Re-runs of the package, 22 October 2026 (from copies of this folder; nothing was run inside the package itself).
Python 3.13 with numpy where needed; Apple clang for the .c programs; every process under nice.

Programs written with the note (outputs recorded here for the first time)
  python3 check_note.py                      -> check_note.out    TOTAL failures: 0   (about 3-4 s)
  python3 check_ranks.py                     -> check_ranks.out   TOTAL failures: 0   (about 140 s)
  (check_note.py was also run twice during writing, and check_ranks.py once, with the same results)

first_run/ (outputs compared with the recorded ones)
  verify_cert.py certificates/cert_G18_18.json --direct   identical to verify_G18_18.log   (1.0 s)
  verify_cert.py certificates/cert_G26_26.json            identical to verify_G26_26.log   (0.2 s)
  check_matrices.py                                       identical to check_matrices.out
  check_thmB.py                                           identical to check_thmB.out      (74 s)
  gf_tables.py 13 TABLE; pairscan TABLE 120               identical to the q=13 block of pairscan_PSL2q_K120.out
  mk_permgroups.py DIR; permscan DIR/A5.txt 60            identical to the A5 block of permscan_K60.out
                        permscan DIR/PSL3_3.txt 60        identical to the PSL3_3 block of permscan_K60.out
                        permscan DIR/A7.txt 80            identical to the A7 block of permscan_K80.out
                        permscan DIR/PSL3_4.txt 80        identical to the PSL3_4 block of permscan_K80.out
  tc (coset enumerator; validation only, the long runs of jobs.log were not repeated):
    A5 = <a,b | a^2, b^3, (ab)^5>, MAXCOSETS 1000           RESULT: index=60 verify=ok
    G(2,2), both commutator conventions (words.py 2 2 1 / 2 2 2)   RESULT: index=1 verify=ok
    G(2,3) over <x> (words.py 2 3 1 IN a), MAXCOSETS 2000000       RESULT: overflow, live=1999999 totaldef=2000147
  Not re-run: permscan for A6, M11, S7, A8, A9, M12 (K = 80) and M22, PSL(3,5), A10, A11 (K = 3); pairscan for
  q != 13; the long coset enumerations.

verification_run_A/
  my_cert_check.py certificates/cert_G26_26.json --dp     identical to my_check_G26.out up to the running time
  my_thmB_fields.py 8 100                                  identical to the q=8 line of my_thmB_fields.out
  Not re-run: my_cert_check.py for G(18,18) with --both; my_thmB_fields.py for the other q.

verification_run_B/
  check_newcert.py on the six newcert_*.json files         CERTIFICATE OK six times, with the same vector counts and
                                                           gcds as check_newcert_out.txt (whose negative controls,
                                                           made on corrupted copies, were not repeated)
  step5_scan_pairs.py 13 13 120                            identical to the first 14 lines of step5_out.txt, the blocks
                                                           for q <= 13 (2 s)
  Not re-run: step5_scan_pairs.py 32 128 120 (the full range).

Final verification run (22 October 2026; from an extracted copy of source.zip, outside the package;
all comparisons up to fields that record running times)
  check_note.py                                            identical to check_note.out (TOTAL failures: 0)
  check_ranks.py                                           identical to check_ranks.out up to running times
  first_run/verify_cert.py G18 --direct, G26               identical to verify_G18_18.log, verify_G26_26.log
  first_run/check_matrices.py, first_run/check_thmB.py     identical to check_matrices.out, check_thmB.out
  first_run/permscan (compiled anew) A6.txt 60, M11.txt 60 identical to the A6 and M11 blocks of permscan_K60.out
  verification_run_A/my_cert_check.py G18 --both, G26 --dp identical to my_check_G18.out, my_check_G26.out
  verification_run_A/my_thmB_fields.py 13 100, 16 100      identical to the q=13 and q=16 lines of my_thmB_fields.out
  verification_run_B/check_newcert.py on the six files     CERTIFICATE OK six times, counts and gcds as recorded
  verification_run_B/step5_scan_pairs.py 13 13 120         identical to the first 14 lines of step5_out.txt
  The programs of independent_run_2/ were run in the working folder of that run; their outputs are the recorded ones.
