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 2606.02455cs.LGcond-mat.mtrl-sciphysics.chem-phphysics.comp-phstat.CO
收藏

用于加速分子动力学的推测采样

Speculative Sampling For Faster Molecular Dynamics

FAIR at Meta(Meta FAIR);School of Computation, Information & Technology, Technical University of Munich(技术大学慕尼黑计算、信息与技术学院);Munich Data Science Institute(慕尼黑数据科学研究所);Munich Center For Machine Learning(慕尼黑机器学习中心)

Arthur Kosmala, Stephan Günnemann, Meng Gao, Brandon Wood

AI总结:

提出Langevin推测动力学(LSD),一种分布式且模型无关的推测采样方法,通过草稿模型快速提议步长并用目标模型并行验证,实现分子动力学模拟的3-9倍加速而不增加相对误差。

摘要:

分子动力学(MD)是模拟原子系统动力学行为的关键工具。然而,MD本质上是串行的,这使得通过并发计算提高单系统吞吐量变得困难。为了解决这个问题,我们引入了Langevin推测动力学(LSD),一种分布式且模型无关的推测采样器,用于在不增加相对误差的情况下加速MD。受语言和扩散建模中推测方法的启发,LSD使用草稿模型提议快速模拟步长,并用较慢的目标模型并行验证,应用从草稿分布到目标分布的传输映射。我们将推测采样扩展到二阶Langevin动力学,推导出作为物理参数函数的可实现加速比,表明LSD在不同系统和草稿-目标组合中实现3-9倍加速,并从理论和实验上证实LSD从其目标模型分布中采样轨迹。

英文摘要:

Molecular dynamics (MD) is a key tool for simulating the dynamical behavior of atomic systems. However, MD is inherently serial, which makes it difficult to increase single-system throughput with concurrent compute. To address this, we introduce Langevin Speculative Dynamics (LSD), a distributed and model-agnostic speculative sampler for accelerating MD without adding relative error. Inspired by speculative methods in language and diffusion modeling, LSD uses a draft model to propose fast simulation steps and verifies them in parallel with a slower target model, applying a transport map from the draft to the target distribution. We extend speculative sampling to second-order Langevin dynamics, derive the achievable speedup as a function of physical parameters, show that LSD generalizes across different systems and draft-target combinations with a 3-9x speedup, and confirm theoretically and empirically that LSD samples trajectories from its target model distribution.

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。

↑