Golden bridge p1 bibliography

docs/golden-bridge/BIBLIOGRAPHY_NOTES.md

@@ -0,0 +1,113 @@
+# Bibliography for Paper 1 — what was found, what was built
+
+## What was there
+
+Paper 1 had **no bibliography**. Not a thin one — none:
+
+- zero `\cite{}` commands anywhere in `chapters/p1-*.tex`
+- no `.bib` file, no `\bibliography`, no bibliography package in `main.tex`
+- instead, 18 `\subsection{References}` blocks, one per chapter
+
+Those 18 blocks were not 18 different lists. They were **the same three lines
+copied verbatim into every chapter**: the Zenodo monograph, the Pellis viXra
+preprint, and the Sherbon HAL preprint. Three preprints, zero peer-reviewed
+sources, repeated 18 times.
+
+The same block appears in **38 files across the whole compendium**, so Papers 2
+and 3 carry the identical hole — 20 files still do, see *Remaining work*.
+
+Meanwhile the prose cited real literature properly — Rissanen, Barron/Rissanen/Yu,
+Grünwald, Chaitin, PDG, CODATA, Planck, Udrescu & Tegmark, Angelis et al. — but
+only as inline text. None of it was in any list.
+
+Appendix D held a source list where **all nine URLs were empty**
+(`- PDG 2024 Physical Constants table: URL:` with nothing after). That template
+was never filled in. It has been replaced by real citations.
+
+## What was built
+
+`references.bib` — 14 entries. Every one verified, none typed from memory:
+
+| Source | How verified |
+|---|---|
+| Rissanen 1978, Barron–Rissanen–Yu 1998, Benjamini–Hochberg 1995, PDG 2024 | Crossref, plus cross-check against the already-audited bibliography of the Pellis–Vasilev–Olsen short paper |
+| Angelis et al. 2023, Udrescu & Tegmark 2020, CODATA 2022, Planck 2018 VI | Crossref API, by DOI or exact-title query |
+| Grünwald 2007, Chaitin 1987, Olsen 2006 | books; publisher, year, ISBN |
+| Pellis viXra, Sherbon HAL, Vasilev Zenodo | marked **preprint / self-published** in a `note` field |
+
+Two Crossref queries initially returned **wrong** matches — CODATA *2014* and
+Planck *2015* — from fuzzy bibliographic search. Both were re-queried by exact
+title and DOI. Do not trust a single fuzzy Crossref hit for these.
+
+The three non-peer-reviewed sources carry an explicit note. They are the claims
+*under test*, not supporting evidence, and the file says so. Do not silently
+upgrade them to journal status.
+
+## Changes to the chapters
+
+- 18 duplicated `\subsection{References}` blocks removed.
+- **34 citations** inserted into the prose (28 in pass 1, 6 in pass 2), via
+  `cite_p1.pl` and `cite_p1_pass2.pl` — kept in the tree so the conversion is
+  reproducible and reviewable rather than a mystery diff.
+- Appendix D's hollow URL list replaced with real citations.
+- `main.tex`: added `\usepackage[numbers,sort&compress]{natbib}`,
+  `\bibliographystyle{plainnat}`, `\bibliography{references}`.
+
+## Verification
+
+All 9 citation keys used in Paper 1 resolve against `references.bib`
+(set difference used-minus-defined is empty). A minimal document citing all 14
+entries compiles with **0 BibTeX warnings** and renders every entry with its
+DOI.
+
+## The compendium now builds — it never did before
+
+`tectonic main.tex` now produces a **139-page PDF** with the bibliography
+typeset at the end. Before this work it did not compile at all.
+
+The source is ASCII-flattened mathematics from a PDF/markdown conversion, and
+it carried five distinct classes of fatal error. `mathcheck.pl` reports the
+first two; all are fixed by `mathfix.pl` plus a few hand edits:
+
+| Class | Count | Example |
+|---|---|---|
+| line ends mid-math, span runs on and dies at the paragraph break | 23 | `...also uses $\varphi` |
+| `^` / `_` in text mode | 103 | `R^+`, `F^*(theta)` |
+| double-escaped underscore — `\\` is a line break, `_` then bare | 2 | `\texttt{uio\\_out}` |
+| markdown `#` table header turned into `\section{}` inside `tabular` | 2 | `\section{\& Item \& Required Answer \\}` |
+| stray closing brace from a mangled superscript | 1 | `{0,1}D}` was `{0,1}^D` |
+
+`mathfix.pl` closes the open spans and escapes the stray `^`/`_`. It does **not**
+reconstruct the intended mathematics: the goal was a document that builds, not
+one that is typeset correctly.
+
+**That distinction matters.** The conversion damage is still there and still
+visible in the output — `$\in$fty` where `infty` was meant (the letters "in"
+became the set-membership symbol), `muT(x)`, `lambda^*L`, `alpha-1`. Rendering
+the flattened notation properly is a separate and much larger job.
+
+### One correction made during this work
+
+The first version of `mathfix.pl` recognised only `$...$` as math and so
+escaped the `^` inside `\ensuremath{{}^{2}}` in `main.tex`, breaking all
+thirteen Unicode superscript definitions (⁻ ⁺ ⁰ ¹ … ⁿ). They would have
+rendered as a literal caret. Reverted; `main.tex` now differs from the
+bibliography commit by nothing. The `equation` and `align` environments were
+checked and were never touched.
+
+## Remaining work
+
+1. **Repair the flattened mathematics.** The document compiles but much of its
+   notation is wrong in the output. This is the large remaining task.
+2. **Unicode characters silently dropped.** The build warns on `…`, `⟂`, `≲`,
+   `ℚ` — absent from the `ec-lmr10` font and missing from the
+   `\newunicodechar` table in `main.tex`. They vanish from the PDF without an
+   error.
+3. **20 files in Papers 2 and 3** still carry the three-preprint boilerplate.
+   `references.bib` already covers the shared sources; the same scripts
+   generalise.
+4. **436 placeholders** (`\_\_C0\_\_`, `\_\_M0\_\_`, …) across 17 files,
+   including the CRediT table in `frontmatter/fm-02-attribution.tex`. They now
+   print literally into the PDF.
+5. Decide **Catalog42 vs Catalog15** — the compendium says Catalog42
+   throughout; the short paper says Catalog15.

docs/golden-bridge/appendix/appx-cat42-B-proof-closure.tex

@@ -14,6 +14,6 @@ \section{Appendix B: Catalog42 Proof-Closure Program}
 
   VERDICT
 
-                                  TRINITY S3AI - the cognitive stack rooted in the identity phi^2 + phi^-2 = 3
+                                  TRINITY S3AI - the cognitive stack rooted in the identity phi\textasciicircum{}2 + phi\textasciicircum{}-2 = 3
 
                                                              APPENDIX D

docs/golden-bridge/appendix/appx-cat42-C-coq-status.tex

@@ -26,6 +26,6 @@ \section{Appendix C: Catalog42 Coq Closure --- Final Status}
 
   STYLE LAW
 
-                                TRINITY S3AI - the cognitive stack rooted in the identity phi^2 + phi^-2 = 3
+                                TRINITY S3AI - the cognitive stack rooted in the identity phi\textasciicircum{}2 + phi\textasciicircum{}-2 = 3
 
                                                                      APPENDIX E

docs/golden-bridge/appendix/appx-cat42-D-style-gate.tex

@@ -27,4 +27,4 @@ \section{Appendix D: Style and PDF Integrity Gate}
 
   PR LOCK
 
-                                      TRINITY S3AI - the cognitive stack rooted in the identity phi^2 + phi^-2 = 3
+                                      TRINITY S3AI - the cognitive stack rooted in the identity phi\textasciicircum{}2 + phi\textasciicircum{}-2 = 3

docs/golden-bridge/appendix/appx-cat42-E-build-ritual.tex

@@ -24,4 +24,4 @@ \section{Appendix E: Build Commands as Reproducible Ritual}
 
                Canonical tail rewrite v22.12: all final pages now use the same white B\&W atlas language.
 
-                              TRINITY S3AI - the cognitive stack rooted in the identity phi^2 + phi^-2 = 3
+                              TRINITY S3AI - the cognitive stack rooted in the identity phi\textasciicircum{}2 + phi\textasciicircum{}-2 = 3

docs/golden-bridge/appendix/appx-dna-A1-integration.tex

@@ -20,7 +20,7 @@ \section{Appendix A.1: v21 TRINITY DNA Integration}
       Status: provenance and reproducibility evidence only; not a completed Standard Model
                                            derivation.
 
-                           TRINITY S3AI - the cognitive stack rooted in the identity phi^2 + phi^-2 = 3
+                           TRINITY S3AI - the cognitive stack rooted in the identity phi\textasciicircum{}2 + phi\textasciicircum{}-2 = 3
 
                                                             APPENDIX A.2
 

docs/golden-bridge/appendix/appx-dna-A2-evidence-matrix.tex

@@ -21,7 +21,7 @@ \section{Appendix A.2: TRI-1 Evidence Matrix}
      Each row is a provenance claim. It can support traceability and reproducibility, but it does not become a
      physics derivation unless the corresponding theorem and measurement comparison are closed.
 
-                                TRINITY S3AI - the cognitive stack rooted in the identity phi^2 + phi^-2 = 3
+                                TRINITY S3AI - the cognitive stack rooted in the identity phi\textasciicircum{}2 + phi\textasciicircum{}-2 = 3
 
                                                        APPENDIX A.3
 

docs/golden-bridge/appendix/appx-dna-A3-claims.tex

@@ -7,7 +7,7 @@ \section{Appendix A.3: Claims Discipline and Corrections Lock}
   A5, H3, E8, RG flow and GOLDEN BRIDGE (Trinity-Pellis Bridge) bridges remain conjectural or programmatic unless the manuscript
   supplies a concrete representation-theoretic or field-theoretic derivation.
 
-                           TRINITY S3AI - the cognitive stack rooted in the identity phi^2 + phi^-2 = 3
+                           TRINITY S3AI - the cognitive stack rooted in the identity phi\textasciicircum{}2 + phi\textasciicircum{}-2 = 3
 
                                                           APPENDIX A.4
 
@@ -16,10 +16,10 @@ \section{Appendix A.3: Claims Discipline and Corrections Lock}
 
           Item                  Core-paper status                      Allowed wording                        Reviewer-safe reason
 
- phi identity              Primary mathematical anchor      Exact identity: phi^2 + phi^-2 = 3.       Closed algebraic statement.
+ phi identity              Primary mathematical anchor      Exact identity: phi\textasciicircum{}2 + phi\textasciicircum{}-2 = 3.       Closed algebraic statement.
 
  Pellis alpha expression   Primary numerical result         High-precision phi-polynomial             Parametrization unless mechanism is
-                                                            parametrization of alpha^-1.              derived.
+                                                            parametrization of alpha\textasciicircum{}-1.              derived.
 
  Trinity monomials         Catalogue result                 Low-complexity monomial                   Search space can be penalized.
                                                             representations over {2,3,phi,pi,e}.

docs/golden-bridge/appendix/appx-dna-A4-ledger.tex

@@ -13,7 +13,7 @@ \section{Appendix A.4: Integration Ledger}
   Every page must distinguish result, conjecture, mechanism proposal and implementation artifact. This is
   the central anti-numerology discipline of the GOLDEN BRIDGE (Trinity-Pellis Bridge) manuscript.
 
-                              TRINITY S3AI - the cognitive stack rooted in the identity phi^2 + phi^-2 = 3
+                              TRINITY S3AI - the cognitive stack rooted in the identity phi\textasciicircum{}2 + phi\textasciicircum{}-2 = 3
 
                                                         APPENDIX A.5
 
@@ -35,4 +35,4 @@ \section{Appendix A.4: Integration Ledger}
 
  Repository overclaim         TRINITY DNA appendix.               Repo artifacts are provenance evidence only.    Managed.
 
-                            TRINITY S3AI - the cognitive stack rooted in the identity phi^2 + phi^-2 = 3
+                            TRINITY S3AI - the cognitive stack rooted in the identity phi\textasciicircum{}2 + phi\textasciicircum{}-2 = 3

docs/golden-bridge/appendix/appx-dna-A5-risk-register.tex

@@ -19,7 +19,7 @@ \section{Appendix A.5: Reviewer Risk Register}
   Move each UnderRevision entry into a formula-specific theorem file, bind it to the numeric tolerance
   table, and keep the article text synchronized with the proof ledger.
 
-                           TRINITY S3AI - the cognitive stack rooted in the identity phi^2 + phi^-2 = 3
+                           TRINITY S3AI - the cognitive stack rooted in the identity phi\textasciicircum{}2 + phi\textasciicircum{}-2 = 3
 
                                                                APPENDIX C
 

docs/golden-bridge/chapters/p1-01-introduction.tex

@@ -9,16 +9,16 @@ \section{P1 §1. Introduction}
 
   1.1 The Open Problem of Dimensionless Constants
   The numerical values of the fundamental dimensionless constants of nature --- the
-  inverse fine-structure constant alpha-1 = 137.035 999 177(21) (NIST CODATA 2022),
-  the strong coupling alphas(mZ) = 0.1180(9) (PDG 2024), the electroweak mixing angle
-  sin2theta(MZ) = 0.23129(4) in the MS scheme (PDG 2024), and the full array of mixing
+  inverse fine-structure constant alpha-1 = 137.035 999 177(21) \citep{codata2022},
+  the strong coupling alphas(mZ) = 0.1180(9) \citep{pdg2024}, the electroweak mixing angle
+  sin2theta(MZ) = 0.23129(4) in the MS scheme \citep{pdg2024}, and the full array of mixing
   angles, mass ratios, and cosmological parameters --- remain empirical inputs to the
   Standard Model. No known symmetry principle or dynamical mechanism within
   quantum field theory determines their specific values; each must be measured and
   inserted by hand. The question of whether these values carry any deeper algebraic or
   structural character is thus not closed by the Standard Model itself, and constitutes a
   legitimate open question at the intersection of mathematical physics and information
-  theory (PDG 2024).
+  theory \citep{pdg2024}.
 
   1.2 A Precisely Bounded Reformulation
   A sharply bounded, operationally well-posed version of this question avoids dynamical
@@ -28,9 +28,8 @@ \section{P1 §1. Introduction}
   question about data compression, not about physics. It belongs to the domain of
   symbolic regression, algorithmic information theory, and statistical model selection
   under structured hypothesis classes --- the same theoretical territory inhabited by the
-  Minimum Description Length principle (Barron, Rissanen \& Yu, IEEE Trans. Inf.
-  Theory, 1998; Grünwald, MIT Press, 2007) and Chaitin's Algorithmic Information
-  Theory (Chaitin, Cambridge University Press, 1987).
+  Minimum Description Length principle \citep{barron1998,grunwald2007} and Chaitin's Algorithmic Information
+  Theory \citep{chaitin1987}.
 
   This paper formalizes the question within a precisely defined statistical framework. A
   symbolic grammar G$\varphi$ is generated by the basis {$\varphi$, pi, e, 3} with integer exponents
@@ -86,8 +85,3 @@ \section{P1 §1. Introduction}
   algebraic properties of the basis, phase-transition structure, and a mapping from the
   PhD monograph to Paper 1.
 
-\subsection{References}
-
-- Vasilev, D. \textit{Flos Aureus Monograph: Trinity Constants}. DOI \href{https://zenodo.org/doi/10.5281/zenodo.19227877}{10.5281/zenodo.19227877}.
-- Pellis, S. "The Fine-Structure Constant and the Golden Ratio." \href{https://vixra.org/pdf/2110.0117v4.pdf}{viXra 2110.0117 v4}.
-- Sherbon, M. A. "Fundamental Nature of the Fine-Structure Constant." \href{https://hal.science/hal-02341850/document}{HAL hal-02341850}.

docs/golden-bridge/chapters/p1-02-background.tex

@@ -11,7 +11,7 @@ \section{P1 §2. Background and Related Work}
   Attempts to express dimensionless physical constants in closed algebraic form have a
   long and largely undisciplined history. Early efforts concentrated on alpha-1 approx
   137, with various proposals involving integers and simple arithmetic --- all post-hoc and
-  statistically uncontrolled (PDG 2024 review articles). The combinatorial proliferation of
+  statistically uncontrolled \citep{pdg2024}. The combinatorial proliferation of
   candidate expressions at increasing complexity was universally ignored, making
   significance claims meaningless.
 
@@ -28,11 +28,10 @@ \section{P1 §2. Background and Related Work}
   Symbolic regression --- the automated discovery of mathematical expressions fitting
   numerical data --- has been developed extensively in the machine learning community,
   with methods ranging from genetic programming to neural-guided search. The AI
-  Feynman algorithm of Udrescu and Tegmark (Science Advances, 2020) demonstrated
+  Feynman algorithm of Udrescu and Tegmark \citep{udrescu2020} demonstrated
   that physics-inspired techniques including dimensional analysis, separability detection,
   and brute-force enumeration can recover 100 equations from the Feynman Lectures on
-  Physics. More recent work in physics-informed symbolic regression (Angelis et al.,
-  Archives of Computational Methods in Engineering, 2023) has extended these methods
+  Physics. More recent work in physics-informed symbolic regression \citep{angelis2023} has extended these methods
   to broader scientific domains.
 
   The present work differs from all these approaches in a fundamental respect: the target
@@ -44,12 +43,12 @@ \section{P1 §2. Background and Related Work}
   hypothesis class.
 
   2.3 Minimum Description Length and Bayesian Model Comparison
-  The Minimum Description Length principle, originating in Rissanen's 1978 paper
-  Modeling by the shortest data description (Automatica, 1978) and systematized in
-  Barron, Rissanen \& Yu (1998), provides a rigorous basis for model comparison that
+  The Minimum Description Length principle, originating in Rissanen's 1978 paper \citep{rissanen1978}
+  Modeling by the shortest data description \citep{rissanen1978} and systematized in
+  Barron, Rissanen \& Yu \citep{barron1998}, provides a rigorous basis for model comparison that
   penalizes complexity in terms of code length. The core insight is that any regularity in
   data can be exploited to compress it, and the best model is the one achieving the
-  shortest description of data and model combined. Grünwald's comprehensive treatment
+  shortest description of data and model combined. Grünwald's comprehensive treatment \citep{grunwald2007}
   (MIT Press, 2007) formalizes the connections between crude two-part codes,
   normalized maximum likelihood, Bayesian marginal likelihoods, and prequential coding,
   establishing asymptotic equivalences between these approaches.
@@ -60,8 +59,7 @@ \section{P1 §2. Background and Related Work}
   deployed in Section 7 with explicit penalization for the size and structure of the
   symbolic hypothesis class.
 
-  The connection to Kolmogorov-Chaitin algorithmic complexity (Chaitin, Cambridge
-  University Press, 1987) is conceptual: the MDL compression score DeltaMDL provides a
+  The connection to Kolmogorov-Chaitin algorithmic complexity \citep{chaitin1987} is conceptual: the MDL compression score DeltaMDL provides a
   computable, statistical lower bound on the potential reduction in Kolmogorov
   complexity achievable by the structured grammar.
 
@@ -131,8 +129,3 @@ \section{P1 §2. Background and Related Work}
   This tier system is deliberately reviewer-facing. If a sentence can be read in more than
   one tier, the lower evidentiary tier controls the interpretation.
 
-\subsection{References}
-
-- Vasilev, D. \textit{Flos Aureus Monograph: Trinity Constants}. DOI \href{https://zenodo.org/doi/10.5281/zenodo.19227877}{10.5281/zenodo.19227877}.
-- Pellis, S. "The Fine-Structure Constant and the Golden Ratio." \href{https://vixra.org/pdf/2110.0117v4.pdf}{viXra 2110.0117 v4}.
-- Sherbon, M. A. "Fundamental Nature of the Fine-Structure Constant." \href{https://hal.science/hal-02341850/document}{HAL hal-02341850}.