Installation¶

Using Conda¶

Conda will install GCG, SCIP and PySCIPOpt automatically, hence everything can be installed in a single command:

conda install --channel conda-forge pygcgopt

Requirements¶

PyGCGOpt requires a working installation of the SCIPOptSuite and the GCG Solver which is usually included in the release. In addition, the Python interface for SCIP, PySCIPOpt, has to be installed. Currently, GCG only runs on Linux and macOS, therefore you can use PyGCGOpt only on these operating systems.

Note that the latest PyGCGOpt version is usually only compatible with the latest major release of the SCIP Optimization Suite. PyGCGOpt requires at least SCIP 8.0 and GCG 3.5. The following table summarizes which versions of PyGCGOpt, GCG, PySCIPOpt, and SCIP are compatible:

SCIP	PySCIPOpt	GCG	PyGCGOpt
8.0	4.0	3.5.x	0.1.x
7.0	3.x	-	-
6.0	2.x	-	-
5.0	1.4, 1.3	-	-
4.0	1.2, 1.1	-	-
3.2	1.0	-	-

We recommend installing PyGCGOpt and its Python dependencies in a Python virtual environment. In short, you can create a virtual environment and activate it with the following commands on Linux and macOS:

python3.9 -m venv venv
source venv/bin/activate

Installing SCIPOptSuite from the binary distribution¶

The SCIPOptSuite can either be installed from the binary distribution or it can be build and installed manually from source. If you want to compile the SCIPOptSuite from source, please skip to the next section.

To install a binary distribution of the SCIPOptSuite, navigate to the download page and select at least version 8.0.0 and your operating system. After the installation, please check that you can run scip and gcg from the command line.

Next, skip to the instructions to install PySCIPOpt.

Compiling SCIPOptSuite from source¶

To install the SCIPOptSuite from source, you need to install SCIP and GCG using CMake. The Makefile system is not compatible with PyGCGOpt and PySCIPOpt!

As an example, from the root directory of the source distribution of the SCIPOptSuite, you can run the following commands to build and install locally:

mkdir build && cd build
cmake .. -DCMAKE_BUILD_TYPE=Debug -DSHARED=on -DCMAKE_INSTALL_PREFIX=./install -DGCG_DEV_BUILD=on -DZIMPL=OFF -DIPOPT=OFF -DPAPILO=OFF
make -j4 && make install

If SCIP and GCG are not installed in the global path, you need to specify the install location using the environment variable SCIPOPTDIR before you can install PySCIPOpt and PyGCGOpt:

export SCIPOPTDIR=<path_to_install_dir>

SCIPOPTDIR needs to have a subdirectory lib that contains the library, e.g. libscip.so and libgcg.so (for Linux) and a subdirectory include that contains the corresponding header files:

SCIPOPTDIR
  > lib
    > libscip.so ...
    > libgcg.so ...
  > include
    > scip
    > gcg
    > lpi
    > nlpi
    > ...

If GCG is installed in a different location than SCIP (this is not the default), you can specify the GCG install directory with the GCGOPTDIR environment variable.

To make the shared libraries available at runtime for PySCIPOpt and PyGCGOpt, you need to add them to the corresponding environment variables:

On Linux: export LD_LIBRARY_PATH=<path_to_install_dir>/lib
On macOS: export DYLD_FALLBACK_LIBRARY_PATH=<path_to_install_dir>/lib

Installing PySCIPOpt from PyPI¶

To install PySCIPOpt using pip, run the following:

pip install PySCIPOpt>=4.0.0

Installing PySCIPOpt from source¶

For detailed instructions, please read the installation instructions of PySCIPOpt.

In short, you can install PySCIPOpt from source with pip. For that, run the following command:

pip install <path/to/PySCIPOpt>

Installing PyGCGOpt from PyPI¶

Simply run the folloing to install PyGCGOpt with pip:

pip install PyGCGOpt

Install PyGCGOpt from source¶

First, install the requirements for building PyGCGOpt:

pip install -r requirements.txt

Furthermore, you need to have the Python development files installed on your system (error message “Python.h not found”):

sudo apt-get install python3-dev  # for Python 3, on Linux

Then, compile and install PyGCGOpt:

pip install .

In case the compiler cannot find SCIP or GCG header files, please make sure that you installed the SCIPOptSuite globally or that you set the SCIPOPTDIR environment variable (see above).

Building with debug information¶

To use debug information in PyGCGOpt you need to build it like this:

python -m pip install --install-option="--debug" .

Be aware that you will need the debug library of the SCIPOptSuite for this to work (cmake .. -DCMAKE_BUILD_TYPE=Debug).

Testing new installation¶

To test your brand-new installation of PyGCGOpt you need pytest on your system.

pip install pytest

Here is the complete installation procedure.

Tests can be run in the PyGCGOpt directory with: :

py.test # all the available tests
py.test tests/test_name.py # a specific tests/test_name.py (Unix)

Ideally, the status of your tests must be passed or skipped. Running tests with pytest creates the __pycache__ directory in tests and, occasionally, a model file in the working directory. They can be removed harmlessly.