Reinforcement learning closures for underresolved partial differential equations using synthetic data

doi:10.1016/j.cma.2026.118767

Reinforcement learning closures for underresolved partial differential equations using synthetic data

2026-01-22

0

PRE

AI

DOI：10.1016/j.cma.2026.118767

Original

Original request for help

Save

Abstract

En 中文

• Introduce a new approach to develop closures for underresolved PDEs in the regime of scarce data. • Generate synthetic training data based on the method of manufactured solutions. • Extend Closure-RL for rewards based on synthetic data. • Develop generalizable learned closure models and verify them for unseen test cases.

Reinforcement learning closures for underresolved partial differential equations using synthetic data

Abstract

Journal

Organization