Input models for synthetic optimization problems

We describe and discuss alternative input models for the coefficients in synthetic optimization problems. Synthetic, or randomly generated, problems are often used in computational studies to establish the efficacy of solution methods or to facilitate comparative evaluations of solution methods. The selection of an input model for the coefficients in synthetic optimization problems is important because such a selection may affect the outcome of a computational study. Understanding how an assumed input model affects the characteristics of test problems can assist researchers in their efforts to accurately quantify and interpret the performance of solution methods