On the ReLU Lagrangian Cuts for Stochastic Mixed Integer Programming
We study stochastic mixed integer programs where both first-stage and recourse decisions can be mixed integers. A new family of Lagrangian cuts, termed “ReLU Lagrangian cuts,” is introduced by reformulating the nonanticipativity constraints using ReLU functions. These cuts can be integrated into scenario decomposition algorithms. Unlike the ordinary Lagrangian cuts, we prove that the inclusion … Read more