back to table · edit · history · where entries came from · files · download
3254 bytes, as of the version from 2026-11-10 03:02. Recorded here, not run.
# Argument selection for T704
This is one scalar invariant of one polynomial family: the raw discriminant
of the ordinary Riemann xi Jensen polynomial. No monic, logarithmic or
rescaled discriminants and no Hermite-combination quantities are included.
The checked source O'Sullivan, arXiv:2007.13582v2, equations (1.2), (1.3),
fixes gamma_m=m! xi^(2m)(1/2)/(2m)! and J^(d,n)(X). The completion has the
factor 1/2, as in Griffin et al., arXiv:1910.01227v3, equations (1.1), (1.2).
The expansion Theta(z)=xi(1/2+sqrt(z)) is defined by its even central series;
it is entire and does not select a square-root branch. The shift n indexes
a derivative of Theta, not the nth derivative of xi with respect to s.
Every parameter has a discrete_index role:
* d is the degree, with natural domain all integers d>=2. Degrees 2 and 3
give the quadratic and cubic Turan expressions discussed by Chen, Jia and
Wang, arXiv:1706.10245v1, section 1. Degrees 4 through 8 include higher
invariants of distinct root separations, beyond the cases whose sign
characterizes hyperbolicity. The complete consecutive degree prefix
includes both parities. Degree 1 is omitted because its discriminant is
the empty product 1 for every shift and carries no coefficient information.
* n is the coefficient shift, with natural domain all integers n>=0. The
consecutive prefix 0 through 20 covers the unshifted central polynomial
and its first twenty derivative shifts. It follows one intrinsically
indexed family without selecting isolated favorable signs or decimal
arguments. It includes early shift variation at every selected degree.
The default enumeration is d=2,...,8, then n=0,...,20 within each degree,
147 values. Both degree breadth and shift breadth were assessed: a table
restricted to quadratic and cubic formulas would omit the higher scalar
invariants; a long single-degree sequence would miss this distinction.
The upper bounds d=8 and n=20 are noncanonical computational cutoffs,
independent of numeral notation. They are not assertions of significance
for each endpoint, asymptotic coverage, or a proof of RH. A modest rectangle
of genuine polynomial indices is appropriate here; there is no continuous
argument grid, free deformation, or mixed exploratory group.
The published T659 happens to cover all these polynomial roots, which
provides a useful full-range comparison. That overlap is a check and is not
the selection's sole justification. T15 supplies related central derivatives;
T381 supplies universal discriminants through degree five. These are
companions rather than duplicate quantities. The unbuilt Hermite-combination
proposal remains separate. The selected values are scalar intervals, and
the measured written length and block size are recorded in numerical evidence.
The completeness note and generator use exactly this finite block. No
independent acceptance is asserted by this build document. The separate
critique must assess both parameters and fingerprint generate.py together
with the other sources on which its selection depends.
Policy evidence: agents/runs/20261110T024815Z-build-skills/manifest.json.
Both archived skill versions have SHA-256
886390324dd1971e381a43cf3b46b822ab0bc67d62f31d8481e29ec79556746b.