Equivariant Flow Matching with Hybrid Probability Transport

TL;DR

This paper introduces geometric flow matching, combining equivariant modeling and stabilized probability dynamics to improve 3D molecule generation, achieving faster sampling and better performance over existing diffusion models.

Contribution

It proposes a hybrid probability path with equivariant optimal transport to enhance molecule generation models, addressing stability and efficiency issues.

Findings

Achieves 4.75× faster sampling on average.

Outperforms existing models on multiple benchmarks.

Improves stability of probability dynamics in diffusion models.

Abstract

The generation of 3D molecules requires simultaneously deciding the categorical features~(atom types) and continuous features~(atom coordinates). Deep generative models, especially Diffusion Models (DMs), have demonstrated effectiveness in generating feature-rich geometries. However, existing DMs typically suffer from unstable probability dynamics with inefficient sampling speed. In this paper, we introduce geometric flow matching, which enjoys the advantages of both equivariant modeling and stabilized probability dynamics. More specifically, we propose a hybrid probability path where the coordinates probability path is regularized by an equivariant optimal transport, and the information between different modalities is aligned. Experimentally, the proposed method could consistently achieve better performance on multiple molecule generation benchmarks with 4.75$\times$ speed up of sampling on average.