Integration of indifference-zone with multi-objective computing budget allocation

Abstract

In this paper, we consider how to address the issues of having designs with close performance in the multi-objective ranking and selection (MORS) problem. To resolve this issue we propose integrating the indifference-zone (IZ) concept into the multi-objective computing budget allocation (MOCBA) framework. In particular, when IZ is introduced into the MOCBA framework, we address how to determine the probability of non-dominance, how to define the Pareto set, and how to derive allocation rules for the simulation replications. Empirical results show that the MOCBA framework with IZ can significantly save simulation budget when designs to be compared have close performance. © 2009 Elsevier B.V. All rights reserved.