This file is auto-generated from config.yml using the generate_readme script. To make changes, please edit config.yml or the generation scripts here and here.
MibS is a solver for stochastic mixed integer bilevel linear optimization problems. For an introduction to bilevel optimization, see this slide deck. A somewhat outdated but still useful introduction to MibS is here. A paper that contains a complete technical description of the algorithms in MibS is here. A paper that discusses the cuts employed by MibS is here.
More detailed documentation is available here.
MibS is written in C++ and is released as open source under the Eclipse Public License 2.0.
It is distributed under the auspices of the COIN-OR Foundation
The MibS development site is https://github.com/coin-or/MibS.
Code:
Paper: https://doi.org/10.1007/s12532-020-00183-6
There is a Docker image that provides MibS, as well as other projects in the COIN-OR Optimization Suite here
For newer releases, binaries will be made available as assets attached to releases in Github here. Older binaries are archived as part of MibS here.
Due to license incompatibilities, pre-compiled binaries may lack some functionality. If binaries are not available for your platform for the latest version and you would like to request them to be built and posted, feel free to let us know in the discussion formum.
Source code can be obtained either by
- Downloading a snapshot of the source code for the latest release version of MibS from the releases page.
- Cloning this repository from Github or
- Using the coinbrew script to get the project and all dependencies (recommended, see below).
Below is a quick start guide for building on common platforms. More detailed build instructions are here.
MibS has a number of dependencies, which are detailed in
config.yml. Dependencies on other COIN-OR projects are
automatically downloaded when obtaining the source with coinbrew. For some
of the remaining third-party dependencies, automatic download scripts and
build wrappers are provided (and will also be automatically run for required
and recommended dependencies), while other libraries that are aeasy to obtain
must be installed using an appropriate package manager (or may come with your
OS by default).
The quick start assumes you are in a bash shell.
To download and build MibS from source, execute the following on the command line.
wget https://raw.githubusercontent.com/coin-or/coinbrew/master/coinbrew
chmod u+x coinbrew
./coinbrew fetch MibS@master
./coinbrew build MibS
For more detailed instructions on coinbrew, see https://coin-or.github.io/coinbrew.
The coinbrew script will fetch the additional projects specified in the Dependencies section of config.yml.
- Download the source code, e.g., by cloning the git repo https://github.com/coin-or/MibS
- Download and install the source code for the dependencies listed in config.yml
- Build the code as follows (make sure to set PKG_CONFIG_PTH to install directory for dependencies).
./configure -C
make
make test
make install
MibS has interfaces to the following modelling systems that allow the user to conveniently build the bilevel model in a high-level modelling language and pass the model to MibS through the interface for solution.
- Pyomo (Python) through the PAO package, see here
- JuMP (Julia) through the BilevelJuMP package, see here for an example.
To solve a deterministic mixed integer bilevel linear optimization problem,
you must provide both an MPS file and an auxiliary information file that
specifies which variables and constraints are associated with the each level
(see a description of the file format
here).
Then call mibs like this:
<build_or_install_dir>/bin/mibs -Alps_instance file.mps -MibS_auxiliaryInfoFile aux_file.txt
It is also possible to specify additional settings in a parameter file with, e.g.,
<build_or_install_dir>/bin/mibs -param <build_or_install_dir>/MibS/src/mibs.par
MibS has many parameters. See the
example parameter file mibs.par and the header file MibParams.hpp for
explanations. You can also find a detailed description of MibS
here.
Furthermore, to conduct the experiments illustrated in this report, see the
README file in the directory scripts.
MibS is also capable of solving two-stage mixed integer stochastic bilevel
linear optimization problems. To solve these problems, there are two ways:
-
Provide an MPS file, a time file and a stoch file in the same way as the SMPS format. The second-stage objective coefficients should be defined at the end of the time file (see here). Then call
mibslike this:<build_or_install_dir>/bin/mibs -Alps_instance file.mps -MibS_auxiliaryTimFile file.tim -MibS_auxiliaryStoFile file.sto -MibS_stochasticityType stochasticWithoutSAA -MibS_isSMPSFormat 1 -MibS_isA2Random 1The parameter
MibS_isA2Randomshould be set to 0 in case the coefficients of the first-stage variables in the second-stage problem are not random variables. -
Provide an MPS file and an auxiliary information file in the same way as deterministic bilevel problems. The probability distributions of the random variables also should be specified by setting the values of corresponding parameters (MibS currently supports only the discrete uniform distribution). For a sample parameter file, see [src/mibsStochastic.par.in].
MibS was developed with support from
- Office of Naval Research (Grant N000141912330)
- National Science Foundation (Grants CMMI-1435453, CMMI-0728011, ACI-0102687)
- Lehigh University
- Zuse Institute Berlin
- Research Campus Modal "Mathematical Optimization and Data Analysis Laboratories" funded by the German Federal Ministry of Education and Research (BMBF Grant 05M14ZAM) and by the DFG SFB/Transregio 154