Edition README
Komorebi Methodology
README.md
1 # The Komorebi Methodology
2 ## Mathematical Discovery, Realization, and the Discipline of Proof
3
4 **Author:** Komorebi P.
5 **Series:** Realization Calculus / Mathematical Essays
6
7 This is the reviewed full essay, not an outline. The reader contains **28 physical
8 pages**, nine chapters, a conclusion, a preface, three reproducible figures, source
9 notes, and references. The main text contains **7,611 words** under
10 `texcount -1 -sum essay.tex`, including countable headings. The approved dark-plum
11 monograph cover, white interior, serif typography, and restrained gold accents remain.
12
13 ## Read and edit
14
15 Start with `Komorebi_Methodology.pdf`.
16
17 The essay follows one argument: realization makes a question accessible, obstruction
18 gives failure mathematical content, admissible correction supplies a constructive
19 response, verification establishes scope, and mathematical memory supports reuse.
20 The returning-coordinate example now continues through an exact memory formula and
21 its all-time induction proof rather than ending at the failed simplification.
22
23 Editable files:
24
25 - `Komorebi_Methodology.tex`: design, cover, title page, preface, contents, assembly.
26 - `essay.tex`: nine chapters and conclusion.
27 - `sources.tex`: source notes, scope, and bibliography.
28 - `figures.py`: three vector diagrams and their PNG previews.
29
30 The PDF contains no administrative checkpoint labels. `PROGRESS_LEDGER.md` and
31 `audit/REVIEW_REPORT.md` preserve the editorial history separately.
32
33 ## Verification and its boundaries
34
35 The recorded run passes **40/40 exact-check groups with 665 finite cases** and
36 **17/17 manifest-validator controls**. The former cover exact matrix calculations,
37 finite factorization and continuation tests, rational memory recurrences, arithmetic
38 transforms, obstruction witnesses, and the declared proof-field illustration. These
39 are cases in a test suite, not counts of new mathematical theorems. The validator
40 controls are separately counted actual command-line executions against isolated
41 fixtures. They include genuine negative controls and check the intended diagnostics.
42
43 The final three-pass typesetting build contains no reported overfull/underfull boxes,
44 LaTeX warnings, or missing-character messages. All 28 pages were rendered and
45 reviewed. The structural audit found no out-of-page text spans, unresolved reference
46 markers, replacement glyphs, invalid internal links, or bookmark-title mismatches.
47 All 27 bookmarks were checked against their destination pages.
48
49 A source-only rebuild under the same installed environment reproduced all 28 pages
50 with matching extracted text, bookmarks, and rendered pixels at 72 dpi using the same
51 renderer. Its exact-calculation and integrity-control records also match. This is
52 reproducibility, not an independent implementation of the tools. PDF byte identity
53 is not asserted. The visual editorial review used separate 145-dpi page renders.
54
55 **No new Lean compilation or corpus-wide formal audit was performed.** The
56 continuation and memory identities have written proofs. The unbounded tower and
57 infinite Hilbert-space example also rely on their written arguments, not on finite
58 tests. Historical Lean records remain attributed to their source documents. The
59 comparison of research workflows is a proposed study, not a completed experiment.
60
61 `audit/CLAIM_MAP.json` connects the major exposition claims to these evidence types.
62 The source package does not bundle the full companion monographs or Lean projects.
63
64 ## Verify the delivered package first
65
66 In the freshly extracted directory, before creating or rebuilding files, run:
67
68 ```text
69 python verify_manifest.py
70 ```
71
72 The verifier requires the **complete file inventory** in `MANIFEST_SHA256.txt` and
73 checks its SHA-256 digests. It rejects empty or incomplete inventories, extra or
74 missing files, modified payloads, duplicates, and unsafe paths. A metadata file added
75 by an operating system can therefore trigger a strict unlisted-file diagnostic.
76 Inspect that diagnostic rather than silently weakening the check.
77
78 The previous validator incorrectly accepted an empty manifest. That defect was
79 reproduced, recorded, and corrected. The original package's nonempty hash check is
80 not thereby invalidated. See `audit/review_baseline_finding.json`.
81
82 Integrity is relative to the supplied manifest. It is **not authentication** of the
83 author and **not a mathematical correctness proof**. Regeneration changes files and
84 creates build artifacts, so a delivery manifest is not expected to pass afterward.
85 Do not overwrite the delivered manifest to disguise those changes.
86
87 ## Run the standard-library checks
88
89 ```text
90 python exact_checks.py
91 python integrity_checks.py
92 ```
93
94 These require Python 3.10 or later but no third-party Python packages or TeX.
95 Both produce readable and machine-readable records in `audit/`. The integrity
96 controls use temporary fixtures and do not mutate the delivered inventory.
97
98 ## Rebuild the PDF
99
100 Install the figure dependencies and use the shared build entry point:
101
102 ```text
103 python -m pip install -r requirements.txt
104 python build.py
105 ```
106
107 PowerShell: `./build.ps1`. Unix-like shell: `./build.sh`.
108
109 `build.py` runs both check suites, regenerates the figures, and performs three
110 `pdflatex` passes. It preserves compiler output in `audit/build_pass_*.log` and
111 returns failure when a command or the checked final-log classes fail. Visually inspect
112 the rebuilt PDF after editorial changes. Command success is not visual review.
113
114 Use a TeX distribution providing `newpxtext`, `newpxmath`, `mathtools`, `microtype`,
115 `tikz`, `tcolorbox`, `titlesec`, `fancyhdr`, `hyperref`, `bookmark`, and dependencies.
116 The display family is TeX EB Garamond (`EBGaramond-LF`) and the sans-serif family is
117 TeX Gyre Heros (`qhv`). Install these using TeX Live or MiKTeX. **No font binaries
118 are distributed in the ZIP.** Fonts embedded in the PDF are part of normal typesetting.
119
120 The actual environment is recorded in `audit/build_environment.json`. Dependency
121 ranges are not a fully locked environment. Different toolchain or font versions may
122 alter pagination or metadata.
123
124 ## Optional PDF and reproduction checks
125
126 ```text
127 python -m pip install -r requirements-review.txt
128 python review_pdf.py
129 python reproduce.py
130 ```
131
132 The optional scripts use PyMuPDF. The first checks PDF structure and navigation.
133 The second starts from sources in a temporary directory with no inherited PDF,
134 figures, or calculation receipts, rebuilds, and compares to the current reader.
135 It still uses the same installed tools. Neither is a proof of the essay's
136 methodological effectiveness or an independent mathematical review.
137
138 ## Audit records and continuity
139
140 `audit/REVIEW_REPORT.md` explains the substantive changes and corrected failures.
141 `audit/QA_REPORT.json` summarizes current checks. `audit/SOURCE_REGISTER.md` identifies
142 bibliographic editions and public documentation. `audit/provenance.json` binds the
143 current source and PDF to the baseline. `PROGRESS_LEDGER.md` states the restart point.
144
145 `audit/history/editorial-01/` preserves the first full-essay receipts as **historical**
146 evidence. Its page counts, hashes, and scope apply to that earlier artifact, not to
147 this reader. Current evidence is directly under `audit/`.
148
149 This package does not change a project-wide licensing decision or relicense
150 third-party material. The public source PDFs and font distributions are not bundled.