This is a collection of codes used to generate challenging mixed integer programming instances as part of my PhD thesis. It's still a work in progress, hoping to extend the generator capabilities over time.
Since the definition of distributions of generated instances quickly becomes
complicated with more parameters, I've intended for this package to allow
distributions to be defined by composing the various generator and constructor
functions. See the examples directory for scripts which automate the instance
generation process.
Installation should be fairly easy:
pip install -r build-requirements.txt # numpy, cython, etc for building
pip install -r requirements.txt # install main packages
Note though that this installs multiple top-level packages (mip_generators, lp_generators, scip_runner, search_algorithms) which is not great. I expect to refine this in future.
Tests are scattered and library uses old numpy conventions, so ...
pytest -W ignore::PendingDeprecationWarning
So far this is a minor extension of the generators in the lp-generators package. If you are using either package please cite our paper published in Mathematical Programming Computation (MPC) where we describe the generator and investigate properties of the generated instances:
@Article{Bowly.Smith-Miles.ea_2020_Generation-techniques,
author = {Bowly, Simon and Smith-Miles, Kate and Baatar, Davaatseren
and Mittelmann, Hans},
title = {Generation techniques for linear programming instances
with controllable properties},
journal = {Mathematical Programming Computation},
year = 2020,
volume = 12,
number = 3,
pages = {389--415},
month = sep,
}