⚠️ Warning ⚠️

Ecole is looking for a new home. It is not being actively developed, only critical issues will be investigated.

.. image:: https://raw.githubusercontent.com/ds4dm/ecole/master/docs/_static/images/ecole-logo.svg :target: https://www.ecole.ai :alt: Ecole logo :width: 30 % :align: right

Ecole

.. image:: https://github.com/ds4dm/ecole/actions/workflows/continuous-testing.yml/badge.svg :target: https://github.com/ds4dm/ecole/actions/workflows/continuous-testing.yml :alt: Test and deploy on Github Actions

Ecole (pronounced [ekɔl]) stands for Extensible Combinatorial Optimization Learning Environments and aims to expose a number of control problems arising in combinatorial optimization solvers as Markov Decision Processes (i.e., Reinforcement Learning environments). Rather than trying to predict solutions to combinatorial optimization problems directly, the philosophy behind Ecole is to work in cooperation with a state-of-the-art Mixed Integer Linear Programming solver that acts as a controllable algorithm.

The underlying solver used is SCIP <https://scip.zib.de/>, and the user facing API is meant to mimic the OpenAI Gym <https://gym.openai.com/> API (as much as possible).

.. code-block:: python

import ecole

env = ecole.environment.Branching( reward_function=-1.5 * ecole.reward.LpIterations() ** 2, observation_function=ecole.observation.NodeBipartite(), ) instances = ecole.instance.SetCoverGenerator()

for _ in range(10): obs, action_set, reward_offset, done, info = env.reset(next(instances)) while not done: obs, action_set, reward, done, info = env.step(action_set[0])

Documentation

Consult the user Documentation <https://doc.ecole.ai>_ for tutorials, examples, and library reference.

Discussions and help

Head to Github Discussions <https://github.com/ds4dm/ecole/discussions>_ for interaction with the community: give and recieve help, discuss intresting envirnoment, rewards function, and instances generators.

Installation

Conda ^^^^^

.. image:: https://img.shields.io/conda/vn/conda-forge/ecole?label=version&logo=conda-forge :target: https://anaconda.org/conda-forge/ecole :alt: Conda-Forge version .. image:: https://img.shields.io/conda/pn/conda-forge/ecole?logo=conda-forge :target: https://anaconda.org/conda-forge/ecole :alt: Conda-Forge platforms

.. code-block:: bash

conda install -c conda-forge ecole

All dependencies are resolved by conda, no compiler is required.

Pip wheel (binary) ^^^^^^^^^^^^^^^^^^ Currently unavailable.

Pip source ^^^^^^^^^^^ .. image:: https://img.shields.io/pypi/v/ecole?logo=python :target: https://pypi.org/project/ecole/ :alt: PyPI version

Building from source requires:

• A C++17 compiler <https://en.cppreference.com/w/cpp/compiler_support>_,
• A SCIP <https://www.scipopt.org/>__ installation.

.. code-block:: bash

pip install ecole

Other Options ^^^^^^^^^^^^^ Checkout the installation instructions <https://doc.ecole.ai/py/en/stable/>_ in the documentation for more installation options.

Related Projects

• OR-Gym <https://github.com/hubbs5/or-gym>_ is a gym-like library providing gym-like environments to produce feasible solutions directly, without the need for an MILP solver;
• MIPLearn <https://github.com/ANL-CEEESA/MIPLearn>_ for learning to configure solvers.

Use It, Cite It

.. image:: https://img.shields.io/badge/arxiv-2011.06069-red :target: https://arxiv.org/abs/2011.06069 :alt: Ecole publication on Arxiv

If you use Ecole in a scientific publication, please cite the Ecole publication

.. code-block:: text

@inproceedings{ prouvost2020ecole, title={Ecole: A Gym-like Library for Machine Learning in Combinatorial Optimization Solvers}, author={Antoine Prouvost and Justin Dumouchelle and Lara Scavuzzo and Maxime Gasse and Didier Ch{'e}telat and Andrea Lodi}, booktitle={Learning Meets Combinatorial Algorithms at NeurIPS2020}, year={2020}, url={https://openreview.net/forum?id=IVc9hqgibyB} }