CISE Research Instrumentation: Instrumentation for Research in Search Technology

CISE研究仪器：用于搜索技术研究的仪器这项研究仪器补助金用于购买若干cpu、交换机、磁盘和磁带，用于以下项目：-异步并行搜索修剪应用，-快速内部点优化算法，-视频压缩的优化运动估计算法，-高性能满意度测试，以及-分布式自动推理，支持探索高性能搜索技术方面的努力。这些项目将涉及：-开发唠叨，一种异步、容错、并行搜索修剪技术，并将其应用于组合优化、计算生物学和定理证明；-开发快速内点优化算法及其在资源分配和类似问题上的应用；-基于优化的高质量视频压缩运动估计技术；-开发有效的可满足性测试算法及其在离散数学约束满足和问题中的应用；-开发基于并行搜索的子句扩散原理的分布式自动推理系统，并应用于验证、约束求解和逻辑编程。除了用于开发和测试粗粒度分布式搜索技术的环境之外，这个新的nsf支持的分布式计算设施的高速大内存CPU构成了唯一可用的专用计算资源，用于算法性能的经验评估，有时可能需要数周的CPU时间。

9729807 Zhang, Hantao University of Iowa CISE Research Instrumentation: Instrumentation for Research in Search Technology This research instrumentation grant contributes to the purchase of several CPUs, switch, disk and tape which will enable the following projects: - Asynchronous Parallel Search Pruning with Applications, - Fast Interior Point Optimization Algorithms, - Optimized Motion Estimation Algorithms for Video Compression, - High Performance Satisfiability Testing, and - Distributed Automated Reasoning, supporting efforts exploring aspects of high-performance search technology. The projects will address: - Development of nagging, an asynchronous, fault-tolerant, parallel search pruning technique with applications to combinatorial optimization, computational biology, and theorem proving; - Development of fast interior-point optimization algorithms and their application to resource allocation and similar problems; - Optimization-based motion estimation techniques for high-quality video compression; - Development of efficient satisfiability testing algorithms and their application to constraint satisfaction and problems in discrete mathematics; - Development of distributed automated-reasoning systems based on the clause- diffusion principle of parallel search with applications to verification, constraint solving, and logic programming. In addition to an environment for the development and testing of coarse-grained distributed search techniques, this new NSF-supported distributed computing facility's high-speed large-memory CPUs constitute the only available dedicated computational resource for empirical evaluations of algorithmic performance, which may sometimes require weeks of CPU time.