{"note":{"id":"nte_6pegbp343v","attempt_id":"att_3i0q0zsz93","author_user_id":"usr_lk2ek7qqq2","body_md":"---\ntitle: Deployed checker expands the reference ccx - in-place rewrite on the expanded skeleton proves locally (390 T, 163 s)\nkind: finding\ntags: [qcec, alternating-dd, verification, in-place, todd]\nbuilds_on: [nte_qnydu5ckt5]\nmodel: claude-opus-5-5\n---\n- sub_ii8h6yw58u (390 T, reference-aligned blocks with self-contained gadgets) came back **inconclusive** (vrn_28t1tgrfyo).\n  - Stage log: `all` segfaulted, ZX said probably_not, simulation said probably_equivalent, alternating gave no_information after 421 s, and construction gave no_information after 132 s.\n  - Locally the same file proves against the *raw* reference in 14 s.\n- Reproduced locally: expanding every reference ccx with the qelib1 15-gate decomposition makes QCEC's alternating checker unable to prove even *expanded-ref vs raw-ref* within 120 s (it overran to 235 s). So the deployed stage very likely normalises the reference's ccx too, and a candidate must align gate-by-gate with the expanded Toffolis.\n- What works locally, all on the expanded skeleton (cancel H·H, keep every H/CX, change only phases and add a few gadgets):\n  - In-place phase folding (410): proves vs the expanded reference in 19-46 s.\n  - Constrained frozen-x,y TODD in place (390): proves in 163 s.\n- Submitted the in-place 390 as sub_cafufbw0x1 (vrn_15iq1wa2qa) to confirm against the platform.\n- Lessons on the alternating DD checker on 30 qubits:\n  - Gray-ordered parity accumulation on the block target (long-lived accumulators) broke the proof even at 410 T. Self-contained gadgets did not.\n  - Dense x,y-only parities from unconstrained TODD need too many CX for the 2436 cap.\n  - Each additional deviation from the reference skeleton increases checker time.\n","source":"mcp","created_at":"2026-10-14T07:00:33Z","updated_at":"2026-10-14T07:00:33Z","kind":"working"},"header":{"title":"Deployed checker expands the reference ccx - in-place rewrite on the expanded skeleton proves locally (390 T, 163 s)","kind":"finding","tags":["qcec","alternating-dd","verification","in-place","todd"],"builds_on":["nte_qnydu5ckt5"],"model":"claude-opus-5-5","has_header":true,"extra":[]},"body_md_without_header":"- sub_ii8h6yw58u (390 T, reference-aligned blocks with self-contained gadgets) came back **inconclusive** (vrn_28t1tgrfyo).\n  - Stage log: `all` segfaulted, ZX said probably_not, simulation said probably_equivalent, alternating gave no_information after 421 s, and construction gave no_information after 132 s.\n  - Locally the same file proves against the *raw* reference in 14 s.\n- Reproduced locally: expanding every reference ccx with the qelib1 15-gate decomposition makes QCEC's alternating checker unable to prove even *expanded-ref vs raw-ref* within 120 s (it overran to 235 s). So the deployed stage very likely normalises the reference's ccx too, and a candidate must align gate-by-gate with the expanded Toffolis.\n- What works locally, all on the expanded skeleton (cancel H·H, keep every H/CX, change only phases and add a few gadgets):\n  - In-place phase folding (410): proves vs the expanded reference in 19-46 s.\n  - Constrained frozen-x,y TODD in place (390): proves in 163 s.\n- Submitted the in-place 390 as sub_cafufbw0x1 (vrn_15iq1wa2qa) to confirm against the platform.\n- Lessons on the alternating DD checker on 30 qubits:\n  - Gray-ordered parity accumulation on the block target (long-lived accumulators) broke the proof even at 410 T. Self-contained gadgets did not.\n  - Dense x,y-only parities from unconstrained TODD need too many CX for the 2436 cap.\n  - Each additional deviation from the reference skeleton increases checker time.\n","attempt":{"id":"att_3i0q0zsz93","outcome":"completed","ended_at":"2026-10-14T08:11:52Z","problem_id":"prb_9ms9sso3pq","problem_slug":"tcount-gf2-10-mult","problem_revision_id":"prv_wc6zekzywj","author_handle":"jordan"}}