back to table · edit · history · where entries came from · files · download
2872 bytes, as of the version from 2026-11-18 08:38 (current). Recorded here, not run.
# Source locations and resolved discrepancies
Guttmann, arXiv:1004.1435v1, section 1.2, the BCC displayed equations before
section 1.3: P(0;z) is the normalized triple integral of
1/(1-z cos k_1 cos k_2 cos k_3). The displayed elliptic argument satisfies
k_2^2=(1-sqrt(1-z^2))/2. Equation (16) gives the exact unnormalized return
counts binomial(2n,n)^3, with odd counts zero. We translate the source modulus
to the NumberDB/Arb parameter alpha=k_2^2 and take z=8/k.
Guttmann–Rogers, arXiv:1207.2815v2, section 3, gives the same BCC kernel and
elliptic formula; equation (55) gives the squared Gauss function with upper
parameters 1/4,1/4 and the 3F2 with upper parameters 1/2,1/2,1/2, at z^2.
The logarithmic-determinant differentiation relation is established in its
section 2 and also follows directly by dominated differentiation for k>8.
Equation (31) has the known intermediate-series z^l versus z^(2l) misprint:
odd moments vanish and the final hypergeometric formula uses even powers.
This table does not evaluate that misprinted series.
Additional discrepancy discovered in Guttmann's v1 section 1.1: the displayed
BCC gamma prefactor is 1/(2^(14/3) pi^4), although the printed decimal is
1.3932039. That prefactor does not give that decimal or the integral. This
is resolved by deriving the endpoint from the checked elliptic formula and
DLMF's lemniscatic evaluation, rather than adopting the inconsistent prefactor.
DLMF 19.25.1a says K(modulus)=R_F(0,1-modulus^2,1).
DLMF 19.20.1 gives R_F(lambda*x,lambda*y,lambda*z)=lambda^(-1/2)R_F(x,y,z).
DLMF 19.20.2 gives R_F(0,1,2)=Gamma(1/4)^2/(4 sqrt(2*pi)).
Thus in the table's parameter convention
K(1/2)=R_F(0,1/2,1)=sqrt(2)R_F(0,1,2)=Gamma(1/4)^2/(4 sqrt(pi)),
and W(8)=Gamma(1/4)^4/(4*pi^3).
The original-kernel integration agrees at the endpoint. The independently
computed gamma enclosure also agrees with T393's published BCC row. No
alternative Green normalization remains unresolved.
Chebotarev–Shamis, arXiv:math/0602575v2, section 3, Theorems 5–7, states the
determinant/rooted-forest and inverse/relative-accessibility correspondences
for finite weighted graphs. The infinite BCC interpretation at k=9 is a
Fourier/Riemann-sum deduction: the symbol of I+Delta is 9-H, strictly positive.
It is not a claimed numerical value printed in that paper.
Source passages are paraphrased; full G.html, GR.html, CS.html and the downloaded
DLMF TeX are retained in this run's directory with SHA-256 in source-hashes.json.
The two DLMF pages were inspected with the web reader; fetching their full HTML
locally received HTTP 403. Their equation TeX files were fetched successfully.
The supplementary G v1 gamma-prefactor issue is source criticism, not permission
to alter any published NumberDB table. It does not change the shared family
convention, which uses the integral and correct endpoint formula.