RERUN_LOG.txt -- re-runs of the programs of this package on 10 October 2026 (OWR-13494-011)

Machine: one laptop (Apple M3 Max, 16 cores, macOS 26.6.2).  All times are UTC.
PYTHON_FLINT = Python 3.12.14 with python-flint 0.9.0, numpy 2.5.3, sympy 1.14.0  (programs of original/)
PYTHON_NUMPY = Python 3.13.5 with numpy 2.5.2                                    (all other programs)

The scripts were run from copies of this folder that were made before this log and the last version of
README.md were written.
- The copy used for run 1b was extracted from an earlier build of the archive.  Its programs and recorded
  outputs (original/, verification_run_A/, verification_run_B/) and its run_quick.sh are byte-identical to
  those of this package.
- The copy used for runs 1a and 2 differed from this package in two comment lines (one in
  original/scripts/lemmaA_examples.py, one in verification_run_B/programs/elementary_range.py; no change of
  code); everything else in the three folders, and run_quick.sh, were byte-identical.
- After run 2, compare_outputs.py and run_long.sh were changed as described in part 2.

============================================================================================================
1. run_quick.sh   (one process, nice -n 10)
============================================================================================================
Run 1a: 12:39:04 - 13:02:29 (23 min 25 s); no other run of this package at the same time.
Run 1b: 13:04:09 - 13:33:32 (29 min 23 s), from the extracted archive; until 13:24:04 run 2 (six
        processes) was running on the same machine.
Command: sh run_quick.sh PYTHON_FLINT PYTHON_NUMPY
The outputs of the two runs are identical:

== original/ (programs with which the results were first obtained; python-flint)
identical                                            original linetools.py (self test)
identical up to times                                original lemmaA_examples.py
identical up to times                                original exhaust_F2_full.py xval 2000
first 12 lines identical up to times                 original exhaust_F2_full.py enum (levels 1..12 of 17)
first 12 lines identical up to times                 original exhaust_F2_full.py check (k = 1..12 of 35)
identical up to times                                original survey_Fp.py 7 2000
identical up to times                                original survey_Fp.py 5 6000
identical up to times                                original survey_Fp.py 3 30000
identical up to times                                original survey_Q.py (output)
identical                                            original survey_Q.py (16376 records)
== verification_run_A/ (numpy)
identical                                            A selftest.py
identical                                            A dims_check.py
identical                                            A capcount_F2.py
identical                                            A curves.py
identical                                            A conductor.py
identical                                            A conductor_explicit.py
identical                                            A crosscheck_claim_reps.py
identical                                            A candidates.py (output)
identical up to times                                A candidates.py (302 records)
12 lines (sizes <= 8) agree with the recorded summary A exhaust_small_F2.py 8 (all sets of <= 8 lines)
== verification_run_B/ (numpy)
identical                                            B test_olib.py
identical up to times                                B f2_group.py (Burnside count)
identical                                            B f2_burnside.json
identical up to times                                B f2_small.py 9 (log)
identical                                            B f2_small.py 9 (6440 records)
first 12 lines 'k=...' identical up to times         B f2_full.py (levels 1..12 of 35)
identical up to times                                B f2_validate.py (Hartshorne-Rao test of the 17 orbits)
identical                                            B f2_count_families.py
identical                                            B compare_claim_f2.py
identical                                            B crosscheck_claim_records.py
first 6 lines identical up to times                  B survey.py cube (k = 1..6 of 8)
identical up to times                                B survey.py pstar
identical up to times                                B survey2.py
identical                                            B lemmaA_toric.py (output)
identical                                            B lemmaA_toric.py (125 records)
identical                                            B lemmaA_bigraded.py
identical                                            B elementary_range.py
== done: 0 comparison(s) failed

============================================================================================================
2. run_long.sh   (six processes, nice -n 10)
============================================================================================================
Run 2: 13:02:29 - 13:24:04 (21 min 35 s); during this run, run 1b was running on the same machine.
Command: sh run_long.sh PYTHON_FLINT PYTHON_NUMPY 6
Steps: L1 exhaust_F2_full.py enum 17: 250 s, check 1 35: 317 s;  L2 f2_group.py, f2_full.py 35,
f2_validate.py: 390 s (f2_full.py: 385 s);  L3 exhaust_small_F2.py 10, five workers: 217 s;
L4 survey.py cube 6 8: 120 s.
Output of the script as it was at the time of the run:

== L1 original: complete enumeration over F_2
identical up to times                                original exhaust_F2_full.py enum 17
identical                                            original orbit representatives (SHA-256 of 36 files)
identical up to times                                original exhaust_F2_full.py check 1 35
identical                                            original the 17 orbits with gap one
== L2 run B: complete enumeration over F_2
identical                                            B f2_burnside.json
35 lines 'k=...' identical up to times               B f2_full.py 35 (orbits and alpha per level)
DIFFERENT                                            B f2_full.py 35 (total)
17 gap-1 orbits identical (canonical keys)           B f2_full.py 35
17 lines 'gap-1 orbit...' identical up to times      B f2_validate.py (on the new representatives)
== L3 run A: all sets of at most 10 lines (five processes)
identical up to times                                A exhaust_small_F2.py 10 (five workers)
== L4 run B: the cube survey
identical up to times                                B survey.py cube 8 (output)
identical                                            B survey.py cube 8 (102817 records)
== done: 1 comparison(s) failed

The comparison reported as DIFFERENT compared the last line ("TOTAL ...") of f2_full_run.log with the
recorded one literally.  The two lines state the same number of orbits (1771439), the same 17 gap-1 orbits
and the same number of gap-1 sets (4763); they differ in the order of two entries of the list of gap-1
orbits, (10, 'pseudostar', 4, 168) and (10, 'pseudostar', 4, 840), and in the running time (385 s; recorded
583 s).  The order in which f2_full.py meets the orbits of one level depends on the scheduling of its
processes (see the Notes in README.md).  Therefore the mode `total` was added to compare_outputs.py, which
compares the TOTAL line with the list sorted, and the corresponding line of run_long.sh was changed to use
it; in the same file two comments were corrected (the size of the output folder, about 40 MB, and the
running times of the steps).  The comparisons of the final run_long.sh, applied to the output files of
run 2 without a new computation, give:

== L1 original: complete enumeration over F_2
identical up to times                                original exhaust_F2_full.py enum 17
identical                                            original orbit representatives (SHA-256 of 36 files)
identical up to times                                original exhaust_F2_full.py check 1 35
identical                                            original the 17 orbits with gap one
== L2 run B: complete enumeration over F_2
identical                                            B f2_burnside.json
35 lines 'k=...' identical up to times               B f2_full.py 35 (orbits and alpha per level)
TOTAL line identical up to order and time            B f2_full.py 35 (total)
17 gap-1 orbits identical (canonical keys)           B f2_full.py 35
17 lines 'gap-1 orbit...' identical up to times      B f2_validate.py (on the new representatives)
== L3 run A: all sets of at most 10 lines (five processes)
identical up to times                                A exhaust_small_F2.py 10 (five workers)
== L4 run B: the cube survey
identical up to times                                B survey.py cube 8 (output)
identical                                            B survey.py cube 8 (102817 records)
== done: 0 comparison(s) failed

============================================================================================================
3. An earlier run of the commands of run_long.sh
============================================================================================================
Run 0: 12:19:15 - 12:38:24 (19 min 9 s), six processes, nice -n 10, no other run of this package at the
same time; the commands of L1 to L4 were issued one after the other, before run_long.sh was written.
The programs used were byte-identical to those of this package.
Steps: L1 enum 17: 181 s, check 1 35: 290 s;  L2: 360 s;  L3: 209 s;  L4: 109 s.
The comparisons of the final run_long.sh, applied to the output files of run 0, give:

== L1 original: complete enumeration over F_2
identical up to times                                original exhaust_F2_full.py enum 17
identical                                            original orbit representatives (SHA-256 of 36 files)
identical up to times                                original exhaust_F2_full.py check 1 35
identical                                            original the 17 orbits with gap one
== L2 run B: complete enumeration over F_2
identical                                            B f2_burnside.json
35 lines 'k=...' identical up to times               B f2_full.py 35 (orbits and alpha per level)
TOTAL line identical up to order and time            B f2_full.py 35 (total)
17 gap-1 orbits identical (canonical keys)           B f2_full.py 35
17 lines 'gap-1 orbit...' identical up to times      B f2_validate.py (on the new representatives)
== L3 run A: all sets of at most 10 lines (five processes)
identical up to times                                A exhaust_small_F2.py 10 (five workers)
== L4 run B: the cube survey
identical up to times                                B survey.py cube 8 (output)
identical                                            B survey.py cube 8 (102817 records)
== done: 0 comparison(s) failed

============================================================================================================
4. Re-runs by the third verification run (the run on the final text of the note)
============================================================================================================
Run 3q: 14:19:15 - 14:58:09 (38 min 54 s), run_quick.sh, one process, nice -n 10, from a copy extracted from
the archive of the package as it was before the third verification run (SHA-256 of that archive:
8c16eb6e28c1f5e1357347f6b54db1229e999b5934c2bff86f5d3d5942badbb6).  The programs and the recorded outputs of
original/, verification_run_A/ and verification_run_B/, and the files compare_outputs.py, run_quick.sh and
run_long.sh, are byte-identical in that archive and in this package.  During the first eight minutes of the
run the programs of independent_run_2/ were running on the same machine with up to five processes.
PYTHON_FLINT = Python 3.12.14 with python-flint 0.9.0, numpy 2.5.3, sympy 1.14.0 (a newly created environment)
PYTHON_NUMPY = Python 3.13.5 with numpy 2.5.2
Command: sh run_quick.sh PYTHON_FLINT PYTHON_NUMPY
Output: identical, line by line, to the output printed in part 1 (37 comparisons, ending with
"== done: 0 comparison(s) failed"; exit code 0).

Run 3r: 15:16:12 - 15:17:01 (49 s), independent_run_2/run_all.sh with five processes, from a copy of
independent_run_2/ whose programs and recorded outputs are byte-identical to those of this package.
Command: sh run_all.sh PYTHON_FLINT 5
Output:

identical            r2_selftest.py
identical            r2_conductor.py (table of the 33 computations)
identical            r2_conductor.py (records)
identical            r2_survey_Q.py (table of the 541 configurations)
identical            r2_survey_Q.py (records)
identical            r2_f2_orbits.py 10 (16672 records)
== done: 0 comparison(s) failed

The recorded outputs of independent_run_2/ were produced by the single commands listed in its README.md
(14:13 - 14:22 and 14:32, with four and five processes).  The long runs (run_long.sh) were not repeated by
the third verification run.
