argument-selection.md

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

3262 bytes, as of the version from 2026-11-18 04:29 (current). Recorded here, not run.

# T743 argument selection

Exact quantity: q(k)=m(Q_k), the ordinary logarithmic, normalized torus
Mahler measure of Boyd hyperelliptic Family 3.5B. The polynomial is written
in generate.py and the table Definition. The only parameter is k, type R,
domain all real numbers, order (k,). The paper's shifted comparison variable
is u=k-2; it is not the stored coefficient. Positive square roots are exact.
There is no evenness assertion for q.

Checked sources: Bertin–Zudilin, journal Introduction (Theorem 1 and polynomial),
The second family (quadratic torus presentation and root analysis), and final
specializations (Q_{-1} genus one); arXiv:1601.07583 pp. 1, 4–7, 14.
The exact Q_5 factorization is a deduction checked by symbolic expansion in
independent_check.py, not a copied numerical coincidence.

The proposed breadth remains appropriate: both real sheets of (k-2)^2=n,
integers 0<=n<=36, n=0 once, total 73. Order is increasing n, plus sheet before
minus sheet, with squares reduced to integer keys. This supplies ordinary
measures beyond the upper-sheet theorem rather than only repetitions of it.
The two sheets are coefficient embeddings only for nonsquare n; square cases
are rational specializations. n and sign enumerate k; they are not extra
parameters or a Cartesian grid of independent couplings.

The complete assessment is mixed, with disjoint groups:

* intrinsic (2): k=-1 (the source's genus-one member), k=5 (factorization).
* documented_application (21): k=2+sqrt(n), 16<=n<=36. Theorem 1 selects
  this comparison family via q(u+2)=p(u), u>=4. The selection contains both
  square and nonsquare n rather than only integer u.
* reference_sample (50): upper sheet n=0,...,15 except n=9 (15 identities),
  and lower sheet n=1,...,36 except n=9 (35 identities).

Reference policy:

* purpose: exploratory identification of ordinary measures of low-height
  real quadratic members, retaining the second embedding and the regime
  outside the proved comparison.
* bounds: exactly those 50 identities, with normalized integer square cases;
  no other rational, decimal, angular or radical grid.
* coverage_rationale: one complete offset-square-root slice tied to the
  source's comparison shift, including integral coefficients, both quadratic
  embeddings, exceptional fibers and different torus-root regimes. The finite
  height 36 is a noncanonical practical quadrature cutoff; all |k|<=8.
  Higher heights extend this table. A dense independent rational coupling
  grid would add cost without a checked application.
* noncanonical_acknowledged: true.
* notation_independent: true, using the algebraic equation (k-2)^2=n.
* sources_checked: true for the polynomial, normalization, shift, theorem and
  degeneration, not for an application of every exploratory input.

enumeration.json records exhaustive normalized memberships generated directly
from the default enumeration; Programs and complete-note describe this same
73-point selection. None of the parameters is a quantile or critical value.

Skill evidence: ../20261118T025446Z-build-skills/manifest.json, matching local/live
bundle d6e85d49873fac5bc748d068b0288cf242b0f9aebf3a9a0fff70295019227420.
This is builder justification, not the mandatory independent argument verdict.