Skip to content

Main: precision follow-ups to the foundational correction (status, scope, pointers) - #925

Merged
amaybaum merged 15 commits into
mainfrom
draft/main-correction-followups
Oct 25, 2026
Merged

amaybaum merged 15 commits into
mainfrom
draft/main-correction-followups

Conversation

@amaybaum

@amaybaum amaybaum commented Oct 25, 2026 •

Copy link
Copy Markdown
Owner

These are the Main.md precision edits held out of #922 (merged as 8dc36d55). The branch is reconciled with main by a merge commit, which keeps the reviewed history.

  • Head: 75a9944487b8b415bd40b5b9002c79650c6f8fde, tree cf5d29cc.
  • Files changed: papers/Main.md, papers/Methodology.md (one mirror line), verification/coverage/LEDGER.json, and their regenerated .tex and PDFs.

Merge: authorized by the owner at exact head 75a99444, as a normal merge commit.

Changes

Commit Site Change
607c23e7 B-F3 Main:16, :530 The Q_fb class is stated as the kernel defines it.
d2a7c4ff B-F4 Main:28 The introduction states the identification at its proved scope.
6909c672 B-F5 Main:608, :690 The characterization's status is stated at its proved strength. No OIBridge module formalizes (ii)⟺(iii).
510df1d1 B-F6 Main:580 Summary claim (2) is stated with its proof route.
0a172590 B-F10 + eb3e5fc5 Main:62, Methodology:146 Double stochasticity of uniform-prior marginals is attributed to counting, not to Birkhoff–von Neumann (the converse direction).
22625cd2 B-F12 Main:167, LEDGER The readback lemma states only the hypothesis its proof uses.
deb1c913 merge main 8dc36d55, carrying #922, merged in. Main.tex regenerated; LEDGER and census reconciled.
045fd344 PDFs Main.pdf and Methodology.pdf rebuilt.
ec6beb61 Main:580, :588, :532 See below.
1cafacf8 PDF Main.pdf rebuilt.
a04443bf Main:86, :522 See below.
2c5d7c30 PDF Main.pdf rebuilt.
4c5012ab Main:530, :580 See below.
75a99444 PDF Main.pdf rebuilt.

ec6beb61

a04443bf

  • The pointers at Main:86 and :522 now read "(§1.2, Remark on the measure as a realization datum)". The former targets, a "§1.3 Remark" and a "§1.4 Remark", do not exist.
  • The LEDGER fingerprint of MAIN:L-partition-relativity-under-canonical-invariant is re-affirmed. Only the pointer text changed.

4c5012ab

  • The Q_fb class names the read map ("the visible record a fixed function of each outcome").
  • Summary claim (2) carries its status word.

Verification

Independent review: ACCEPT WITH NITS at 2c5d7c30, with the scope addition reviewed on its own as ACCEPT.

The two required nits, N1 (the read map) and N2 (the status word), are applied in 4c5012ab in the reviewer's wording. The word diff 2c5d7c30..75a99444 is exactly those two insertions plus the regenerated .tex and PDF.

Checks at the head all pass:

  • staleness, voice, claims, duplicate, coverage (128), census, dependency-label --strict;
  • mirror, citation (112/0), architecture (259/0);
  • legacy_records_check (303 intact);
  • edge-rigidity, finite-horizon equivalence and equivalence-chain probes.

Pre-merge integration check on the trial merge with main ebf98d83 (clean, tree c1cc4b57):

  • the checks above, plus artifact-placement, build_migration_manifest --check, and the canonical-measure and Bohr-frequency probes;
  • legacy records: 303 intact;
  • V3 self-test.

All pass. The PR's CI at 75a99444 is green: 25 pass, 1 skipped.

Some commit messages of the original seven say "(local draft, not pushed)", and eb3e5fc5 cites a superseded draft SHA (568d22b7). These are left unchanged so that no history is rewritten.

Optional observations, not applied:

  • O1: a (C1)/(C3) "which condition is primitive" remark at Main:608.
  • O2: book ch01:159 "hidden register prepared in the corresponding prior" could be harmonized with Main:532, in both files.
  • O3: Substratum:94 cites "[Main §2.2]" for an argument in §1.2. This predates the PR.

Related: #922 (merged foundational correction), #924 (F13 ≤ n³ planning), #926 (manuscript consistency for the book, GR and Explainer; leaves Main.md untouched).

🤖 Generated with Claude Code

https://claude.ai/code/session_013T6pfKuv5mi3MXtKjH2tZb

claude added 15 commits October 25, 2026 12:41
Part B, finding F3 (local draft, not pushed). The abstract (Main 1) and the
equivalence theorem's class (Q_fb) now say what OIBridge/Equivalence.lean's
QfbReal defines: a classical (diagonal) initial law on the fixed basis and a
projective Born-rule readout in that basis at every step (QfbReal.chain).
The proof of D => Q_fb already uses exactly this. No claim strengthened.

Main.tex regenerated (pandoc 3.1.3). STALE: papers/Main.pdf not rebuilt --
xelatex unavailable; rebuild with sh ./build.sh Main before release.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013T6pfKuv5mi3MXtKjH2tZb
Part B, finding F4, Main 1.1 only (local draft, not pushed). "quantum
mechanics is identified as its form" exceeded the paper's own scope remarks
(abstract; 3.4 Remarks on what the equivalence does and does not say; the
operational-extension boundary; Conclusion). The sentence now states the
identification proved in 3.4 -- finite observable laws with fixed-basis
unitary Born representations -- and names the full operational theory as the
separate completion question. The title is not changed here: it is the
deposited title cited with a DOI in Methodology and by Structure and
main_repro; options are listed in the proposal for an owner decision.

Main.tex regenerated (pandoc 3.1.3). STALE: papers/Main.pdf not rebuilt --
xelatex unavailable; rebuild with sh ./build.sh Main before release.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013T6pfKuv5mi3MXtKjH2tZb
Part B, finding F5 (local draft, not pushed).
- 3.4, characterization proof: "Both directions of the equivalence are
  theorems of this paper" now records that both directions are prose
  proofs, not formalized, and that, (C4) coinciding with (ii) in every
  faithful realization (the paper's own Remark), (ii) => (iii) adds only the
  existence of a realization and (iii) => (ii) is the readback lemma.
- 4.4, relation to prior programs: "an equivalence with necessary and
  sufficient conditions ... with representability as a one-way consequence"
  replaced: representability is universal (holds for every finite law, so it
  is not a consequence of (ii)), and the conditions are diagnostics.

Main.tex regenerated (pandoc 3.1.3). STALE: papers/Main.pdf not rebuilt --
xelatex unavailable; rebuild with sh ./build.sh Main before release.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013T6pfKuv5mi3MXtKjH2tZb
Part B, finding F6 (local draft, not pushed). 3.4 Summary (2) labelled
"M_t > 0 <=> unavoidable readout-relevant hidden predictive memory --
proved, every faithful completion". Checked against
OIBridge/HiddenMemory.lean: clauses (a)-(c) are kernel-proved for any joint
law and any next-visible map; the biconditional is not a kernel statement.
It restates clause (a): forward, M_t > 0 gives two positive-probability
histories with different next-step laws, which by (a) are different
pushforwards in every completion; converse, (a) in one completion, which
exists by S => D. Both directions use the vanishing criterion for
conditional mutual information, which is not formalized. The summary now
says so. Consistent with the corrected Layer-1 sentence of 1.5 (Part A).

Main.tex regenerated (pandoc 3.1.3). STALE: papers/Main.pdf not rebuilt --
xelatex unavailable; rebuild with sh ./build.sh Main before release.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013T6pfKuv5mi3MXtKjH2tZb
…ounting (B-F10)

Part B, finding F10 (local draft, not pushed). Main 1.2's status Remark
attributed the double stochasticity of a uniform-prior bijection marginal to
Birkhoff-von Neumann. It follows from bijectivity by counting: column j sums
|phi^{-1}({x_j} x C_H)| / |C_H| = 1. Birkhoff-von Neumann is the converse
decomposition and stays where 3.2's one-step lemma uses it correctly.

Main.tex regenerated (pandoc 3.1.3). STALE: papers/Main.pdf not rebuilt --
xelatex unavailable; rebuild with sh ./build.sh Main before release.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013T6pfKuv5mi3MXtKjH2tZb
…-F10 (F10-fix)

Fix commit for B-F10 (568d22b7), from the independent fidelity review, N3
(local draft, not pushed). papers/Methodology.md 'The hidden-sector margin,
closed' repeated Main 1.2's misattribution of the double stochasticity of a
uniform-prior bijection marginal to Birkhoff-von Neumann. It follows by
counting (each visible value has |C_H| preimages); Birkhoff-von Neumann is
the converse decomposition. Corrected the same way as Main.

Corpus re-scan (Birkhoff / von Neumann / 'doubly stochastic' near
bijection, permutation or marginal) across papers/, book/, FULL.md and
READMEs: every other hit uses Birkhoff-von Neumann for the converse
decomposition (Main 3.2 lemma; appendix-b B.2.2 and its FULL mirror) or is
an acknowledgment, bibliography or preface mention, or states the double
stochasticity without attribution (Main 2.3/3.2/3.4, Substratum 94/136,
ch03 214 and FULL mirror). None needs change.

Methodology.tex regenerated (pandoc 3.1.3; reproduces the checked-in file
byte for byte from the unmodified source). STALE: papers/Methodology.pdf
not rebuilt -- xelatex unavailable; rebuild with sh ./build.sh Methodology
before release.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013T6pfKuv5mi3MXtKjH2tZb
Part B, finding F12 (local draft, not pushed). Main 2.3's Lemma (accessible
backflow from readback) assumed (C1)-(C3); the proof (mixture-divergence form
of the conditional mutual information, Pinsker, triangle inequality) uses
only the gap: two positive-probability histories with a common endpoint,
each of probability at least p_0, with next-step laws at least delta apart.
The hypothesis now reads as the gap (C4) asserts. The statement is weakened
in hypotheses only, so the conclusion holds for every process with such a
gap; no dependent claim used (C1)-(C3) through this lemma.

Coverage ledger: MAIN:L-accessible-backflow-from-readback fingerprint
re-affirmed (449e3df3147485ef -> 1bf64869e7938d80), assumptions field
updated; level P and checks unchanged (c4_backflow_probes.py).

Main.tex regenerated (pandoc 3.1.3). STALE: papers/Main.pdf not rebuilt --
xelatex unavailable; rebuild with sh ./build.sh Main before release.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013T6pfKuv5mi3MXtKjH2tZb
Reconciles the Main.md follow-ups (B-F3..B-F12, F10-fix) onto main at
8dc36d5. Main.md, Methodology.md, LEDGER.json and the census merged
cleanly; the only conflict was the generated papers/Main.tex, regenerated
from the merged Main.md with the build.sh pandoc recipe (source-sha256
recorded). Main.pdf is carried from main and rebuilt in the next commit.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013T6pfKuv5mi3MXtKjH2tZb
PDF-only commit, built with sh ./build.sh Main Methodology (pandoc 3.1.3,
xelatex, tools/unicode-fix.tex); no dropped glyphs reported.

- papers/Main.pdf: source papers/Main.md sha256 0f89ebcb41a3c021...a9add8,
  90 pages, pdf sha256 25557bd118095a88995be11e62dbf0ad81158ea8ff7d47930864fb850260d9f3
- papers/Methodology.pdf: source papers/Methodology.md sha256
  5c5b9741b8a4d217...4c93be, 54 pages, pdf sha256
  12d55d8adff5dc07740bf77333ea06866b0eb506af8aa7e3689dbff55cc8ad86

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013T6pfKuv5mi3MXtKjH2tZb
…initial law

The summary claim (2) at Main:580 and the kernel pointer of the
unavoidable-hidden-predictive-memory theorem at Main:588 now carry the
scope Main:99 gives the same identifiers: pushforward_identity,
distinguishability_floor, capacity_floor and
unavoidable_hidden_predictive_memory are stated for a joint law R.w of
history, present and hidden state, and reading a realization's time-t law
as that joint law is not formalized. Main:532's D => Q_fb proof diagonalizes
the initial law on C_V x C_H (class (D) may correlate the coordinates), not
a hidden prior. Main.tex regenerated with the build.sh pandoc recipe.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013T6pfKuv5mi3MXtKjH2tZb
PDF-only commit, built with sh ./build.sh Main (pandoc 3.1.3, xelatex,
tools/unicode-fix.tex); no dropped glyphs reported; Main.tex unchanged by
the rebuild.

- source papers/Main.md sha256
  9f5b3dcac7b65f7777335b688770b66354ecf0a2ee8c8c31147183244e71b71a
- 90 pages
- papers/Main.pdf sha256
  c5ca60f8b22ee27449d90429942f319f99fc6e3e91ab57f34657c22c0667a0d7

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013T6pfKuv5mi3MXtKjH2tZb
Main:86 cited a "(§1.3 Remark)" and Main:522 a "(§1.4 Remark)"; neither
section has such a remark. The remark that the measure is a realization
datum is Main:60, in §1.2. Both pointers now read "(§1.2, Remark on the
measure as a realization datum)". Main.tex regenerated. The coverage
entry MAIN:L-partition-relativity-under-canonical-invariant is
re-affirmed (only the pointer text changed; mapping, level and delta
unchanged), fingerprint 277d94b19b9a10f2 -> 8e8ec36fdf09a19e.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013T6pfKuv5mi3MXtKjH2tZb
papers/Main.pdf rebuilt with `sh ./build.sh Main` from a04443b
(Main.md sha256 3de4ad921722071d4e3219ddb75ba543687a46c6a20803c4dea52d1cbced7dec). 90 pages, no
missing glyphs; the build leaves Main.tex byte-identical. PDF sha256
c3c351ccd5780b970c0abb08c511e1e49a669bfa7d66cfb50e7c19f50a2cb4d8. PDF file only.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013T6pfKuv5mi3MXtKjH2tZb
…atus

Review of the follow-ups at 2c5d7c3 (ACCEPT WITH NITS):
- N1, Main:530: the Q_fb class description adds "the visible record a
  fixed function of each outcome", the kernel's `read : Bas -> V` (with
  read = id every Q_fb law is Markov; D_imp_Qfb uses read = Prod.fst).
- N2, Main:580: claim (2) states its status word ("proved:"), as the
  Summary's "each with its exact status" requires.
Main.tex regenerated; no coverage fingerprint changes.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013T6pfKuv5mi3MXtKjH2tZb
papers/Main.pdf rebuilt with `sh ./build.sh Main` from 4c5012a
(Main.md sha256 0088a82a5b01add9c6b4f45ded93f9a017d6458dce7ec2361a3cf7da541ff702). 90 pages, no missing glyphs; the build leaves Main.tex
byte-identical. PDF sha256 d286627434356f5af8d2f668925df513c8630deed6a633f3fed43f841861a632. PDF file only.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013T6pfKuv5mi3MXtKjH2tZb
@amaybaum
amaybaum marked this pull request as ready for review October 25, 2026 19:14
@amaybaum
amaybaum merged commit 8b5e868 into main Oct 25, 2026
26 checks passed
@amaybaum
amaybaum deleted the draft/main-correction-followups branch October 25, 2026 19:14
amaybaum pushed a commit that referenced this pull request Oct 26, 2026
…e2db4)

r4 re-bases the site references and fidelity notes to main 52e2db4
(after #925-#928, #916 and #930), records each fidelity note as fixed,
changed or open with its fixing commit, adds F28 (fixed by #930) and the
candidate F29, and applies first-hand full-text reading: the XOR example
is known (RHP 2014 Eq. 15), C3 has full-text support, the divisibility
credit is RHP 2014 Thm 2.4, C4 and C5 stay partly verified, and C8-R
stays a candidate with no priority sentence. Unread primary texts are
listed with the reason each was not read. Correctness and originality
are recorded separately.

Retained: AUDIT-r3.md (byte-identical to the committed r3),
AUDIT-r4-prerecheck.md (the object the re-check read) and
review/RECHECK-r4.md (ACCEPT WITH NITS; its R1, R2 and nits applied).
The parent README's carried-forward findings are updated to their
current status. No PDFs or extractions are committed.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013T6pfKuv5mi3MXtKjH2tZb
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants