arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2601.21985cs.LG
收藏

Elign:从基础机器学习力场中等效扩散模型对齐

Elign: Equivariant Diffusion Model Alignment from Foundational Machine Learning Force Fields

Department of Computer Science, Yale University, New Haven, USA(计算机科学系,耶鲁大学,新 Haven,USA);Program in Computational Biology and Biomedical Informatics, Yale University, New Haven, USA(计算生物学与生物医学信息学项目,耶鲁大学,新 Haven,USA);IQuestLab

Yunyang Li, Lin Huang, Luojia Xia, Wenhe Zhang, Mark Gerstein

AI总结:

Elign通过结合基础机器学习力场和强化学习优化,实现了等效扩散模型的对齐,提升分子构象生成的准确性和效率。

摘要:

生成3D分子构象的模型必须尊重欧几里得对称性,并将概率质量集中在热力学有利、机械稳定的结构上。然而,E(3)-等效扩散模型往往复现了半经验训练数据中的偏差,而不是捕捉高保真哈密顿量的平衡分布。虽然基于物理的指导可以纠正这一点,但它面临两个计算瓶颈:昂贵的量子化学评估(例如DFT)以及在每次采样步骤中重复此类查询的需要。我们提出了Elign,一种后训练框架,旨在 amortize 两种成本。首先,我们用更快的预训练基础机器学习力场(MLFF)替代昂贵的DFT评估,以提供物理信号。其次,我们通过将物理引导转移到训练阶段来消除重复的运行时查询。为了实现第二种amortization,我们将反向扩散公式化为强化学习问题,并引入Force--Energy Disentangled Group Relative Policy Optimization(FED-GRPO)来微调去噪策略。FED-GRPO包括基于势能的奖励和基于力的稳定性奖励,这些奖励被独立优化和分组归一化。实验表明,Elign生成的构象具有更低的黄金标准DFT能量和力,同时提高了稳定性。关键的是,推断速度仍与无指导采样一样快,因为在生成过程中不需要能量评估。

英文摘要:

Generative models for 3D molecular conformations must respect Euclidean symmetries and concentrate probability mass on thermodynamically favorable, mechanically stable structures. However, E(3)-equivariant diffusion models often reproduce biases from semi-empirical training data rather than capturing the equilibrium distribution of a high-fidelity Hamiltonian. While physics-based guidance can correct this, it faces two computational bottlenecks: expensive quantum-chemical evaluations (e.g., DFT) and the need to repeat such queries at every sampling step. We present Elign, a post-training framework that amortizes both costs. First, we replace expensive DFT evaluations with a faster, pretrained foundational machine-learning force field (MLFF) to provide physical signals. Second, we eliminate repeated run-time queries by shifting physical steering to the training phase. To achieve the second amortization, we formulate reverse diffusion as a reinforcement learning problem and introduce Force--Energy Disentangled Group Relative Policy Optimization (FED-GRPO) to fine-tune the denoising policy. FED-GRPO includes a potential-based energy reward and a force-based stability reward, which are optimized and group-normalized independently. Experiments show that Elign generates conformations with lower gold-standard DFT energies and forces, while improving stability. Crucially, inference remains as fast as unguided sampling, since no energy evaluations are required during generation.

机构、总结、中文摘要均由AI分析而来,可能存在错误,请以论文正文为准。

↑