GRIFDIR: Graph Resolution-Invariant FEM Diffusion Models in Function Spaces over Irregular Domains

James Rowbottom, Elizabeth L. Baker, Nick Huang, Ben Adcock, Carola-Bibiane Schönlieb, Alexander Denker · 2026

Read GRIFDIR: Graph Resolution-Invariant FEM Diffusion Models in Function Spaces over Irregular Domains by James Rowbottom, Elizabeth L. Baker, Nick Huang, Ben Adcock, Carola-Bibiane Schönlieb, Alexander Denker (2026) 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-2605-03497.txt.

Description

James Rowbottom, Elizabeth L. Baker, Nick Huang, Ben Adcock, Carola-Bibiane Schönlieb, and Alexander Denker introduce GRIFDIR, a diffusion model for irregular domains. It works on meshes over complex domains. Function-space diffusion promises resolution invariance. But standard backbones favor regular grids and fail on awkward shapes.

GRIFDIR builds its score network from graph kernels. Those kernels are written as finite element functions, so the network works directly on unstructured meshes of any shape. Sampling tests, plain and conditional, covered non-convex and multiply connected domains. The method kept resolution invariance and captured distributions with high fidelity.

Catalog record

CREATORJames Rowbottom, Elizabeth L. Baker, Nick Huang, Ben Adcock, Carola-Bibiane Schönlieb, Alexander Denker
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE66 KB
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2605.03497

Browse the full library · About the Tower of Babel