# Heterogeneous Ca Dynamics Eval

> Evaluates the long-term phenotypic and genotypic dynamics of a heterogeneous cellular automaton with age constraints and local evolution, testing its ability to sustain open-ended innovation without stagnation. Use when the user wants to benchmark on Heterogeneous Life-Like CA Simulation, or asks about evaluating this task. Reports quantitative metrics.

- Skill: `qhjqhj00/heterogeneous-ca-dynamics-eval` (Agent Skill)
- Install (CLI): `npx skillmds add qhjqhj00/heterogeneous-ca-dynamics-eval`
- Raw SKILL.md: https://api.skillmd.com/api/skills/qhjqhj00/heterogeneous-ca-dynamics-eval/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Productivity
- Author: qhjqhj00 (https://skillmd.com/u/qhjqhj00)
- Updated: 2026-09-08
- Page: https://skillmd.com/skills/qhjqhj00/heterogeneous-ca-dynamics-eval

---


# heterogeneous-ca-dynamics-eval

> Emergent Dynamics in Heterogeneous Life-Like Cellular Automata — Shrestha et al. (2024) (arXiv:2406.13383, 2024)

## What this evaluates

Evaluates the long-term phenotypic and genotypic dynamics of a heterogeneous cellular automaton with age constraints and local evolution, testing its ability to sustain open-ended innovation without stagnation.

## Datasets

- **Heterogeneous Life-Like CA Simulation** — total ?; splits: test (-1)

## Metrics

- `quantitative metrics` **(primary)** — range: other
  - Averaged results over 10 independent runs, with standard deviation calculated. Specific measures (e.g., population count, diversity) are not explicitly named in the text.

## Input / output format

**Input**: 2D grid (50x50 or 500x500). Initial state: 50% alive, 50% quiescent. All alive cells initialized with GoL genome (B3S23). Phenotypic states split 50/50 between 0 and 1. Age parameters: amax (10 or 50), adec (15 or 70).

**Output**: Simulation state evolution over generations. Final reported values are averaged quantitative metrics across 10 runs.

## Scoring recipe

```python
# Pseudo-code based on protocol
results = []
for run in range(10):
    grid = initialize_grid(size, alive_ratio=0.5)
    for gen in range(generations):
        grid = apply_ca_rules(grid, amax, adec, P_inh=0.125, P_mut=0.2)
    results.append(compute_quantitative_metrics(grid))
mean_val = sum(results) / len(results)
std_val = sqrt(sum((x - mean_val)**2 for x in results) / len(results))
return {"mean": mean_val, "std": std_val}
```

## Common pitfalls

- Confusing the age-constrained GoL variant (P_mut=0.0, amax=50, adec=70) with the classical GoL baseline (no age limits, all cells initially alive).
- Using different grid sizes (50x50 for 10k steps vs 500x500 for 1k steps) without adjusting expectations for population dynamics.
- Forgetting that only 'Alive' cells can inherit genomes; decay/quiescent cells cannot reproduce.

## Evidence (verbatim from paper)

> The experiments are run on a 50 × 50 grid for 10,000 generations and on a 500 × 500 grid for 1,000 steps. The experiments were repeated 10 times and for the quantitative metrics, results were averaged and the deviation calculated.

## Citation

```bibtex
@misc{shrestha2024emergent,
  title={Emergent Dynamics in Heterogeneous Life-Like Cellular Automata},
  author={Shrestha et al. (2024)},
  year={2024},
  note={arXiv:2406.13383}
}
```

- arXiv: 2406.13383