This private research repository tests a narrow, falsifiable question:
Can local particle-dynamics laws produce mesoscopic organization that remains phenotypically continuous while its material constituents turn over, and that may later recover after a controlled perturbation?
The current substrate is periodic two-dimensional particle dynamics (CORE V0). The project does not claim life, consciousness, autopoiesis, memory, or identity from visually interesting patterns.
- deterministic positions, velocities, types, asymmetric interactions, short-range repulsion, finite-range response, damping, and periodic boundaries;
- scalar O(N²) and independent vectorized force paths;
- periodic proximity-component entity detection;
- geometry/size lineage association with explicit birth, disappearance, split, merge, and ambiguity events;
- ID-independent phenotype descriptor
Phiand phenotype continuityP(tau); - diagnostic-ID Jaccard material retention
M(tau); - first-class ID-permutation, static-material-flux, and tracker/cadence controls;
- reproducible CLI experiments with manifests, raw/indexed tables, summaries, and audit figures.
P and M remain separate. There is no theseus_score or composite memory_score. The initial P > 0.8, M < 0.5 quadrant is an exploratory probe, not an identity definition.
py -m venv .venv
.\.venv\Scripts\python.exe -m pip install -e '.[dev]'
.\.venv\Scripts\python.exe -m pytest -q
.\.venv\Scripts\python.exe -m edlab.cli validate-forces
.\.venv\Scripts\python.exe -m edlab.cli nullsBefore autonomous work, read AGENTS.md and the durable state files it lists.