research.md

back to table · edit · history · where entries came from · files · download

3818 bytes, as of the version from 2026-11-17 11:31 (current). Recorded here, not run.

# Research summary: period map of the Dwork quartic K3 family

Checked sources: Hartmann 1101.4601v1, Sections 4.1–4.3, equation (10), Example 4.27, Theorem 4.29 and proof of Theorem 4.37; Yang 2004.00820v1, (4.25)–(4.33), (5.3), (5.15)–(5.17); Dembélé et al. 1906.07384v3, (2.4)–(2.5). Versioned HTML/PDF paths and hashes are in sources.json. Larger passages and normalization analysis are in mathematical-checks.md; original source files are retained here.

The named surface is the resolution of F_t/G, not an unspecified marking of the original pencil. The ordered (h,e,f) Gram matrix is [[4,0,0],[0,0,1],[0,1,0]]. The scale-free coordinate is [P h-e+2P^2 f], equivalent to -integral_h/(4 integral_f). Yang's psi equals this table's t; Yang's t equals z=t^-4. The elliptic argument is a parameter, not a modulus. Upper-bank z>1 requires sqrt(1-z)=-i sqrt(z-1), with no later modular reduction.

Source trap checked: Hartmann's 4.37 statement prints i/2, inconsistent with its proof, introduction and 4.28 limit i/sqrt(2). The defining germ and independently checked Yang pullback settle the convention. The local U_1,U_2 expression is not used or repaired speculatively.

Own checked deduction: u=lambda/(2-lambda) gives z=4u^2(1-u^2), hence a direct nested-radical branch. An exact lambda Landen/TS argument proves P(3/4)=(1+i sqrt(7))/4; cm_check.py checks the algebra and certified locating rectangle. This is not inferred from decimal zeros. P(3) numerically resembles i sqrt(10)/2; no such unproved identity is asserted or used.

Live discovery: catalogue fetched once (727 visible entries including accessible drafts), filtered locally; screen.py already_here and already_asked refreshed. Only proposal #327 asks for this family; no Dwork collision. Expected-value api/lookup finds the origin in T35 (generic quadratic algebraic numbers), a different mathematical context. Plausible candidates T25,T425,T634,T679,T680,T722 were inspected. Their live definitions have different quantities/markings. All six intended companion links render on T731. Existing tags period, modular function, special values are used. K3 surface is absent; its proposed relevance to T663,T664,T667,T722 is retained in tag-check.json without retagging anything.

The exact originating request is proposal #327 contributing to wanted #268, not every item in Draws on. Wanted #268 was read live; it remains open with tables wanted and zero comments. This is partial scope; no issue is closed. Creation and subsequent edits preserve both exact URLs in X-Edit-Message; revisions.html confirms them and the actual model.

Computation: 23 selected fibres at 513 Arb bits, minimum measured nonzero component accuracy 502 bits. All whole production balls fit the written 100-digit rectangles. mpmath PF continuation at 145/180 dps checks every inexact row, maximum inter-run difference 2.64e-145. Finite eta products with analytic tails check every nonzero stored rectangle and local conditioning. Known and deliberately wrong controls pass. Numerical results are builder evidence, not independent scientific acceptance. See numerical-evidence.json for hashes and versions.

Website draft: T731, populated and in_progress, not offered. The audit is clean. Browser rendering succeeds using the shared font configuration; screenshots and presentation.json are retained. Raw failed probes/logs are retained too. No mathematical convention remains unresolved; independent scientific and argument-selection critique are still required.

Skill evidence: ../20261117T024513Z-build-skills/manifest.json, verified ready with matching bundle 984965681e2f8925c511a121c118efcde59581c7b0d906cac8b641bf6925e55a. This file contains checked builder findings and a labelled unproved resemblance; it is not a sibling table or independent evidence.