Continuous Cellular Engine

A real-time, interactive continuous cellular-automaton sketch built with Python and Taichi. It evolves a scalar field with a ring-shaped Gaussian neighborhood kernel and a smooth Lenia-style growth rule, then renders the result as a procedural RGB palette.

Hardware and performance: the program initializes Taichi with ti.gpu; a compatible GPU backend and working graphics/window environment are needed to run the interactive demo. Performance depends heavily on the Taichi backend and device. The 300+ FPS figure sometimes associated with this algorithm has not been benchmarked or guaranteed by this repository.

Install

python -m pip install continuous-cellular-engine

Run

continuous-cellular-engine
# or
python -m continuous_cellular_engine

Use the mouse to paint seeds. Press Space to pause/resume, R to reseed, C to clear, and Esc to exit. See the usage guide for source installation, controls, parameters, and platform notes.

What it does

• Evolves a continuous state S(x, y) clamped to [0, 1] rather than binary cells.
• Measures local density with a normalized radial Gaussian ring and toroidal boundary wrapping.
• Applies a Gaussian growth function and an explicit Euler integration step.
• JIT-compiles the grid work through Taichi and visualizes the field with a phase-shifted RGB palette.

Read the architecture and usage guide for equations, execution flow, implementation details, limitations, and examples. The runnable implementation is continuous_cellular_engine.py.

Development

python -m pip install -e '.[test]'
python -m compileall -q continuous_cellular_engine.py tests
python -m pytest

The tests validate Python syntax and the source-level simulation structure without requiring a GPU. GitHub Actions runs these checks and validates the package build on pushes and pull requests.

Releases

Version tags use vMAJOR.MINOR.PATCH. Pushing a v* tag builds distributions and attempts publication to PyPI through Trusted Publishing. The repository's workflow and one-time PyPI publisher setup are described in release setup.

License

This project is licensed under the Apache License, Version 2.0. See the license file for terms, conditions, and notices.