Skip to main content

Real-World Multi-Agent Reinforcement Learning problems, potentially yielding a high positive impact on society when solved.

Project description

HiveX header image

HiveX

Real-World Multi-Agent Reinforcement Learning problems, potentially yielding a high positive impact on society when solved.

About

The motivation of the HiveX suite is to provide advanced reinforcement learning benchmarking environments with an emphasis on: (1) real-world scenarios, (2) multi-agent systems, (3) investigating problems and solutions with high impact on society, (4) cooperation and communication mechanisms.

Available Environments

Thumbnail Title Domain Paper Env Name
WindFarm Windfarm Orientation Optimisation Distributed Energy Grids Link
(Neurips'21)
"WindFarm"
Wildfire Wildfire-Management
Resource Distribution
Catastrophe Management Link
(ICLR'22)
"Wildfire"

Installation

Manual install (Windows)

The installation steps are as follows (see install.sh for an example installation script):

  1. Create and activate a virtual environment, e.g.:

    conda create -n hivex python=3.10
    conda activate hivex
    
  2. Install pip and pip-tools:

    conda install pip
    pip install pip-tools
    pip install pathlib
    
  3. Install HiveX:

    git clone -b main https://github.com/philippds/HiveX
    cd HiveX
    pip install .
    
  4. Test the HiveX installation:

    pip install pytest
    pytest hivex
    

Example Usage

Training agents with Stable-Baselines3 (Environment Interface | Learning Framework)

cd <hivex_root>
pip install -e .[stable-baselines3]

Stable-Baselines3 VecEnv | Stable-Baselines3

cd <hivex_root>/examples/stable_baselines3
python sb3_train.py

Gym | Stable-Baselines3

cd <hivex_root>/examples/gym
python gym_train.py

SuperSuit | Stable-Baselines3

cd <hivex_root>/examples/super_suit
python super_suit_train.py

Training agents with ML-Agents (Environment Interface | Learning Framework)

UnityEnvironment | ML-Agents

cd <hivex_root>/examples/ml_agents
python mlagents_train.py

Training agents with RLLib (Linux Only) (Environment Interface | Learning Framework)

cd <hivex_root>
pip install -e .[rllib]
pip install -e .[pettingzoo]

PettingZoo ParallelEnv (AEC) | RLLib

cd <hivex_root>/examples/pettingzoo
python pettingzoo_rllib_train.py

Documentation

Full documentation is available here

Citing HiveX

If you use HiveX in your work, please cite:

@inproceedings{siedler2022hivex,
    title={},
    author={Philipp D. Siedler},
    year={2022},
    journal={},
    organization={}
}

Project details


Download files

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

Source Distribution

hivex-1.0.1.tar.gz (3.9 kB view hashes)

Uploaded Source

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page