Unfitted finite element interpolated neural networks

Wei Li 1, Alberto F. Martín 2, Santiago Badia 1,∗ · 2025

Read Unfitted finite element interpolated neural networks by Wei Li 1, Alberto F. Martín 2, Santiago Badia 1,∗ (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-2501-17438.txt.

Description

Wei Li, Alberto Martín, and Santiago Badia blend neural networks with finite elements to solve equations on tricky shapes in a paper from 2025. Many physics problems live on domains with curved or messy borders. Standard meshes must hug those borders, which is slow to build.

Their method lays a simple grid over the shape and lets cells cut through the border. A neural network is trained so that, once placed on the grid, it nearly solves the equation. Border rules are enforced weakly by a classic trick due to Nitsche. The approach stays robust when cut cells are tiny. Tests in two and three dimensions cover both forward and inverse problems well.

Catalog record

CREATORWei Li 1, Alberto F. Martín 2, Santiago Badia 1,∗
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE82 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2501-17438

Browse the full library · About the Tower of Babel