From 82a49011de15287583d8cfec11ac5cca7efee747 Mon Sep 17 00:00:00 2001 From: YurenHao0426 Date: Sat, 13 Jun 2026 04:33:08 -0500 Subject: Add master technical reference; retire unverified e0 headline Note 42 is a self-contained write-from document: full statements and proofs for Thm 1-4 / Prop 1-3 / Cor 1-2, every result with exact numbers and configs cross-checked against the CSVs, the negative results, and the honesty boundaries. Audit finding while building it: the "E_B[e0]=0.4378 +/- 0.008 vs 0.4403" headline (cited in notes 29/35/36/40/41 and the old draft) is not reproducible from the current moments CSV. Retired in favor of the reproducible 6x6 table (max |error| <= 0.008, predicted span 0.15-0.45). Living docs updated; note 37 sec 4 records the retirement. Co-Authored-By: Claude Fable 5 --- notes/37_wrapup_corrections.md | 19 ++++++++++++++++++- 1 file changed, 18 insertions(+), 1 deletion(-) (limited to 'notes/37_wrapup_corrections.md') diff --git a/notes/37_wrapup_corrections.md b/notes/37_wrapup_corrections.md index 671c237..ad2d60b 100644 --- a/notes/37_wrapup_corrections.md +++ b/notes/37_wrapup_corrections.md @@ -68,7 +68,24 @@ form — still zero fitted parameters — tracks the ramp at 0.982 and isolates 0.977; the figure now shows measured (error bars), matched-init curve, ensemble geometric mean, and the init-seed band. -## 3. Related cleanup executed with this audit +## 4. e0 moment headline "0.4378/0.4403" not reproducible (retired) + +Found 2026-06-13 while building the master technical reference (note 42). +The figure `E_B[e_0] = 0.4378 +/- 0.008` vs hidden share `0.4403`, cited in +notes 29/35/36/40/41 and the old `paper/body.tex`, does NOT appear anywhere in +the current `actual_fa_initial_operator_moments/initial_operator_moment_rows.csv` +(6 widths {16,24,32,48,64,96} x 6 init seeds; no row or aggregate is ~0.44; +closest is width 16 init 0: predicted 0.4519 / empirical 0.4443). It is +presumed to come from a superseded run that was overwritten, or from the +AI-written draft. + +Retired. The reproducible Thm 3 evidence is the full 6x6 table (note 42 §6): +the no-fit prediction sits within Monte-Carlo error of the measurement, **max +|error| <= 0.008**, with predicted erosion spanning 0.15-0.45 (a wide-range +calibration, stronger than a single point). Living docs (36, 40, 41, README) +updated to this statement; historical notes 29/35 left as-is with this pointer. + +## 5. Related cleanup executed with this audit - `scripts/alignment_recovery_dynamics.py`: superseded probe (note 35); its divide-by-zero guard was fixed earlier but the rerun was never completed -- cgit v1.2.3