research.md

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

3957 bytes, as of the version from 2026-11-15 20:42 (current). Recorded here, not run.

# T719 concise research

Checked facts:
- Goldmakher, *Legendre, Jacobi, and Kronecker symbols*, Theorem 1.1 and
  following remarks, PDF page 2: primitive quadratic characters correspond
  to nontrivial fundamental discriminants and have conductor |D|. The squarefree
  congruence enumeration gives exactly the selected 26 with |D|<=40.
- Zagier, *Modular Forms of One Variable*, Appendix A3, printed page 60,
  PDF page 62: explicitly gives (q/pi)^(s/2) Gamma((s+a)/2)L(s,chi),
  quadratic root number +1, and Lambda(s)=Lambda(1-s). The dossier's earlier
  locator/title (Appendix A4) is erroneous; the mathematical normalization
  is correct. The actual source was inspected in this run.
- Kedlaya 6.1.1 and section 6.2: even/odd theta Mellin integrals. For odd
  characters his completed function includes an extra (q/pi)^(1/2);
  multiplying by C=(pi/q)^(a/2) gives the specified normalization.
- DLMF 25.15.3: q^(-s) sum chi_D(r) zeta(s,r/q), with the last term zero.
- Griffin-Ono-Rolen-Zagier, arXiv:1902.07321v2 equation (1): exponential
  even-series indexing z^(2m)/m! for a pole-removed zeta object. This supports
  the indexing analogy only; no hyperbolicity theorem is transferred.
- FLINT acb_poly/acb_calc documentation: rigorous power-series coefficient
  balls and validated finite quadrature. Runtime versions are in versions.json.

Normalization decisions: gamma_D(m)=m! Lambda_D^(2m)(1/2)/(2m)!;
parameter order D,m; real z, no imaginary-axis (-1)^m; no pole-removing
factor; no central-value division; no unknown branch. Natural domains are
all nontrivial fundamental D and integers m>=0. Finite bounds and rationale
are in argument-selection.md, the generator and completeness note.

Live discovery: catalogue cached once, screen.already_here/asked executed,
synonyms searched and plausible candidate metadata/documents inspected before
claiming. Six api/lookup intervals for D=-3,-4,5 at m=0,1 had no matches.
L-function, Taylor series and number theory tags exist. Main already_asked
returned []; secondary family query failed with HTTPError, recorded rather
than interpreted as absence. Issues #319 and #274 were directly refreshed;
proposal #319 contributes to #274, with no claim to close it.

Related tables inspected: T15 (xi central derivatives), T142 (Gauss sums and
completion), T145 (uncompleted integer L-values), T718 (quadratic Li data),
T659/T704 (xi Jensen data), T380 (moment arithmetic factors). T4 and T128
metadata supplied additional context. These are distinct quantities, and
none supplies the independent numerical oracle. The table links T15,
T142, T145 and the assigned family's T718; the rendered Similar tables
section was checked after converting it to supported table/relation records.
No companion is edited.

Computation: every one of 546 Hurwitz/Gamma-jet values was checked against
new theta moments. 640-bit balls; production minimum 427 bits; checker
minimum 448 bits. Known and wrong controls passed. Stored intervals and
verify(sample=None) both checked all 546 with no failures at 100 digits.
No universal sign, positivity, hyperbolicity or GRH conclusion is asserted.
No unresolved convention alternative remains; independent actor validation
is still pending.

Artifacts in this directory: generate.py, theta_check.py, stored_check.py,
theta-checkpoints.json, check-results.json, control-results.json,
stored-check-results.json, verify-results.json and numerical-evidence.md.
Original source bytes and SHA-256s: sources/manifest.json. Exact instruction
hashes: skill-manifest.json and ../20261115T202357Z-build-skills/manifest.json.
The final evidence-manifest.json records final file SHA-256s.

Reuse: discriminant/binary-ball conversion and elementary Kronecker components
were read from the family dossier's documented T718 generator/checker,
not from transcripts. Their original hashes are retained in the dossier;
this run's generator/checker and all central coefficient results are new.