Statistical methods in physics
Delip Rao, Chris Callison-Burch
For researchers using LLM judges in rubric-based evaluation, this work clarifies metric redundancy and offers guidelines to avoid misleading reporting.
Tingjia Miao, Wenkai Jin, Muhua Zhang et al.
For AI researchers and physicists, this benchmark provides a realistic testbed for assessing LLMs' ability to perform autonomous scientific discovery, highlighting current limitations.
Oz Amram, Darius A. Faroughy, Tjarko Gerdes et al.
For researchers training large generative models for collider physics, this work provides the first empirical evidence that scaling laws for jet generation differ from language models, highlighting fundamental limits in data and compute scaling.
Gregor Krzmanc, Vinicius Mikuni, Benjamin Nachman et al.
This work demonstrates that particle physics foundation models can generalize across vastly different energy scales and detector technologies, enabling detector-agnostic inference for the particle physics community.
Zeyu Xia, Tyler Kim, Trevor Reed et al.
This work addresses unreliable convergence diagnostics in generative models for nuclear physics, with incremental improvements in evaluation protocols applicable to domains like medical imaging and finance.
Nihanth W. Cherukuru, Matt Rehme, Kirsten J. Mayer et al.
For climate scientists and meteorologists, this workbench addresses the need to inspect and verify embedding-based similarity searches against physical evidence, enabling more trustworthy discovery in large datasets.
Sascha Diefenbacher, Sofia Palacios Schweitzer, Gregor Kasieczka
This work addresses the problem of validating generative models for physicists using them as fast surrogates and density estimators.
Shu Zhou, K. Y. Michael Wong, Juntao Wang et al.
This work addresses the lack of theoretical insights for practitioners using analog solvers, though it is incremental as it builds on existing dynamical systems approaches.
Shi-Yuan Ma, Jérémie Laydevant, Mandar M. Sohoni et al.
This addresses the problem of enabling accurate machine vision in photon-starved scenarios for applications like consumer devices and scientific instruments, representing a novel method rather than an incremental improvement.
Cole Granger, James Giroux, Richard Tyson et al.
For high-energy physics experiments requiring fast, high-fidelity detector simulation, this work offers a practical deep generative alternative to computationally expensive Monte-Carlo methods.
Itay Lavie, Noam Levi, Yonatan Kahn
For researchers applying deep learning to physics, this review clarifies how scaling laws and inductive biases affect model performance, but it is primarily a survey without new results.
Maria Kaselimi, Anna Belehaki
For Earth system modellers, this review provides a structured foundation for understanding AI's role in advancing coupled modelling, though it is a conceptual overview without concrete results.
Lorenzo Livi
This provides theoretical foundations for understanding temporal learning limits in RNNs, which is important for researchers working on sequence modeling.
Gaia Grosso, Vinicius Mikuni, Lukas Heinrich
For physicists using ML in particle physics, astrophysics, and cosmology, this paper provides a framework for verifying AI reliability before discovery claims, but it is a review without concrete results.
Biwei Dai, Po-Wen Chang, Wahid Bhimji et al.
This work addresses the need for standardized benchmarks and robust ML methods to handle systematic uncertainties and distribution shifts in weak lensing analysis, which is critical for upcoming cosmological surveys.
Kensuke Muto, Hirotaka Sakamoto, Kenji Nagata et al.
This work addresses inefficiency in neutron-scattering experiments for researchers, though it is incremental as it builds on prior bin-width optimization methods.
Farouk Mokhtar, Joosep Pata, Michael Kagan et al.
This work provides a concrete demonstration of a foundation model for collider physics, enabling shared representations across reconstruction and analysis tasks, which is a step toward end-to-end pipelines for high-energy physics.
Addis Fuhr, Zachary R. Fox, David Parker et al.
This work provides a physically interpretable machine-learning representation for predicting magnetic properties in 2D materials, addressing a key bottleneck in materials discovery for spintronics and quantum technologies.
Sobhi Saeed, Mehmet Müftüoglu, Glitta R. Cheeran et al.
For researchers in physical reservoir computing, this study provides practical training principles to optimize performance and reduce overfitting.
Rikab Gambhir, Luisa Lucie-Smith, Jesse Thaler
Provides conceptual clarity for physicists and ML practitioners navigating model transparency in scientific applications.