This repository contains scripts for generating partition-based formulations for trained ReLU neural networks and several Optimal Sparse Input Features (OSIF) test instances implemented in Gurobi. More details on the problems and methods can be found here: https://arxiv.org/abs/2202.05198.
Please cite this work as:
@article{kronqvist2022p,
title = {{P-split formulations: A class of intermediate formulations between big-M and convex hull for disjunctive constraints}},
author = {Kronqvist, Jan and Misener, Ruth and Tsay, Calvin},
journal = {arXiv},
volume = {2202.05198},
year = {2022}
}
The solver software Gurobi is required to run the examples. Gurobi is a commercial mathematical optimization solver and free of charge for academic research. It is available on Linux, Windows and Mac OS.
Please follow the instructions to obtain a free academic license. Once Gurobi is installed on your system, follow the steps to setup the Python interface gurobipy.
- Jan Kronqvist (jkronqvi) - KTH Royal Institute of Technology
- Ruth Misener (rmisener) - Imperial College London
- Calvin Tsay (tsaycal) - Imperial College London
This repository is released under the Apache License 2.0. Please refer to the LICENSE file for details.
This work was supported by Engineering & Physical Sciences Research Council (EPSRC) Fellowships to CT and RM (grants EP/T001577/1 and EP/P016871/1), an Imperial College Research Fellowship to CT, a Royal Society Newton International Fellowship (NIF\R1\182194) to JK, a grant by the Swedish Cultural Foundation in Finland to JK, and a grant by the Swedish Research Council (2022-03502) to JK. The project was also in-part financially sponsored by Digital Futures at KTH through JK and CT.