327 papers
This section explores the intersection where physics meets data analysis, a rapidly evolving frontier where complex datasets reveal hidden patterns in the universe. From tracking particle collisions to modeling cosmic structures, these studies rely on advanced statistical methods to turn raw numbers into fundamental insights about how reality works.
Gist.Science monitors every new preprint in this category as it appears on arXiv, ensuring you never miss a breakthrough. We process each entry to provide both plain-language overviews for general understanding and detailed technical summaries for experts, bridging the gap between dense research and clear comprehension.
Below are the latest papers in physics and data analysis, organized for easy reading and discovery.
This paper proposes a hierarchical topological clustering algorithm that can be applied with any distance metric to identify clusters of arbitrary shapes and persistent outliers across diverse datasets, including images, medical, and economic data.
The paper proposes the Poisson Log-Normal (PoLoN) process, a non-parametric framework that utilizes Gaussian processes to model Poisson log-rates, enabling effective prediction, signal detection, and parameter extraction for discrete count data.
This paper introduces `PoissonRatioUQ`, an R package designed for Bayesian modeling and uncertainty quantification of count ratios by treating them as ratios of Poisson means, offering various methods for spatial and non-linear intensity ratio problems.
By employing Tensor Programs and singular perturbation theory to analyze the fast-slow dynamics of infinite-width two-layer neural networks, this paper identifies the specific mechanisms and conditions—such as data nonlinearity and initial weight scales—that drive the phenomenon of feature unlearning during stochastic gradient descent.
DerivKit is a Python package that provides stable numerical differentiation tools to bridge the gap between fast Fisher-matrix forecasting and computationally intensive MCMC sampling by enabling higher-order likelihood approximations.
This paper proposes a novel method for estimating task-relevant latent dimensionality by framing it as an Information Bottleneck problem and introducing a hybrid neural critic that prevents dimension inflation, allowing for efficient, one-shot estimation across synthetic, geometric, and physics-based datasets.
This paper proposes an efficient method for certifying the performance of quantum resources in non-i.i.d. scenarios, demonstrating that only a constant average number of spot-checking trials is required to provide asymptotically tight performance guarantees.