Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
108 changes: 108 additions & 0 deletions audits/dropbox_import/2026-09-27/MANIFEST.csv

Large diffs are not rendered by default.

27 changes: 27 additions & 0 deletions audits/dropbox_import/2026-09-27/NOT_IMPORTED.csv
Original file line number Diff line number Diff line change
@@ -0,0 +1,27 @@
dropbox_path,bytes,dropbox_content_hash,reason
Github/00.2_CANONICAL_MASTER.zip,815523,b8ebbebcd075f622ef870e8caf92913692c5dd37534b2b4ea6d87a00c2a37ee5,binary_or_archive
Github/00_CANONICAL_MASTERS_AND_SOURCE.zip,780110,4d1afb7de078426f44d8657d5bf7cdc3c4ce53e09f9bfd94e36ac61757699774,binary_or_archive
Github/00_MASTER_INDEX_AND_ROUTING.zip,45850,165caa58497a235e3594496c565bc62809bd4d7b711e6141569f226c86cc3cdc,binary_or_archive
Github/01_TIER-2_CLASS_A_B_AUTHORIZATION_AND_PILOT.zip,11532,68b8699b662bab976308148e5505ad5550798024e0c3c2d51a0a2954f87b766b,binary_or_archive
Github/05_MASTER_SETUP.zip,53048,27ae358e501c299a88d85108aecf6239516e7b505e8ae9b575ce187891734fa5,binary_or_archive
Github/64_FORMAL_CORE_R2_P02_LM008_LM009.zip,23277,51d20e5d1c67d2701dac396865d791be7a017f6c1e81a2d324a45fcff281e57a,binary_or_archive
Github/K3_SIDE24_LB.zip,19175951,3ab6ecdb08c1fca987b52e96b1a9a5b96c7ce6c1dea6abe96d50fc4a14ed5ba9,binary_or_archive
Github/K3_SIDE24_LB/K3_SIDE24_LB/W4_independent/maps/wedge_fine.tsv,13480081,893f7f42a3b9c6e2d14befe977d1a56afcef58f36038cc50865ef2e111e93997,text_over_256KiB
Github/K3_SIDE24_LB/K3_SIDE24_LB/W4_independent/maps/wedge_finer.tsv,13204497,ec33f210b1055121f612c4b91b1c01d33a9c20cce05119df31a521fca72c2537,text_over_256KiB
Github/LPW_Return_04_Operator_2026-09-12.zip,529626,d56db11321ec9d858e999602f98d1d0c088a0756ee6de898832105dea103564e,binary_or_archive
Github/LPW_Return_05_Archive_Audit_2026-09-12 (1)/receipts/MANIFEST_AUDIT.json,397889,fda16840421026a2576cf10f5638609f66328fe93b97cdb0b3e48a42cf198605,text_over_256KiB
Github/LPW_Return_05_Archive_Audit_2026-09-12.zip,1244032,d1c280f25dbb6506f558cb082d2eb05f5ba07eef7074c646dd2c9a08c18cab0f,binary_or_archive
Github/LPW_Review_Reconciliation_2026-09-12.zip,137941,ecb6966da0f7865fff59098a3b043b4fb064542ed72638ba55a4765c951fed47,binary_or_archive
Github/T2 — SARD-G Transversality.zip,16894,18c3319e265595cd87a7ccdad92208cbfbc9c0001cba6d301cd2820c33cb9caa,binary_or_archive
Github/UPPER2D.zip,2052290,568e7c738571809f1680968f28aef2c4239af098fcbcf22a6700d836b986dd92,binary_or_archive
Github/fig B renyi profiles13M+.png,282714,89b26fc1b0d486f128a941f2ecff3e0d5c4ad4a60f638fbe096ae255f73fa9a3,binary_or_archive
Github/fig C crossover line13M+.png,537140,c4d7c2bb8eaadc54937bc3e15d6637ec9108754a49bf22f1ad4f786e40fd05cd,binary_or_archive
Github/fig D unified framework13M+.png,626048,97cc86c6f2a676eb0dfb5282caf25ba16e0d4d45908f5a4ec23db6d9436597b3,binary_or_archive
Github/master-final-v1.zip,695348,491ef5d0abdf7e1ef682eb4e1aecb089e07c3b063fb52086a4bbd1d91b5a118c,binary_or_archive
Github/master-final-v1/master-final-v1/07.log,511685,06ce40c154051d1cb9b5f1e914aea60f2c9d1fd01410129a8f368228265c8d37,text_over_256KiB
Github/master-final-v1/master-final-v1/pytest.xml,504961,8318a0a06c29e5ec63199b763d684302ab26f06462f35d2be285dbcf19a57ef0,text_over_256KiB
Github/master-final-v1/master-final-v1/report.json,2962992,ba05ee84b0affaf5f3aa1bb37e5b3f9a7864ab246fc92f47746498b62881b0ed,text_over_256KiB
Github/rn-sector-review-master-20260921.zip,700996,ede92971e304ef7cee13f5776c152e7a2e02d308cb924f9d7fab432e9a6c434a,binary_or_archive
Github/rn-sector-review-master-20260921/rn-sector-review-master-20260921/07.log,560309,d4f21d14a65b54583c9242c8c2507dff3aefda53b713c40f3811cfe635df789f,text_over_256KiB
Github/rn-sector-review-master-20260921/rn-sector-review-master-20260921/pytest.xml,505348,0450a4466bc24657c4c4a87aa66851e1f8a39532c60b455e74dcd4749b19d9db,text_over_256KiB
Github/rn-sector-review-master-20260921/rn-sector-review-master-20260921/report.json,2971824,bf5421f9911e8d803e1e996dc3c581ab057e76efef8eaa2f5908db4425de53e9,text_over_256KiB
101 changes: 101 additions & 0 deletions audits/dropbox_import/2026-09-27/README.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,101 @@
# Dropbox import — 2026-09-27

Provenance record of project material that existed only in Dylan's Dropbox.
**Scientific effect: NONE.** Nothing here promotes a claim or a status. Files keep
the claims their authors made at the time, and current Git and Drive carriers
govern present status. Treat every file as historical input for review.

This import follows the owner's standing rule
[OP-PRIVACY-20260927](../../../governance/OP-PRIVACY-20260927.md): only material
related to the project is published here, and nothing personal is.

## What was compared

The Dropbox was listed through the Dropbox connector (5,105 files). Each file
was matched against every blob reachable from any ref of the ten account
repositories and against the Drive source map (snapshot 2026-09-17), using
`tools/dropbox_reconcile.py`:

| tier | files | size |
|---|---:|---:|
| `GIT_EXACT` | 244 | 2.9 MB |
| `DRIVE_EXACT` | 277 | 3.0 MB |
| `GIT_NAME_SIZE` | 909 | 35.6 MB |
| `DRIVE_NAME_SIZE` | 2,529 | 46.4 MB |
| `DUP_IN_DROPBOX` | 275 | 5.6 MB |
| `SKIP_SOFTWARE_BUNDLE` | 235 | 30.7 MB |
| `SKIP_CACHE` | 83 | 2.6 MB |
| `SKIP_RUN_DEBRIS` | 180 | 59.2 MB |
| `MISSING` | 373 | 119.1 MB |

Byte-exact tiers use the Dropbox `content_hash` (SHA-256 over the SHA-256
digests of 4 MiB blocks). Name+size tiers are probable matches, not verified
bytes.

## What happened to the 373 unmatched files

| outcome | files |
|---|---:|
| Published here, byte-identical to the Dropbox original | 93 |
| Published here as extracted text of a PDF or Word file | 14 |
| Project material that could not be imported here (listed in `NOT_IMPORTED.csv`) | 26 |
| Kept private: personal information, keyed share links or confidential review material | 65 |
| Kept private: the owner's other research, not part of this project | 157 |
| Kept private: relation to the project not confirmed | 6 |
| Not imported: other archives, binaries or unreadable files | 9 |
| Empty | 3 |

Only published files are named here. Files kept private are not listed, because
their names could themselves reveal private or unrelated material. They remain in
the owner's Dropbox, and a full manifest is in the owner's private Drive.

- `MANIFEST.csv`: one row per published file, with the Dropbox path, size,
Dropbox content hash, disposition, archive path, SHA-256 of the archived bytes,
a suggested home and a one-line summary.
- `PUBLISHED_EXACT_ORIGINAL`: `files/<Dropbox path>` is byte-identical to the
Dropbox file. The connector returned its text, and those bytes were accepted
only because they reproduce the file's Dropbox `content_hash`.
- `PUBLISHED_EXTRACTED_TEXT`: `files/<Dropbox path>.txt` is the connector's
extracted text of a PDF or Word file, not the original. Layout, figures and
equations may be lost.
- `suggested_home`: 11 files are marked `Math-imports-candidate`: a
mathematical proof, calculation or code that a Math- reviewer should look at.
The rest are program history, operations and reviews.
- `NOT_IMPORTED.csv`: project files known to exist only in Dropbox that this
container could not import. They are archives, figures, and logs or maps over
256 KiB, and the reason is given per row. Many archives' contents are already
here as their extracted loose files.

Paths are relative to the Dropbox root, except that `Github/` is the shared folder
at `2026/Research 07-13-26/Research topology 07-14-26/Github`.

## How files were chosen

1. **Privacy.** Each recovered text went to an agent instructed to read it in full
and propose publish, hold or skip. Two independent adversarial verifiers then
re-read every publish proposal. One checked personal data, secrets and keyed
links; the other checked third-party copyright and provenance. Regex scans for
credentials, emails, phone numbers, keyed URLs and GPS pairs ran first.
2. **Relation to the project.** A second pass classified each privacy-cleared
file against the project's scope ([README](../../../README.md),
[research index](../../../docs/RESEARCH_INDEX.md)). A skeptic re-checked every
"related" call, and only files both agreed on are published.
3. **Final scan.** The exact published bytes were scanned for the personal values
found in held files, and for credential, email, phone, social-profile and
share-link patterns. No match remained.

Agents were instructed to read in full; that is an instruction, not something this
record can prove. Any doubt meant the file was kept private.

## Limits

- The Drive side is the 2026-09-17 source-map snapshot.
- Name+size matches were not re-verified byte for byte.
- The container could not download from Dropbox (`*.dl.dropboxusercontent.com` and
`www.dropbox.com` were refused by its network policy). Only text the connector
could extract was recovered.
- Automated review can miss things. If anything here should not be public, remove
it; the originals remain in Dropbox.

Generated 2026-09-27 by Anthropic / Claude (Claude Code); the Dropbox was treated as
read-only.
Original file line number Diff line number Diff line change
@@ -0,0 +1,18 @@
Let me start by searching the project knowledge for context on this framework, then systematically verify the claims and identify the strongest directions to build on.​​​​​​​​​​​​​​​​

Good — the project knowledge covers your broader theoretical program but the topological fingerprint work lives primarily in the uploaded documents. Let me now do a proper independent verification by running the code, testing the mathematical claims, and then building on what survives.​​​​​​​​​​​​​​​​

Now let me reconstruct the production code from the documents and run a thorough independent verification.​​​​​​​​​​​​​​​​

All core claims verified. Now let me build on the two most promising extensions: the RG-like smoothing flow and the redshift evolution test.​​​​​​​​​​​​​​​​

Excellent results. Now let me compile everything into a rigorous synthesis document.​​​​​​​​​​​​​​​​

Here’s the bottom line from a clean independent run:
All 9 core claims verified. The 4-coordinate fingerprint (J₀, J₁, ΔJ, K_avg) is solid — Gaussian null (p = 0.88), massive effect sizes (Cohen’s d up to 5.4), kernel discrimination (d = 1.94), and clean 3D extension.
The thermodynamic formalism is mathematically correct — F = U - S/β holds to machine precision, Fisher = C/β² is exact, and C(β) shows real structure (identifiable peaks at β* ≈ 5–12).
The two big hypotheses both hold in the test regime:
The Gaussian fixed point (Hypothesis A): smoothing drives ΔJ → 0 across multiple seeds. Breaks down at very large σ due to feature depletion, which is a finite-size issue, not a conceptual one.
The topological second law (Hypothesis B): 8/8 evolution steps show ΔJ monotonically decreasing, from +0.0004 (Gaussian) to −0.086 (strong nonlinearity). J₀ drops while J₁ stays flat — exactly the “peak condensation, trough spread” picture.
The honest caveats: Log-normal is a proxy, not actual gravitational evolution. The ~50% new information from C(β) is real but comes at the cost of a free parameter. Finite-size scaling needs more grid sizes. The RG flow gets noisy when feature counts drop below ~15.
The report, reconstructed production code, and both test scripts are in the files above. Priority next step: get this onto actual N-body simulation snapshots.​​​​​​​​​​​​​​​​
Original file line number Diff line number Diff line change
@@ -0,0 +1,40 @@
Subject: Request for private pre-submission technical review: short-lifetime H0 persistence for a three-dimensional Gaussian field

Dear Professor [Name],

My name is Dylan Roy, and I am an independent researcher seeking a candid pre-submission technical assessment of a mathematical result concerning persistent homology of a smooth Gaussian random field.

The fixed-scope claim is that, for the normalized periodized three-dimensional Bargmann--Fock field on the flat torus of side 24, the first-moment density of finite nonessential superlevel-set H0 persistence lifetimes satisfies

nu_{3,24}(ell) = c_{3,24} ell^(-1/3) (1 + o(1)),
as ell decreases to 0.

The proof combines two-point Kac--Rice analysis, a local cubic fold relation, determinant-weighted Palm conditioning, and an argument that elder-rule selection changes only the lower-order term. It also derives an explicit coefficient and an analytic estimate of the side-24 periodization correction.

I am not asking for endorsement, publication approval, or institutional association. I am asking whether the argument appears mathematically sound and worth developing toward formal peer review. The most valuable response would identify the earliest unsupported inference, incorrect normalization, missing hypothesis, or invalid scope transition, if one exists.

AI systems were used extensively during derivation, checking, criticism, and editing. No AI-generated verdict should be treated as independent validation, and all mathematical responsibility remains with me. The purpose of contacting you is to obtain the first genuinely independent expert assessment.

The attached packet is self-contained; it does not require access to my Google Drive. A short initial assessment can be made from `00_READ_ME_FIRST.pdf`, `02_PRE_REVIEW_DOSSIER.pdf`, and the main theorem in `03_SIDE24_MANUSCRIPT.pdf`. The exact technical appendix and small reproducibility files are supplied for a deeper audit.

I would especially value your assessment of:

1. the two-point Kac--Rice normalization;
2. the cubic lifetime pushforward and the exponent -1/3;
3. the elder-selection estimate 1 - p_r = O(r^3);
4. the deterministic capture/escape argument;
5. the ordered-pair and directed-sphere multiplicities;
6. the claimed finite-side correction; and
7. whether the result appears novel relative to literature you know.

Please feel free to review only the portions for which you have appropriate expertise. The response form is optional; marginal notes or an informal reply would be equally welcome. If the project spans areas outside your expertise, it would also help to know which components require a specialist in probability, stochastic geometry, Morse theory, or persistent homology.

Thank you for considering it.

Sincerely,

Dylan Roy
Independent researcher
[Email]
[Telephone, optional]

Loading
Loading