Approximate Multiparametric Mixed-integer Convex Programming

Figure: control evaluation time. Bars show the mean while error bars shown the minimum and maximum values. The explicit implementation is up to three orders of magnitude faster than on-line optimization.

General Description

This repository implements the algorithm for generatic suboptimal explicit solutions of multiparametric mixed-integer convex programs, submitted to IEEE Control Systems Letters. The algorithm can be run either locally or on a cluster via mpirun.

@ARTICLE{Mayuta2019,
       author = {{Malyuta}, Danylo and {A\c{c}{\i}kme\c{s}e}, Beh\c{c}et},
        title = {Approximate Multiparametric Mixed-integer Convex Programming},
      journal = {arXiv e-prints},
     keywords = {Mathematics - Optimization and Control},
         year = "2019",
        month = "Feb",
          eid = {arXiv:1902.10994},
        pages = {arXiv:1902.10994},
archivePrefix = {arXiv},
       eprint = {1902.10994},
 primaryClass = {math.OC},
       adsurl = {https://ui.adsabs.harvard.edu/abs/2019arXiv190210994M},
      adsnote = {Provided by the SAO/NASA Astrophysics Data System}
}

Requirements

To run the code, you must have Python 3.7.2 and MOSEK 9.0.87 installed. To install Python and other dependenies (except MOSEK) on Ubuntu, we recommend that you install Anaconda for Python 3.7 and then execute (from inside this repository's directory):

$ conda create -n py372 python=3.7.2 anaconda # Answer yes to everything
$ source activate py372
$ pip install -r requirements.txt

Instructions

Partitioning jobs are created through make_jobs.sh. Run

bash make_jobs.sh -h

for more information. The job files are stored in the ./runtime directory.