COMPOL: A Unified Neural Operator Framework for Scalable Multi-Physics Simulations

TL;DR

COMPOL is a novel neural operator framework that enhances multi-physics simulations by effectively modeling complex interdependencies among coupled physical processes, leading to improved predictive accuracy across diverse scientific domains.

Contribution

It introduces a flexible, architecture-agnostic framework incorporating recurrent and attention mechanisms to better capture interactions in multiphysics systems.

Findings

Outperforms existing methods in diverse benchmarks

Achieves higher predictive accuracy in complex simulations

Seamlessly integrates with various neural operator architectures

Abstract

Multiphysics simulations play an essential role in accurately modeling complex interactions across diverse scientific and engineering domains Although neural operators especially the Fourier Neural Operator FNO have significantly improved computational efficiency they often fail to effectively capture intricate correlations inherent in coupled physical processes To address this limitation we introduce COMPOL a novel coupled multiphysics operator learning framework COMPOL extends conventional operator architectures by incorporating sophisticated recurrent and attentionbased aggregation mechanisms effectively modeling interdependencies among interacting physical processes within latent feature spaces Our approach is architectureagnostic and seamlessly integrates into various neural operator frameworks that involve latent space transformations Extensive experiments on diverse benchmarksincluding biological reactiondiffusion systems patternforming chemical reactions multiphase geological flows and thermohydromechanical processes demonstrate that COMPOL consistently achieves superior predictive accuracy compared to stateoftheart methods.