Run ledger

The run kept no ledger. The first three entries reconstruct its computations on landing (2026-10-21) from RESULTS.md, results.json and HANDBACK.json; their start and finish times were not recorded. The last two entries are the landing's own checks.

R-ABF47B-CALIBRATION

id: r_abf47b-calibration
hunt: r_abf47b
started: unknown
finished: unknown
ran:
  - hunts/r_044dd2/audit.py on orbit18-support-07.json and audit_laurent.py on orbit18-laurent-01.json, as reported
  - hunts/r_044dd2/polynomial_sieve.py at relation caps 3 and 4 on Support 7, as reported
  - probe_6x3.py, a copy of hunts/r_31b6c1/probe.py, for the 6x3 gate and the 8x3 symmetric quotient
outcome: reruns of main's own scripts, not independent routes; cap 3 kept as cap3.json, cap 4 reported without an artifact, and the audit replays left none
artifacts:
  - cap3.json
  - probe_results.json
  - PROBE_RESULTS.md

R-ABF47B-ORBIT-MAP

id: r_abf47b-orbit-map
hunt: r_abf47b
started: unknown
finished: unknown
ran:
  - map_orbits.py, the 57 H x S2 pair orbits against the 31 S8 x S3 triple orbits over all 105^3 triples
outcome: the run's text records the argument, not the script's output; rerun on landing, the 57 pair orbits map onto all 31 triple orbits, printed to stdout with no artifact
artifacts:

R-ABF47B-GROEBNER

id: r_abf47b-groebner
hunt: r_abf47b
started: unknown
finished: unknown
ran:
  - run_groebner.py, sympy.groebner in grevlex on the three target brackets entered as zero generators plus 6,558 relations in 138 variables, under a 5-minute alarm
outcome: results.json reports a stop at its own alarm at 300.03 s with 32.59 MB resident there, and a landing rerun also stopped at the alarm but read 600.54 MB, so the memory figure is not reproduced (r_abf47b-landing-groebner-rerun); the generators are homogeneous with no constant term, so no outcome could have excluded Support 7; no algebraic conclusion
artifacts:
  - results.json

R-ABF47B-LANDING-CHECKS

id: r_abf47b-landing-checks
hunt: r_abf47b
started: 2026-10-21T15:33Z
finished: 2026-10-21T15:42Z
ran:
  - map_orbits.py rerun unchanged, 131 s of CPU and 205 MB peak resident set on a 32-core Linux server under load
  - polynomial_sieve.py at relation cap 3 on Support 7, compared byte for byte with cap3.json
  - build_system on Support 7, the census of the generators run_groebner.py hands sympy
  - test_orbit_refinement.py, five tests, and five planted faults that each turn it red
outcome: map_orbits.py reproduces 57 onto 31; cap3.json is byte-identical to the rerun; every generator is homogeneous of degree 3 or 4 with no constant term; the slice route agrees with map_orbits.py and with Hunt #71's orbit sizes
artifacts:
  - test_orbit_refinement.py

R-ABF47B-LANDING-GROEBNER-RERUN

id: r_abf47b-landing-groebner-rerun
hunt: r_abf47b
started: 2026-10-21T21:30Z
finished: 2026-10-21T21:37Z
ran:
  - run_groebner.py unchanged under /usr/bin/time -v, from a scratch directory whose hunts/r_044dd2 is this branch's copy, so the committed results.json is untouched
  - run_groebner.py's setup steps without the Groebner call, reading the resident set size after import sympy, after build_system and after building the 6,561 polynomials
outcome: the rerun stopped at its alarm at 304.45 s and read 600.54 MB resident there, with 75 s of CPU for the whole process at load average 49 to 144, against 59.03 MB after import sympy and 84.41 MB before the Groebner call, so results.json's 32.59 MB is not reproduced
artifacts:
  - landing_groebner_rerun.json