Artificial-life: A 300-line Python reproduction of Computational Life research

What this is

Artificial-life is a 300-line Python reproduction of the Computational Life research paper 'How Well-formed, Self-replicating Programs Emerge from Simple Interaction.' It simulates emergent self-replication in a grid of simple programs.

How it works

The simulation uses a 240x135 grid containing 64 instruction-length Brainfuck-like programs that are randomly initialized. Each iteration follows this process:

• Neighboring programs are randomly paired
• Their instruction tapes are concatenated together
• The combined program runs for a maximum of 2¹³ steps
• After execution, the tapes are split back apart

The instructions can loop and mutate the instruction tapes themselves. As described in the original paper, self-replicating programs that copy themselves over their neighbor's tape often spontaneously emerge and spread to take over the entire grid.

Visual representation and running

Each pixel represents an instruction with unique colors, while black indicates raw data storage (not an instruction). Every 8x8 pixel section represents a single program.

To run a simulation with seed 1:

uv run main.py --seed 1

In this specific run, a self-replicator emerges relatively early and takes over most of the grid until a more efficient self-replicator evolves and dominates everything.

The repository includes example outputs: universe.gif and universe.mp4 showing the simulation progress.

Technical details

The project uses Python exclusively (100.0% according to GitHub language detection) and includes standard Python project files: pyproject.toml, uv.lock, and .python-version.

This implementation demonstrates how complex emergent behavior can arise from simple interaction rules, specifically showing how self-replicating programs can evolve without explicit programming for replication.