数据与计算发展前沿 ›› 2024, Vol. 6 ›› Issue (1): 3-11.

CSTR: 32002.14.jfdc.CN10-1649/TP.2024.01.001

doi: 10.11871/jfdc.issn.2096-742X.2024.01.001

• 专题:超算互联网及应用 • 上一篇    下一篇

聚合物分子动力学软件PyGAMD及在国产异构计算平台的适配

于卓辰1(),蒲鑫1,韩秉豫2,朱有亮1,*(),吕中元1   

  1. 1.吉林大学，化学学院，超分子结构与材料国家重点实验室，吉林 长春 130012
    2.河南师范大学，软件学院，河南 新乡 453007
  • 收稿日期:2023-09-20 出版日期:2024-02-20 发布日期:2024-02-21
  • 通讯作者: * 朱有亮（E-mail: youliangzhu@jlu.edu.cn）
  • 作者简介:于卓辰，吉林大学化学学院博士研究生在读。研究兴趣为数据驱动的材料设计。
    本文中负责撰写和汇总。
    YU Zhuochen is a Ph.D. candidate in College of Chemistry, Jilin University. His research interest is data-driven design of new materials.
    In this paper, he is responsible for article writing and summarizing.
    E-mail: zcyu@pygamd.com|朱有亮，吉林大学化学学院研究员，博士生导师。2013年获得吉林大学理学博士学位，师从吕中元教授，随后进入中国科学院长春应化所工作，2021年人才引进回到吉林大学。致力于发展聚合物分子动力学模拟方法和软件，提出了在粗粒化模型中描述可逆反应、氢键和配位键等的“动态键”方法，开发了GALAMOST(J. Comput. Chem. 34, 2197, 2013)和PYGAMD(http://pygamd.com/)等软件，软件应用结果已在Science等国际著名期刊发表论文220余篇。本人也在Nat. Electron., Adv. Mater., ACS Nano，Nano Lett.等杂志上发表论文90余篇，被引用超过1700次，获批软件著作权4项，主持国家自然科学基金青年项目和面上项目（2项）, 主持企业或高校横向项目多项。
    本文中负责文章修改。
    ZHU Youliang is currently a Research Fellow and Doctoral Supervisor at the College of Chemistry, Jilin University. In 2013, he received his PhD under the supervision of Professor LU Zhongyuan from Jilin University. He then joined the Changchun Institute of Applied Chemistry, Chinese Academy of Sciences, and moved back to Jilin University in 2021. He is committed to developing polymer molecular dynamics simulation methods and software. He proposed a "dynamic bond" method to describe reversible reactions, hydrogen bonds, coordination bonds, etc. at the coarse-grained level for polymers, and developed GALAMOST (J. Compute. Chem. 34, 2197, 2013) and PYGAMD (http://pygamd.com/) software. More than 220 papers of the software applications have been published in famous journals such as Science. He has published over 90 papers in refereed journals including Nat. Electron., Adv. Mater., ACS Nano and Nano Lett. with citations of 1700. He obtained 4 software copyrights and hosted the Youth Program and General Program (2 projects) of the National Natural Science Foundation, as well as multiple projects from enterprises or universities.
    In this paper, he is responsible for article modification.
    E-mail: youliangzhu@jlu.edu.cn
  • 基金资助:
    国家重点研发计划“关键聚合物结构材料集成计算设计方法与应用”(2022YFB3707300);国家自然科学基金“支持粗粒化-细粒化建模和二次开发的分子模拟软件”(22273031);国家自然科学基金“新型二维高分子的设计制备及其结构演化机制”(22133002);吉林大学计算交叉创新平台项目

Polymer Molecular Dynamics Simulation Software PyGAMD and Its Adaptation to the Domestic Heterogeneous Computing Platform

YU Zhuochen1(),PU Xin1,HAN Bingyu2,ZHU Youliang1,*(),LYU Zhongyuan1   

  1. 1. State Key Laboratory of Supramolecular Structure and Materials, College of Chemistry, Jilin University, Changchun, Jilin 130012, China
    2. College of Software, Henan Normal University, Xinxiang, Henan 453007, China
  • Received:2023-09-20 Online:2024-02-20 Published:2024-02-21

RichHTML

25

PDF

174

摘要：

【目的】分子动力学模拟是研究微/介观尺度下物质内部运动和相互作用的强大工具，使科研人员可以观察到实验难以捕捉的细节。开发并优化分子动力学模拟软件可以使科研人员能够分析和预测复杂系统的微观行为，推动材料和医药等领域的科学发现与创新。【方法】传统的分子动力学模拟软件主要基于Fortran、C或C++等编程语言开发，其复杂的代码结构使得大多数用户难以增加新的功能。为此，我们汲取自主开发聚合物分子动力学模拟软件的经验，基于解释型语言Python开发了分子动力学模拟解释器PyGAMD。【结果】该解释器支持用户自定义分子动力学模拟功能，还包括了聚合物建模工具、异构加速库、数据处理等工具，用户可以通过PyGAMD轻松组合所需模块完成分子动力学模拟工作。在计算效率方面，PyGAMD的表现在多数场合都非常出色。未来，PyGAMD的发展将侧重于构建高度自由的多模块化框架，同时开发机器学习势能和细粒化等功能。

关键词: 分子动力学, 粗粒化模拟, 解释器

Abstract:

[Objective] Molecular dynamics simulation is a powerful tool for studying the internal movement and interactions of matter at micro/meso-scales, allowing researchers to observe details that are difficult to capture experimentally. The development and optimization of the molecular dynamics software enable researchers to analyze and predict the microscopic behaviors of complex systems, driving the discovery and innovation of materials and drugs, etc. [Methods] Traditional molecular simulation software is mainly written in programming languages such as Fortran, C, and C++, and their complex code structures make it a challenge for most users to customize new features. For this reason, based on our experiences in developing the polymer molecular dynamics simulation software, we developed a molecular dynamics interpreter PyGAMD using the interpreted language Python. [Results] PyGAMD supports customized molecular dynamics functions. In addition, it includes polymer modeling tools, heterogeneous acceleration libraries, and data processing tools. With PyGAMD users can easily compose the required modules and run molecular dynamics simulations. In terms of computational efficiency, PyGAMD performs exceptionally well in most cases. The further development of PyGAMD will focus on a highly flexible multi-modular framework, as well as machine learning potentials and fine-graining functions.

Key words: molecular dynamics, coarse-grained simulation, interpreter