Skip to content

Companion note, version 1.34

Choose a tag to compare

@carlok carlok released this 23 Oct 10:05

tex/diaz_prove2me.tex, 42 pages. The PDF is attached as diaz_prove2me-v1.34.pdf.

New: Proposition 5.29, conjugation hulls of every z rational in u. Let u ∉ Q̄ with uū algebraic, z ∈ Q̄(u),
and W_z = span(1, u, ū, z, z̄). W_z carries a 2×3 configuration exactly when dim W_z = 4 and
z ∈ span(1, u, ū, 1/(u − a)) for some algebraic a with aā = uū, or dim W_z = 5 and W_z has one of three shapes:
degree one (D1), two-orbit (D2a) or cyclic (D2b).

  • (a) Theorem NF: a configuration whose ratio is p(u)/q(u), with p, q coprime of degree 2, gives one of the three
    shapes, in any five-dimensional W containing 1, u and u⁻¹ (DiazModulus.degree_two_configuration_normal_form).
    The proof is elementary: a non-zero resultant, freeness over Q̄(p(u)/q(u)), a count of bad λ and a 2×2 Jordan
    form. The statement is Theorem 2 of the project's research notes, proved there with vector bundles on P¹.
  • (b) In dimension 5 conjugation plays no part (DiazModulus.rational_space_two_by_three_configuration_iff).
  • (c) Theorem F (DiazModulus.rational_conj_hull_two_by_three_configuration_iff).
  • (d) The degree-one shape, and its form under conjugation: visible to Theorem 5.5 or a real pencil space W(λ, f)
    (DiazModulus.degree_one_multiplier_space_visible_or_degree_one_space,
    DiazModulus.conj_stable_degree_one_multiplier_space_visible_or_real_pencil_space).

Remark 5.30 gives the credit. The four-dimensional configuration is Diaz's Théorème 7(1) of 2007 at v = 1/(u − a),
with the exclusion at a candidate in his Corollaire 5(4). The critical case is Bachoc–Serra–Zémor's Lemma 4.
Not found in the sources read: Theorem NF outside the research notes, and the "only if" halves. No source on scrolls
or on rank-two vector bundles on P¹ was among the sources read.

The remark also says what is not proved:

  • For Laurent z, agreement with Proposition 5.25 (Theorem E) was checked by exact computation, not proved, and its
    explicit criteria do not follow.
  • No exclusion at a candidate is stated formally.
  • The conjugation-stable families of the research notes, for a general involution, stay unformalised (Remark 5.13
    now says so).

Appendix A. One new row with five identifiers and 17 supporting results; 213 identifiers in all. The mirror holds
497 results, all building in CI on Lean's three standard axioms.