Finding feasible systems in the presence of constraints on multiple performance measures

Abstract

We consider the problem of finding a set of feasible or near-feasible systems among a finite number of simulated systems in the presence of constraints on secondary performance measures. We first present a generic procedure that detects the feasibility of one system in the presence of one constraint and extend it to the case of two or more systems and constraints. To accelerate the elimination of infeasible systems, a method that reuses collected observations and its varianceupdating version are discussed. Experimental results are presented to compare the performance of the proposed procedures.

Simulation Optimization: A Review and Exploration in the New Era of Cloud Computing and Big Data

“…This represents a stochastic constraint extension of problem (1). There has been recent work on extending indifference-zone R&S procedures as well as OCBA procedures (Batur and Kim, 2010;Park and Kim, 2011;Lee et al, 2012;Pasupathy et al, 2014) to handle stochastic constraints. Nagarajan and Pasupathy (2013) presented a simulation optimization algorithm that supports stochastic constraints.…”

S-model

“…Andradóttir and Kim (2010) developed an IZ procedure for feasibility check in the presence of one stochastic constraint, which was extended to multiple stochastic constraints by Batur and Kim (2010) using the Bonferroni inequality, which makes the procedure conservative for large k and s. To lessen the conservativeness, they introduce an accelerated feasibility check which features an artificial constraint in addition to original s constraints. The artificial constraint is obtained by aggregation (or linear combination) of all secondary performance measures Y im a , the aggregated tolerance level e a , and the aggregated target value q a .…”

Ranking and selection for multiple performance measures using incomplete preference information

“…The R&S procedures currently available for selecting non-dominated designs are variants of MOCBA , 2009Lee, Chew, & Teng, 2007;Lee et al, 2004;Lee, Chew, Teng, & Goldsman, 2010;. In addition, one line of research considers R&S procedures for maximizing or minimizing a primary performance measure under constraints on a number of secondary performance measures (Andradottir, Goldsman, & Kim, 2005;Andradottir & Kim, 2010;Batur & Kim, 2010;Hunter & Pasupathy, 2013;Morrice & Butler, 2006).…”