Learn More
We introduce a new randomized method called Model Reference Adaptive Search (MRAS) for solving global optimization problems. The method works with a parameterized probabilistic model on the solution space and generates at each iteration a group of candidate solutions. These candidate solutions are then used to update the parameters associated with the(More)
We consider a variation of the subset selection problem in ranking and selection, where motivated by recently developed global optimization approaches applied to simulation optimization, our objective is to identify the top-m out of k designs based on simulated output. Using the optimal computing budget framework, we formulate the problem as that of(More)
Simultaneous perturbation stochastic approximation (SPSA) algorithms have been found to be very effective for high-dimensional simulation optimization problems. The main idea is to estimate the gradient using simulation output performance measures at only <i>two</i> settings of the <i>N</i>-dimensional parameter vector being optimized rather than at the(More)
A major assumption in the analysis of (s; S) inventory systems with stochastic lead times is that orders are received in the same sequence as they are placed. Even under this assumption, much of the work to date has focused on the unconstrained optimization of the system, in which a penalty cost for unsatissed demand is assigned. The literature on(More)
In this paper, we consider Simultaneous Perturbation Stochastic Approximation (SPSA) for function minimization. The standard assumption for convergence is that the function be three times differentiable, although weaker assumptions have been used for special cases. However, all work that we are aware of at least requires differentiability. In this paper, we(More)
A number of Monte Carlo simulation-based approaches have been proposed within the past decade to address the problem of pricing American-style derivatives. The purpose of this paper is to empirically test some of these algorithms on a common set of problems in order to be able to assess the strengths and weaknesses of each approach as a function of the(More)
Based on recent results for multiarmed bandit problems, we propose an adaptive sampling algorithm that approximates the optimal value of a finite-horizon Markov decision process (MDP) with finite state and action spaces. The algorithm adaptively chooses which action to sample as the sampling process proceeds and generates an asymptotically unbiased(More)