Fast Randomized Algorithms for Optimization and Other Applications of Geometric Random Walks

用于几何随机游走优化和其他应用的快速随机算法

• 批准号: 9528215
• 负责人: Ravindran Kannan
• 金额: $ 5万
• 依托单位: Carnegie-Mellon University
• 依托单位国家: 美国
• 项目类别: Standard Grant
• 财政年份: 1996
• 资助国家: 美国
• 起止时间: 1996-02-15 至 1997-01-31
• 项目状态: 已结题
• 来源: https://www.nsf.gov/awardsearch/showAward?AWD_ID=9528215&HistoricalAwards=false
• 关键词: Fast; Randomized; Algorithms; Optimization; Other

This is a companion project with CCR-95-28973 (entitled ``Optimization and Learning over Convex Sets'') and is an award under the Software Capitalization Program. The project's goal is to upgrade and update software that had been produced. These upgrades and updates are based on algorithms developed under support of CCR-92-08597. The software is for executing random walks that produce a sample from certain multi-variate probability distributions. An update of the software involves simply changing the ``number of steps'' parameter in the random walk. Applications of this discrete probabilistic software (i) to contingency tables and (ii) to the membership oracles for optimization over convex sets, are investigated. With good planned dissemination (for example, on the world wide-web)and well thought-out testing, other applications of this software are anticipated. ***

这是CCR-95-28973（题为“优化和学习凸集”）的同伴项目，是软件资本化计划下的奖项。 该项目的目标是升级和更新已经生产的软件。 这些升级和更新基于CCR-92-08597支持下开发的算法。该软件用于执行随机游走，从某些多变量概率分布中产生样本。 软件的更新只涉及改变随机游走中的“步数”参数。这个离散的概率软件的应用（一）列联表和（二）的成员资格神谕凸集上的优化，进行了研究。 如果有良好的传播计划（例如，在万维网上）和经过深思熟虑的测试，预计该软件将用于其他方面。 ***