Re-run of programs from an extracted copy of source.zip (2026-10-16), each in its own folder, at most three at a
time, Python 3.13.5 / NumPy 2.5.2 / SciPy 1.18.1 / SymPy 1.14.0 / mpmath 1.3.0 (laptop, macOS, Apple silicon).
Outputs were written to a separate folder and compared with the recorded outputs (cmp / diff).

program (folder)                                   command                                   wall time   result
check_note.py (.)                                  python3 check_note.py                     35.1 s      identical to check_note_output.txt
                                                                                                         except the "(... s)" timing fields
chebyshev_sharpness.py (original/)                 python3 chebyshev_sharpness.py            1.1 s       byte-identical
lemma_checks.py (original/)                        python3 lemma_checks.py 7 300             25.3 s      byte-identical
near_extremal.py (original/)                       python3 near_extremal.py 11               32.9 s      byte-identical
check_covering.py (original/)                      python3 check_covering.py 7 40            99.5 s      byte-identical
vA_prop2.py (verification_run_A/)                  python3 vA_prop2.py                       3.9 s       byte-identical
B1_exact.py (verification_run_B/)                  python3 B1_exact.py                       1.6 s       byte-identical
B5_crosscheck_claim_maxima.py (verification_run_B/) python3 B5_crosscheck_claim_maxima.py    1.0 s       byte-identical
B6_norm_check.py (verification_run_B/)             python3 B6_norm_check.py                  1.7 s       byte-identical

Not re-run here (running times from the recorded outputs where available): check_covering.py 2026 60;
vA_theorem_test.py (221 s, 215 s, 183 s); vA_coplanar.py; B2b_onedim_slsqp.py; B3_planar.py; B4_space.py (429 s,
300 s); B4b_space_targeted.py; B7_zero_free_component.py. Verification run B re-ran all five programs of
original/ from a copy with byte-identical outputs (claim_rerun/).

Complete re-run by the third verification run (2026-10-16, from an extracted copy of the source archive; all 22
commands of README.md, including the slow ones listed above, at most four at a time): all exit codes 0; 16 outputs
byte-identical, 5 identical up to the lines that print running times, and vA_coplanar.py (run with seed 2026, the
recorded seed being unknown) again finds 0 of 150 configurations with R3 > R2 + 1e-7. Table with commands and times:
independent_run_2/package_rerun/COMPARISON.txt.
