Use of surrogates in derivative-free optimization

在无导数优化中使用代理

• 批准号: 418250-2012
• 负责人: LeDigabel, Sébastien
• 金额: $ 1.89万
• 依托单位: École Polytechnique de Montréal
• 依托单位国家: 加拿大
• 项目类别: Discovery Grants Program - Individual
• 财政年份: 2015
• 资助国家: 加拿大
• 起止时间: 2015-01-01 至 2016-12-31
• 项目状态: 已结题
• 来源: https://www.nserc-crsng.gc.ca/ase-oro/Details-Detailles_eng.asp?id=573557
• 关键词: Use; surrogates; derivative; free; optimization

The research project described in this proposal concerns derivative-free optimization (DFO). More precisely, it focuses on blackbox optimization, which occurs when the objective(s) and constraints of an engineering optimization problem are obtained by a computer code seen as a blackbox. Several characteristics make these codes impractical for optimization: they may be expensive to evaluate, be contaminated with noise, or fail to return a value. No derivative information is available and even approximations can not be exploited for the optimization. In this context, derivative-based methods cannot be used, and DFO methods may be considered. The present proposal discusses extensions of the mesh adaptive direct search (MADS) method, and in particular the use of surrogates to improve its efficiency. My research is equally divided into three categories: algorithmic developments, optimization software design, and applications. The present application proposes six projects in the first category. Support for the two other categories has been requested from other organizations. Most of these six projects concern the use of surrogates within a direct search framework. Each project, in terms of objectives and time frame, has been specifically defined for an MSc or PhD student. As a consequence, approximately 86% of the requested budget is devoted to graduate-student salaries.

该提案中描述的研究项目涉及无衍生品优化（DFO）。更确切地说，它的重点是黑框优化，这是在目标（S）和工程优化问题的约束时发生的，该计算机代码被视为黑框。这些特征使这些代码不切实际以进行优化：它们可能很昂贵，评估，被噪声污染或无法返回值。没有可用的衍生信息，甚至无法利用近似值进行优化。在这种情况下，不能使用基于导数的方法，并且可以考虑DFO方法。本提案讨论了网格自适应直接搜索方法（MADS）方法的扩展，尤其是使用替代物来提高其效率。 我的研究同样分为三类：算法开发，优化软件设计和应用程序。本申请提出了第一类的六个项目。其他组织已要求对其他两个类别的支持。 这六个项目中的大多数涉及在直接搜索框架内使用替代物。每个项目就目标和时间范围而言，都是为MSC或博士生专门定义的。结果，要求的预算中约有86％用于研究生薪水。