Generative Latent Neural PDE Solver using Flow Matching

Zijie Li †, Anthony Zhou †, Amir Barati Farimani † · 2025

Read Generative Latent Neural PDE Solver using Flow Matching by Zijie Li †, Anthony Zhou †, Amir Barati Farimani † (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-2503-22600.txt.

Description

Zijie Li, Anthony Zhou, and Amir Barati Farimani build a new solver for equations that track change over time. Their twenty twenty-five paper deals with partial differential equations, used for flow, heat, and waves. Step-by-step models can drift, while noise-based training can steady them at a cost.

Their solver works in a small hidden space, not on the full grid. A coder shrinks each state, flow matching learns how states move, and a decoder rebuilds them. This cuts cost and works on odd shapes, not just neat grids. Tests show stable long forecasts and a way to gauge doubt in the results.

Catalog record

CREATORZijie Li †, Anthony Zhou †, Amir Barati Farimani †
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE35 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2503-22600

Browse the full library · About the Tower of Babel