OmniArch: Building Foundation Model For Scientific Computing

TL;DR

OmniArch is a pioneering foundation model designed for scientific computing, integrating multi-scale, multi-physics problems with physical alignment, achieving new benchmarks and adaptability in PDE solving.

Contribution

It introduces a unified architecture with Fourier encoding, Transformer backbone, and PDE-Aligner for physics-informed fine-tuning, pioneering multi-dimensional pre-training on PDEBench.

Findings

Sets new performance benchmarks for 1D, 2D, and 3D PDEs.

Demonstrates exceptional adaptability to new physics via in-context and zero-shot learning.

Supports realistic engineering applications and physics discovery.

Abstract

Foundation models have revolutionized language modeling, while whether this success is replicated in scientific computing remains unexplored. We present OmniArch, the first prototype aiming at solving multi-scale and multi-physics scientific computing problems with physical alignment. We addressed all three challenges with one unified architecture. Its pre-training stage contains a Fourier Encoder-decoder fading out the disharmony across separated dimensions and a Transformer backbone integrating quantities through temporal dynamics, and the novel PDE-Aligner performs physics-informed fine-tuning under flexible conditions. As far as we know, we first conduct 1D-2D-3D united pre-training on the PDEBench, and it sets not only new performance benchmarks for 1D, 2D, and 3D PDEs but also demonstrates exceptional adaptability to new physics via in-context and zero-shot learning approaches, which supports realistic engineering applications and foresight physics discovery.