Collaborative: SI2-CHE: Development and Deployment of Chemical Software for Advanced Potential Energy Surfaces

一个由Teresa Head-Gordon和Martin Head-Gordon(加州大学伯克利分校)、Paul Nerenberg(克莱蒙特·麦肯纳学院)、David Case(罗格斯大学)、Jay Ponder(华盛顿大学)、Mark Tuckerman(纽约大学)组成的国际团队与他们的英国合作者：Lorna Smith和Neil Chue Hong(爱丁堡大学)、Chris-Kriton Skylaris和Jonathan W.Essex(南安普顿大学)、Ilian Todorov(Daresbury实验室)、Mario Antonioletti(EPCC)通过SI2-CHE计划得到支持，为先进的潜在能源表面开发和部署健壮且可持续的软件。新一代势能面的改进带来了更高的精度，这给它们在当前或新兴硬件平台上作为算法和软件的表现带来了几个挑战，进而限制了它们在计算化学界的广泛采用。研究团队正在通过多个但综合的方向克服这些障碍：(1)在多核和GPU支持的系统中以最佳方式实现先进的势能面，(2)开发一系列先进的可极化模型，改变精度和计算速度之间的权衡，(3)创造新的多时间推进方法；(4)编写完全支持相互极化的量子力学/分子力学(QM/MM)应用程序编程接口(API)；(5)采用软件最佳实践以确保自我维持社区的发展；(6)在几个新兴的应用领域利用新软件提供样本计算。分子模拟和量子化学软件是化学和化学生物学的组成部分，已被学术研究人员和工业科学家广泛采用。利用化学软件的下一代科学突破将通过部署最先进的理论模型和算法来实现，这些模型和算法将转化为可持续的软件框架，在新兴的高性能计算平台上快速实施。势能面描述了原子之间的相互作用。先进和高精度的势能面遇到了与软件相关的障碍，阻碍了它们在重大挑战化学问题中的应用。这个代表计算化学软件界广泛领域的英美财团，正致力于直接解决这些障碍。美国和英国的大学和高性能计算中心之间的合作努力证明，在先进的势能表面模型中进行的化学软件投资在社区可持续发展和下一代科学家培训方面具有长期回报。推广和培训研讨会是围绕先进的势能软件的出现而组织的，其中包括为本科生设立的介绍性分子模拟软件新兵训练营。美国的调查人员得到了NSF内CHE和ACI部门的支持；英国的调查人员得到了EPSRC的支持。

An international team consisting of Teresa Head-Gordon and Martin Head-Gordon (University of California, Berkeley), Paul Nerenberg (Claremont McKenna College), David Case (Rutgers University), Jay Ponder (Washington University), Mark Tuckerman (New York University) with their UK collaborators: Lorna Smith and Neil Chue Hong (University of Edinburgh), Chris-Kriton Skylaris and Jonathan W. Essex (University of Southampton), Ilian Todorov (Daresbury Laboratory), Mario Antonioletti (EPCC) are are supported through the SI2-CHE program to develop and deploy robust and sustainable software for advanced potential energy surfaces. The greater accuracy introduced by improvements in the new generation of potential energy surfaces opens up several challenges in their manifestation as algorithms and software on current or emergent hardware platforms that in turn limits their wide adoption by the computational chemistry community. The research team is overcoming these obstacles via multiple but integrated directions: (1) to optimally implement advanced potential energy surfaces across multi-core and GPU enabled systems, (2) to develop a hierarchy of advanced polarizable models that alter the tradeoff between accuracy and computational speed,(3) to create new multiple time stepping methods; (4) to write a Quantum Mechanics/Molecular Mechanics (QM/MM ) application programing interface (API) that fully supports mutual polarization, (5) to adopt software best practices to ensure growth of a self-sustaining community and (6) to provide exemplar calculations with the new software in the several emerging application areas.Molecular simulation and quantum chemistry software is an integral part of chemistry and chemical biology, and has been broadly adopted by academic researchers and industry scientists. Next generation scientific breakthroughs that utilize chemical software will be enabled by the deployment of state of the art theoretical models and algorithms that are translated into a sustainable software framework rapidly implemented on emergent high performance computing platforms. Potential energy surfaces describe the interactions between atoms. Advanced and highly accurate potential energy surfaces encounter software-related obstacles that inhibit their application to grand challenge chemistry problems. This UK and US consortium, representing a broad cross section of the computational chemistry software community, is working to directly tackle these obstacles. This US and UK collaboration between universities and High Performance Computing centers works to endure that chemical software investments made in advanced potential energy surface models has a long term payoff in community sustainability and the training of the next generation of scientists. Outreach and training workshops are organized around the emergence of the advanced potential energy software including an introductory molecular simulation software boot camp for undergraduate students.The US based investigators are supported by the CHE and ACI divisions within NSF; the UK based investigators are supported by the EPSRC.