Reactive Scattering Codes for Massively Parallel Architecture Supercomputers

大规模并行架构超级计算机的反应散射码

• 批准号: 9211104
• 负责人: Edward Hayes
• 金额: --
• 依托单位: Ohio State University Research Foundation -DO NOT USE
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 1992
• 资助国家: 美国
• 起止时间: 1992-09-01 至 1995-02-28
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=9211104&HistoricalAwards=false
• 关键词: Reactive; Scattering; Codes; Massively; Parallel

It has recently become possible to solve quantum reactive scattering problems within their full three-dimensional context for three-atom systems with modest numbers of energetically-open initial-and-final quantum states. While a high level of accuracy has been achieved for a few simple systems, there remain major impediments to the extension of these methods to the vast majority of interesting chemical reactions. The basic theoretical and computational issues are easily characterized on a qualitative level. First, the complexity and computational intensity of current methods even for simple three-atom systems challenge current generations of high-performance computers. Second, as one moves beyond these simple chemical reactions, the complexity of the calculations increases tremendously because the number of open initial-and-final quantum states increases by orders of magnitude. For this reason such problems are currently beyond the realm of consideration for even the most optimistic of researchers. This research will focus on developments that will provide steps towards solving such problems on massively parallel computers.

最近，对于具有适度能量开放的初始和最终量子态的三原子系统，在完整的三维环境中解决量子反应散射问题已经成为可能。虽然在一些简单的系统中已经达到了很高的准确度，但将这些方法扩展到绝大多数有趣的化学反应中仍然存在主要障碍。基本的理论和计算问题很容易定性表征。首先，即使是简单的三原子系统，当前方法的复杂性和计算强度也对当前几代高性能计算机构成了挑战。其次，当我们超越这些简单的化学反应时，计算的复杂性会大大增加，因为开放的初始和最终量子态的数量会以数量级增加。由于这个原因，即使是最乐观的研究人员目前也无法考虑这些问题。这项研究将侧重于在大规模并行计算机上解决这些问题的进展。