# Reproducibility statement

`python -m alexander_grobner verify` performs a frozen-input regression
comparison. Each selected case is executed in a fresh Python process by
default, with `PYTHONHASHSEED=0`, and the computed reduced Gröbner basis is
compared with the archived expected basis. Selected collision examples also
compare normalized common divisors.

This verifies

```text
archived input + documented conventions + this software version
    -> archived expected output.
```

It does not independently verify the association between an external table
label and the archived PD code, decide equivalence of diagrams, reproduce an
unlisted exhaustive table scan, or replace the mathematical proofs.

For long-term archiving, retain the input JSON, output JSON, release wheel,
source archive, and SHA-256 manifest together.