/machines/1RB1RF_1LC0LE_1RD0LB_1RA0RC_1RE1LB_0RD---/rocq/log.md · raw

Rocq log: 1RB1RF_1LC0LE_1RD0LB_1RA0RC_1RE1LB_0RD--- (machine #25)

Round 1 (2026-10-10)

Result: check_kill PASS, exit 0 (full report in rocq/kill.out). Checked at 2026-10-10T16:57:17Z: - compile 69.3 s, native on; - coqchk -norec passed in 24.6 s, no timeout; - Print Assumptions of nonhalt: closed, 0 assumptions; - 1,305 lines, 1,484 sentences, sha256 766d227080425b8550e793e3f13cacf99bae3d7626a6b17b490cce51d7f979fc.

How the file was made (proof.md §5 delta, applied mechanically)

The source was the parent kill file machines/1RB0LD_1RC0RA_1RD1RF_1LA0LE_1RE1LD_0RB---/…---.v (sha256 91861259…fcbf). A Python script made these changes, each asserting exactly one match: 1. A new header comment. It keeps the house-style disclosure line. 2. A simultaneous σ-swap of the state names in the whole body, by regex over {{X}}: A→C, B→D, D→B (C→A would also apply, but C does not occur). - Parent counts A 4, B 12, D 19; twin counts C 4, D 12, B 19, as proof.md §5 predicts. - Only the 14 es sweep lemmas and K name states. 3. The TM string. 4. init0 : c0 -->* K 2 [(11,13,false)] TN, 1,088 steps. 5. base: last pair (1, 26, true) instead of (1, 24, true). 6. The closing rule-table comment: <C replaces <A, and the start line now reads c0 -->(1,088) K0' -->(78,026) B'.

Unchanged: every es proof, the macro rules, J/PN/PI/TC, every ob_*/inv_* lemma, Jstep, the J_base script (cbn; repeat split; lia, which closed with q_4 = 26), init (esx, now 79,114 steps) and nonhalt.

Compiles

What failed

Nothing. Lemma R from proof.md (σ maps transitions to transitions) is why the es proofs re-close verbatim. The base differs only in q_4, and J_base's arithmetic absorbs it.

Packaging (2026-10-10)