[{"data":1,"prerenderedAt":430},["ShallowReactive",2],{"content-query-5ejaOuETdw":3},{"_path":4,"_dir":5,"_draft":6,"_partial":6,"_locale":7,"title":8,"description":9,"date":10,"cover":11,"type":12,"state":13,"body":14,"_type":424,"_id":425,"_source":426,"_file":427,"_stem":428,"_extension":429},"/activities/zh/251","zh",false,"","MindSpore社区8月新社区群组介绍系列之三","WG-MM的主要职责将会是集中梳理分子动力学研究对MindSpore的需求、在每个版本周期同相关SIG合作落地相应诉求、以及形成相关的教程性文档。","2020-09-07","https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2020/09/21/c9d74d8d78bc4ea58bd7051e8158cc6a.png","activities",1,{"type":15,"children":16,"toc":418},"root",[17,25,32,38,43,48,63,68,73,78,101,106,111,116,121,126,131,136,141,146,154,159,164,192,197,225,230,239,244,249,257,262,267,274,279,286,291,296,301,306,311,322,327,332,342,349,354,358,369,374,379,390,401,406,411],{"type":18,"tag":19,"props":20,"children":22},"element","h1",{"id":21},"mindspore社区8月新社区群组介绍系列之三",[23],{"type":24,"value":8},"text",{"type":18,"tag":26,"props":27,"children":29},"h2",{"id":28},"wg-molecular-modeling",[30],{"type":24,"value":31},"WG- Molecular Modeling",{"type":18,"tag":33,"props":34,"children":35},"p",{},[36],{"type":24,"value":37},"什么是分子动力学",{"type":18,"tag":33,"props":39,"children":40},{},[41],{"type":24,"value":42},"分子动力学[0]最早在20世纪50年代由物理学家提出，是一套分子模拟方法，该方法主要是依靠计算机来模拟分子、原子体系的运动，是一种多体模拟方法。",{"type":18,"tag":33,"props":44,"children":45},{},[46],{"type":24,"value":47},"分子动力学如今被广泛应用于广泛应用于物理、化学、生物、材料、医学等领域：",{"type":18,"tag":49,"props":50,"children":51},"ul",{},[52,58],{"type":18,"tag":53,"props":54,"children":55},"li",{},[56],{"type":24,"value":57},"通过研究石墨烯中的离子输运来了解石墨烯的机械性质，从而为创造新材料做好基础；",{"type":18,"tag":53,"props":59,"children":60},{},[61],{"type":24,"value":62},"在新冠肺炎病毒变异研究中利用分子动力学来预测药物靶位，从而研制出高效的疫苗。",{"type":18,"tag":33,"props":64,"children":65},{},[66],{"type":24,"value":67},"可以说是当之无愧的基础核心技术。",{"type":18,"tag":33,"props":69,"children":70},{},[71],{"type":24,"value":72},"具体来说分子动力学是通过对分子、原子在一定时间内运动状态的模拟，从而以动态观点考察系统随时间演化的行为。通常分子、原子的动向是通过数值求解牛顿运动方程得到，势能（或其对笛卡尔坐标的一阶偏导数，即力）通常可以由分子间相互作用势能函数、分子力学力场、全始计算给出。",{"type":18,"tag":33,"props":74,"children":75},{},[76],{"type":24,"value":77},"势能函数的计算方式通常有两种：",{"type":18,"tag":49,"props":79,"children":80},{},[81,86,91,96],{"type":18,"tag":53,"props":82,"children":83},{},[84],{"type":24,"value":85},"量子力学：",{"type":18,"tag":53,"props":87,"children":88},{},[89],{"type":24,"value":90},"又称为“从头算”（ab initio）方法或“第一性原理”（ First Principle ）方法，指基于量子力学基本原理，直接求解薛定谔方程（Schrödinger Equation ）的量子化学计算方法。其特点是没有经验参数，精度高，可以描述化学键的断裂和生成等涉及电子结构变化的过程，但计算耗时极长。",{"type":18,"tag":53,"props":92,"children":93},{},[94],{"type":24,"value":95},"分子力学：",{"type":18,"tag":53,"props":97,"children":98},{},[99],{"type":24,"value":100},"使用分子力场（Force Field）来描述体系的势能函数。传统的分子力场使用一系列经验性的数学函数，对用QM放法计算或者实验得到的数据进行参数拟合，从而得到体系的势能函数。该方法的特点是计算速度快，但精度受限于所拟合的对象，可迁移性差，且通常无法模拟化学键的断裂和生成，也即无法模拟化学反应的过程。",{"type":18,"tag":33,"props":102,"children":103},{},[104],{"type":24,"value":105},"分子动力学模拟的现实与需求",{"type":18,"tag":33,"props":107,"children":108},{},[109],{"type":24,"value":110},"优化的对象→MD的可达到的时间尺度：",{"type":18,"tag":33,"props":112,"children":113},{},[114],{"type":24,"value":115},"能量函数的计算复杂度",{"type":18,"tag":33,"props":117,"children":118},{},[119],{"type":24,"value":120},"积分步长",{"type":18,"tag":33,"props":122,"children":123},{},[124],{"type":24,"value":125},"体系粒子数",{"type":18,"tag":33,"props":127,"children":128},{},[129],{"type":24,"value":130},"优化的目标→如何拓展MD的时间尺度:",{"type":18,"tag":33,"props":132,"children":133},{},[134],{"type":24,"value":135},"特殊硬件: ×10^3",{"type":18,"tag":33,"props":137,"children":138},{},[139],{"type":24,"value":140},"增强抽样方法：×10^3~6",{"type":18,"tag":33,"props":142,"children":143},{},[144],{"type":24,"value":145},"能够快速计算的替代势能函数(light-speed surrogate energy function)：×10^3~6",{"type":18,"tag":33,"props":147,"children":148},{},[149],{"type":18,"tag":150,"props":151,"children":153},"img",{"alt":7,"src":152},"https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2020/09/07/c104a460e0f541b39350ba536b2e76d2.jpg",[],{"type":18,"tag":33,"props":155,"children":156},{},[157],{"type":24,"value":158},"传统分子动力学软件的问题及应对策略",{"type":18,"tag":33,"props":160,"children":161},{},[162],{"type":24,"value":163},"问题：",{"type":18,"tag":49,"props":165,"children":166},{},[167,172,177,182,187],{"type":18,"tag":53,"props":168,"children":169},{},[170],{"type":24,"value":171},"多年前开始使用C/C++或Fortran编写，历史包袱沉重",{"type":18,"tag":53,"props":173,"children":174},{},[175],{"type":24,"value":176},"需要对所有函数手工写其对应的导数，以计算势能和力",{"type":18,"tag":53,"props":178,"children":179},{},[180],{"type":24,"value":181},"需要手工处理并行（MPI）操作",{"type":18,"tag":53,"props":183,"children":184},{},[185],{"type":24,"value":186},"在专用加速设备（如GPU）上运行需要单独编写对应代码，不同设备之间的迁移非常麻烦",{"type":18,"tag":53,"props":188,"children":189},{},[190],{"type":24,"value":191},"传统分子力场模型存在“精度”和“速度”之间的矛盾",{"type":18,"tag":33,"props":193,"children":194},{},[195],{"type":24,"value":196},"应对策略：",{"type":18,"tag":49,"props":198,"children":199},{},[200,205,210,215,220],{"type":18,"tag":53,"props":201,"children":202},{},[203],{"type":24,"value":204},"使用新的库重新编写MD软件",{"type":18,"tag":53,"props":206,"children":207},{},[208],{"type":24,"value":209},"使用自动微分技术处理势能和力",{"type":18,"tag":53,"props":211,"children":212},{},[213],{"type":24,"value":214},"使用自动并行技术实现MD并行操作",{"type":18,"tag":53,"props":216,"children":217},{},[218],{"type":24,"value":219},"使用TVM等技术实现代码在不同设备间的迁移",{"type":18,"tag":53,"props":221,"children":222},{},[223],{"type":24,"value":224},"使用深度分子模型替代传统分子力场模型",{"type":18,"tag":33,"props":226,"children":227},{},[228],{"type":24,"value":229},"深度分子模型：深度学习与分子动力学的结合[1]",{"type":18,"tag":33,"props":231,"children":232},{},[233],{"type":18,"tag":234,"props":235,"children":236},"strong",{},[237],{"type":24,"value":238},"现有的深度分子模型架构：",{"type":18,"tag":33,"props":240,"children":241},{},[242],{"type":24,"value":243},"基于描述符（Descriptor）的模型",{"type":18,"tag":33,"props":245,"children":246},{},[247],{"type":24,"value":248},"基于图神经网络（Graph Neural Network）的模型",{"type":18,"tag":33,"props":250,"children":251},{},[252],{"type":18,"tag":234,"props":253,"children":254},{},[255],{"type":24,"value":256},"基于图神经网络的分子模型：",{"type":18,"tag":33,"props":258,"children":259},{},[260],{"type":24,"value":261},"SchNet: arXiv:1706.08566",{"type":18,"tag":33,"props":263,"children":264},{},[265],{"type":24,"value":266},"PhysNet: arXiv:1902.08408",{"type":18,"tag":33,"props":268,"children":269},{},[270],{"type":18,"tag":150,"props":271,"children":273},{"alt":7,"src":272},"https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2020/09/07/2777d70f1de0486d8342403e27a0e9dc.jpg",[],{"type":18,"tag":33,"props":275,"children":276},{},[277],{"type":24,"value":278},"为什么要发起WG-MM（分子动力学工作组）",{"type":18,"tag":33,"props":280,"children":281},{},[282],{"type":18,"tag":150,"props":283,"children":285},{"alt":7,"src":284},"https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2020/09/07/63a71a590ec04997a63801382490b538.jpg",[],{"type":18,"tag":33,"props":287,"children":288},{},[289],{"type":24,"value":290},"如前所述，深度分子模型的实现，需要在框架层面进行多方面的适配，而在非常适合科学计算场景的MindSpore社区成立协调多个SIG协作的工作组，成为了深圳湾实验室分子动力学团队进行相关开源协作的首选。",{"type":18,"tag":33,"props":292,"children":293},{},[294],{"type":24,"value":295},"WG-MM的主要职责将会是集中梳理分子动力学研究对MindSpore的需求、在每个版本周期同相关SIG合作落地相应诉求、以及形成相关的教程性文档。我们在此特别感谢深圳湾实验室团队选择MindSpore社区进行这样卓有远见的研究和开发。",{"type":18,"tag":33,"props":297,"children":298},{},[299],{"type":24,"value":300},"如何参与WG-MM",{"type":18,"tag":33,"props":302,"children":303},{},[304],{"type":24,"value":305},"1. 参加社区开发非常简单：",{"type":18,"tag":33,"props":307,"children":308},{},[309],{"type":24,"value":310},"请先查看WG-MM的立项文档：",{"type":18,"tag":33,"props":312,"children":313},{},[314],{"type":18,"tag":315,"props":316,"children":320},"a",{"href":317,"rel":318},"https://gitee.com/mindspore/community/blob/master/design/meps/mep-mm/MEP-MM.md",[319],"nofollow",[321],{"type":24,"value":317},{"type":18,"tag":33,"props":323,"children":324},{},[325],{"type":24,"value":326},"如果你有任何问题，欢迎联系文档开头提到的几位来自深圳湾实验室分子动力学团队的专家老师",{"type":18,"tag":33,"props":328,"children":329},{},[330],{"type":24,"value":331},"2. 在社区提交ISSUE或者PR，正式开始提出你的问题或者做出你的贡献",{"type":18,"tag":33,"props":333,"children":334},{},[335,337],{"type":24,"value":336},"3. 微信添加好友搜索：",{"type":18,"tag":234,"props":338,"children":339},{},[340],{"type":24,"value":341},"mindspore0328",{"type":18,"tag":33,"props":343,"children":344},{},[345],{"type":18,"tag":150,"props":346,"children":348},{"alt":7,"src":347},"https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2020/09/07/4c1e4a64977145da9725ae2dbccbc1d1.jpg",[],{"type":18,"tag":33,"props":350,"children":351},{},[352],{"type":24,"value":353},"参考文献",{"type":18,"tag":26,"props":355,"children":356},{"id":353},[357],{"type":24,"value":353},{"type":18,"tag":33,"props":359,"children":360},{},[361,363],{"type":24,"value":362},"[0]",{"type":18,"tag":315,"props":364,"children":367},{"href":365,"rel":366},"https://zh.wikipedia.org/wiki/%E5%88%86%E5%AD%90%E5%8A%A8%E5%8A%9B%E5%AD%A6",[319],[368],{"type":24,"value":365},{"type":18,"tag":33,"props":370,"children":371},{},[372],{"type":24,"value":373},"[1]Zhang, J.; Lei, Y.-K.; Zhang, Z.; et al. A Perspective on Deep Learning for Molecular Modeling and Simulations. J. Phys. Chem. A 2020, DOI: 10.1021/acs.jpca.0c04473",{"type":18,"tag":33,"props":375,"children":376},{},[377],{"type":24,"value":378},"MindSpore官方资料",{"type":18,"tag":33,"props":380,"children":381},{},[382,384],{"type":24,"value":383},"GitHub:",{"type":18,"tag":315,"props":385,"children":388},{"href":386,"rel":387},"https://github.com/mindspore-ai/mindspore",[319],[389],{"type":24,"value":386},{"type":18,"tag":33,"props":391,"children":392},{},[393,395],{"type":24,"value":394},"Gitee:",{"type":18,"tag":315,"props":396,"children":399},{"href":397,"rel":398},"https://gitee.com/mindspore/mindspore",[319],[400],{"type":24,"value":397},{"type":18,"tag":33,"props":402,"children":403},{},[404],{"type":24,"value":405},"官方QQ群: 871543426",{"type":18,"tag":33,"props":407,"children":408},{},[409],{"type":24,"value":410},"扫描下方二维码关注我们",{"type":18,"tag":33,"props":412,"children":413},{},[414],{"type":18,"tag":150,"props":415,"children":417},{"alt":7,"src":416},"https://obs-mindspore-file.obs.cn-north-4.myhuaweicloud.com/file/2020/09/07/a2cd3c00095641bb83a26811ca9fca29.jpg",[],{"title":7,"searchDepth":419,"depth":419,"links":420},4,[421,423],{"id":28,"depth":422,"text":31},2,{"id":353,"depth":422,"text":353},"markdown","content:activities:zh:251.md","content","activities/zh/251.md","activities/zh/251","md",1776506018803]