cat _posts/2026-11-02-diaz-modulus-lean-a-companion-note-shows-roy-s-rank-bound-cannot-reach-u-at-rank-at-most-three.md
diaz-modulus-lean: a companion note shows Roy's rank bound cannot reach u⁴ at rank at most three
A new companion note in diaz-modulus-lean, tex/diaz-roy-rank-three.tex with its checks under scripts/roy_rank/, closes the question a scoping study of 23 October had left open and guessed would take days: whether matrices larger than the rank-one case — that is, beyond the strong six exponentials theorem stopping at u³ — can break Roy’s rank bound (1992, Cor. 1) far enough to exclude u⁴ (f5b4d6c). The answer is no at rank at most three. Nothing here is formal: no Lean file changed and the library stays at 618 results, with note v1.41 still the newest release.
The reduction is the load-bearing part. If a matrix over a monomial space breaks the bound, so does a graded stable one of no larger size and rank, and the limit is taken in the moduli space of semistable Kronecker modules rather than by degenerating the matrix itself, which can create zero blocks. Then the count: no matrix over the model of the hull of u⁴ breaks the bound at any size, graded or not, and the same holds for the hull of u⁴ + cu³ — although its initial space does carry one explicit 5 × 8 failure, which does not lift. No failure appears at structural rank at most three with free entries, and the full graded rank-three search takes about four minutes rather than the days the earlier study feared.
The checks are recorded with the note: each part was rederived by one referee agent and the finished note by another that saw only the note, no theorem was found wrong, two slips in an example were corrected, and one citation — Draisma 2006 — was wrong and was withdrawn; the graded rank-three case has a SAT run over the whole proved range under two solvers plus separate exhaustive enumerations. The note is explicit about the edges: rank four is open, so is the generic model with actual logarithms, results with free entries concern formal rank only, the good-quotient property from GIT was checked only through later papers, the rank-three and hull theorems are computer-assisted, and no novelty is claimed — the reduction assembles standard parts. It is a barrier on one method, not progress on Diaz’s conjecture, and the prove2me-logs board carries the same conclusion from the research side (e6e3a9a).