Collaborative Research: Implementation of Tabu Search for the Design of Novel Molecules using Parallel Computers

合作研究：使用并行计算机实现新型分子设计的禁忌搜索

• 批准号: 0224887
• 负责人: David Miller
• 金额: $ 5万
• 依托单位: Rose-Hulman Institute of Technology
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 2002
• 资助国家: 美国
• 起止时间: 2002-08-01 至 2003-10-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=0224887&HistoricalAwards=false
• 关键词: Collaborative; Research; Implementation; Tabu; Search

Research:Product design methodology is a mechanism that permits the synthesis of new materials with specifically tailored properties. Two challenges in computational molecular design are (1) the ability to predict the physical and chemical properties of a given molecule and (2) the ability to solve the large optimization problem which arises from the search for the best molecule for a given application. Research in molecular characterization can predict many physical and chemical properties to a reasonable accuracy for common molecules. A missing technology is a set of parallel algorithms which can efficiently solve the large mixed-integer optimization problems using parallel computing hardware. This project focuses the application of tabu search, a stochastic technique for the solution of such optimization problems, to the molecular design of novel polymers.Tabu search uses a neighborhood search algorithm which is efficient at searching the entire solution space, thus enabling globally optimal solutions to be found within a reasonable time. These algorithms are amenable to parallelization since each member of a solution set can be updated independently. As part of this project, the PIs plan to develop a parallel implementation suitable for use on clusters of workstations. The work will focus on the development of improvements to handle the types of constraint that occur when predicting the physical properties of polymers, and the application of these improvements within a parallel implementation for use on clusters of multiprocessor workstations.Impact:-Tailor-made molecules have huge potential in the chemical process industries. -This Small Grant for exploratory research will permit two young PIs to venture into a totally new research area for them. -This collaborative project between an undergraduate institution, Rose-Hulman, and the University of Kansas, will provide opportunities to undergraduates to be involved in research otherwise not open to them.

研究：产品设计方法是一种允许合成具有特定特性的新材料的机制。 计算分子设计中的两个挑战是（1）预测给定分子的物理和化学性质的能力和（2）解决大的优化问题的能力，该优化问题产生于针对给定应用搜索最佳分子。 分子表征的研究可以预测许多物理和化学性质，以合理的精度为常见的分子。 遗漏技术是一组能够利用并行计算硬件高效求解大型混合整数优化问题的并行算法。 本项目主要研究如何将禁忌搜索（Tabu Search）应用于新型聚合物的分子设计。禁忌搜索是一种随机的优化技术，它使用邻域搜索算法，能够有效地搜索整个解空间，从而在合理的时间内找到全局最优解。 这些算法是服从并行化，因为每个成员的解决方案集可以独立地更新。 作为这个项目的一部分，PI计划开发一个适合在工作站集群上使用的并行实现。 这项工作将集中在开发的改进，以处理类型的约束时，发生预测的聚合物的物理性质，并在一个并行的实现中使用的集群的多处理器workstations.Impact：-量身定制的分子在化学加工行业有巨大的潜力。- 这项探索性研究的小额补助金将允许两名年轻的PI冒险进入一个全新的研究领域。 - 本科院校罗斯-休曼学院和堪萨斯大学之间的这一合作项目将为本科生提供参与研究的机会，否则他们将无法参与。