ecojax 1.1

A framework for simulating ecosystems in JAX.

EcoJAX

This repository aims to provide a simple and modular codebase to run experiments on large neural-agents ecological models using JAX.

Your browser does not support the video tag.

Installation

Clone the repository and create a virtual environment. The repo work (at least) in python 3.10 to 3.12.

git clone [e-mail]:tboulet/EcoJAX.git
cd EcoJAX
python -m venv venv
source venv/bin/activate   # on Windows, use `venv\Scripts\activate.bat`

Install JAX

On linux (note your CUDA version may vary) :

pip install jax[cuda12_pip]==0.4.24 -f https://storage.googleapis.com/jax-releases/jax_cuda_releases.html

On Windows :

pip install jax[cpu]==0.4.24

Install the requirements

pip install -r requirements.txt

Eventually install the package in editable mode

pip install -e .

Usage

To run any experiment, run the run.py file with the desired configuration. We use Hydra as our configuration manager.

python run.py

To use a specific configuration, use the --config-name argument.

python run.py --config-name=my_config

You can modify a specific argument thanks to Hydra's override system.

python run.py path.to.arg.in.config=new_value