Phi-Genesis

Reproducibility and Computational Workflow

The repository’s source code, tests, fixed-seed protocol data and versioned records are the primary reproducibility artifacts. This guide summarizes the commands in the project README; use a clean local clone and follow the cautions below.

Environment

The project uses Python with NumPy and SciPy for numerical work.

pip install numpy scipy
python -c "import phi_genesis; print('OK')"

For editable installation and tests:

pip install -e . pytest
pytest -q -m "not nightly"

The README reports that the fast suite can take roughly 2–3 minutes on one CPU core, with the level-8 test dominating. Nightly tests are excluded by this command.

Reproduce protocol v1.2.1

python -m phi_genesis.selection_protocol_v121 stepA
python -m phi_genesis.selection_protocol_v121 stepB

stepA regenerates the null pools using seeds fixed in the code; the README estimates about 15 minutes.
stepB runs the evaluation and should be compared with protocol_data/results_v1_2_1.json.
Caution: stepA rewrites protocol_data/null_pools_v1_2_1.json. If you want to preserve the committed file exactly, perform the reproduction on a separate working copy.

Reproduce Open Problem #2 estimator studies

python protocol_data/op2_window_bias.py table
python protocol_data/op2_window_bias.py sliding
python protocol_data/op2_chord_estimator.py 8 9

The window-bias script uses the cached numerical inertia counts documented by the project. The chord-estimator command compares levels 8 and 9 for the documented estimator.

What to record

For a reproducible report, include:

Git commit SHA and Python / NumPy / SciPy versions;
exact command and test selection;
graph level, operator convention, boundary condition and numerical tolerance;
protocol version, random seeds and null ensemble;
whether an output was regenerated or read from committed data;
any deviation from preregistered settings.

A passing test suite validates the tested code paths, not every scientific interpretation in the repository. Numerical results should always be reported with their level, estimator, parameter range and known limitations.

Source artifacts

Explore Phi-Genesis
Interactive portal · Research map · Verified results · Open problems · Reproducibility · Dissertation · Source repository