\(\)
We analyze the inverse of the Gomory corner relaxation (GCR) of a pure integer program (IP). We prove the inverse GCR is equivalent to the inverse of a shortest path problem, yielding a polyhedral representation of the GCR inverse-feasible region. We present a linear programming (LP) formulation for solving the inverse GCR under the \(L_{1}\) and \(L_{\infty}\) norms, with significantly fewer variables and constraints than existing LP formulations for solving the inverse IP in literature. We show that the inverse GCR bounds the inverse IP optimal value as tightly as the inverse LP relaxation under mild conditions. We provide sufficient conditions for the inverse GCR to exactly solve the inverse IP.
Citation
https://doi.org/10.48550/arXiv.2410.22653
Article
View Inverse of the Gomory Corner Relaxation of Integer Programs