Special Issue for INFORMS Simulation Society (I-Sim) Workshop, 2021

“

The papers in this special issue focus on two main areas. The first area explored appropriate probability models to use as “input data” to drive stochastic, dynamic simulations of systems – such simulations are used to model call centers, hospitals, supply chains, airports, and port operations, and many other systems. The first two papers explore the use of neural networks to create input data generators for such models. The third paper presents a new method for uncertainty characterization of dynamic (functional) data, either as simulation input or output. Simulation optimization was the second focus area. Ranking and selection refers to optimization defined as a choice over a fixed set of system configurations. The fourth paper presents a way to incorporate contextual information that can modify the choice of a best system. When the optimization task is to find a set of system design parameters that optimizes some measure of performance, derivative-based stochastic optimization methods are often used. The fifth paper employs such methods to determine equilibrium efficiencies in multi-agent simulations, where simulated agents take stochastic but rule-based actions based on their environment. The last paper explores optimization strategies when decisions are made over multiple time periods and misspecified parameters must be learned from streaming and non-stationary data as time progresses. The workshop was successful in meeting its goals. Collectively, the papers in this issue present the state of the art in managing streaming and non-stationary data processes and addressing concerns associated with input and model uncertainty and making decisions under such uncertainty.

”