0:004:43

Recursos

Resumen

Take a nef klt adjoint K_X + Δ on a compact Kähler fourfold that has one nonzero plurisection. This class shows how to explain, step by step, why some multiple of it must be globally generated, and why it is trivial when the Iitaka dimension is zero.

The statement [0:27]

Theorem 1.1 assumes X is normal, compact Kähler and four-dimensional, (X, Δ) is klt, and D = K_X + Δ is Q-Cartier, analytically nef, with κ(X, D) ≥ 0. It concludes that some mD is Cartier and generated by global sections, and that O_X(mD) ≃ O_X when κ = 0. It does not assume projectivity or Q-factoriality.

The route of the proof [0:55]

  • κ ≥ 1: a crepant Q-factorial dlt model puts the pair inside Höring–Lazić–Lehn, Theorem 4.1, and generation descends back to X.
  • κ = 0: take s₀ with divisor M = (1/m₀) div s₀ [1:12]. If M = 0, you are done. The rest is a proof by contradiction that rules out M ≠ 0.
    1. Supported model [1:36]: from a log resolution with coefficients raised to one, build a dlt fourfold (V, B) with A nef, κ(V, A) = 0, A ~ P ≠ 0 and Supp P = Supp ⌊B⌋. This puts the section's divisor on the floor, where generation can be proved, and keeps the pair klt away from it [2:07].
    2. Generation on the whole floor S [2:29]: Das–Ou handles each component. Theorem 6.1 then makes the sections agree on every intersection and stratum, like patches of a quilt.
    3. Supported lifting [2:55]: a cyclic cover gives a reduced E with ideal I. The conormal layers I^j/I^(j+k) lift at every finite order, so h⁰ of NG modulo constants is at least N − C. The loss is at most h¹(O_V), so sections grow and κ(V, A) ≥ 1 [3:22].
  • The two results contradict each other, so M = 0 and s₀ trivializes O_X(m₀D) [3:58].

The move to keep [4:25]: push the section's divisor onto a dlt floor, generate there, then lift until sections grow. The open question is nonvanishing itself, which the theorem only assumes.

Abundance after nonvanishing for compact Kähler fourfolds

OpenAI Math Preprints

Clase 69 / 100

Ver clases