ReSHOP

Short Technical Overview

ReSHOP is a solver that implements reformulations. Its general scope is to solve hierarchical optimization problems through reformulations, leveraging existing mature solvers for a given problem class. One of its preeminent feature is that it is modeling-language agnostic: there is an internal representation of the optimization problem that is independent of the input or output. A by-product of this is also the ability to translate a optimization problem from one language to the other. Since the solution methodology involves performing transformation on the problem, the consistency of the internal model representation at any stage is ensured. That is a big departure from previous approaches that rely on writing out a new model in a file, which would get used as input to the next pipeline. This involves parsing the problem again and due to the need to represent the data in a text format, might provide challenges to ensure that problem is not changed by those operations.

The solver interface considers hierarchical problems such as bilevel programs and mathematical programs with equilibirium constraints, equilibrium and compleletmentarity problems, stochastic and dynamic equilibria, and composite optimization problems among the classes it can process.

Team Members

Related Papers

[reshop]
Michael C. Ferris, Olivier Huber, Johannes O. Royset Nonconvex, Nonsmooth, and Nonregular Optimization: A Computational Framework. INFORMS TutORials in Operations Research:189-223, 2024.

Model files

The files mentioned in the tutorial are available at: [github repo]

Acknowledgements

We thank Jong-Shi Pang, University of Southern California, for his relentless focus on ``non-problems,'' which inspired some of this work. We also thank Welington de Oliveira, Ecole Nationale Superieure des Mines de Paris, for reviewing the manuscript. We acknowledge financial support from the GAMS Corporation and the Office of Naval Research under grants N00014-24-1-2318 and N00014-24-1-2277.

News Releases, etc

None

* Copyright 2024. Personal use of this material is permitted. However, permission to reprint/republish this material for advertising or promotional purposes or for creating new collective works for resale or redistribution to servers or lists, or to reuse any copyrighted component of this work in other works must be obtained from the IEEE.