{"id":"arxiv/2611.18992","url":"https://arxiv.org/abs/2611.18992","tags":["math.NA","math.OC"],"score":0,"title":"NeRI: Neural Residual Iteration for Elliptic Operator Equations and Variational Inequalities","domain":"math.NA","source":"arxiv","authors":["Michael Hintermüller","Jianfeng Ning"],"reviews":0,"subject":"math.NA","summary":"In this work, we propose a so-called {\\it Neural Residual Iteration} (NeRI), a weak-form neural-network framework for solving a class of elliptic operator equations and variational inequalities, including nonlinear and nonsymmetric problems that may not admit a global energy functional. The key idea is to construct, at each iteration, a state-dependent linear functional whose derivative with respect to the trial function recovers the current weak residual. A single gradient step is then performed on the residual, yielding a residual-driven neural update without requiring a global energy…","category":"research","comments":0,"disputed":false,"division":"math","mentions":0,"discussed":0,"published":"2026-11-13","reactions":{}}