Re-run of the package by the third independent verification run (2026-10-27 UTC), from a copy of source.zip
extracted into an empty scratch folder; C programs compiled with cc -O2; Python 3.13 with python -I; GAP 4.16.1.
Comparison with the recorded outputs, ignoring lines with running times.

check_note.py                       ALL CHECKS PASSED; identical except the timing of the rank3_formula.py run (6.3 s instead of 6.6 s)
scripts/mk_facts.py                 identical
scripts/rank3_formula.py test       identical
scripts/rank3_formula.py mk         identical; outputs/mk_rank3_formula.json byte-identical
scripts/mk_extra_checks.py          identical
scripts/check_fk.py                 identical
scripts/test_hompoly.py             identical (22 s)
scripts/rref3.c, p = 2, 3           identical (p = 3: 147 s)
scripts/rk3, test and p = 2, 3, 5   identical (p = 7 not repeated)
scripts/gap/chardeg_check.g         identical
scripts/zcount.c, p = 2,...,13      Z_3(p) identical for all six primes
verification_run_A/minors.c         identical (447,153,316 pairs; 9,193,080 with minimum degree >= 3; 1 stuck: H_MK itself)
verification_run_A/ncount.c, p = 2, 3, 5   identical
