research.md

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

3700 bytes, as of the version from 2026-11-16 22:11 (current). Recorded here, not run.

# T728 research and build notes

Checked sources: Stopple, arXiv:1508.05870v2, p.2 §2 defines xi(s)=s(s-1)pi^(-s/2)Gamma(s/2)zeta(s)/2; pp.5–6 §4 defines Pf=f''/f' and gives priority to P over a following prime; p.6 equation (9) defines the strong-pair left side and Theorem 1 gives its Lehmer implication. RH is assumed throughout that paper. Simonic, arXiv:1612.08627v2, p.5 §4 independently restates the same xi criterion; p.7 Tables 1–2 select n=34,6709; p.6 equation (19) gives Cauchy derivative extraction.

Normalization checked: derivatives are in height t for Xi(t)=xi(1/2+i*t). The sole stored parameter is lower index n; both selected roots must be simple. S_n=Delta_n^2[-Q_Xi(gamma_n)-Q_Xi(gamma_(n+1))], Q_f=(f''/f')'. Source implication uses RH; finite evaluation does not. Simonic p.4 defines H with s(1-s), opposite Stopple's s(s-1); this overall constant sign cancels in Q. We retain Stopple/T660's xi convention. Stopple's auxiliary G numerator and later repeated h shorthand are inconsistent with his earlier definitions; neither is used. Equation (9) and Simonic §4 agree unambiguously.

Derived and checked: at a zero Xi=a Z, a=-0.5(t^2+1/4)pi^(-1/4)|Gamma(1/4+i*t/2)|. With ell=a'/a, -Q_Xi=-Q_Z+ell*(Z''/Z')-3ell'+ell^2. Generator uses this ball identity. The separate mpmath formula differentiates xi=h*zeta: -Q_Xi=w-v^2-u*v+3u'-u^2, where u=h'/h, v=zeta''/zeta', w=zeta'''/zeta'. Its imaginary discrepancy is measured before taking the real part. Exact polynomial controls test the derivative-ratio and envelope algebra. Numerical agreement is not proof of a universal identity; these identities also follow directly from the product rule.

Live discovery: 724 accessible catalogue records, fetched once for this pass; already_here/already_asked rerun for xi/Lehmer/pre-Schwarzian synonyms; plausible definitions T3,T584,T585,T586,T660,T726,T727 inspected and refreshed before creation. No duplicate scalar xi family found. Corrected exact-value lookups for n=1,34,6709 return no public matches. T726 is now published; T727 is offered. Neither was edited. Existing tags L-function, number theory, special values, inequality are used. Riemann zeta function remains absent; possible cross-family tag suggestion is deferred. #324 is the exact proposal, contributing partially to #276. #230/#263 are background, not direct origins.

Prior research screening report location supplied by the family is ideas/BATCH-2026-11-16T1947.md.gz in numberdb/numberdb-runs; the preceding dossier records its 404, and this build independently received HTTP 404 from the GitHub contents endpoint (screening-report-access.json). Current discovery supplies fresh corpus evidence instead. No conclusions from sibling scalars are treated as independent checks. Reused FLINT bridge and certified finite-prefix input have origin hashes in finite-prefix-origin.json; the current run rechecks count, ordering, all selected endpoint overlaps and adjacent counts. All original family evidence remains preserved.

Canonical skill manifest: ../20261116T213041Z-build-skills/manifest.json, complete matching local/live bundle SHA-256 984965681e2f8925c511a121c118efcde59581c7b0d906cac8b641bf6925e55a. Archived files were verified against all manifest resource hashes before work. Only one matching core and relevant references were read.

Larger numerical, source and render artifacts are indexed by evidence-index.json. Current computation outcomes and unresolved presentation/access limitations are recorded in handoff.md after readback. Independent scientific/argument-selection verdicts and family consistency approval belong to the controller's separate sessions; none is issued by this builder.