Perspective reformulations of mixed integer nonlinear programs with indicator variables (Q2638370)

In this paper the authors study mixed integer nonlinear programs (MINLP) that are driven by a collection of indicator variables where each indicator variable controls a subset of decision variables. Based on this work some concepts that have been applied successfully in the case of mixed integer linear programs (MILP) can be applied for MINLP. In order to apply these ideas, the authors analyse simple sets that form the structures of many practical MINLPs and than the results are extended to more general sets. Finally, the described ideas are applied to three problems: a quadratic uncapacitated facility location problem, a network design problem with nonlinear congestion constraints and a portfolio optimization problem with buy-in thresholds.