back to table · edit · history · where entries came from · files · download
4203 bytes, as of the version from 2026-11-18 04:29 (current). Recorded here, not run.
# Boyd Family 3.2, checked research
The live source and ordinary torus/Jensen normalization agree. One quantity:
p(k)=m((x²+x+1)y²+kx(x+1)y+x(x²+x+1)), real k≥0. Natural log, torus
mass one, convergent torus-zero singularities. Positive square roots and the
permanent parameter k:R agree with family #336. No L-value conjecture is an
input. Source-derived evenness is exact by y→−y.
Inspected sources: Bertin–Zudilin, DOI 10.1007/s40316-016-0068-4, journal HTML
Introduction/Theorems 1–2, The first family, and final specializations;
arXiv:1601.07583 PDF pp. 1–4 and 12–13. The introduction names this Family 3.2,
prints the polynomial and elliptic quotient E_n, and explicitly discusses
k² integral. The first-family section gives the Jensen numerator integral,
the algebraic contact roots and elliptic derivative integral. Equation (11)
gives p(3)=L′(χ_{−15},−1)/6. The conversion to 5sqrt(15)L(2,χ)/(8π) follows
from the odd-character functional equation and is separately checked with
Hurwitz zeta sums. The PDF's isolated |4c²−1| denominator transcription differs
from |4c−1| derived from the polynomial and used in the journal HTML; the
implemented final numerator formula is consistent in both sources. No
intermediate printed denominator is used as an oracle.
Primary method: real log of the Jensen numerator with both discriminant
branches retained, algebraic breakpoints and a sin² endpoint map. Exact
input intervals are propagated numerically by monotonicity, at 140 and 180
decimal digits. Output keeps 100 significant digits and exact p(0)=0.
Rigour is heuristic (agreement-checked): quadrature lacks an error theorem.
The alternative check integrates the complete-elliptic derivative from zero,
using Carlson RF and a split at 3; it shares the source and mpmath library,
but does not use the primary numerator integration or any sibling values.
See independent-results.json for actual outcomes, not a verdict by a critic.
Live discovery: one cached authenticated catalogue (738 entries), text searches,
screen.already_here/already_asked and precise numeric lookups at k=1,3,4 found
no same family. Candidate metadata was refreshed immediately before claiming
T742. T664 is a different three-variable Bertin Q; T658 is the reciprocal
cubic; T282 is the diamond family. Other candidates store Hessian, higher or
multiple measures, not this P_k. The signed-in catalogue had no matching draft.
T145 entry (15,14,2) explicitly identifies χ=(−15/·), odd; it is a companion
control, not an independent input. T25 uses elliptic parameter m, not modulus.
Current relation records point only to T145 and T25. Future Q/R builders should
add their real identifiers and exact theorem-range relations; no fictitious
T-numbers were inserted. The sibling tables are not numerical evidence.
Registry: exact API lookups confirmed number theory, integral, special values,
elliptic curves and period. Mahler measure (also lowercase lookup) and
hyperelliptic curves were absent. Both are justified distinct concepts and are
used as new tags; no alias or unrelated-table retagging was attempted. The
period tag is supported by the explicit elliptic derivative and regulator
context. No statistical-mechanics application was checked or claimed.
Provenance: only proposal #336 contributes to wanted #263 for this table.
Other issue numbers in the family's Draws on list are background. Creation
and every amendment retain those two exact URLs in X-Edit-Message. Wanted
#263 was read live; family #336 has no comments. The preceding screening
report path ideas/BATCH-2026-11-18T0151.md.gz in numberdb/numberdb-runs was
unreachable through the connected GitHub API (404, including main ref).
The live proposal, dossier and refreshed website/source research were used;
no transcript cache was consulted.
Skill: ../20261118T021708Z-build-skills/manifest.json records matching local
and live complete bundles. Every resource hash was revalidated; only one copy
of the core and the six applicable references was read. Bundle SHA-256:
d6e85d49873fac5bc748d068b0288cf242b0f9aebf3a9a0fff70295019227420.
See evidence-hashes.json for larger artifacts, sources, code and results.