Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Kim, Doyoung, Lee, Donghee, Lee, Hye-Sung, Lee, Jiheon, Yi, Jaeok
Format: Preprint
Veröffentlicht: 2026
Schlagworte:
Online-Zugang:https://arxiv.org/abs/2601.05889
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866914274395291648
author Kim, Doyoung
Lee, Donghee
Lee, Hye-Sung
Lee, Jiheon
Yi, Jaeok
author_facet Kim, Doyoung
Lee, Donghee
Lee, Hye-Sung
Lee, Jiheon
Yi, Jaeok
contents In the analysis of complex physical systems, the objective often extends beyond merely computing a numerical solution to capturing the precise crossover between different regimes and extracting parameters containing meaningful information. However, standard numerical solvers and conventional deep learning approaches, such as Physics-Informed Neural Networks (PINNs), typically operate as black boxes that output solution fields without disentangling the solution into its interpretable constituent parts. In this work, we propose GlueNN, a physics-informed learning framework that decomposes the global solution into interpretable, patchwise analytic components. Rather than approximating the solution directly, GlueNN promotes the integration constants of local asymptotic expansions to learnable, scale-dependent coefficient functions. By constraining these coefficients with the differential equation, the network effectively performs regime transition, smoothly interpolating between asymptotic limits without requiring ad hoc boundary matching. We demonstrate that this coefficient-centric approach reproduces accurate global solutions in various examples and thus directly extracts physical information that is not explicitly available through standard numerical integration.
format Preprint
id arxiv_https___arxiv_org_abs_2601_05889
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle GlueNN: gluing patchwise analytic solutions with neural networks
Kim, Doyoung
Lee, Donghee
Lee, Hye-Sung
Lee, Jiheon
Yi, Jaeok
Machine Learning
Cosmology and Nongalactic Astrophysics
Computational Physics
In the analysis of complex physical systems, the objective often extends beyond merely computing a numerical solution to capturing the precise crossover between different regimes and extracting parameters containing meaningful information. However, standard numerical solvers and conventional deep learning approaches, such as Physics-Informed Neural Networks (PINNs), typically operate as black boxes that output solution fields without disentangling the solution into its interpretable constituent parts. In this work, we propose GlueNN, a physics-informed learning framework that decomposes the global solution into interpretable, patchwise analytic components. Rather than approximating the solution directly, GlueNN promotes the integration constants of local asymptotic expansions to learnable, scale-dependent coefficient functions. By constraining these coefficients with the differential equation, the network effectively performs regime transition, smoothly interpolating between asymptotic limits without requiring ad hoc boundary matching. We demonstrate that this coefficient-centric approach reproduces accurate global solutions in various examples and thus directly extracts physical information that is not explicitly available through standard numerical integration.
title GlueNN: gluing patchwise analytic solutions with neural networks
topic Machine Learning
Cosmology and Nongalactic Astrophysics
Computational Physics
url https://arxiv.org/abs/2601.05889