Coupling-informed operator inference: Structure-preserving reduced-order modeling for multi-physics equations

Tengfei Zhang

doi:10.1016/j.jcp.2026.115145

Coupling-informed operator inference: Structure-preserving reduced-order modeling for multi-physics equations

2026-06-18

0

PRE

AI

DOI：10.1016/j.jcp.2026.115145

Original

Original request for help

Save

Abstract

En 中文

• Proposes coupling-informed operator inference, embedding governing equation structure/sparsity as constraints to build structure-preserving multi-physics reduced-order models. • Multi-basis POD per physics field is implemented to explicitly construct inter-field interactions from governing equations. • Employs masks to constrain regression to non-zero entries and introduces dominant component truncation for complexity reduction. • Demonstrates significantly improved accuracy and robustness over traditional operator inference on three multi-physics benchmarks with nonlinear and time-varying parametric features.

Keywords:

Coupling-informed operator inference

Structure-preserving reduced-order model

Physics-informed learning

Multi-physics dynamical system

Coupling-informed operator inference: Structure-preserving reduced-order modeling for multi-physics equations

Abstract

Journal

Organization