focus on the $\mathcal{R}^4$-operator in minimal supergravity theories in $d=10,9$, and on the leading $\mathcal{R}^2$-term in setups with 8 supercharges in $d=6,5,4$. Finally, we provide a symmetry-based criterion for determining when the relevant elliptic operator should be the Laplacian. A bottom-up rationale for this constraint remains to be fully understood, and we conclude by outlining some compelling possibilities. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:00:02 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Aoufia",
"Christian",
""
],
[
"Castellano",
"Alberto",
""
],
[
"Ibáñez",
"Luis",
""
]
] |
2506.03254 | Daniel Burgarth | Daniel Burgarth, Paolo Facchi | Orbital angular momentum can take non-integer values in a closed universe | 5 pages, 5 figures | null | null | null | quant-ph gr-qc math-ph math.MP | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We show that the spectrum of orbital angular momentum in quantum mechanics consists of two parts when the underlying space has periodic boundaries. While the first part consists of the usual textbook integer quantized values, the second is a continuous band arising from regions at the `edge' of space with respect to the center of rotation. The spectrum thus contains not only half-integer values, previously thought impossible for orbital angular momentum, but even irrational ones. Remarkably, this effect is independent of the size of space. While these spectral components remain undetectable in laboratory experiments, they could still produce observable effects on cosmological scales, for instance in the Cosmic Microwave Background Radiation. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:00:04 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Burgarth",
"Daniel",
""
],
[
"Facchi",
"Paolo",
""
]
] |
2506.03255 | Oleksandr Tomalak | Oleksandr Tomalak | Reducing Hadronic Uncertainty in Low-Energy Neutral-Current Processes | 6 pages, 2 figures | null | null | null | hep-ph hep-ex nucl-ex nucl-th | http://creativecommons.org/licenses/by/4.0/ | We analyze the dominant hadronic uncertainty in low-energy neutral-current (anti)neutrino scattering, estimated at the $3$-$4$ per mille level. This uncertainty arises from limited knowledge of the charge-isospin correlation function of quark currents. We study the charge-charge and charge-isospin correlators within $\mathrm{SU}(2)$ and $\mathrm{SU}(3)$ chiral perturbation theory (ChPT). In $\mathrm{SU}(2)$ ChPT, the two correlators are identical to all orders in the chiral and electromagnetic expansions. We further perform a one-loop $\mathrm{SU}(3)$ ChPT calculation and discuss the relevant counterterms. Our findings reduce the hadronic uncertainty in neutral-current processes such as (anti)neutrino-electron and coherent elastic (anti)neutrino-nucleus scattering to the sub-permille level, $\lesssim 1$ per mille. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:00:05 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Tomalak",
"Oleksandr",
""
]
] |
2506.03256 | Maximilien Franco | Maximilien Franco, Caitlin M. Casey, Anton M. Koekemoer, Daizhong Liu, Micaela B. Bagley, Henry Joy McCracken, Jeyhan S. Kartaltepe, Hollis B. Akins, Olivier Ilbert, Marko Shuntov, Santosh Harish, Brant E. Robertson, Rafael C. Arango-Toro, Andrew J. Battisti, Nima Chartab, Nicole E. Drakos, Andreas L. Faisst, Carter Flayhart, Ghassem Gozaliasl, Michaela Hirschmann, Richard Massey, Jason Rhodes, Zahra Sattari, Diana Scognamiglio, John R. Weaver, Lilan Yang, Jorge A. Zavala, Edward M. Berman, Fabrizio Gentile, Steven Gillman, Arianna S. Long, Georgios Magdis, Jacqueline E. McCleary, Jed McKinney, Bahram Mobasher, Louise Paquereau, Armin Rest, David B. Sanders, Sune Toft, Si-Yue Yu | COSMOS-Web: Comprehensive Data Reduction for Wide-Area JWST NIRCam Imaging | 22 pages, 11 figures, Submitted to ApJ | null | null | null | astro-ph.IM astro-ph.GA | http://creativecommons.org/licenses/by-nc-nd/4.0/ | We present the data reduction methodology used for the COSMOS-Web survey JWST NIRCam data. Covering 0.54 deg^2 with four broadband filters (F115W, F150W, F277W, F444W) and a total exposure time of approximately 270 hours, COSMOS-Web represents the largest contiguous field surveyed during JWST Cycle 1, posing unique data reduction challenges due to its extensive scale. By combining the official JWST Calibration Pipeline with custom improvements for noise removal, background subtraction, and astrometric alignment, we achieve high fidelity science-ready mosaics. We detail the systematic approach employed in the three stages of the JWST Calibration Pipeline. The data, collected in three epochs from January 2023 to January 2024, encompass 152 visits and have been processed into 20 mosaic tiles to optimize computational efficiency and data processing. The final data products achieve 5 sigma depths of 26.7-28.3 AB mag in 0.15" apertures. The processed and calibrated datasets are made available to the public. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:00:05 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Franco",
"Maximilien",
""
],
[
"Casey",
"Caitlin M.",
""
],
[
"Koekemoer",
"Anton M.",
""
],
[
"Liu",
"Daizhong",
""
],
[
"Bagley",
"Micaela B.",
""
],
[
"McCracken",
"Henry Joy",
""
],
[
"Kartaltepe",
"Jeyhan S.",
""
],
[
"Akins",
"Hollis B.",
""
],
[
"Ilbert",
"Olivier",
""
],
[
"Shuntov",
"Marko",
""
],
[
"Harish",
"Santosh",
""
],
[
"Robertson",
"Brant E.",
""
],
[
"Arango-Toro",
"Rafael C.",
""
],
[
"Battisti",
"Andrew J.",
""
],
[
"Chartab",
"Nima",
""
],
[
"Drakos",
"Nicole E.",
""
],
[
"Faisst",
"Andreas L.",
""
],
[
"Flayhart",
"Carter",
""
],
[
"Gozaliasl",
"Ghassem",
""
],
[
"Hirschmann",
"Michaela",
""
],
[
"Massey",
"Richard",
""
],
[
"Rhodes",
"Jason",
""
],
[
"Sattari",
"Zahra",
""
],
[
"Scognamiglio",
"Diana",
""
],
[
"Weaver",
"John R.",
""
],
[
"Yang",
"Lilan",
""
],
[
"Zavala",
"Jorge A.",
""
],
[
"Berman",
"Edward M.",
""
],
[
"Gentile",
"Fabrizio",
""
],
[
"Gillman",
"Steven",
""
],
[
"Long",
"Arianna S.",
""
],
[
"Magdis",
"Georgios",
""
],
[
"McCleary",
"Jacqueline E.",
""
],
[
"McKinney",
"Jed",
""
],
[
"Mobasher",
"Bahram",
""
],
[
"Paquereau",
"Louise",
""
],
[
"Rest",
"Armin",
""
],
[
"Sanders",
"David B.",
""
],
[
"Toft",
"Sune",
""
],
[
"Yu",
"Si-Yue",
""
]
] |
2506.03257 | Gustavo Alves G.H.S. Alves | Gustavo H. S. Alves, Celso C. Nishi | Effective description of Nelson-Barr models and the theta parameter | 42 pages, 7 figures | null | null | null | hep-ph | http://creativecommons.org/licenses/by/4.0/ | Nelson-Barr theories solve the strong CP problem with CP spontaneously broken at \Lambda_{CP} and transmitted to the CP conserving version of the SM through vector-like quarks (VLQs). For an arbitrary number of CP breaking scalars and VLQs, we perform the full one-loop matching calculations at \Lambda_{CP} up to dimension five operators and the relevant matching calculations at the VLQ scale relevant to tracking the contributions to the parameter \bar{\theta}. In the EFT after the CP breaking scalars have been integrated out, we confirm that the running of \bar{\theta} at one-loop is induced by CP violating dimension five operators and similarly by dimension six operator after matching to the SMEFT. In the latter, an additional contribution enters in the matching to \bar{\theta} at tree-level due to a dimension five operator. Analyzing the experimental constraint from \bar{\theta} for generic settings, we see that a large separation between \Lambda_{CP} and the VLQ scale already suppresses the one-loop contribution to \bar{\theta} sufficiently. We confirm that a simple extension based on a nonconventional CP has a vanishing one-loop contribution to \bar{\theta}. Part of our results can be also applied to generic VLQ extensions of the SM. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:00:07 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Alves",
"Gustavo H. S.",
""
],
[
"Nishi",
"Celso C.",
""
]
] |
2506.03258 | April Cheng | April Qiu Cheng, Shion Elizabeth Andrew, Haochen Wang, and Kiyoshi W. Masui | Exploring selection biases in FRB dispersion-galaxy cross-correlations with magnetohydrodynamical simulations | 15 pages, 10 figures, comments are welcome | null | null | null | astro-ph.CO astro-ph.HE | http://creativecommons.org/licenses/by/4.0/ | The dispersion of fast radio bursts (FRBs) in conjunction with their redshifts can be used as powerful probes of the distribution of extragalactic plasma, and with a large enough sample, the free-electron--galaxy power spectrum $P_{eg}$ can be measured by cross-correlating FRB dispersions with galaxy positions. However, a precise measurement of $P_{eg}$ requires a careful investigation of the selection effects -- the fact that the probability of both observing the FRB dispersion measure and obtaining a host galaxy redshift depends on observed properties of the FRB and its host. We use ray tracing simulations with IllustrisTNG300-1 to investigate the impact of expected observational selection effects on FRB dispersion-galaxy angular cross-correlations with a sample of 3000 FRBs at redshift range of $0.3 \leq z \leq 0.4$ . Our results show that cross-correlations with such an FRB sample are robust to properties of the FRB host galaxy: this includes DM contributions from the FRB host and optical followup selection effects biased against FRBs with dim galaxy hosts. We also find that such cross-correlations are robust to DM dependent and scattering selection effects specific to the CHIME/FRB survey. However, a DM dependent selection effect that cuts off the 10\% most dispersed FRB at a fixed redshift shell can bias the amplitude of the cross-correlation signal by over 50\% at angular scales of $\sim 0.1^\circ$ (corresponding to Mpc physical scales). Our findings highlight the importance of both measuring and accounting for selection effects present in existing FRB surveys as well as mitigating DM dependent selection effects in the design of upcoming FRB surveys aiming to use FRBs as probes for large-scale structure. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:00:08 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Cheng",
"April Qiu",
""
],
[
"Andrew",
"Shion Elizabeth",
""
],
[
"Wang",
"Haochen",
""
],
[
"Masui",
"Kiyoshi W.",
""
]
] |
2506.03259 | Michael Garcia-Alcoser | Michael E. Garcia-Alcoser, Mobina GhojoghNejad, Fakrul Islam Tushar, David Kim, Kyle J. Lafata, Geoffrey D. Rubin and Joseph Y. Lo | Evaluating Large Language Models for Zero-Shot Disease Labeling in CT Radiology Reports Across Organ Systems | 23 pages, 10 figures, to be submitted in Radiology: Artificial Intelligence | null | null | null | cs.CL | http://creativecommons.org/licenses/by-nc-sa/4.0/ | Purpose: This study aims to evaluate the effectiveness of large language models (LLMs) in automating disease annotation of CT radiology reports. We compare a rule-based algorithm (RBA), RadBERT, and three lightweight open-weight LLMs for multi-disease labeling of chest, abdomen, and pelvis (CAP) CT reports.
Materials and Methods: This retrospective study analyzed 40,833 CT reports from 29,540 patients, with 1,789 CAP reports manually annotated across three organ systems. External validation was conducted using the CT-RATE dataset. Three open-weight LLMs were tested with zero-shot prompting. Performance was evaluated using Cohen's Kappa and micro/macro-averaged F1 scores.
Results: In 12,197 Duke CAP reports from 8,854 patients, Llama-3.1 8B and Gemma-3 27B showed the highest agreement ($\kappa$ median: 0.87). On the manually annotated set, Gemma-3 27B achieved the top macro-F1 (0.82), followed by Llama-3.1 8B (0.79), while the RBA scored lowest (0.64). On the CT-RATE dataset (lungs/pleura only), Llama-3.1 8B performed best (0.91), with Gemma-3 27B close behind (0.89). Performance differences were mainly due to differing labeling practices, especially for lung atelectasis.
Conclusion: Lightweight LLMs outperform rule-based methods for CT report annotation and generalize across organ systems with zero-shot prompting. However, binary labels alone cannot capture the full nuance of report language. LLMs can provide a flexible, efficient solution aligned with clinical judgment and user needs. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:00:08 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Garcia-Alcoser",
"Michael E.",
""
],
[
"GhojoghNejad",
"Mobina",
""
],
[
"Tushar",
"Fakrul Islam",
""
],
[
"Kim",
"David",
""
],
[
"Lafata",
"Kyle J.",
""
],
[
"Rubin",
"Geoffrey D.",
""
],
[
"Lo",
"Joseph Y.",
""
]
] |
2506.03260 | Jeonghyeon Song | Soojin Lee, Dongjoo Kim, Jin-Hwan Cho, Jinheung Kim, and Jeonghyeon Song | Multi-step Strong First-Order Electroweak Phase Transitions in the Inverted Type-I 2HDM: Parameter Space, Gravitational Waves, and Collider Phenomenology | 54 pages with 15 figures and 2 tables | null | null | KIAS-P25030 | hep-ph hep-ex | http://creativecommons.org/licenses/by/4.0/ | We investigate the nature of the electroweak phase transition (EWPT) in the Type-I two-Higgs-doublet model under the inverted Higgs scenario, where the observed 125 GeV Higgs boson corresponds to the heavier \textit{CP}-even scalar. We systematically map the parameter space viable for strong first-order EWPTs (SFOEWPTs) under current theoretical and experimental constraints, revealing that such transitions can occur through one-step, two-step, and rarely three-step pathways with distinct phenomenological characteristics. One-step transitions confine $m_{H^\pm}$ to $[295, 441]{\;{\rm GeV}}$ and $\tan\beta$ to $[4.2, 8.8]$, whereas two-step transitions allow broader ranges: $m_{H^\pm} \in [100, 350]{\;{\rm GeV}}$ and $\tan\beta \in [2.5, 45.4]$. Notably, SFOEWPTs involving negative $\sin (\beta-\alpha)$ appear almost exclusively in the one-step case, offering a potential discriminator. This study also provides the first detailed calculation of LISA gravitational wave (GW) signal-to-noise ratios (SNRs) for multi-step SFOEWPTs in this scenario, finding that detectable GW signals (SNR $>10$) predominantly arise from two-step transitions. Furthermore, we show that while a correlation between the vacuum uplifting measure $\Delta {F}_0$ and EWPT strength $\xi_c$ is affirmed for one-step transitions, this correlation breaks down for multi-step pathways. Finally, we conduct the first dedicated collider analysis at the 1.5 TeV CLIC for representative benchmark points supporting an SFOEWPT, identifying $e^+ e^- \to H^+H^- \to W^+ W^- hh$ as a promising discovery channel. We emphasize the enhancement of $\text{Br}(h \to \gamma\gamma)$ for negative $\sin(\beta-\alpha)$ and propose two golden final states, $W^+ W^- b\bar{b}\tau^+\tau^-$ and $W^+ W^- b\bar{b}\gamma\gamma$, with negligible backgrounds. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:00:09 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Lee",
"Soojin",
""
],
[
"Kim",
"Dongjoo",
""
],
[
"Cho",
"Jin-Hwan",
""
],
[
"Kim",
"Jinheung",
""
],
[
"Song",
"Jeonghyeon",
""
]
] |
2506.03261 | Maria Daghofer | Maria Daghofer and Krzysztof Wohlfeld and Jeroen van den Brink | Altermagnetic polarons | 9 pages and 8 figures including supplemental material | null | null | null | cond-mat.str-el | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | While a spin-dependent band splitting is one of the characteristic features of altermagnets, the conventional band picture itself breaks down in the many altermagnets that are correlated Mott materials. We employ two numerical many-body methods, the self-consistent Born approximation and variational cluster approach, to explore this strongly correlated regime and investigate hole motion in Mott altermagnets. Our results reveal that spin-dependent spectral-weight transfer is the dominant signature of Mott altermagnetism. This pronounced spin-momentum locking of the quasiparticle spectral weight arises from the formation of altermagnetic polarons, whose dynamics are governed by the interplay between free hole motion and the coupling of the hole to magnon excitations in the altermagnet. We demonstrate this effect by calculating ARPES spectra for two canonical altermagnetic systems: the checkerboard $J$-$J'$ model and the Kugel-Khomskii spin-orbital altermagnet based on cubic vanadates $R$VO$_3$ ($R$=La, Pr, Nd, Y). | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:00:09 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Daghofer",
"Maria",
""
],
[
"Wohlfeld",
"Krzysztof",
""
],
[
"Brink",
"Jeroen van den",
""
]
] |
2506.03262 | David Saykin | David R. Saykin, Martin Gonzalez, Jennifer Fowlie, Steven A. Kivelson, Harold Hwang, Aharon Kapitulnik | Spin-glass state in nickelate superconductors | 10 pages, 11 figures | null | null | null | cond-mat.str-el cond-mat.mtrl-sci cond-mat.supr-con | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Magneto-optical measurements in La${}_{0.8}$Sr${}_{0.2}$NiO${}_2$ and Nd${}_{0.825}$Sr${}_{0.175}$NiO${}_2$ reveal an intriguing new facet of infinite-layer nickelate superconductors: the onset of spin-glass behavior at a temperature far exceeding the superconducting critical temperature $T_c$. This discovery sharply contrasts with copper oxide superconductors, where magnetism and superconductivity remain largely exclusive. Moreover, the magnitude and onset temperature of the polar Kerr effect in Nd${}_{0.825}$Sr${}_{0.175}$NiO${}_2$ fabricated on SrTiO${}_3$ and (LaAlO${}_3$)${}_{0.3}$(Sr${}_2$TaAlO${}_6$)${}_{0.7}$ substrates differ dramatically, while $T_c$ does not. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:00:09 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Saykin",
"David R.",
""
],
[
"Gonzalez",
"Martin",
""
],
[
"Fowlie",
"Jennifer",
""
],
[
"Kivelson",
"Steven A.",
""
],
[
"Hwang",
"Harold",
""
],
[
"Kapitulnik",
"Aharon",
""
]
] |
2506.03263 | Lorenzo Lanzini | Lorenzo Lanzini, Purnendu Das, Michael Knap | Theory of Angle Resolved Photoemission Spectroscopy of Altermagnetic Mott Insulators | 5+5 pages, 4+5 figures | null | null | null | cond-mat.str-el cond-mat.stat-mech quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Altermagnetism has emerged as an unconventional form of collinear magnetism with spatial rotational symmetries, that give rise to strongly spin-split bands despite of an underlying fully-compensated antiferromagnetic order. Here, we develop a theory for the Angle Resolved Photoemission Spectroscopy (ARPES) response of altermagnetic Mott insulators. Crucially, the spectrum does not simply reflect the non-interacting band structure, but instead a magnetic polaron is formed at low energies, that can be interpreted as a spinon-holon bound state. We develop a spinon-holon parton theory and predict a renormalized bandwidth that we confirm by tensor network simulations. We analyze the characteristic spin-split spectrum and identify a spin-dependent spectral weight of the magnetic polaron, resulting from the altermagnetic symmetry. Our work paves the way for a systematic study of doping effects and correlation phenomena in altermagnetic Mott insulators. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:00:14 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Lanzini",
"Lorenzo",
""
],
[
"Das",
"Purnendu",
""
],
[
"Knap",
"Michael",
""
]
] |
2506.03264 | Lucian Harland-Lang | L. A. Harland-Lang | Modelling Coincident Particle Production in Ultraperipheral Heavy Ion Collisions | 15 pages, 6 figures | null | null | null | hep-ph hep-ex nucl-ex nucl-th | http://creativecommons.org/licenses/by/4.0/ | In this paper, we present an analysis of coincident particle production in ultraperipheral heavy ion collisions. In particular, we present the first detailed and differential predictions for coincident $\rho^0$ meson production in association with muon pair production, motivated by the recent ATLAS measurement of this process. These are found to describe the data well, including the dependence of the coincidence fraction on the ZDC selection and/or other kinematic constraints on the muons. Differential predictions at the level of various kinematic variables are presented, and the calculation is made publicly available in the SuperChic MC generator. We also present general results for coincident two photon-initiated production, focussing on muon and electron pair production; while the former is rather suppressed, the latter will be ubiquitous near threshold. The impact of coincident production on exclusivity vetos in ultraperipheral measurements is discussed. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:00:15 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Harland-Lang",
"L. A.",
""
]
] |
2506.03265 | Sugata Kaviraj | S. Kaviraj, B. Bichang'a, I. Lazar, A. E. Watkins, G. Martin and R. A. Jackson | Nearby dwarf galaxies with extreme star formation rates: a window into dwarf-galaxy evolution in the early Universe | 9 pages, 6 figures, published in MNRAS | 2025, MNRAS, 540, 594 | 10.1093/mnras/staf716 | null | astro-ph.GA | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We study a sample of nearby (z~0.2) low-luminosity dwarf (10^7 MSun < M* < 10^8 MSun) galaxies which have extreme (0.1 - 3 MSun/yr) star formation rates (SFRs) for this mass regime, making them plausible analogues of dwarfs at z~5.5. We compare the properties of these analogues to control samples of 'normal' dwarfs, which reside on the star formation main sequence (SFMS) at z~0.2 and are matched in their stellar mass and redshift distributions to the analogue population. The analogue and normal populations do not show differences, either in their half-light radii or the projected distances to nodes, filaments and massive galaxies. This suggests that the comparatively extreme SFRs in the analogues are not driven by them being anomalously compact or because they reside in specific environments which might provide a larger gas supply. However, the fractions of interacting galaxies and those that have early-type morphology are significantly elevated (by factors of ~5.6 and ~9 respectively) in the analogues compared to the normal population. Extrapolation of the redshift evolution of the star formation main sequence into our mass range of interest appears to underestimate the SFRs of observed dwarfs at z~5.5. Since current SFMS measurements remain dominated by low and intermediate redshift data (especially at low stellar masses), our study suggests that this underestimation may be driven by interactions (which are more frequent at earlier epochs) boosting the SFRs in the high-redshift dwarf population. Our results are consistent with a picture where higher gas availability, augmented by interactions, drives much of the stellar mass assembly of dwarf galaxies in the early Universe. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:00:17 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Kaviraj",
"S.",
""
],
[
"Bichang'a",
"B.",
""
],
[
"Lazar",
"I.",
""
],
[
"Watkins",
"A. E.",
""
],
[
"Martin",
"G.",
""
],
[
"Jackson",
"R. A.",
""
]
] |
2506.03266 | Ethan Lake | Ethan Lake | Fast offline decoding with local message-passing automata | 47+18 pages, 19 figures | null | null | null | quant-ph cond-mat.stat-mech cond-mat.str-el nlin.CG | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | We present a local offline decoder for topological codes that operates according to a parallelized message-passing framework. The decoder works by passing messages between anyons, with the contents of received messages used to move nearby anyons towards one another. We prove the existence of a threshold, and show that in a system of linear size $L$, decoding terminates with an $O((\log L)^\eta)$ average-case runtime, where $\eta$ is a small constant. For i.i.d Pauli noise, our decoder has $\eta=1$ and a threshold at a noise strength of $p_c\approx 7.3\%$. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:00:26 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Lake",
"Ethan",
""
]
] |
2506.03267 | Shahbaz Rezaei | Shahbaz Rezaei, Avishai Halev, Xin Liu | On the Necessity of Multi-Domain Explanation: An Uncertainty Principle Approach for Deep Time Series Models | null | null | null | null | cs.LG cs.HC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | A prevailing approach to explain time series models is to generate attribution in time domain. A recent development in time series XAI is the concept of explanation spaces, where any model trained in the time domain can be interpreted with any existing XAI method in alternative domains, such as frequency. The prevailing approach is to present XAI attributions either in the time domain or in the domain where the attribution is most sparse. In this paper, we demonstrate that in certain cases, XAI methods can generate attributions that highlight fundamentally different features in the time and frequency domains that are not direct counterparts of one another. This suggests that both domains' attributions should be presented to achieve a more comprehensive interpretation. Thus it shows the necessity of multi-domain explanation. To quantify when such cases arise, we introduce the uncertainty principle (UP), originally developed in quantum mechanics and later studied in harmonic analysis and signal processing, to the XAI literature. This principle establishes a lower bound on how much a signal can be simultaneously localized in both the time and frequency domains. By leveraging this concept, we assess whether attributions in the time and frequency domains violate this bound, indicating that they emphasize distinct features. In other words, UP provides a sufficient condition that the time and frequency domain explanations do not match and, hence, should be both presented to the end user. We validate the effectiveness of this approach across various deep learning models, XAI methods, and a wide range of classification and forecasting datasets. The frequent occurrence of UP violations across various datasets and XAI methods highlights the limitations of existing approaches that focus solely on time-domain explanations. This underscores the need for multi-domain explanations as a new paradigm. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:00:28 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Rezaei",
"Shahbaz",
""
],
[
"Halev",
"Avishai",
""
],
[
"Liu",
"Xin",
""
]
] |
2506.03268 | Cristiano Chesi | Cristiano Chesi | A conclusive remark on linguistic theorizing and language modeling | null | null | null | null | cs.CL | http://creativecommons.org/licenses/by/4.0/ | This is the final remark on the replies received to my target paper in the Italian Journal of Linguistics | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:00:29 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Chesi",
"Cristiano",
""
]
] |
2506.03269 | Jae Hyeok Chang | Jae Hyeok Chang, Chang Sub Shin, and James Unwin | Parametric Coincidence in the Baryon to Dark Matter Ratio from Affleck-Dine Baryogenesis and UV Freeze-in Dark Matter | 13 pages + appendices and references, 3 figures | null | null | FERMILAB-PUB-25-0376-T-V | hep-ph | http://creativecommons.org/licenses/by/4.0/ | We highlight that the observed concurrence between the baryon and dark matter relic densities can be explained via a parametric coincidence between two distinct production mechanisms: Affleck-Dine baryogenesis and dark matter UV freeze-in. In the Affleck-Dine mechanism, the baryon asymmetry is naturally proportional to the inflationary reheating temperature $T_{\rm rh}$, which also plays a critical role in setting the relic abundance of UV freeze-in dark matter. Since Affleck-Dine baryogenesis requires flat directions in the potential, the framework is inherently supersymmetric, offering compelling UV freeze-in dark matter candidates such as the gravitino. We outline scenarios in which $T_{\rm rh}$ simultaneously determines both relic abundances, resulting in a baryon-to-dark matter ratio of order unity that is largely insensitive to $T_{\rm rh}$. We also discuss the conditions required to avoid Q-ball formation or dark matter production by other mechanisms, such as NLSP decays, to preserve the parametric coincidence between baryon and dark matter abundances. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:00:30 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Chang",
"Jae Hyeok",
""
],
[
"Shin",
"Chang Sub",
""
],
[
"Unwin",
"James",
""
]
] |
2506.03270 | Keisuke Shirai | Jeremy Siburian, Keisuke Shirai, Cristian C. Beltran-Hernandez, Masashi Hamaya, Michael G\"orner, Atsushi Hashimoto | Grounded Vision-Language Interpreter for Integrated Task and Motion Planning | Project website: https://omron-sinicx.github.io/ViLaIn-TAMP/ | null | null | null | cs.RO cs.AI | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | While recent advances in vision-language models (VLMs) have accelerated the development of language-guided robot planners, their black-box nature often lacks safety guarantees and interpretability crucial for real-world deployment. Conversely, classical symbolic planners offer rigorous safety verification but require significant expert knowledge for setup. To bridge the current gap, this paper proposes ViLaIn-TAMP, a hybrid planning framework for enabling verifiable, interpretable, and autonomous robot behaviors. ViLaIn-TAMP comprises three main components: (1) ViLaIn (Vision-Language Interpreter) - A prior framework that converts multimodal inputs into structured problem specifications using off-the-shelf VLMs without additional domain-specific training, (2) a modular Task and Motion Planning (TAMP) system that grounds these specifications in actionable trajectory sequences through symbolic and geometric constraint reasoning and can utilize learning-based skills for key manipulation phases, and (3) a corrective planning module which receives concrete feedback on failed solution attempts from the motion and task planning components and can feed adapted logic and geometric feasibility constraints back to ViLaIn to improve and further refine the specification. We evaluate our framework on several challenging manipulation tasks in a cooking domain. We demonstrate that the proposed closed-loop corrective architecture exhibits a more than 30% higher mean success rate for ViLaIn-TAMP compared to without corrective planning. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:00:32 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Siburian",
"Jeremy",
""
],
[
"Shirai",
"Keisuke",
""
],
[
"Beltran-Hernandez",
"Cristian C.",
""
],
[
"Hamaya",
"Masashi",
""
],
[
"Görner",
"Michael",
""
],
[
"Hashimoto",
"Atsushi",
""
]
] |
2506.03271 | Lucas Johns | Lucas Johns and Anson Kost | Local-equilibrium theory of neutrino oscillations | 16 pages | null | null | LA-UR-25-25004 | hep-ph astro-ph.HE | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Miscidynamics describes coarse-grained neutrino transport under the assumption that flavor mixing is in local equilibrium. Here we introduce the concept of turbulent flavor-wave viscosity and develop techniques for including it in miscidynamics. This extension of the theory is necessary when neutrinos develop weak flavor instabilities as a result of astrophysical driving. The flavor-wave dispersion relation is obtained using a new and more general form of linear analysis, which does not require small flavor coherence. Flavor-wave transport is then approximated using wave kinetics and geometric optics. We hypothesize that the dynamical emergence of local oscillation invariants restricts the extent of flavor thermalization. In this view, flavor evolution is sculpted by both entropy-increasing and order-promoting factors. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:01:04 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Johns",
"Lucas",
""
],
[
"Kost",
"Anson",
""
]
] |
2506.03272 | Achraf Toufah | My Youssef El Hafidi, Achraf Toufah and Mohamed Achraf Kadim | Investigating Quantum Feature Maps in Quantum Support Vector Machines for Lung Cancer Classification | 14 pages, 7 figures | null | null | null | quant-ph cs.LG | http://creativecommons.org/licenses/by/4.0/ | In recent years, quantum machine learning has emerged as a promising intersection between quantum physics and artificial intelligence, particularly in domains requiring advanced pattern recognition such as healthcare. This study investigates the effectiveness of Quantum Support Vector Machines (QSVM), which leverage quantum mechanical phenomena like superposition and entanglement to construct high-dimensional Hilbert spaces for data classification. Focusing on lung cancer diagnosis, a concrete and critical healthcare application, we analyze how different quantum feature maps influence classification performance. Using a real-world dataset of 309 patient records with significant class imbalance (39 non-cancer vs. 270 cancer cases), we constructed six balanced subsets for robust evaluation. QSVM models were implemented using Qiskit and executed on the qasm simulator, employing three distinct quantum feature maps: ZFeatureMap, ZZFeatureMap, and PauliFeatureMap. Performance was assessed using accuracy, precision, recall, specificity, and F1-score. Results show that the PauliFeatureMap consistently outperformed the others, achieving perfect classification in three subsets and strong performance overall. These findings demonstrate how quantum computational principles can be harnessed to enhance diagnostic capabilities, reinforcing the importance of physics-based modeling in emerging AI applications within healthcare. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:01:29 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Hafidi",
"My Youssef El",
""
],
[
"Toufah",
"Achraf",
""
],
[
"Kadim",
"Mohamed Achraf",
""
]
] |
2506.03273 | Sergio Ernesto Aguilar Gutierrez | Sergio E. Aguilar-Gutierrez, Hugo A. Camargo, Viktor Jahnke, Keun-Young Kim, and Mitsuhiro Nishida | Krylov operator complexity in holographic CFTs: Smeared boundary reconstruction and the dual proper radial momentum | 55 pgs | null | null | null | hep-th | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Motivated by bulk reconstruction of smeared boundary operators, we study the Krylov complexity of local and non-local primary CFT$_d$ operators from the local bulk-to-bulk propagator of a minimally-coupled massive scalar field in Rindler-AdS$_{d+1}$ space. We derive analytic and numerical evidence on how the degree of non-locality in the dual CFT$_d$ observable affects the evolution of Krylov complexity and the Lanczos coefficients. Curiously, the near-horizon limit matches with the same observable for conformally-coupled probe scalar fields inserted at the asymptotic boundary of AdS$_{d+1}$ space. Our results also show that the evolution of the growth rate of Krylov operator complexity in the CFT$_d$ takes the same form as to the proper radial momentum of a probe particle inside the bulk to a good approximation. The exact equality only occurs when the probe particle is inserted in the asymptotic boundary or in the horizon limit. Our results capture a prosperous interplay between Krylov complexity in the CFT, thermal ensembles at finite bulk locations and their role in the holographic dictionary. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:01:32 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Aguilar-Gutierrez",
"Sergio E.",
""
],
[
"Camargo",
"Hugo A.",
""
],
[
"Jahnke",
"Viktor",
""
],
[
"Kim",
"Keun-Young",
""
],
[
"Nishida",
"Mitsuhiro",
""
]
] |
2506.03274 | Cedric Quint | Cedric Quint, Fabian Hei{\ss}e, Joerg Jaeckel, Lutz Leimenstoll, Christoph H. Keitel, Zolt\'an Harman | Stringent Constraints on New Pseudoscalars from Precision Hyperfine Splitting Measurements | 9 pages, 2 figures | null | null | null | hep-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Axion-like particles and similar new pseudoscalar bosons coupled to nucleons and electrons are predicted to lead to spin-dependent forces in atoms and ions. We argue that hyperfine structure measurements in the hydrogen- and lithium-like charge states are a sensitive probe to this effect. Employing specific differences of these splittings reduces uncertainties due to nuclear effects in hyperfine structure calculations and measurements. Using this, we show that existing measurements on Be provide competitive limits in the region $m_{\phi}\gtrsim 100\,{\rm keV}$, improving upon existing constraints by up to a factor of $8$. We also find that future measurements on Cs and In have a further order of magnitude improved discovery potential when compared with existing bounds. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:01:33 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Quint",
"Cedric",
""
],
[
"Heiße",
"Fabian",
""
],
[
"Jaeckel",
"Joerg",
""
],
[
"Leimenstoll",
"Lutz",
""
],
[
"Keitel",
"Christoph H.",
""
],
[
"Harman",
"Zoltán",
""
]
] |
2506.03275 | Soham Govande | Austin Silveria, Soham V. Govande, Daniel Y. Fu | Chipmunk: Training-Free Acceleration of Diffusion Transformers with Dynamic Column-Sparse Deltas | 10 pages, 4 figures | null | null | null | cs.CV cs.AI | http://creativecommons.org/licenses/by/4.0/ | Diffusion Transformers (DiTs) have achieved state-of-the-art performance in high-quality image and video generation but incur substantial compute cost at inference. A common observation is that DiT latent noise vectors change slowly across inference steps, which suggests that the DiT compute may be redundant across steps. In this paper, we aim to speed up inference by reducing this redundancy, without additional training. We first study how activations change between steps in two state-of-the-art open-source DiTs. We find that just 5-25% of the values in attention and MLP explain 70-90% of the change in activations across steps. This finding motivates our approach, Chipmunk, which uses dynamic sparsity at inference time to recompute only the fastest-changing intermediate activations, while caching the rest. Dynamic sparsity introduces two systems challenges: (1) sparse attention and MLP operations tend to underutilize GPU tensor cores; and (2) computing dynamic sparsity patterns at runtime and caching activations both introduce overhead. To address these challenges, Chipmunk first uses a voxel-based reordering of input tokens to introduce column-wise sparsity. We implement column-sparse kernels utilizing efficient sparse gathers from global to shared GPU memory, achieving a 9.3x speedup at 93% sparsity compared to highly-optimized dense baselines. Second, Chipmunk overlaps the computation of sparsity patterns and cache updates with other parts of the computation (e.g., second layer of the MLP) to hide the extra latency. Chipmunk achieves up to 2.16x speedup on HunyuanVideo and 1.41x on FLUX.1-dev without compromising generation quality. Furthermore, we show that Chipmunk can be stacked on top of full step caching, achieving a 3.72x speedup on HunyuanVideo, a 2.67x speedup on WAN2.1, and a 2.25x speedup on FLUX.1-dev with minimal quality impact. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:03:32 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Silveria",
"Austin",
""
],
[
"Govande",
"Soham V.",
""
],
[
"Fu",
"Daniel Y.",
""
]
] |
2506.03276 | Michael Neunteufel | Jay Gopalakrishnan and Michael Neunteufel | Guided modes of helical waveguides | null | null | null | null | physics.optics cs.NA math.NA | http://creativecommons.org/licenses/by/4.0/ | This paper studies guided transverse scalar modes propagating through helically coiled waveguides. Modeling the modes as solutions of the Helmholtz equation within the three-dimensional (3D) waveguide geometry, a propagation ansatz transforms the mode-finding problem into a 3D quadratic eigenproblem. Through an untwisting map, the problem is shown to be equivalent to a 3D quadratic eigenproblem on a straightened configuration. Next, exploiting the constant torsion and curvature of the Frenet frame of a circular helix, the 3D eigenproblem is further reduced to a two-dimensional (2D) eigenproblem on the waveguide cross section. All three eigenproblems are numerically treated. As expected, significant computational savings are realized in the 2D model. A few nontrivial numerical techniques are needed to make the computation of modes within the 3D geometry feasible. They are presented along with a procedure to effectively filter out unwanted non-propagating eigenfunctions. Computational results show that the geometric effect of coiling is to shift the localization of guided modes away from the coiling center. The variations in modes as coiling pitch is changed are reported considering the example of a coiled optical fiber. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:03:47 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Gopalakrishnan",
"Jay",
""
],
[
"Neunteufel",
"Michael",
""
]
] |
2506.03277 | Laurel White | Laurel White, Michael McDonald, Francesco Ubertosi, Massimo Gaspari, Julie Hlavacek-Larrondo, Helen Russell, and Taweewat Somboonpanyakul | The Onset of Feedback in Abell 1885: Evidence for Large-Scale Quenching Despite a Young Central AGN | 11 pages, 6 figures | null | null | null | astro-ph.CO | http://creativecommons.org/publicdomain/zero/1.0/ | We present a new 8.5 ks Chandra observation of Abell 1885, obtained as part of the Cluster Evolution Reference Ensemble At Low-z (CEREAL) survey of ~200 low-z galaxy groups and clusters. These data reveal that Abell 1885 is a strong cool core, with a central cooling time of 0.43 Gyr, and that the central galaxy hosts an X-ray luminous point source at its center (L=2.3x10^42 erg/s), indicative of a rapidly accreting supermassive black hole. In the context of the larger CEREAL sample, we constrain the fraction of clusters at z~0.15 with X-ray bright central AGN to be no more than 4.1%. Including radio data from LOFAR, GMRT, ASKAP, and the VLA and optical integral field unit data from SDSS MaNGA, we probe the details of cooling, feeding, and feedback in this system. These data reveal that cooling of the intracluster medium is highly suppressed on large (>10 kpc) scales despite a central supermassive black hole that is in the early stages of the self-regulation cycle (characterized by rapid accretion, physically small jets, and no large-scale low-frequency radio emission). To reconcile the large-scale quenching with a lack of visible large-scale feedback, we propose that the timescale on which energy is dissipated on large scales is significantly longer than the timescale on which black hole feeding operates on small (~pc) scales. These observations suggest that there are two separate timescales characterizing AGN feedback in clusters: the short timescale of small-scale feeding and feedback processes and a longer timescale by which energy is dissipated on large physical scales in the intracluster medium. This interpretation disfavors a model in which the energy is rapidly dissipated (e.g. shocks), which would synchronize the feeding and feedback timescales, and favors a model in which the heating effects of AGN feedback can linger long after the outburst has passed (e.g. turbulent mixing). | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:03:55 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"White",
"Laurel",
""
],
[
"McDonald",
"Michael",
""
],
[
"Ubertosi",
"Francesco",
""
],
[
"Gaspari",
"Massimo",
""
],
[
"Hlavacek-Larrondo",
"Julie",
""
],
[
"Russell",
"Helen",
""
],
[
"Somboonpanyakul",
"Taweewat",
""
]
] |
2506.03278 | Christodoulos Constantinides | Christodoulos Constantinides, Dhaval Patel, Shuxin Lin, Claudio Guerrero, Sunil Dagajirao Patil, Jayant Kalagnanam | FailureSensorIQ: A Multi-Choice QA Dataset for Understanding Sensor Relationships and Failure Modes | null | null | null | null | cs.CL | http://creativecommons.org/licenses/by/4.0/ | We introduce FailureSensorIQ, a novel Multi-Choice Question-Answering (MCQA) benchmarking system designed to assess the ability of Large Language Models (LLMs) to reason and understand complex, domain-specific scenarios in Industry 4.0. Unlike traditional QA benchmarks, our system focuses on multiple aspects of reasoning through failure modes, sensor data, and the relationships between them across various industrial assets. Through this work, we envision a paradigm shift where modeling decisions are not only data-driven using statistical tools like correlation analysis and significance tests, but also domain-driven by specialized LLMs which can reason about the key contributors and useful patterns that can be captured with feature engineering. We evaluate the Industrial knowledge of over a dozen LLMs-including GPT-4, Llama, and Mistral-on FailureSensorIQ from different lens using Perturbation-Uncertainty-Complexity analysis, Expert Evaluation study, Asset-Specific Knowledge Gap analysis, ReAct agent using external knowledge-bases. Even though closed-source models with strong reasoning capabilities approach expert-level performance, the comprehensive benchmark reveals a significant drop in performance that is fragile to perturbations, distractions, and inherent knowledge gaps in the models. We also provide a real-world case study of how LLMs can drive the modeling decisions on 3 different failure prediction datasets related to various assets. We release: (a) expert-curated MCQA for various industrial assets, (b) FailureSensorIQ benchmark and Hugging Face leaderboard based on MCQA built from non-textual data found in ISO documents, and (c) LLMFeatureSelector, an LLM-based feature selection scikit-learn pipeline. The software is available at https://github.com/IBM/FailureSensorIQ. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:05:10 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Constantinides",
"Christodoulos",
""
],
[
"Patel",
"Dhaval",
""
],
[
"Lin",
"Shuxin",
""
],
[
"Guerrero",
"Claudio",
""
],
[
"Patil",
"Sunil Dagajirao",
""
],
[
"Kalagnanam",
"Jayant",
""
]
] |
2506.03279 | Santiago Lamata-Ot\'in | Santiago Lamata-Ot\'in, Octavian C. Rotita-Ion, Alex Arenas, David Soriano-Pa\~nos and Jes\'us G\'omez-Garde\~nes | Genotype networks drive oscillating endemicity and epidemic trajectories in viral evolution | 13 pages, 5 figures. SI: 11 pages, 8 figures | null | null | null | q-bio.PE physics.soc-ph | http://creativecommons.org/licenses/by/4.0/ | Rapidly evolving viruses use antigenic drift as a key mechanism to evade host immunity and persist in real populations. While traditional models of antigenic drift and epidemic spread rely on low-dimensional antigenic spaces, genomic surveillance data reveal that viral evolution produces complex antigenic genotype networks with hierarchical modular structures. In this study, we present an eco-evolutionary framework in which viral evolution and population immunity dynamics are shaped by the structure of antigenic genotype networks. Using synthetic networks, we demonstrate that network topology alone can drive transitions between stable endemic states and recurrent seasonal epidemics. Furthermore, our results show how the integration of the genotype network of the H3N2 influenza in our model allows for estimating the emergence times of various haplotypes resulting from its evolution. Our findings underscore the critical role of the topology of genotype networks in shaping epidemic behavior and, besides, provide a robust framework for integrating real-world genomic data into predictive epidemic models. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:12:09 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Lamata-Otín",
"Santiago",
""
],
[
"Rotita-Ion",
"Octavian C.",
""
],
[
"Arenas",
"Alex",
""
],
[
"Soriano-Paños",
"David",
""
],
[
"Gómez-Gardeñes",
"Jesús",
""
]
] |
2506.03280 | David E. Mertz | David. E Mertz (1) ((1) Fermilab) | Design Didn't Prevent These Hazard Exposures | 2025 IEEE Electrical Safety Workshop | null | null | FERMILAB-CONF-25-0060-ESH | physics.ins-det | http://creativecommons.org/licenses/by/4.0/ | Two recent electrical incidents demonstrate how the design of equipment encouraged electrical workers to take actions that violated NFPA 70E principles. The features that encouraged non-compliant work execution will be described, as well as how the equipment was improved to facilitate safe work practices. Design-stage processes that help identify features that will foster rather than compromise safe work practices will be identified as well. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:12:42 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Mertz",
"David. E",
"",
"Fermilab"
]
] |
2506.03281 | Andrea Ferrari | M. Tiberi, A. Montanaro, C. Wen, J. Zhang, O. Balci, S. M. Shinde, S. Sharma, A. Meersha, H. Shekhar, J. E. Muench, B. R. Conran, K. B. K. Teo, M. Ebert, X. Yan, Y. Tran, M. Banakar, C. Littlejohns, G. T. Reed, M. Romagnoli, A. Ruocco, V. Sorianello, A. C. Ferrari | Graphene Electro-Absorption Modulators for Energy-Efficient and High-Speed Optical Transceivers | null | null | null | null | cond-mat.mtrl-sci cond-mat.mes-hall | http://creativecommons.org/licenses/by/4.0/ | The increasing demand for energy-efficient hardware for artificial intelligence (AI) and data centres requires integrated photonic solutions delivering optical transceivers with Tbit/s data rates and energy consumption$<$1pJ/bit. Here, we report double single-layer graphene electro-absorption modulators on Si optimized for energy-efficient and ultra-fast operation, demonstrating 67GHz bandwidth and 80Gbit/s data rate, in both O and C bands, using a fabrication tailored for wafer-scale integration. We measure a data rate$\sim$1.6 times larger than previously reported for graphene. We scale the modulator's active area down to 22$\mu$m$^2$, achieving a dynamic power consumption$\sim$58fJ/bit, $\sim$3 times lower than previous graphene modulators and Mach-Zehnder modulators based on Si or lithium niobate. We show devices with$\sim$0.037dB/V$\mu$m modulation efficiency,$\sim$16 times better than previous demonstrations based on graphene. This paves the way to wafer-scale production of graphene modulators on Si useful for Tbit/s optical transceivers and energy-efficient AI | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:13:44 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Tiberi",
"M.",
""
],
[
"Montanaro",
"A.",
""
],
[
"Wen",
"C.",
""
],
[
"Zhang",
"J.",
""
],
[
"Balci",
"O.",
""
],
[
"Shinde",
"S. M.",
""
],
[
"Sharma",
"S.",
""
],
[
"Meersha",
"A.",
""
],
[
"Shekhar",
"H.",
""
],
[
"Muench",
"J. E.",
""
],
[
"Conran",
"B. R.",
""
],
[
"Teo",
"K. B. K.",
""
],
[
"Ebert",
"M.",
""
],
[
"Yan",
"X.",
""
],
[
"Tran",
"Y.",
""
],
[
"Banakar",
"M.",
""
],
[
"Littlejohns",
"C.",
""
],
[
"Reed",
"G. T.",
""
],
[
"Romagnoli",
"M.",
""
],
[
"Ruocco",
"A.",
""
],
[
"Sorianello",
"V.",
""
],
[
"Ferrari",
"A. C.",
""
]
] |
2506.03282 | Iarley P. Lobo Dr | Iarley P. Lobo, Gislaine Var\~ao, Giulia Gubitosi, Moises Rojas, Valdir B. Bezerra | Coherence and Entanglement in a Quantum Spacetime | 12 pages, 8 figures | null | null | null | quant-ph gr-qc | http://creativecommons.org/licenses/by/4.0/ | We investigate the emergence of quantum coherence and quantum correlations in a two-particle system with deformed symmetries arising from the quantum nature of spacetime. We demonstrate that the deformation of energy-momentum composition induces a momentum-dependent interaction that counteracts the decoherence effects described by the Lindblad equation in quantum spacetime. This interplay leads to the formation of coherence, entanglement and other correlations, which we quantify using concurrence, the $l_1$-norm of coherence, quantum discord and Local Quantum Fisher Information. Our analysis reveals that while the openness of quantum spacetime ultimately degrades entanglement, it also facilitates the creation and preservation of both classical and quantum correlations. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:13:56 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Lobo",
"Iarley P.",
""
],
[
"Varão",
"Gislaine",
""
],
[
"Gubitosi",
"Giulia",
""
],
[
"Rojas",
"Moises",
""
],
[
"Bezerra",
"Valdir B.",
""
]
] |
2506.03283 | Yannic Noller | Viola Campos, Ridwan Shariffdeen, Adrian Ulges, Yannic Noller | Empirical Evaluation of Generalizable Automated Program Repair with Large Language Models | null | null | null | null | cs.SE | http://creativecommons.org/licenses/by/4.0/ | Automated Program Repair (APR) proposes bug fixes to aid developers in maintaining software. The state of the art in this domain focuses on using LLMs, leveraging their strong capabilities to comprehend specifications in natural language and to generate program code. Recent works have shown that LLMs can be used to generate repairs. However, despite the APR community's research achievements and several industry deployments in the last decade, APR still lacks the capabilities to generalize broadly. In this work, we present an intensive empirical evaluation of LLMs for generating patches. We evaluate a diverse set of 13 recent models, including open ones (e.g., Llama 3.3, Qwen 2.5 Coder, and DeepSeek R1 (dist.)) and closed ones (e.g., o3-mini, GPT-4o, Claude 3.7 Sonnet, Gemini 2.0 Flash). In particular, we explore language-agnostic repairs by utilizing benchmarks for Java (e.g., Defects4J), JavaScript (e.g., BugsJS), Python (e.g., BugsInPy), and PHP (e.g., BugsPHP). Besides the generalization between different languages and levels of patch complexity, we also investigate the effects of fault localization (FL) as a preprocessing step and compare the progress for open vs closed models. Our evaluation represents a snapshot of the current repair capabilities of the latest LLMs. Key results include: (1) Different LLMs tend to perform best for different languages, which makes it hard to develop cross-platform repair techniques with single LLMs. (2) The combinations of models add value with respect to uniquely fixed bugs, so a committee of expert models should be considered. (3) Under realistic assumptions of imperfect FL, we observe significant drops in accuracy from the usual practice of using perfect FL. Our findings and insights will help both researchers and practitioners develop reliable and generalizable APR techniques and evaluate them in realistic and fair environments. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:15:14 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Campos",
"Viola",
""
],
[
"Shariffdeen",
"Ridwan",
""
],
[
"Ulges",
"Adrian",
""
],
[
"Noller",
"Yannic",
""
]
] |
2506.03284 | Guanglei Hong | Guanglei Hong | Ratio of Mediator Probability Weighting for Estimating Natural Direct and Indirect Effects | 15 pages | JSM Proceedings, Biometrics Section. Alexandria, VA: American Statistical Association, pp.2401-2415 (2010) | null | null | stat.ME | http://creativecommons.org/licenses/by-nc-nd/4.0/ | Decomposing a total causal effect into natural direct and indirect effects is central to revealing causal mechanisms. Conventional methods achieve the decomposition by specifying an outcome model as a linear function of the treatment, the mediator, and the observed covariates under identification assumptions including the assumption of no interaction between treatment and mediator. Recent statistical advances relax this assumption typically within the linear or nonlinear regression framework. I propose a non-parametric approach that also relaxes the assumption of no treatment-mediator interaction while avoiding the problems of outcome model specification that become particularly acute in the presence of a large number of covariates. The key idea is to estimate the marginal mean of each counterfactual outcome by assigning a weight to every experimental unit such that the weighted distribution of the mediator under the experimental condition approximates the counterfactual mediator distribution under the control condition. The weight is a ratio of the conditional probability of a mediator value under the control condition to that of the same mediator value under the experimental condition. A non-parametric approach to estimating the weight on the basis of propensity score stratification promises to increase the robustness of the direct and indirect effect estimates. The outcome is modeled as a function of the direct and indirect effects with minimal model-based assumptions. This method applies regardless of the distribution of the outcome or the functional relationship between the outcome and the mediator, and is suitable for handling a large number of pretreatment covariates. RMPW software packages are available in Stata (https://ideas.repec.org/c/boc/bocode/s458301.html) and R (https://cran.r-project.org/web/packages/rmpw/index.html). | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:16:32 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Hong",
"Guanglei",
""
]
] |
2506.03285 | Pierdomenico Duttilo | Pierdomenico Duttilo, Stefano Antonio Gattone, Alfred Kume | Constrained mixtures of generalized normal distributions | 34 pages, 12 figures, 7 tables | null | null | null | stat.ME stat.CO | http://creativecommons.org/licenses/by-nc-nd/4.0/ | This work introduces a family of univariate constrained mixtures of generalized normal distributions (CMGND) where the location, scale, and shape parameters can be constrained to be equal across any subset of mixture components. An expectation conditional maximisation (ECM) algorithm with Newton-Raphson updates is used to estimate the model parameters under the constraints. Simulation studies demonstrate that imposing correct constraints leads to more accurate parameter estimation compared to unconstrained mixtures, especially when components substantially overlap. Constrained models also exhibit competitive performance in capturing key characteristics of the marginal distribution, such as kurtosis. On a real dataset of daily stock index returns, CMGND models outperform constrained mixtures of normals and Student's t distributions based on the BIC criterion, highlighting their flexibility in modelling nonnormal features. The proposed constrained approach enhances interpretability and can improve parametric efficiency without compromising distributional flexibility for complex data. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:16:55 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Duttilo",
"Pierdomenico",
""
],
[
"Gattone",
"Stefano Antonio",
""
],
[
"Kume",
"Alfred",
""
]
] |
2506.03286 | Tanay Roy | Taeyoon Kim, Tanay Roy, Xinyuan You, Andy C. Y. Li, Henry Lamm, Oleg Pronitchev, Mustafa Bal, Sabrina Garattoni, Francesco Crisa, Daniel Bafia, Doga Kurkcuoglu, Roman Pilipenko, Paul Heidler, Nicholas Bornman, David van Zanten, Silvia Zorzetti, Roni Harnik, Akshay Murthy, Shaojiang Zhu, Changqing Wang, Andre Vallieres, Ziwen Huang, Jens Koch, Anna Grassellino, Srivatsan Chakram, Alexander Romanenko, and Yao Lu | Ultracoherent superconducting cavity-based multiqudit platform with error-resilient control | 20 pages with 3 main figures and 8 supplementary figures | null | null | FERMILAB-PUB-25-0379-SQMS | quant-ph | http://creativecommons.org/licenses/by/4.0/ | Superconducting radio-frequency (SRF) cavities offer a promising platform for quantum computing due to their long coherence times and large accessible Hilbert spaces, yet integrating nonlinear elements like transmons for control often introduces additional loss. We report a multimode quantum system based on a 2-cell elliptical shaped SRF cavity, comprising two cavity modes weakly coupled to an ancillary transmon circuit, designed to preserve coherence while enabling efficient control of the cavity modes. We mitigate the detrimental effects of the transmon decoherence through careful design optimization that reduces transmon-cavity couplings and participation in the dielectric substrate and lossy interfaces, to achieve single-photon lifetimes of 20.6 ms and 15.6 ms for the two modes, and a pure dephasing time exceeding 40 ms. This marks an order-of-magnitude improvement over prior 3D multimode memories. Leveraging sideband interactions and novel error-resilient protocols, including measurement-based correction and post-selection, we achieve high-fidelity control over quantum states. This enables the preparation of Fock states up to $N = 20$ with fidelities exceeding 95%, the highest reported to date to the authors' knowledge, as well as two-mode entanglement with coherence-limited fidelities reaching up to 99.9% after post-selection. These results establish our platform as a robust foundation for quantum information processing, allowing for future extensions to high-dimensional qudit encodings. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:18:46 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Kim",
"Taeyoon",
""
],
[
"Roy",
"Tanay",
""
],
[
"You",
"Xinyuan",
""
],
[
"Li",
"Andy C. Y.",
""
],
[
"Lamm",
"Henry",
""
],
[
"Pronitchev",
"Oleg",
""
],
[
"Bal",
"Mustafa",
""
],
[
"Garattoni",
"Sabrina",
""
],
[
"Crisa",
"Francesco",
""
],
[
"Bafia",
"Daniel",
""
],
[
"Kurkcuoglu",
"Doga",
""
],
[
"Pilipenko",
"Roman",
""
],
[
"Heidler",
"Paul",
""
],
[
"Bornman",
"Nicholas",
""
],
[
"van Zanten",
"David",
""
],
[
"Zorzetti",
"Silvia",
""
],
[
"Harnik",
"Roni",
""
],
[
"Murthy",
"Akshay",
""
],
[
"Zhu",
"Shaojiang",
""
],
[
"Wang",
"Changqing",
""
],
[
"Vallieres",
"Andre",
""
],
[
"Huang",
"Ziwen",
""
],
[
"Koch",
"Jens",
""
],
[
"Grassellino",
"Anna",
""
],
[
"Chakram",
"Srivatsan",
""
],
[
"Romanenko",
"Alexander",
""
],
[
"Lu",
"Yao",
""
]
] |
2506.03288 | Nilaksha Barman | Nilaksha Barman, Debarati Chatterjee | $f$-mode oscillations in hot Neutron Stars: Effect of hyperons and neutrino trapping | 16 pages, 30 figures | null | null | null | astro-ph.HE nucl-th | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | In this work, we present an equation of state formalism for hot Neutron Stars (NSs) which consistently includes the effects of finite temperature, hyperons as well as neutrino trapping, relevant for the study of proto-neutron stars, binary neutron star mergers and supernova explosions. Within a non-linear relativistic mean field description, the framework allows for a systematic variation of nuclear parameters within the range allowed by uncertainties in nuclear experimental data, ensuring compatibility with nuclear theory, astrophysical and heavy-ion data. We then investigate the role of nuclear and hypernuclear parameters on NS macroscopic properties and $f$-mode oscillations in hot neutron stars within Cowling approximation. Our results reveal that in hyperonic neutron stars with trapped neutrinos, the saturation nuclear density shows moderate to strong correlation with NS astrophysical observables. We also investigate the $C-Love$ universal relations for this model and provide fit relations for hot NS configurations in the neutrino-trapped regime. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:21:32 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Barman",
"Nilaksha",
""
],
[
"Chatterjee",
"Debarati",
""
]
] |
2506.03289 | Marco Polo Moreno De Souza | Marco P. M. de Souza, Alex B. Siqueira | Exploring Heat Exchanges with the Calorimetry Simulator -- SimuF\'isica | null | null | null | null | physics.ed-ph | http://creativecommons.org/licenses/by/4.0/ | This article presents the Calorimetry - SimuF\'isica simulator, an interactive computational tool designed for teaching heat exchange processes. The simulator enables dynamic and audiovisual exploration of phenomena such as the heating of liquids and the establishment of thermal equilibrium between bodies at different temperatures. We describe its interface and the underlying physics, based on the equations of specific heat, latent heat of vaporization, and Newton's law of cooling. Two educational application examples are analyzed, in which the simulation results are compared with calculations typically performed at the high school and early undergraduate levels, highlighting the consistency between theory and simulation and the pedagogical potential of the tool. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:29:06 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"de Souza",
"Marco P. M.",
""
],
[
"Siqueira",
"Alex B.",
""
]
] |
2506.03290 | Jisoo Jeong | Leyla Mirvakhabova, Hong Cai, Jisoo Jeong, Hanno Ackermann, Farhad Zanjani, Fatih Porikli | Learning Optical Flow Field via Neural Ordinary Differential Equation | CVPRW 2025 | null | null | null | cs.CV | http://creativecommons.org/licenses/by/4.0/ | Recent works on optical flow estimation use neural networks to predict the flow field that maps positions of one image to positions of the other. These networks consist of a feature extractor, a correlation volume, and finally several refinement steps. These refinement steps mimic the iterative refinements performed by classical optimization algorithms and are usually implemented by neural layers (e.g., GRU) which are recurrently executed for a fixed and pre-determined number of steps. However, relying on a fixed number of steps may result in suboptimal performance because it is not tailored to the input data. In this paper, we introduce a novel approach for predicting the derivative of the flow using a continuous model, namely neural ordinary differential equations (ODE). One key advantage of this approach is its capacity to model an equilibrium process, dynamically adjusting the number of compute steps based on the data at hand. By following a particular neural architecture, ODE solver, and associated hyperparameters, our proposed model can replicate the exact same updates as recurrent cells used in existing works, offering greater generality. Through extensive experimental analysis on optical flow benchmarks, we demonstrate that our approach achieves an impressive improvement over baseline and existing models, all while requiring only a single refinement step. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:30:14 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Mirvakhabova",
"Leyla",
""
],
[
"Cai",
"Hong",
""
],
[
"Jeong",
"Jisoo",
""
],
[
"Ackermann",
"Hanno",
""
],
[
"Zanjani",
"Farhad",
""
],
[
"Porikli",
"Fatih",
""
]
] |
2506.03291 | Wasilij Barsukow | Wasilij Barsukow | An Active Flux method for the Euler equations based on the exact acoustic evolution operator | null | null | null | null | math.NA cs.NA | http://creativecommons.org/licenses/by-nc-sa/4.0/ | A new Active Flux method for the multi-dimensional Euler equations is based on an additive operator splitting into acoustics and advection. The acoustic operator is solved in a locally linearized manner by using the exact evolution operator. The nonlinear advection operator is solved at third order accuracy using a new approximate evolution operator. To simplify the splitting, the new method uses primitive variables for the point values and for the reconstruction. In order to handle discontinuous solutions, a blended bound preserving limiting is used, that combines a priori and a posteriori approaches. The resulting method is able to resolve multi-dimensional Riemann problems as well as low Mach number flow, and has a large domain of stability. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:30:26 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Barsukow",
"Wasilij",
""
]
] |
2506.03292 | Jiuding Sun | Jiuding Sun, Sidharth Baskaran, Zhengxuan Wu, Michael Sklar, Christopher Potts, Atticus Geiger | HyperSteer: Activation Steering at Scale with Hypernetworks | null | null | null | null | cs.CL cs.AI cs.LG | http://creativecommons.org/licenses/by/4.0/ | Steering language models (LMs) by modifying internal activations is a popular approach for controlling text generation. Unsupervised dictionary learning methods, e.g., sparse autoencoders, can be scaled to produce many steering vectors, but lack guarantees on the individual efficacy of each vector and control over the coverage of relevant steering tasks. In contrast, supervised methods for constructing steering vectors are targeted and effective, but require more data collection and training for each additional steering vector produced. In this work, we introduce HyperSteer, a family of hypernetwork-based architectures which are trained end-to-end to generate steering vectors conditioned on the natural language steering prompts and the internals of the steered LM. In our evaluations, we show that scaling HyperSteer with thousands of steering prompts exceeds the performance of state-of-the-art activation steering methods, even on steering prompts never seen during training. Moreover, HyperSteer performs on par with steering-via-prompting. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:32:01 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Sun",
"Jiuding",
""
],
[
"Baskaran",
"Sidharth",
""
],
[
"Wu",
"Zhengxuan",
""
],
[
"Sklar",
"Michael",
""
],
[
"Potts",
"Christopher",
""
],
[
"Geiger",
"Atticus",
""
]
] |
2506.03293 | Polina Turishcheva | Nina S. Nellen, Polina Turishcheva, Michaela Vystr\v{c}ilov\'a, Shashwat Sridhar, Tim Gollisch, Andreas S. Tolias, Alexander S. Ecker | Learning to cluster neuronal function | null | null | null | null | q-bio.NC | http://creativecommons.org/licenses/by/4.0/ | Deep neural networks trained to predict neural activity from visual input and behaviour have shown great potential to serve as digital twins of the visual cortex. Per-neuron embeddings derived from these models could potentially be used to map the functional landscape or identify cell types. However, state-of-the-art predictive models of mouse V1 do not generate functional embeddings that exhibit clear clustering patterns which would correspond to cell types. This raises the question whether the lack of clustered structure is due to limitations of current models or a true feature of the functional organization of mouse V1. In this work, we introduce DECEMber -- Deep Embedding Clustering via Expectation Maximization-based refinement -- an explicit inductive bias into predictive models that enhances clustering by adding an auxiliary $t$-distribution-inspired loss function that enforces structured organization among per-neuron embeddings. We jointly optimize both neuronal feature embeddings and clustering parameters, updating cluster centers and scale matrices using the EM-algorithm. We demonstrate that these modifications improve cluster consistency while preserving high predictive performance and surpassing standard clustering methods in terms of stability. Moreover, DECEMber generalizes well across species (mice, primates) and visual areas (retina, V1, V4). The code is available at https://github.com/Nisone2000/sensorium/tree/neuroips_version. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:33:54 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Nellen",
"Nina S.",
""
],
[
"Turishcheva",
"Polina",
""
],
[
"Vystrčilová",
"Michaela",
""
],
[
"Sridhar",
"Shashwat",
""
],
[
"Gollisch",
"Tim",
""
],
[
"Tolias",
"Andreas S.",
""
],
[
"Ecker",
"Alexander S.",
""
]
] |
2506.03294 | Travis Gagie | Diego Diaz-Dominguez, Travis Gagie, Veronica Guerrini, Ben Langmead, Zsuzsanna Liptak, Giovanni Manzini, Francesco Masillo and Vikram Shivakumar | Prefix-free parsing for merging big BWTs | null | null | null | null | cs.DS | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | When building Burrows-Wheeler Transforms (BWTs) of truly huge datasets, prefix-free parsing (PFP) can use an unreasonable amount of memory. In this paper we show how if a dataset can be broken down into small datasets that are not very similar to each other -- such as collections of many copies of genomes of each of several species, or collections of many copies of each of the human chromosomes -- then we can drastically reduce PFP's memory footprint by building the BWTs of the small datasets and then merging them into the BWT of the whole dataset. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:34:01 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Diaz-Dominguez",
"Diego",
""
],
[
"Gagie",
"Travis",
""
],
[
"Guerrini",
"Veronica",
""
],
[
"Langmead",
"Ben",
""
],
[
"Liptak",
"Zsuzsanna",
""
],
[
"Manzini",
"Giovanni",
""
],
[
"Masillo",
"Francesco",
""
],
[
"Shivakumar",
"Vikram",
""
]
] |
2506.03296 | Jiakun Fan | Jiakun Fan, Yanglin Zhang, Xiangchen Li, Dimitrios S. Nikolopoulos | Parallel CPU-GPU Execution for LLM Inference on Constrained GPUs | Preprint, under review | null | null | null | cs.DC | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Deploying large language models (LLMs) for online inference is often constrained by limited GPU memory, particularly due to the growing KV cache during auto-regressive decoding. Hybrid GPU-CPU execution has emerged as a promising solution by offloading KV cache management and parts of attention computation to the CPU. However, a key bottleneck remains: existing schedulers fail to effectively overlap CPU-offloaded tasks with GPU execution during the latency-critical, bandwidth-bound decode phase. This particularly penalizes real-time, decode-heavy applications (e.g., chat, Chain-of-Thought reasoning) which are currently underserved by existing systems, especially under memory pressure typical of edge or low-cost deployments.
We present APEX, a novel, profiling-informed scheduling strategy that maximizes CPU-GPU parallelism during hybrid LLM inference. Unlike systems relying on static rules or purely heuristic approaches, APEX dynamically dispatches compute across heterogeneous resources by predicting execution times of CPU and GPU subtasks to maximize overlap while avoiding scheduling overheads.We evaluate APEX on diverse workloads and GPU architectures (NVIDIA T4, A10), using LLaMa-2-7B and LLaMa-3.1-8B models. Compared to GPU-only schedulers like VLLM, APEX improves throughput by 84% - 96% on T4 and 11% - 89% on A10 GPUs, while preserving latency. Against the best existing hybrid schedulers, it delivers up to 49% (T4) and 37% (A10) higher throughput in long-output settings.APEX significantly advances hybrid LLM inference efficiency on such memory-constrained hardware and provides a blueprint for scheduling in heterogeneous AI systems, filling a critical gap for efficient real-time LLM applications. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:35:56 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Fan",
"Jiakun",
""
],
[
"Zhang",
"Yanglin",
""
],
[
"Li",
"Xiangchen",
""
],
[
"Nikolopoulos",
"Dimitrios S.",
""
]
] |
2506.03297 | Runhan Liu | Runhan Liu, Hui Ren, Wei Fan | Dynamics and Control of Vision-Aided Multi-UAV-tethered Netted System Capturing Non-Cooperative Target | null | null | null | null | cs.RO cs.SY eess.SY | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | As the number of Unmanned Aerial Vehicles (UAVs) operating in low-altitude airspace continues to increase, non-cooperative targets pose growing challenges to low-altitude operations. To address this issue, this paper proposes a multi-UAV-tethered netted system as a non-lethal solution for capturing non-cooperative targets. To validate the proposed system, we develop mySim, a multibody dynamics-based UAV simulation environment that integrates high-precision physics modeling, vision-based motion tracking, and reinforcement learning-driven control strategies. In mySim, the spring-damper model is employed to simulate the dynamic behavior of the tethered net, while the dynamics of the entire system is modeled using multibody dynamics (MBD) to achieve accurate representations of system interactions. The motion of the UAVs and the target are estimated using VINS-MONO and DETR, and the system autonomously executes the capture strategy through MAPPO. Simulation results demonstrate that mySim accurately simulates dynamics and control of the system, successfully enabling the multi-UAV-tethered netted system to capture both non-propelled and maneuvering non-cooperative targets. By providing a high-precision simulation platform that integrates dynamics modeling with perception and learning-based control, mySim enables efficient testing and optimization of UAV-based control policies before real-world deployment. This approach offers significant advantages for simulating complex UAVs coordination tasks and has the potential to be applied to the design of other UAV-based systems. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:36:25 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Liu",
"Runhan",
""
],
[
"Ren",
"Hui",
""
],
[
"Fan",
"Wei",
""
]
] |
2506.03298 | Kosar Behnia | Kosar Behnia, H.A. Talebi, and Farzaneh Abdollahi | Online Detection and Mitigation of Robust Zero Dynamics Anomaly Behavior in MIMO Nonlinear Control Systems | null | null | null | null | eess.SY cs.SY | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | This paper presents a methodology to detect robust zero dynamics anomaly behavior and mitigate the impacts in general multi-input multi-output (MIMO) nonlinear systems. The proposed method guarantees the resiliency and stability of the closed-loop system without relying on an accurate dynamical model. The presented method operates in two stages. First, it measures the difference between the system input and that of the model as a residual signal to detect the anomaly behavior. After detecting the attack, a recovery signal is generated to restore the system to its nominal condition. In this stage, a neural network model is used to estimate the anomaly signal and recover the closed-loop system. The weights of the neural network model are updated online using adaptation rules without needing prior data for training. The accuracy and performance of the proposed methods are verified by simulating various scenarios on a fourtank system. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:36:31 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Behnia",
"Kosar",
""
],
[
"Talebi",
"H. A.",
""
],
[
"Abdollahi",
"Farzaneh",
""
]
] |
2506.03299 | Jessie Christiansen | Jessie L. Christiansen, Douglas L. McElroy, Marcy Harbut, David R. Ciardi, Megan Crane, John Good, Kevin K. Hardegree-Ullman, Aurora Y. Kesseli, Michael B. Lund, Meca Lynn, Ananda Muthiar, Ricky Nilsson, Toba Oluyide, Michael Papin, Amalia Rivera, Melanie Swain, Nicholas D. Susemiehl, Raymond Tam, Julian van Eyken and Charles Beichman | The NASA Exoplanet Archive and Exoplanet Follow-up Observing Program: Data, Tools, and Usage | 18 pages, 7 figures, 4 tables. Replaces Akeson et al. (2013) as the citation for the NASA Exoplanet Archive and Exoplanet Follow-up Observing Program (ExoFOP), accepted in PSJ | null | null | null | astro-ph.EP astro-ph.IM astro-ph.SR | http://creativecommons.org/licenses/by/4.0/ | The NASA Exoplanet Archive and the Exoplanet Follow-up Observing Program service are two widely used resources for the exoplanet community. The NASA Exoplanet Archive provides a complete and accurate accounting of exoplanetary systems published by NASA missions and by the community in the refereed literature. In anticipation of continued exponential growth in the number of exoplanetary systems, and the increasing complexity in our characterization of these systems, the NASA Exoplanet Archive has restructured its primary tables and interfaces, as well as extending and standardizing their modes of access. The Exoplanet Follow-up Observing Program service provides the exoplanet community with a venue for coordinating and sharing follow-up and precursor data for exoplanets, their host stars, and stars that might eventually be targets for future planet searches, and recently reached one million files uploaded by the community. In this paper we describe the updates to our data holdings, functionality, accessibility, and tools, as well as future priorities for these two services. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:36:52 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Christiansen",
"Jessie L.",
""
],
[
"McElroy",
"Douglas L.",
""
],
[
"Harbut",
"Marcy",
""
],
[
"Ciardi",
"David R.",
""
],
[
"Crane",
"Megan",
""
],
[
"Good",
"John",
""
],
[
"Hardegree-Ullman",
"Kevin K.",
""
],
[
"Kesseli",
"Aurora Y.",
""
],
[
"Lund",
"Michael B.",
""
],
[
"Lynn",
"Meca",
""
],
[
"Muthiar",
"Ananda",
""
],
[
"Nilsson",
"Ricky",
""
],
[
"Oluyide",
"Toba",
""
],
[
"Papin",
"Michael",
""
],
[
"Rivera",
"Amalia",
""
],
[
"Swain",
"Melanie",
""
],
[
"Susemiehl",
"Nicholas D.",
""
],
[
"Tam",
"Raymond",
""
],
[
"van Eyken",
"Julian",
""
],
[
"Beichman",
"Charles",
""
]
] |
2506.03300 | Enoch Ofosu | Enoch Ofosua, Kevin Bradley Dsouzaa, Daniel Chukwuemeka Amaogud, Jerome Pigeond, Richard Boudreaultc, Juan Moreno-Crubb, Pooneh Maghoule, Yuri Leonenko | Climate benefits of afforestation and reforestation with varying species mixtures and densities in the north-western boreal lands | null | null | null | null | q-bio.PE q-bio.QM | http://creativecommons.org/licenses/by-nc-nd/4.0/ | The boreal forest plays a crucial role as a global carbon sink. This study uses two 250-year simulations of Canada's Taiga Plains, an area targeted by the 2 Billion Trees Program to evaluate afforestation and reforestation strategies that vary by species mix, planting density, and surface albedo. Medium density stands, 600 to 1400 trees per hectare, composed of mixed species with approximately 25 to 40 percent deciduous trees sequestered 15 to 30 percent more net ecosystem carbon than conifer monocultures. These benefits stem from a combination of rapid early growth, long-term carbon retention, and enhanced resilience to disturbance. Replanting understocked stands with such mixtures increased long-term carbon storage by 18 to 30 percent relative to prevailing scenarios. When surface albedo was considered, pure evergreen or deciduous stands showed a reduction in climate benefit by 6 to 20 percent, while mixed stands maintained net cooling and achieved the highest sequestration rates, approximately 4.6 to 4.7 tons of carbon dioxide equivalent per hectare per year. Scenarios involving partial harvesting followed by replanting sustained or improved ecosystem carbon stocks, about 300 to 340 tons of carbon per hectare, and productivity, roughly 1.6 to 2.0 tons of carbon per hectare per year, without increasing ecological risk. Overall, integrating fast-growing deciduous species with long-lived conifers at moderate planting densities enhances the climate mitigation potential of boreal afforestation and reforestation efforts and offers guidance for reforestation policy in similar high latitude ecosystems. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:37:10 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Ofosua",
"Enoch",
""
],
[
"Dsouzaa",
"Kevin Bradley",
""
],
[
"Amaogud",
"Daniel Chukwuemeka",
""
],
[
"Pigeond",
"Jerome",
""
],
[
"Boudreaultc",
"Richard",
""
],
[
"Moreno-Crubb",
"Juan",
""
],
[
"Maghoule",
"Pooneh",
""
],
[
"Leonenko",
"Yuri",
""
]
] |
2506.03301 | Daham Mustafa | Daham M. Mustafa, Abhishek Nadgeri, Diego Collarana, Benedikt T. Arnold, Christoph Quix, Christoph Lange, Stefan Decker | From Instructions to ODRL Usage Policies: An Ontology Guided Approach | The paper is accepted at LLM+KG: International Workshop on Data Management Opportunities in Unifying Large Language Models + Knowledge Graphs, VLDB 2024, August 26, 2024, Guangzhou, China. https://vldb.org/workshops/2024/proceedings/LLM+KG/LLM+KG-15.pdf | null | null | null | cs.CL | http://creativecommons.org/licenses/by/4.0/ | This study presents an approach that uses large language models such as GPT-4 to generate usage policies in the W3C Open Digital Rights Language ODRL automatically from natural language instructions. Our approach uses the ODRL ontology and its documentation as a central part of the prompt. Our research hypothesis is that a curated version of existing ontology documentation will better guide policy generation. We present various heuristics for adapting the ODRL ontology and its documentation to guide an end-to-end KG construction process. We evaluate our approach in the context of dataspaces, i.e., distributed infrastructures for trustworthy data exchange between multiple participating organizations for the cultural domain. We created a benchmark consisting of 12 use cases of varying complexity. Our evaluation shows excellent results with up to 91.95% accuracy in the resulting knowledge graph. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:38:41 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Mustafa",
"Daham M.",
""
],
[
"Nadgeri",
"Abhishek",
""
],
[
"Collarana",
"Diego",
""
],
[
"Arnold",
"Benedikt T.",
""
],
[
"Quix",
"Christoph",
""
],
[
"Lange",
"Christoph",
""
],
[
"Decker",
"Stefan",
""
]
] |
2506.03302 | James Bagrow | James Bagrow and Josh Bongard | Multi-Exit Kolmogorov-Arnold Networks: enhancing accuracy and parsimony | 14 pages, 7 figures, 2 tables | null | null | null | cs.LG cs.NE physics.data-an stat.ML | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Kolmogorov-Arnold Networks (KANs) uniquely combine high accuracy with interpretability, making them valuable for scientific modeling. However, it is unclear a priori how deep a network needs to be for any given task, and deeper KANs can be difficult to optimize. Here we introduce multi-exit KANs, where each layer includes its own prediction branch, enabling the network to make accurate predictions at multiple depths simultaneously. This architecture provides deep supervision that improves training while discovering the right level of model complexity for each task. Multi-exit KANs consistently outperform standard, single-exit versions on synthetic functions, dynamical systems, and real-world datasets. Remarkably, the best predictions often come from earlier, simpler exits, revealing that these networks naturally identify smaller, more parsimonious and interpretable models without sacrificing accuracy. To automate this discovery, we develop a differentiable "learning to exit" algorithm that balances contributions from exits during training. Our approach offers scientists a practical way to achieve both high performance and interpretability, addressing a fundamental challenge in machine learning for scientific discovery. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:41:30 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Bagrow",
"James",
""
],
[
"Bongard",
"Josh",
""
]
] |
2506.03303 | Mustafa Eyceoz | Mustafa Eyceoz, Nikhil Shivakumar Nayak, Hao Wang, Ligong Han, Akash Srivastava | Hopscotch: Discovering and Skipping Redundancies in Language Models | 10 pages, 4 figures, 9 tables | null | null | null | cs.CL cs.AI cs.LG | http://creativecommons.org/licenses/by/4.0/ | Modern causal language models stack many attention blocks to improve performance, but not all blocks are necessary for every task. We propose Hopscotch, a simple yet effective method that identifies and skips attention blocks with least contributions to a task and adapts to preserve output quality. Hopscotch jointly optimizes which blocks to skip and how to scale the outputs of the remaining layers. By introducing lightweight, trainable scaling parameters to attention and MLP blocks, it mitigates distribution shifts in hidden states caused by removing attention blocks. Hopscotch does not modify model weights or require access to pretraining or instruction-tuning data, and is compatible with existing model compression techniques. When applied to $\texttt{Llama-3.1-8B}$ and $\texttt{Qwen2.5-7B}$, Hopscotch achieves less than a 2% drop in performance even after skipping four attention blocks. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:43:00 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Eyceoz",
"Mustafa",
""
],
[
"Nayak",
"Nikhil Shivakumar",
""
],
[
"Wang",
"Hao",
""
],
[
"Han",
"Ligong",
""
],
[
"Srivastava",
"Akash",
""
]
] |
2506.03304 | Matthew Hennefarth | Matthew R. Hennefarth, Younghwan Kim, Bhavnesh Jangid, Jacob Wardzala, Matthew R. Hermes, Donald G. Truhlar and Laura Gagliardi | MC-PDFT Nuclear Gradients and L-PDFT Energies with Meta and Hybrid Meta On-Top Functionals for Ground- and Excited-State Geometry Optimization and Vertical Excitation Energies | 27 pages, 8 figures | null | null | null | physics.chem-ph quant-ph | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ | Multiconfiguration pair-density functional theory (MC-PDFT) is a post-MCSCF multireference electronic-structure method that explicitly models strong electron correlation, and linearized pair-density functional theory (L-PDFT) is a recently developed multi-state extension that can accurately model conical intersections and locally-avoided crossings. Because MC-PDFT and L-PDFT rely on an on-top energy functional, their accuracy depends on the quality of the on-top functional used. Recent work has introduced translated meta-gradient-approximation (meta-GA) on-top functionals, and specifically the MC23 hybrid meta-GA on-top functional, which is the first on-top functional specifically optimized for MC-PDFT. Here we report the derivation and implementation of analytic nuclear gradients for MC-PDFT calculations using meta-GA and hybrid meta-GA on-top functionals. This development also enables analytic nuclear gradients for the widely successful tPBE0 hybrid on-top functional. Because MC-PDFT nuclear-gradient calculations involve the derivative of the on-top functional, this development also enables the use of meta-GA on-top functionals in L-PDFT single-point energy calculations. We use the new capabilities to test MC23 for ground-state geometries, excited-state geometries, and vertical excitation energies of s-trans-butadiene and benzophenone as well as to test MC23, another hybrid meta-GA, and seven other meta-GA on-top functionals for 441 vertical excitation energies. We find MC23 performs the best of all nine meta and hybrid meta functionals for vertical excitation energies and is comparable in accuracy to tPBE0 and to the NEVPT2 multireference wave function method. Additionally, we directly compare our MC-PDFT vertical excitation results to previously computed TD-DFT values and find that MC-PDFT outperforms even the best performing Kohn-Sham density functional. | [
{
"version": "v1",
"created": "Tue, 03 Jun 2025 18:44:07 GMT"
}
] | 2025-06-05T00:00:00 | [
[
"Hennefarth",
"Matthew R.",
""
],
[
"Kim",
"Younghwan",
""
],
[
"Jangid",
"Bhavnesh",
""
],
[
"Wardzala",
"Jacob",
""
],
[
"Hermes",
"Matthew R.",
""
],
[
"Truhlar",
"Donald G.",
""
],
[
"Gagliardi",
"Laura",
""
]
] |
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.