Workshop on Simulation Optimization

University Of Maryland, College Park, College Park MD

Investigators

Abstract

The research objective of this workshop is to further methodological advances in simulation optimization for the setting in which evaluation of the objective function to be optimized is computationally expensive. The major complicating factors in simulation optimization are nonlinearities (and non-convexity) in the objective function and high dimensionality (e.g., from stochastic models or from discretizing nonlinear dynamics). A further key feature that makes simulation optimization more difficult than ordinary optimization is that the computational burden must be allocated between both the search and the evaluation, i.e., there is a basic tradeoff between exploring the search space and obtaining a better estimate of the performance (objective function) at the current candidate solutions. Simulation is used throughout the US industry, frequently cited as the most widely used operations research technique, so if successful, this workshop will lead to advances in simulation optimization methodologies and techniques that would provide a competitive advantage to those companies that can integrate them in their long-term strategic and tactical planning, as well as in their more immediate execution and daily operations. Applications are numerous, from improved efficiencies in manufacturing operations, supply chain management, and telecommunications and transportation networks/systems (air, rail, road) to innovations in financial engineering products, pricing, and hedging.