# Installation

# Prerequisites

# Minizinc 2.6+

You need to install minizinc (opens new window) (version greater than 2.6) and update the PATH environment variable so that it can be found by Python. See minizinc documentation (opens new window) for more details.

TIP

You can easily install minizinc from the command line, which can be useful when on cloud. In order to make life easier to cloud users, we reproduce below the necessary lines. Please be careful that this is not an official documentation for minizinc and that the following lines can stop working without notice as we do not test them automatically.

# Linux command line

On a Linux distribution, you can use the bundled minizinc AppImage (opens new window).

If FUSE (opens new window) is available:

mkdir minizinc_install
curl -o minizinc_install/minizinc -L https://github.com/MiniZinc/MiniZincIDE/releases/download/2.6.3/MiniZincIDE-2.6.3-x86_64.AppImage
chmod +x minizinc_install/minizinc
export PATH="$(pwd)/minizinc_install/":$PATH
minizinc --version

Else, this is still possible by extracting the files:

mkdir minizinc_install
cd minizinc_install
curl -o minizinc.AppImage -L https://github.com/MiniZinc/MiniZincIDE/releases/download/2.6.3/MiniZincIDE-2.6.3-x86_64.AppImage
chmod +x minizinc.AppImage
./minizinc.AppImage --appimage-extract
cd ..
export LD_LIBRARY_PATH="$(pwd)/minizinc_install/squashfs-root/usr/lib/":$LD_LIBRARY_PATH
export PATH="$(pwd)/minizinc_install/squashfs-root/usr/bin/":$PATH
minizinc --version

# MacOs command line

mkdir minizinc_install
curl -o minizinc.dmg -L https://github.com/MiniZinc/MiniZincIDE/releases/download/2.6.3/MiniZincIDE-2.6.3-bundled.dmg
hdiutil attach minizinc.dmg
cp -R /Volumes/MiniZinc*/MiniZincIDE.app minizinc_install/.
export PATH="$(pwd)/minizinc_install/MiniZincIDE.app/Contents/Resources":$PATH
minizinc --version

# Windows command line

Works on Windows Server 2022 with bash shell:

mkdir minizinc_install
curl -o minizinc_setup.exe -L https://github.com/MiniZinc/MiniZincIDE/releases/download/2.6.3/MiniZincIDE-2.6.3-bundled-setup-win64.exe
cmd //c "minizinc_setup.exe /verysilent /currentuser /norestart /suppressmsgboxes /sp"
export PATH="~/AppData/Local/Programs/MiniZinc":$PATH
minizinc --version

# Skipping minizinc version check

It may happen that you do not want to install minzinc. For instance if you need to use only a part of the library which is not relying on minizinc at all, especially when not relying on discrete-optimization which is the actual library depending on minizinc.

This can be troublesome as the minizinc binary version is checked by discrete-optimization (opens new window) at library import. However discrete-optimization provides a way to bypass this check by setting the environment variable DO_SKIP_MZN_CHECK:

export DO_SKIP_MZN_CHECK=1

Please note however that the library is never tested without minizinc (or minizinc versions < 2.6).

# Python 3.8+ environment

The use of a virtual environment for scikit-decide is recommended, and you will need to ensure that the environment use a Python version greater than 3.8. This can be achieved either by using conda (opens new window) or by using pyenv (opens new window) (or pyenv-win (opens new window) on windows) and venv (opens new window) module.

The following examples show how to create a virtual environment with Python version 3.8.13 with the mentioned methods.

# With conda (all platforms)

conda create -n skdecide python=3.8.13
conda activate skdecide

# With pyenv + venv (Linux/MacOS)

pyenv install 3.8.13
pyenv shell 3.8.13
python -m venv skdecide-venv
source skdecide-venv/bin/activate

# With pyenv-win + venv (Windows)

pyenv install 3.8.13
pyenv shell 3.8.13
python -m venv skdecide-venv
skdecide-venv\Scripts\activate

# Install scikit-decide library

# Full install [Recommended]

Install scikit-decide library from PyPI with all dependencies required by domains/solvers in the hub (scikit-decide catalog).

pip install -U pip
pip install -U scikit-decide[all]

# Minimal install

Alternatively you can choose to only install the core library, which is enough if you intend to create your own domain and solver.

pip install -U pip
pip install -U scikit-decide

# Known issues

# pygrib

When installing pygrib (opens new window) on MacOS ARM (in dependencies of scikit-decide[all]), no wheel exists on PyPI and there is issues when pip tries to build it. You can overcome this by first installing eccodes which provides GRIB header files required to build the pygrib wheel:

brew install eccodes

Then, reinstall scikit-decide[all] with pip.

If the issue persists, you can try to install the pygrib package available on conda-forge:

conda install -c conda-forge pygrib