arXivDaily arXiv每日学术速递 周一至周五更新
arXiv 2402.03789cond-mat.mtrl-sci
收藏

分子动力学模拟中图神经网络原子间势的可扩展并行算法

Scalable Parallel Algorithm for Graph Neural Network Interatomic Potentials in Molecular Dynamics Simulations

Yutack Park, Jaesun Kim, Seungwoo Hwang, Seungwu Han

AI总结:

本文提出兼容消息传递 GNN 原子间势的并行方案,开发基于 NequIP 并对接 LAMMPS 的 SevenNet,在 32 GPU 基准中实现良好扩展,并利用 SevenNet-0 完成十万原子级非晶 Si$_3$N$_4$ 模拟。

摘要:

消息传递图神经网络原子间势(GNN-IP),尤其是诸如 NequIP 等具有等变表示的模型,因其数据效率和高精度正受到广泛关注。然而,GNN-IP 的并行化面临挑战,因为多个消息传递层使空间分解方法中的数据通信变得复杂,而该方法受到许多分子动力学(MD)软件包的青睐。本文提出一种与 GNN-IP 兼容的高效并行化方案,并基于 NequIP 架构开发了名为 SevenNet(Scalable EquiVariance-Enabled Neural NETwork)的软件包。对于 MD 模拟,SevenNet 与 LAMMPS 软件包对接。通过在含 32 块 GPU 的集群上以 SiO$_2$ 为例进行基准测试,SevenNet 在弱扩展情形下实现超过 80% 的并行效率,并且只要 GPU 得到充分利用,就表现出接近理想的强扩展性能。但是,当 GPU 利用率不佳时,强扩展性能显著下降,尤其会影响轻量模型或原子数较少的模拟中的并行效率。我们还使用来自 Materials Project 的大规模数据集预训练 SevenNet(称为 SevenNet-0),并评估其在生成包含超过 100,000 个原子的非晶 Si$_3$N$_4$ 时的性能。通过开发可扩展的 GNN-IP,本工作旨在弥合先进机器学习模型与大规模 MD 模拟之间的差距,为研究人员提供以高精度和高效率探索复杂材料系统的强大工具。

英文摘要:

Message-passing graph neural network interatomic potentials (GNN-IPs), particularly those with equivariant representations such as NequIP, are attracting significant attention due to their data efficiency and high accuracy. However, parallelizing GNN-IPs poses challenges because multiple message-passing layers complicate data communication within the spatial decomposition method, which is preferred by many molecular dynamics (MD) packages. In this article, we propose an efficient parallelization scheme compatible with GNN-IPs and develop a package, SevenNet (Scalable EquiVariance-Enabled Neural NETwork), based on the NequIP architecture. For MD simulations, SevenNet interfaces with the LAMMPS package. Through benchmark tests on a 32-GPU cluster with examples of SiO$_2$, SevenNet achieves over 80% parallel efficiency in weak-scaling scenarios and exhibits nearly ideal strong-scaling performance as long as GPUs are fully utilized. However, the strong-scaling performance significantly declines with suboptimal GPU utilization, particularly affecting parallel efficiency in cases involving lightweight models or simulations with small numbers of atoms. We also pre-train SevenNet with a vast dataset from the Materials Project (dubbed `SevenNet-0') and assess its performance on generating amorphous Si$_3$N$_4$ containing more than 100,000 atoms. By developing scalable GNN-IPs, this work aims to bridge the gap between advanced machine learning models and large-scale MD simulations, offering researchers a powerful tool to explore complex material systems with high accuracy and efficiency.

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。

↑