cat _posts/2026-10-29-diaz-modulus-lean-notes-v1-36-v1-38-machine-check-roy-s-six-exponentials-from-a-multiplicity-estimate-then-discharge-its-one-open-hypothesis.md
diaz-modulus-lean: notes v1.36–v1.38 machine-check Roy's six exponentials from a multiplicity estimate, then discharge its one open hypothesis
Three rounds on 29 October took diaz-modulus-lean from 503 to 564 proved results, each released as its own version of the companion note: v1.36 at 08:37, v1.37 at 18:38 and v1.38 at 19:46, with the appendix running from 221 to 224 checked identifiers (7de8588, 7be2496, 9fcf00d).
C1.4 — Roy’s theorem, from a multiplicity estimate. Twenty-six ported nodes prove the strong six exponentials theorem out of the elementary multiplicity estimate hME, whose architecture is the point: a coordinate complement of a rational span, a matrix whose vanishing maximal minors force a kernel vector, a left inverse for an integral-to-rational transfer, and the interpolation determinant that does the tangent counting. 529 of 529 results.
C1.1 — the Hilbert side, under unmixedness. Thirty nodes give the finite-level Hilbert-function layer of hME on the boxes of K[X, Y, Z] modulo the torus ideal of 𝔾ₐ^{A₀} × 𝔾ₘ^{A₁} — the box filtration, its chain bound by a telescoping argument, and the torus reduction — with Macaulay’s unmixedness carried as the explicit hypothesis hU, because nothing in the library proved it. 559 of 559.
C1.2 — and then it is proved. Five nodes settle the inequality half of Macaulay’s unmixedness theorem for polynomial rings over an algebraically closed field (DiazModulus.unmixedHeight_mvPolynomial, after Macaulay 1916 and Bruns–Herzog Th. 2.1.6) by a local route through Matsumura Th. 17.2, Rees’ characterisation of regular sequences via Ext, and the Xᵢ − aᵢ from the Nullstellensatz, so torus_hilbert_side_complex states the Hilbert side over ℂ with hU gone. 564 of 564.
What is left is stated rather than hidden: the multiplicity side of hME, and the assembly of hME from its two sides, so hME remains a hypothesis and Roy’s theorem is still machine-checked only under it. The prove2me-logs board carries the same day from the research side, one entry per round (43e542a, 1401299, 6123d3f).