Strong permanent contraction on four-by-seven bipartite supports
Scott Schubr, Luminary Oddities LLC

Strong_Permanent_K47.pdf is the paper; manuscript/main.tex is its complete
editable source. Compile with Tectonic or a LaTeX distribution providing
amsmath, amssymb, amsthm, geometry, booktabs and hyperref. For example:
  tectonic manuscript/main.tex
or run pdflatex twice to resolve references:
  pdflatex -output-directory manuscript manuscript/main.tex

The exact certificate uses Python 3.10+ and its standard library only:
  python -B verify_k47_certificate.py
  python -O -B verify_k47_certificate.py
The checker regenerates all 720 small-row corner/pairing evaluations and
all 38,416 rational Bernstein coefficients across the 256 balanced boxes.
It uses direct budget evaluation and exact interpolation, with no private
inputs, network access, floating tolerances, optimizer or output-file writes.
The manuscript supplies the analytic reductions and positive-basis argument;
the checker supplies their explicitly identified finite arithmetic step.

The proved domain is every complex 4-by-7 operator contraction, equivalently
the strong self inequality on Hermitian PSD K4,7 supports. Arbitrary phases,
missing edges, zero diagonals and singular endpoints are included. The
all-column heavy-entry lemma is a precise published dependency cited in the
paper, from Strong permanent contraction on four-row bipartite supports
through six columns. That predecessor is not reproduced in the ZIP.

This is one specialist continuation in the existing structural result family.
It does not prove K4,8, K5,5, all bipartite supports of order ten or eleven,
arbitrary second factors, all K4,q, or unrestricted Chollet. No absolute
priority, optimal threshold or full equality classification is claimed.

MANIFEST.json lists the explicit public payload hashes; SHA256SUMS covers
the payload and manifest, omitting itself. Assessments, correspondence,
internal reviews, private coefficient arrays and privacy/release records
are excluded from this package.
