source-passages.md

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

3464 bytes, as of the version from 2026-11-19 21:04 (current). Recorded here, not run.

# Inspected primary source locations

- Bertin–Zudilin, *On the Mahler measure of hyperelliptic families*,
  DOI 10.1007/s40316-016-0068-4, introduction equation (1), and paragraph
  following Theorem 1: the displayed model is
  y^2=x^3+(k^2-24)x^2-16(k^2-9)x. That paragraph also discusses integral
  k^2, hence the arithmetic parameter n=k^2 used here. The main theorem
  compares measures; it does not establish all rational L-value ratios.
  Introduction after Theorem 2 gives the conductor-224 k=4 example. The
  final specializations identify its minimal curve as
  y^2=x^3+x^2-8x-8, Cremona 224a2 (LMFDB 224.a1).
  https://link.springer.com/article/10.1007/s40316-016-0068-4
- Sutherland, MIT 18.783 Lecture 24 (December 7, 2023), slides 14–20:
  arithmetic good and bad Euler factors, minimal versus displayed models,
  conductor (including wild contributions at 2 and 3), weight-two modularity,
  and completion N^(s/2)(2*pi)^(-s)Gamma(s)L(s) with s <-> 2-s.
  PDF was inspected through the web reader; direct artifact download returned
  an HTTP error. This is an archive limitation, not missing source inspection.
  https://math.mit.edu/classes/18.783/2023/LectureSlides24.pdf
- PARI L-functions manual, introductory caveat and `lfun(L,s,D)` section:
  D is derivative order. Generic accuracy assumptions do not establish a
  rigorous bound for arbitrary L-data. Our certification instead comes from
  the explicit elliptic coefficient bound and split Mellin enclosure.
  https://pari.math.u-bordeaux.fr/dochtml/html/_L_minusfunctions.html
- Sage elliptic L-series reference, `dokchitser` examples: the default backend
  is PARI, so it would not provide an independent second implementation.
  The 37a L(2) example is 0.381575408260711..., used only as a low-precision
  known-answer control. Inspected through web reader; direct download failed.
  https://doc.sagemath.org/html/en/reference/arithmetic_curves/sage/schemes/elliptic_curves/lseries_ell.html

# Deductions checked in this build

For a2=n-24 and a4=-16(n-9), the model discriminant is
16*a4^2*(a2^2-4*a4)=2^12*n*(n+16)*(n-9)^2, verified in QQ[n].
It is not generally the minimal discriminant. Exact substitutions from the
original model to each global minimal model are checked in QQ[x,y].

The Euler product is positive and convergent at 2. Taking the limit of the
completed functional equation at zero gives D=L'(0)=w*N*L(2)/(4*pi^2).
The zero at zero is simple, with derivative sign w, independently of BSD.
No central rank or numerical inference of exact central zeros is used.

With f(z)=sum a_m exp(i*2*pi*m*z), the Fricke relation at positive height is
f(i/(sqrt(N)*y))=w*y^2*f(i*y/sqrt(N)). Splitting the Mellin integral at y=1
therefore gives N*L(2)/(4*pi^2)=integral_1^infinity f(i*y/sqrt(N))*(y+w/y)dy.
Termwise integration yields the E1 sum in generate.py. Its truncation bound
follows by |a_m|<=d(m)*sqrt(m)<=2*m and E1(x)<=exp(-x)/x:
each absolute summand is at most exp(-c*m)*(4/c+2/(c^2*m)); replace 1/m by
1/(M+1) for m>M and sum the geometric series. Absolute convergence justifies
this bound and termwise integration. Both the sum and the tail are enclosed
with Arb, including pi and sqrt(N). The finite coefficient recurrence is exact.

Numerical agreement is evidence about implementation, not proof of a universal
new relation. No Mahler/L-value ratio is claimed. The required fresh scientific
critic has not yet evaluated these deductions or the populated draft.