/machines/1RB---_0LC1RF_1LE0RD_0RB1RC_1RD0LE_0RC1RA/rocq/log.md · raw

Rocq log: 1RB---_0LC1RF_1LE0RD_0RB1RC_1RD0LE_0RC1RA (F3, delta of kill #4)

Formalization round 1 (2026-10-10)

Result: check_kill PASS on the first run (exit 0; compile 6.5 s native, coqchk -norec 2.1 s, 0 assumptions, 9 s wall). Kill file: 1RB---_0LC1RF_1LE0RD_0RB1RC_1RD0LE_0RC1RA.v, 510 lines, sha256 0c89b5109f87c793105cadaa9b157902254e5b8e33d1f9bc610d76b9f7a43e37. Full check_kill output: rocq/check_kill.out.

How it was built: python3 -B rocq/make_kill.py (reads #4's .v read-only, asserts its sha256 51e3dea1...e923). - Header: replaced whole with a house-style header (machine, "Adapted from the proof of ... (kill #4), which is adapted from verify/SBCv7.v, Module TM4", sigma and start state, the delta, the AI line, the Checked line with "busycoq BB6" and "Closed under the global context"). Asserts no repo-local path prefix in the output. - Code: identical to rocq/plan_check.v (verified by diff; only the closing comment differs): TM string; sigma = EDBCFA on every {{X}} in one re.sub (asserted counts A 37, C 6, E 35; parent uses only A, C, E); init0 RHS ... bA^^14 <* bB^^4 <* <[0;0] {{B}}> 0inf by esx (16,590 steps); init : c0 -->* CfgF 5 0 14 4 by apply init0; nonhalt index (5,0,14,4)%nat, final unfold Inv. cbn. lia. unchanged (closes the tight facet). - Closing comment block: sigma line added, <A 001 / 001 E> -> <E 001 / 001 F>, 00 C> -> 00 B> in P/Q, start line updated (with #4's start for comparison).

Nothing failed. Remaining for packaging (not this round): status.json / notes.md / STATUS.md update, submission bundle.

Packaging (2026-10-10)