Skip to main content

Gurobi backend for Sage MixedIntegerLinearProgram

Project description

sage-numerical-backends-gurobi: Gurobi mixed integer linear programming backend for SageMath


GurobiBackend has previously been available as part of the SageMath source tree, from which it is built as an "optional extension" if the proprietary Gurobi library and header files have been symlinked into $SAGE_LOCAL manually.

Because of the proprietary nature of the Gurobi software, GurobiBackend is not available in any binary distributions of SageMath.

The present standalone Python package sage-numerical-backends-gurobi has been created from the SageMath sources, version 9.0.beta10. It can be installed on top of various Sage installations using pip, including older versions of Sage such as 8.1 (as shipped by Ubuntu bionic 18.04LTS).

Sage ticket uses this package to remove the in-tree version of GurobiBackend.


Install Gurobi according to the instructions on the website, which includes obtaining a license key.

  • On a Linux system, after unpacking the Gurobi archive in the desired location, such as /opt, set the environment variable GUROBI_HOME to the directory containing the subdirectories bin, lib, ...:

      $ export GUROBI_HOME=/opt/gurobi900/linux64

    Then adjust your PATH (or create symbolic links) so that the interactive Gurobi shell can be found from your PATH:

      $ export PATH="$GUROBI_HOME/bin:$PATH"
  • On macOS, the Gurobi installer should make the interactive Gurobi shell available in /usr/local/bin and therefore from your PATH.

Verify this by typing the shell command

Python 3.7.4 (default, Aug 27 2019, 11:27:39)
Gurobi Interactive Shell (mac64), Version 9.0.0
Copyright (c) 2019, Gurobi Optimization, LLC
Type "help()" for help

If this does not work, adjust your PATH (or create symbolic links) so that is found.

This package finds the Gurobi installation using the GUROBI_HOME environment variable. (On macOS, it suffices to have in your PATH.) An alternative method of build configuration is to set compiler/linker flags appropriately.

Install this package from PyPI using

$ sage -python -m pip install sage-numerical-backends-gurobi

or from a checked out source tree using

$ sage -python -m pip install .

or from GitHub using

$ sage -python -m pip install git+

(See build.yml in the related package sage-numerical-backends-coin package for details about package prerequisites on various systems.)

Using this package

To obtain a solver (backend) instance:

sage: from sage_numerical_backends_gurobi.gurobi_backend import GurobiBackend
sage: GurobiBackend()
<sage_numerical_backends_gurobi.gurobi_backend.GurobiBackend object at 0x7fb72c2c7528>


sage: from sage_numerical_backends_gurobi.gurobi_backend import GurobiBackend
sage: from sage.numerical.backends.generic_backend import get_solver
sage: get_solver(solver=GurobiBackend)
<sage_numerical_backends_gurobi.gurobi_backend.GurobiBackend object at 0x7fe21ffbe2b8>

To use this solver (backend) with MixedIntegerLinearProgram:

sage: from sage_numerical_backends_gurobi.gurobi_backend import GurobiBackend
sage: M = MixedIntegerLinearProgram(solver=GurobiBackend)
sage: M.get_backend()
<sage_numerical_backends_gurobi.gurobi_backend.GurobiBackend object at 0x7fb72c2c7868>

To make it available as the solver named 'Gurobi', we need to make the new module known as sage.numerical.backends.gurobi_backend (note dots, not underscores), using the following commands:

sage: import sage_numerical_backends_gurobi.gurobi_backend as gurobi_backend, sage.numerical.backends as backends, sys
sage: sys.modules['sage.numerical.backends.gurobi_backend'] = backends.gurobi_backend = gurobi_backend

If these commands are executed in a Sage session before any MixedIntegerLinearProgram is created, then the new 'Gurobi' solver wins over the 'GLPK' solver in the selection of the default MIP backend. To select the 'Gurobi' solver explicitly as the default MIP backend, additionally use the following command.

sage: default_mip_solver('Gurobi')

To make these settings permanent, add the above 2 + 1 commands to your ~/.sage/init.sage file. Note that this setting will not affect doctesting (sage -t) because this file is ignored in doctesting mode.

Running doctests

To run the (limited) testsuite of this package, use:

$ sage test

If no Gurobi license is available, the testing is skipped without error.

To run the Sage testsuite with the default MIP solver set to the backend provided by this package, use:

$ sage check_sage_testsuite

If no Gurobi license is available, the testing is skipped without error.

Running tests with tox

The doctests can also be invoked using tox:

$ tox -e local
$ tox -e local

Testing multiple installed Gurobi versions in parallel (see tox.ini):

$ tox -p auto

Overriding the default solver by patching the Sage installation

Another method is to patch the module in permanently to the sage installation (at your own risk). This method will affect doctesting.

$ sage -c 'import os; import sage.numerical.backends as dm; import sage_numerical_backends_gurobi.gurobi_backend as sm; s = sm.__file__; f = os.path.basename(s); d = os.path.join(dm.__path__[0], f); (os.path.exists(d) or os.path.lexists(d)) and os.remove(d); os.symlink(s, d);'

Or use the script in the source distribution that does the same:

$ sage -c 'load("")'
Success: Patched in the module as sage.numerical.backends.gurobi_backend

Verify with that the patching script has worked:

$ sage -c 'load("")'
Success: get_solver(solver='gurobi') gives <sage_numerical_backends_gurobi.gurobi_backend.GurobiBackend object at 0x7f8f20218528>

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Files for sage-numerical-backends-gurobi, version 9.0.0
Filename, size File type Python version Upload date Hashes
Filename, size sage_numerical_backends_gurobi-9.0.0.tar.gz (26.0 kB) File type Source Python version None Upload date Hashes View hashes

Supported by

Elastic Elastic Search Pingdom Pingdom Monitoring Google Google BigQuery Sentry Sentry Error logging AWS AWS Cloud computing DataDog DataDog Monitoring Fastly Fastly CDN DigiCert DigiCert EV certificate StatusPage StatusPage Status page