New Computational Approaches to the Quantum Dynamics of Chemical Reactions

化学反应量子动力学的新计算方法

• 批准号: 8713555
• 负责人: Robert Wyatt
• 金额: $ 8.2万
• 依托单位: University of Texas at Austin
• 依托单位国家: 美国
• 项目类别: Continuing Grant
• 财政年份: 1987
• 资助国家: 美国
• 起止时间: 1987-09-15 至 1990-09-30
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=8713555&HistoricalAwards=false
• 关键词: New; Computational; Approaches; Quantum; Dynamics

This grant is being made jointly by the Theoretical and Computational Chemistry Program in the Chemistry Division and the New Technologies Program in the Division for Advanced Scientific Computing. The purpose of the work is to develop new methods for computing the outcomes of collisions between molecules, suitable for modern supercomputer architectures. The work involves interdisciplinary collaboration between Professor Wyatt, a computational chemist, and Dr. David Scott, a computer scientist at Intel Corporation. Discretization of the coupled integral equations for molecular collisions leads to a very large system of linear algebraic equations. The multilevel adaptive technique (MLAT), now widely used in computational fluid mechanics, will be combined with recursive and iterative techniques. First, a multichannel solution is obtained on a low level (coarse) grid. The basis (or grid) is then adapted to this solution and the coarse solution is projected or interpolated onto the new (adapted) basis. The scattering amplitudes on the fine grid are then developed through use of either the recursion method or the iterative technique. Applications to very large systems are possible because it is not necessary to store or invert a large matrix. The potential of these methods for parallel processing with be explored, and applications of the methods to several interesting chemical reactions will be undertaken.

这笔赠款是由化学部的理论和计算化学项目和高级科学计算部门的新技术项目联合提供的。这项工作的目的是开发新的方法来计算分子之间碰撞的结果，适用于现代超级计算机体系结构。这项工作涉及计算化学家怀亚特教授和英特尔公司计算机科学家大卫·斯科特博士之间的跨学科合作。分子碰撞耦合积分方程组的离散化导致了一个非常大的线性代数方程组。目前在计算流体力学中广泛使用的多层自适应技术(MLAT)将与递归和迭代技术相结合。首先，在低层(粗)网格上得到多通道解。然后将基础(或网格)适应于该解，并将粗略解投影或内插到新的(适应的)基础上。然后用递推法或迭代法求出细网格上的散射幅度。应用于非常大的系统是可能的，因为它不需要存储或求大矩阵的逆。将探索这些方法用于并行处理的潜力，并将这些方法应用于几个有趣的化学反应。