back to table · edit · history · where entries came from · files · download
3596 bytes, as of the version from 2026-11-17 13:46 (current). Recorded here, not run.
# T739 build research
Checked sources: DPVZ June 2020 author PDF, Table 2.10 p.5 (fifteen rational
rows, eight templates); (3.2)--(3.3) p.6 (Gauss/finite hypergeometric sums),
(3.8) p.7 (exp of negative traces), (3.16), Example 5.5, and (5.7) (full
polynomial controls). Files and exact SHA256: source-hashes.json. The archived
canonical skill manifest is ../20261117T132607Z-build-skills/manifest.json;
local/live bundles match, digest 889383d95f882353fc09d210af003abe6edcf0331f27c377d7acdabafa344208.
Normalization: arithmetic beta=0^5 (same residues as 1^5), rank 5, weight 4,
Hodge numbers (1,1,1,1,1), no extra Tate twist. Classical coordinate is z;
PARI uses 1/z (its Motives documentation). P(T)=det(1-T Frob) has constant 1,
not its monic reciprocal and not 1/P. Frobenius is the cohomological convention
with |eigenvalues|=p^2. Sage hgmparams confirms twist zero on these templates.
Independent checker source: BCM Theorem 1.3, including gcd-polynomial
multiplicities s(m), epsilon=(-1)^sum(q_j) and M. Native Arb polynomial
multiplication implements the Gauss-sum DFT through Bluestein convolution.
Watkins, known.pdf, 11.1.1 pp.21--22 gives the determinant character:
tilde_t=1/z and all five lower cyclotomic factors Phi_1 have discriminant 1,
so u=-tilde_t*(1-tilde_t)=(1-z)/z^2. The coefficient sign is -chi_u(p),
i.e. -Legendre(1-z,p), independently evaluated using Euler's criterion.
Production: Sage 10.9 Gross--Koblitz, explicit p-adic precisions N satisfying
p^N>10*p^(2*r), r=1,2, reconstruct uniquely inside the weight-four Weil bound.
Newton plus local duality gives all coefficients. Exact representations need
no artificial decimal digits. prepared-values.json preserves all 120 results.
Live discovery: catalogue-complete.json (paginated lightweight catalogue),
catalogue-candidates.json, discovery-summary.json. T674/T675 are rank four,
weight three; T142/T143 supply trace ingredients; T729 concerns primitive
Fermat cohomology; T738 is the analytic realization of these (template,z).
No same arithmetic family found. Polynomial lookups on three published controls
returned no matches. Tags number theory, polynomial, L-function exist;
hypergeometric motive does not. No new tag or published-table edit is made.
The December 2023 errata requires central characters in Hilbert-side
quadratics. No Asai factorization is used here. The broader family recognition
experiment is deferred: these numerical local factors do not resolve missing
Hilbert-form identifications, bad factors or global L-values. In particular,
no Hecke values recovered from these same polynomials count as evidence.
Sources from sibling research are leads, not independent checks. The prior
screening blob was reported unreachable (404) in the shared dossier; use
fresh authoritative discovery. No convention remains guessed. Independent
scientific/argument-selection verdicts belong to a later controller session.
New checked finding: DPVZ (3.16) p11 quadratic prints a minus sign for 30T.
The precise finite-sum definition instead gives +30T and trace91; an exact
Q(zeta_110) calculation independently resolves it (exact_control.py,
exact-control-results.json), supported by both coordinate checks in
investigation.log and all stored-value checks. This is disclosed in Comments,
not silently forced into the printed control. Three other published full-factor
controls agree. Initial checker API errors and the exact-control multiplicity
repair are retained in the attempt logs; only actually executed final results
are cited as checks. No independent verdict was written by this builder.