evidence/v126/g2_odd_complement/Manuscript v1.26 — CERTIFICATE bundle: the complete odd-sector sign at lambda=4 — a direction-complement certificate (768/896 bits), a directional M=8192 certificate (1024/1280 bits) and the binding certificate complete_lambda4_v126_certificate.json. Together with ../g2_simultaneous/ this gives W_4 >= 0 in both parities (prop:v126-odd-complement, prop:v126-full-window). Marked RESTORED in ../../MISSING.md.
Start here: README_odd_complement.md, review_rank_one_trial.md
Manifest: v1.26_manifest.json, v1.26_evidence_recovery.json, README_v126_evidence_recovery.md
README_odd_complement.md (3 KB) — v1.26: complete positivity at lambda = 4build_complete_pdf.py (1 KB) — Compile in a fresh directory; install only a closed, parseable full PDF.certify_odd_full_window.py (3 KB) — Bind the complete complement and directional certificates to one Schur form.complete_lambda4_v126_certificate.json (2 KB) — (no description in file)direction_complement_certificate.py (3 KB) — Complete old-metric complement bound for an exact frozen dyadic direction. All matrices are outward-certified complete residual ingredients, not cut tails.direction_complement_lemma.tex (2 KB) — Proposed proof insertion only; no numerical gate is asserted here.direction_complement_progress.txt (1 KB) — (no description in file)direction_complement_replay_progress.txt (1 KB) — (no description in file)direction_complement_report_b768.json (1 KB) — (no description in file)direction_complement_report_b896.json (1 KB) — (no description in file)directional_M8192_progress.txt (1 KB) — 8192 {‘step’: 4, ‘recursive_residual2’: ‘[0.009948444871571556819380617683084827237970 +/- 2.56e-43]’} 11.7directional_M8192_verify_b1024_progress.txt (2 KB) — verified exact head 83.9directional_M8192_verify_b1280_progress.txt (2 KB) — verified exact head 119.2directional_M8192_verify_progress.txt (586 B) — Traceback (most recent call last):directional_odd_M8192_s24_report_b1024_J65536.json (2 KB) — (no description in file)directional_odd_M8192_s24_report_b1280_J65536.json (2 KB) — (no description in file)directional_odd_M8192_s24_witness.json.gz (604 KB) — (no description in file)directional_tail_repair.py (7 KB) — Frozen one-direction finite trials; complete residual verification is separate. The old head and normalization remain exact. PCG is proposal-only.finish_v126_sign.py (9 KB) — (no description in file)joint_remote_gate.py (3 KB) — Whole-head certificate with one joint remote Gram, retaining its cross block.localized_error_insert.tex (2 KB) — (no description in file)mixed_refinement.py (6 KB) — Directional mixed inverse diagnostic, with exact frozen inputs and Arb actions. Finite-prefix mixed values are explicitly NOT complete-tail certificates.odd_complement_insert.tex (6 KB) — A complete complement bound and a reduced odd-sector targetodd_s16_ordinary_error_progress.txt (4 KB) — (no description in file)package_v126.py (8 KB) — (no description in file)posterior_trial.py (6 KB) — Finite preconditioned trials, verified by the COMPLETE new Weil residual. Candidate iteration is not a positivity proof. All gates include remote rows.quantify_head_error.py (2 KB) — Certified ordinary-head lower errors, never positivity from an inconclusive bound.rank_one_joint_progress.txt (938 B) — (no description in file)rank_one_odd_verify_b768_progress.txt (4 KB) — verified column odd 1 / 16 seconds 13.9rank_one_trial.py (3 KB) — One finite residual repair of a proposed bad direction; full verification separate.rank_one_trial_progress.txt (1 KB) — rank-one step 1 [0.0001828208367839724611749556276781974985955 +/- 4.97e-44] 6.3refine_inverse.py (5 KB) — Reproducible Arb experiments for the complete lambda=4 tail inverse. Finite-prefix experiments are not certificates of the complete Schur sign.render_v126.py (1 KB) — (no description in file)review_rank_one_trial.md (18 KB) — Adversarial review: one repaired direction and the complete complementsimultaneous_odd_M4096_s16_normalized_ingredients_b768.json (5.5 MB) — (no description in file)simultaneous_odd_M4096_s16_normalized_ordinary_error_b768.json (4 KB) — (no description in file)simultaneous_odd_M4096_s16_normalized_report_b768.json (3 KB) — (no description in file)simultaneous_odd_M4096_s16_normalized_witness.json.gz (5.1 MB) — (no description in file)simultaneous_odd_M4096_s32_normalized_ingredients_b768.json (5.5 MB) — (no description in file)simultaneous_odd_M4096_s32_normalized_joint_lower_b768.json (105 KB) — (no description in file)simultaneous_odd_M4096_s32_normalized_joint_report_b768.json (938 B) — (no description in file)simultaneous_odd_M4096_s32_normalized_report_b768.json (3 KB) — (no description in file)simultaneous_odd_M4096_s32_normalized_witness.json.gz (5.1 MB) — (no description in file)simultaneous_trial.py (9 KB) — Complete matrix Schur verification of independently frozen finite trials. No finite-solve convergence or positivity is used as a certificate.window_scaling_probe.py (7 KB) — Growing-window diagnostics. Finite corrections are NOT complete tail uppers. Fresh analytic far-cut bounds are independently certified at every window.