A Learning-Based Ansatz Satisfying Boundary Conditions in Variational Problems

Rafael Florencio, Julio Guerrero · 2025

Read A Learning-Based Ansatz Satisfying Boundary Conditions in Variational Problems by Rafael Florencio, Julio Guerrero (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2505-12430.txt.

Description

Rafael Florencio and Julio Guerrero fix a weak point in a learning-based solver for physics problems in a 2025 paper. The Deep Ritz method trains a neural network to solve a variational problem, the kind that asks for the function that makes an energy-like quantity smallest. A network does not naturally obey the problem's boundary conditions, the fixed values at the edges.

The usual patch fines the network for breaking the boundary rule, but that fine depends on a tunable weight and can mislead training. The authors instead design the trial form so the boundary conditions hold by construction, with no fine at all. They prove their supporting theorems in the Sobolev norms that fit variational problems, where both the function and its slopes enter the score. The result is a cleaner solver for a wide class of problems.

Catalog record

CREATORRafael Florencio, Julio Guerrero
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE45 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2505-12430

Browse the full library · About the Tower of Babel