Massive Parallelization of Exact Quantum Dynamics Calculations: Computing (ro)vibrational states for real molecular applications

德克萨斯理工大学的Lionel (Bill) Poirier获得了化学部门理论、模型和计算方法项目的奖励，开发和测试大规模并行高性能计算代码，使精确量子动力学（QD）旋转振动光谱计算能够用于比目前更大、更具有挑战性的分子应用。该奖项由保险业监理处的一项试点计划共同资助，该计划旨在促进网络基础设施（CI）元素的再利用。这种能力对于各种动力学、动力学和统计建模工作是有价值的，对于这些工作，有关分子种类的旋转振动状态和配分函数的准确知识通常是至关重要的。与以往的量子动力学研究相比，这项研究的一个新特点是使用了大规模并行化。许多必要的网络基础设施，即PI之前开发的并行线性代数代码，已经到位。为了评估并行代码的性能、可扩展性和广泛适用性，我们考虑了两个极具挑战性但又非常不同的基准分子应用程序，这两个应用程序都远远超出了当前精确量子点分析方法的能力。第一个是氖簇的异构化。二是分子内氢原子在丙二醛中的转移，作为质子转移在中间屏障酶催化和更复杂的生物系统中的原型。所建议的计算规模是可调整的，并且将根据PI可用的1000多个核心化学计算集群（CCC）的最大容量而变化。PI和他的研究小组正投入大量精力，通过开发一个用户友好的“前端”，使他们的计算机代码和精确的量子动力学更容易被“非专家”使用。他们还在德克萨斯理工大学化学和生物化学系为德克萨斯理工大学、区域和其他用户举办培训讲习班。正在通过中央协调委员会更新和分发平行量子码。这些软件工具极大地扩展了可以研究量子动态的系统的复杂性，同时使化学动力学社区的更广泛的横截面能够自己执行这样的计算。

Lionel (Bill) Poirier of Texas Tech University is supported by an award from the Theory, Models and Computational Methods program in the Chemistry division to develop and test massively parallel high performance computing codes that enable exact quantum dynamics (QD) rovibrational spectroscopy calculations for larger and more challenging molecular applications than are currently possible. The award is co-funded by a pilot program in OCI promoting the reuse of Cyberinfrastructure (CI) elements. This capability is valuable for various dynamical, kinetic, and statistical modeling endeavors, for which accurate knowledge of rovibrational states and partition function of the molecular species involved is often vitally important. A novel feature of this research over previous quantum dynamics efforts is use of massive parallelization. Much of the requisite cyberinfrastructure, i.e., parallel linear algebra codes, developed by the PI previously, is already in place. To assess the performance, scalability, and broad applicability of the parallel codes, two extremely challenging, but very different, benchmark molecular applications are being considered, both well beyond the capabilities of the current state of the art in exact QD methods. The first is isomerization in neon clusters. The second is intramolecular hydrogen atom transfer in malonaldehyde, as a prototype for proton transfer in intermediate-barrier enzyme catalysis, and more complex biological systems. The scale of the proposed calculations is designed to be adjustable, and will be varied up to the maximum capacity of the 1000-plus-core Chemistry Computation Cluster (CCC) available to the PI. The PI and his research group are devoting much effort to make their computer codes, and exact quantum dynamics in general, easier for "non experts" to use, by developing a user-friendly "front end". They are also hosting training workshop at the Texas Tech (TTU) Department of Chemistry and Biochemistry, for TTU, regional and other users. The parallelQD codes are being updated and disseminated through the CCC. These software tools greatly extend the complexity of systems that can be studied quantum dynamically, while simultaneously cyber-enabling a broader cross section of the chemical dynamics community to perform such calculations on their own.