arXivDaily arXiv每日学术速递 周一至周五更新
arXiv周末暂无论文更新,休息一下吧,周末愉快~~
arXiv 2503.14569cs.LGcs.AI
收藏

势能分数匹配:利用势能引导消除分子结构采样的偏差

Potential Score Matching: Debiasing Molecular Structure Sampling with Potential Energy Guidance

Yau Mathematical Sciences Center, Tsinghua University(清华大学姚数学科学中心);Department of Mathematical Sciences, Tsinghua University(清华大学数学科学系);Microsoft Research AI4Science, Beijing, China(微软研究院AI4Science,北京,中国);School of Computer Science, Peking University(北京大学计算机科学学院);Beijing, China(中国北京)

Liya Guo, Zun Wang, Chang Liu, Junzhe Li, Pipi Hu, Yi Zhu

AI总结:

本文提出势能分数匹配(PSM),利用势能梯度引导生成模型,无需精确能量函数即可在有限且有偏差的数据上消除采样偏差,在LJ势及MD17/22数据集上优于现有最先进模型,生成的分子分布更接近玻尔兹曼分布。

摘要:

分子物理性质的系综平均与分子构象的分布密切相关,而对此类分布进行采样是物理学和化学中的一个基本挑战。传统方法如分子动力学(MD)模拟和马尔可夫链蒙特卡洛(MCMC)采样虽常用,但可能耗时且成本高昂。近年来,扩散模型通过学习训练数据的分布,已成为高效的替代方案。然而,获得无偏的目标分布仍然是一项昂贵的任务,主要因为它需要满足遍历性。为应对这些挑战,我们提出了势能分数匹配(PSM),一种利用势能梯度来引导生成模型的方法。PSM不需要精确的能量函数,即使在有限且有偏差的数据上训练,也能消除样本分布的偏差。我们的方法在常用的玩具模型伦纳德-琼斯(LJ)势上优于现有的最先进(SOTA)模型。此外,我们使用MD17和MD22数据集将PSM的评估扩展到高维问题。结果表明,与传统扩散模型相比,PSM生成的分子分布更接近玻尔兹曼分布。

英文摘要:

The ensemble average of physical properties of molecules is closely related to the distribution of molecular conformations, and sampling such distributions is a fundamental challenge in physics and chemistry. Traditional methods like molecular dynamics (MD) simulations and Markov chain Monte Carlo (MCMC) sampling are commonly used but can be time-consuming and costly. Recently, diffusion models have emerged as efficient alternatives by learning the distribution of training data. Obtaining an unbiased target distribution is still an expensive task, primarily because it requires satisfying ergodicity. To tackle these challenges, we propose Potential Score Matching (PSM), an approach that utilizes the potential energy gradient to guide generative models. PSM does not require exact energy functions and can debias sample distributions even when trained on limited and biased data. Our method outperforms existing state-of-the-art (SOTA) models on the Lennard-Jones (LJ) potential, a commonly used toy model. Furthermore, we extend the evaluation of PSM to high-dimensional problems using the MD17 and MD22 datasets. The results demonstrate that molecular distributions generated by PSM more closely approximate the Boltzmann distribution compared to traditional diffusion models.

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。

↑