SI2-SSE: Enabling Chemical Accuracy in Computer Simulations: An Integrated Software Platform for Many-Body Molecular Dynamics

SI2-SSE：实现计算机模拟中的化学准确性：多体分子动力学集成软件平台

• 批准号: 1642336
• 负责人: Francesco Paesani
• 金额: $ 49.99万
• 依托单位: University of California-San Diego
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2017
• 资助国家: 美国
• 起止时间: 2017-04-01 至 2020-03-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=1642336&HistoricalAwards=false
• 关键词: SI2; SSE; Enabling; Chemical; Accuracy

This project is jointly funded by the Office of Advanced Cyberinfrastructure and and the Division of Chemistry within the Directorate of Mathematical and Physical Sciences. As attested by the 2013 Nobel Prize in Chemistry awarded to Martin Karplus, Michael Levitt, and Arieh Warshel, molecular-level computer simulations have become indispensable in many research areas, including chemistry, physics, materials science, and biochemistry, and often provide fundamental insights that are otherwise difficult to obtain. Nowadays, computer simulations are used to complement, guide, and sometimes replace experimental measurements, reducing the amount of time and money spent on research to bring ideas from the lab to practical applications. In the pharmaceutical industry computer simulations play a key role in structure-based drug design as demonstrated by the development of HIV protease inhibitors. In the chemical industry, computer modeling guides the design of new catalysts as well as novel materials for applications in more efficient batteries, and fuel and solar cells. More recently, there has been significant success in using of computer simulations to design more effective chemical processes as well as to provide information on safety issues. However, both the realism and the predictive power of a molecular-level computer simulation directly depend on the accuracy with which the interactions between molecules are described. To address the limitations of existing simulation approaches, the project group has recently developed a new theoretical/computational methodology that has been shown to display unprecedented accuracy when applied to a variety of molecular systems. The overarching goal of the proposed research is the implementation of this new methodology into an integrated and publicly available software platform that will allow the scientific community to address a broad range of problems through computer simulations. Potential applications include, but are not limited to, the rational design of new drugs as well as novel materials for water purification and the detection of toxic compounds and explosives, the virtual screening of catalysts for more efficient chemical processes, the development of new batteries, solar and fuel cells, and biomolecular structure prediction. A diverse group of high school, undergraduate, and graduate students will be directly involved in different aspects of the proposed research. The students will thus acquire critical knowledge about computer simulations and programming, which will significantly enhance their competitiveness in today's computer-driven job market. Given its multidisciplinary and multifaceted nature, the proposed research will promote scientific progress at different levels and contribute to the development of new technologies that will advance the national health, prosperity and welfare, as well as secure the national defense.The proposed research focuses on the development and implementation of unique software elements that will enable computer simulations on both CPU and GPU architectures using the so-called many-body molecular dynamics (MB-MD) methodology developed by the Paesani group. These software elements will be made publicly available to the scientific community through an integrated platform. MB-MD is a new simulation methodology that has already been shown to provide unprecedented accuracy in molecular simulations of a variety of molecular systems from the gas to the condensed phase. The new software elements comprise three components integrated in a unique software platform: a suite of publicly available computational tools for the automated generation of many-body potential energy functions from electronic structure data; a client-server architecture for the calculation of the required electronic structure data through volunteer computing; independent CPU and GPU plugins for the OpenMM toolkit which will enable MB-MD simulations of generic molecular systems across different phases. In parallel with the proposed research and software engineering projects, outreach and mentoring activities to promote STEM disciplines among students from underprivileged and underrepresented minorities through the PI and Co-PI direct involvement in several outreach programs at UC San Diego and the San Diego Supercomputer Center. These activities are specifically designed to increase the involvement and advancement of women, minorities, and economically disadvantaged groups across different education levels, from high school to undergraduate and graduate students.

该项目由先进网络基础设施办公室和数学与物理科学局化学部共同资助。正如 2013 年诺贝尔化学奖授予 Martin Karplus、Michael Levitt 和 Arieh Warshel 所证明的那样，分子级计算机模拟已成为化学、物理、材料科学和生物化学等许多研究领域不可或缺的一部分，并且常常提供基本见解否则很难获得。如今，计算机模拟被用来补充、指导，有时甚至取代实验测量，减少了将实验室想法转化为实际应用的研究所花费的时间和金钱。在制药行业中，计算机模拟在基于结构的药物设计中发挥着关键作用，HIV 蛋白酶抑制剂的开发就证明了这一点。在化学工业中，计算机建模指导新催化剂以及新材料的设计，用于更高效的电池、燃料和太阳能电池。最近，使用计算机模拟来设计更有效的化学工艺以及提供有关安全问题的信息已取得重大成功。然而，分子水平计算机模拟的真实性和预测能力直接取决于描述分子之间相互作用的准确性。为了解决现有模拟方法的局限性，该项目组最近开发了一种新的理论/计算方法，该方法已被证明在应用于各种分子系统时显示出前所未有的准确性。拟议研究的总体目标是将这种新方法实施到一个集成的、公开的软件平台中，该平台将使科学界能够通过计算机模拟解决广泛的问题。潜在的应用包括但不限于新药的合理设计以及用于水净化和有毒化合物和爆炸物检测的新型材料，虚拟筛选更高效化学过程的催化剂，新型电池的开发，太阳能和燃料电池以及生物分子结构预测。高中生、本科生和研究生等不同群体将直接参与拟议研究的不同方面。因此，学生将获得有关计算机模拟和编程的关键知识，这将显着提高他们在当今计算机驱动的就业市场中的竞争力。鉴于其多学科和多方面的性质，拟议的研究将促进不同层面的科学进步，并有助于新技术的发展，从而促进国民健康、繁荣和福祉，以及确保国防安全。拟议的研究重点是开发和实施独特的软件元素，使用 Paesani 小组开发的所谓多体分子动力学 (MB-MD) 方法，在 CPU 和 GPU 架构上进行计算机模拟。这些软件元素将通过集成平台向科学界公开。 MB-MD 是一种新的模拟方法，已被证明可以为从气体到凝聚相的各种分子系统的分子模拟提供前所未有的精度。新的软件元素包括集成在独特软件平台中的三个组件：一套公开可用的计算工具，用于根据电子结构数据自动生成多体势能函数；客户端-服务器架构，用于通过志愿者计算计算所需的电子结构数据； OpenMM 工具包的独立 CPU 和 GPU 插件将能够对不同阶段的通用分子系统进行 MB-MD 模拟。在拟议的研究和软件工程项目的同时，还开展了外展和指导活动，通过 PI 和 Co-PI 直接参与加州大学圣地亚哥分校和圣地亚哥超级计算机中心的多个外展项目，在贫困和代表性不足的少数族裔学生中推广 STEM 学科。这些活动专门旨在提高从高中到本科生和研究生等不同教育级别的妇女、少数族裔和经济弱势群体的参与度和进步。

项目成果

Active learning of many-body configuration space: Application to the Cs + –water MB-nrg potential energy function as a case study

多体构型空间主动学习：以Cs→水MB-nrg势能函数为例

• DOI: 10.1063/5.0002162
• 发表时间: 2020-04
• 期刊: The Journal of Chemical Physics
• 作者: Zhai, Yaoguang;Caruso, Alessandro;Gao, Sicun;Paesani, Francesco
• 通讯作者: Paesani, Francesco

Comparison of permutationally invariant polynomials, neural networks, and Gaussian approximation potentials in representing water interactions through many-body expansions

通过多体展开表示水相互作用时置换不变多项式、神经网络和高斯近似势的比较

• DOI: 10.1063/1.5024577
• 发表时间: 2018-06
• 期刊: The Journal of Chemical Physics
• 作者: Nguyen, Thuong T.;Székely, Eszter;Imbalzano, Giulio;Behler, Jörg;Csányi, Gábor;Ceriotti, Michele;Götz, Andreas W.;Paesani, Francesco
• 通讯作者: Paesani, Francesco