Equivariant Asynchronous Diffusion: An Adaptive Denoising Schedule for Accelerated Molecular Conformation Generation

TL;DR

The paper introduces Equivariant Asynchronous Diffusion (EAD), a novel 3D molecular generation model that combines asynchronous and synchronous diffusion advantages with adaptive scheduling, achieving state-of-the-art results.

Contribution

EAD is a new diffusion model that captures molecular hierarchy through asynchronous denoising and uses dynamic scheduling for improved 3D molecular generation.

Findings

EAD outperforms existing methods in 3D molecular generation tasks.

Adaptive denoising schedule improves model performance.

EAD effectively captures hierarchical molecular structures.

Abstract

Recent 3D molecular generation methods primarily use asynchronous auto-regressive or synchronous diffusion models. While auto-regressive models build molecules sequentially, they're limited by a short horizon and a discrepancy between training and inference. Conversely, synchronous diffusion models denoise all atoms at once, offering a molecule-level horizon but failing to capture the causal relationships inherent in hierarchical molecular structures. We introduce Equivariant Asynchronous Diffusion (EAD) to overcome these limitations. EAD is a novel diffusion model that combines the strengths of both approaches: it uses an asynchronous denoising schedule to better capture molecular hierarchy while maintaining a molecule-level horizon. Since these relationships are often complex, we propose a dynamic scheduling mechanism to adaptively determine the denoising timestep. Experimental results show that EAD achieves state-of-the-art performance in 3D molecular generation.