FlowLenia

Official implementation of Exploring Flow-Lenia Universes with a Curiosity-driven AI Scientist: Discovering Diverse Ecosystem Dynamics (ALIFE 2025).

Flow-Lenia is a continuous cellular automaton with mass conservation and flow-based dynamics, enabling complex self-organizing patterns and artificial life simulations. This codebase implements Flow-Lenia in JAX for high-performance computation and includes all experiments from the paper, featuring Intrinsically Motivated Goal Exploration Processes (IMGEP) for automated discovery of diverse ecosystem dynamics.

Companion Website | Paper

Installation

Requires Python 3.10.

pip install -r requirements.txt
pip install -e .

For GPU support, follow the JAX installation instructions.

Quick Start

Basic Examples

# Basic Flow Lenia simulation
python examples/example_fl.py

# Flow Lenia with spatial parameters
python examples/example_flp.py

# Flow Lenia with walls/environments
python examples/example_walls.py

# Generate video from simulation
python examples/example_generate_video.py

Running Paper Experiments

Key experiments from the paper are located in experiments/:

• IMGEP (Curiosity-Driven Discovery): experiments/imgep_original/, experiments/imgep_corner/

Additional experiments exploring various aspects of Flow-Lenia dynamics:

• experiments/ea_walls/ - Evolutionary activity with walls
• experiments/mixing/ - Parameter mixing experiments
• experiments/distribution/ - Distribution analysis
• experiments/complexity_analysis/ - Complexity metrics
• experiments/metamorph/ - Metamorphosis experiments
• experiments/large_map/ - Large-scale simulations
• experiments/switch_params/ - Parameter switching experiments

Project Structure

flowlenia/
├── flowlenia_vanilla.py         # Basic Flow Lenia
├── flowlenia_params.py          # Flow Lenia with spatial parameters
├── flowlenia_food.py            # Flow Lenia with food dynamics
├── environments.py              # Environment initialization utilities
├── reintegration_tracking.py   # Particle reintegration mechanisms
├── utils.py                     # Core utilities (kernels, growth functions)
├── vizutils.py                  # Visualization utilities
├── simutils.py                  # Simulation runner and logging
├── analysis/                    # Analysis tools (complexity, distribution, etc.)
├── autodisc/                    # Quality-diversity algorithms (IMGEP)
└── optimization/                # Evolutionary optimization

examples/                        # Example scripts
experiments/                     # Paper experiments and explorations

Citation

If you use this code in your research, please cite:

@article{michel2025exploring,
  title={Exploring Flow-Lenia Universes with a Curiosity-driven AI Scientist: Discovering Diverse Ecosystem Dynamics},
  author={Michel, Thomas and Cvjetko, Marko and Hamon, Gautier and Oudeyer, Pierre-Yves and Moulin-Frier, Cl{\'e}ment},
  journal={arXiv preprint arXiv:2505.15998},
  year={2025}
}

Contact

Thomas Michel - https://thomick.github.io/

For questions or issues, please open an issue on GitHub or contact via the website above.

License

This project is licensed under the MIT License - see the LICENSE file for details.

Acknowledgments

This work builds upon Lenia by Bert Chan and related work on continuous cellular automata and artificial life systems.