← Session 33 — ii′-3b/3c crux + F-D finding; Lagarias letter package
# Package ready — committed (2f33290)
**`research/correspondence/lagarias-F-D-package/`** — everything the letter's question 5 references:
| File | Purpose |
|---|---|
| **finding-note.pdf** | The detailed finding note (4 pages, compiled clean, verified rendering) — the claim, the e/2 > 1 issue, evidence tables, the mechanism (peak σ ≈ 0.36√n, height ~e^{0.066√n}), the repair verification, assessment, questions, reproduction |
| `FINDING-NOTE.md` | Same content in Markdown |
| `finding-note.tex` | LaTeX source (pdflatex, 2 passes) |
| **E-031-right-tail-stepping.py** | The mpmath script: **V1** product profiles + peak tracking across n ∈ {10³, 10⁴, 10⁵}; **V2** right-tail envelope integrals (~5.9×10¹⁰ at n = 10⁵); **V3** product log-profiles at T = c√n showing the repair works at c = 1.5, 2. Runs with `--force`, deterministic (dps 25), Python 3 + mpmath only |
| `evidence.csv` | All evidence tables in tidy long form (peak products, envelope integrals, the V3 profiles at three heights, direct factor evaluations) |
| `README.md` | Package contents, reproduction instructions, one-paragraph summary |
## Notes
- The package is **self-contained**: no references to internal project docs — Lagarias can read the note and run the script with zero context.
- The evidence in the note matches the script output verbatim (peak 4.40 / 494 / 1.56×10⁹ at σ = 11 / 35 / 111, envelope 1.05×10⁴ / 5.87×10¹⁰, direct factors 4.39×10³⁹ and 1.8×10⁻³¹).
- LaTeX build artifacts (.aux/.log/.out) are in the folder but untracked; the PDF is committed.
The letter (`research/paper/letter-to-lagarias.md`, commit 2dfb919) references this package in its question 5. When you've reviewed both and filled in affiliation/contact details, they're ready to send.
raw content_json (database row)
[{"type":"text","text":"# Package ready — committed (2f33290)\n\n**`research/correspondence/lagarias-F-D-package/`** — everything the letter's question 5 references:\n\n| File | Purpose |\n|---|---|\n| **finding-note.pdf** | The detailed finding note (4 pages, compiled clean, verified rendering) — the claim, the e/2 > 1 issue, evidence tables, the mechanism (peak σ ≈ 0.36√n, height ~e^{0.066√n}), the repair verification, assessment, questions, reproduction |\n| `FINDING-NOTE.md` | Same content in Markdown |\n| `finding-note.tex` | LaTeX source (pdflatex, 2 passes) |\n| **E-031-right-tail-stepping.py** | The mpmath script: **V1** product profiles + peak tracking across n ∈ {10³, 10⁴, 10⁵}; **V2** right-tail envelope integrals (~5.9×10¹⁰ at n = 10⁵); **V3** product log-profiles at T = c√n showing the repair works at c = 1.5, 2. Runs with `--force`, deterministic (dps 25), Python 3 + mpmath only |\n| `evidence.csv` | All evidence tables in tidy long form (peak products, envelope integrals, the V3 profiles at three heights, direct factor evaluations) |\n| `README.md` | Package contents, reproduction instructions, one-paragraph summary |\n\n## Notes\n- The package is **self-contained**: no references to internal project docs — Lagarias can read the note and run the script with zero context.\n- The evidence in the note matches the script output verbatim (peak 4.40 / 494 / 1.56×10⁹ at σ = 11 / 35 / 111, envelope 1.05×10⁴ / 5.87×10¹⁰, direct factors 4.39×10³⁹ and 1.8×10⁻³¹).\n- LaTeX build artifacts (.aux/.log/.out) are in the folder but untracked; the PDF is committed.\n\nThe letter (`research/paper/letter-to-lagarias.md`, commit 2dfb919) references this package in its question 5. When you've reviewed both and filled in affiliation/contact details, they're ready to send."}]