Rigorous Error Certification for Neural PDE Solvers: From Empirical Residuals to Solution Guarantees

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Mukherjee, Amartya, Fitzsimmons, Maxwell, Fernández, David C. Del Rey, Liu, Jun
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915875945185280
author Mukherjee, Amartya
Fitzsimmons, Maxwell
Fernández, David C. Del Rey
Liu, Jun
author_facet Mukherjee, Amartya
Fitzsimmons, Maxwell
Fernández, David C. Del Rey
Liu, Jun
contents Uncertainty quantification for partial differential equations is traditionally grounded in discretization theory, where solution error is controlled via mesh/grid refinement. Physics-informed neural networks fundamentally depart from this paradigm: they approximate solutions by minimizing residual losses at collocation points, introducing new sources of error arising from optimization, sampling, representation, and overfitting. As a result, the generalization error in the solution space remains an open problem. Our main theoretical contribution establishes generalization bounds that connect residual control to solution-space error. We prove that when neural approximations lie in a compact subset of the solution space, vanishing residual error guarantees convergence to the true solution. We derive deterministic and probabilistic convergence results and provide certified generalization bounds translating residual, boundary, and initial errors into explicit solution error guarantees.
format Preprint
id arxiv_https___arxiv_org_abs_2603_19165
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Rigorous Error Certification for Neural PDE Solvers: From Empirical Residuals to Solution Guarantees
Mukherjee, Amartya
Fitzsimmons, Maxwell
Fernández, David C. Del Rey
Liu, Jun
Machine Learning
Analysis of PDEs
Functional Analysis
Uncertainty quantification for partial differential equations is traditionally grounded in discretization theory, where solution error is controlled via mesh/grid refinement. Physics-informed neural networks fundamentally depart from this paradigm: they approximate solutions by minimizing residual losses at collocation points, introducing new sources of error arising from optimization, sampling, representation, and overfitting. As a result, the generalization error in the solution space remains an open problem. Our main theoretical contribution establishes generalization bounds that connect residual control to solution-space error. We prove that when neural approximations lie in a compact subset of the solution space, vanishing residual error guarantees convergence to the true solution. We derive deterministic and probabilistic convergence results and provide certified generalization bounds translating residual, boundary, and initial errors into explicit solution error guarantees.
title Rigorous Error Certification for Neural PDE Solvers: From Empirical Residuals to Solution Guarantees
topic Machine Learning
Analysis of PDEs
Functional Analysis
url https://arxiv.org/abs/2603.19165