What is it about?
This paper introduces a novel method of testing the first-order optimality conditions through sequential statistical hypothesis tests. The test is derived for black-box simulation optimization problems with multiple responses, yet it is also possible to use it with white-box simulation optimization problems.
Featured Image
Why is it important?
The derived stopping rule is very general and can be used with any simulation-based optimization algorithm. If the problem is convex, the heuristic stopping rule stops at an estimated optimum solution.
Perspectives
Simulation-based optimization problems are usually difficult to solve. Two main approaches to deal with such problems are black-box (Response Surface Methdology, etc) and white-box (Infinitesimal Perturbation Analysis, etc) approaches. Eventually, an optimization algorithm is applied to solve the resulting surrogate model(s); hence, a stopping rule that stops the algorithm at some pre-defined neighborhood of an estimated optimum solution is required. This paper derives asymptotically such a rule.
Ebru Angun
Galatasaray Universitesi
Read the Original
This page is a summary of: An Asymptotic Test of Optimality Conditions in Multiresponse Simulation Optimization, INFORMS Journal on Computing, February 2012, INFORMS,
DOI: 10.1287/ijoc.1100.0438.
You can read the full text:
Contributors
The following have contributed to this page
