0:005:40

Recursos

Resumen

Take a compact Kähler manifold T with a known algebraic dimension. You should come away able to say which branch of the induction handles it, and what that branch has to prove about the nef summand M.

Why the conclusion is numerical

On an elliptic curve, a non-torsion degree-zero line bundle M is nef, but no positive power of it has a section, so M itself is not semiample [0:36]. Its Chern class is zero, though, and that is the class of the semiample trivial bundle. The theorem [1:17] says exactly this. If K_X + B is pseudo-effective, M is nef and D = K_X + B + M is nef, then some semiample L has c₁(L) = c₁(D) in real Bott–Chern cohomology. In other words, D is semiample up to a flat twist.

The engine of the proof

  • Boucksom's divisorial negative part N(α) [1:39] collects the minimal generic Lelong multiplicities along prime divisors. A nef class has N = 0.
  • Theorem 1.2 [2:10]: after a modification μ, μ*(J + M) ≡ P + R, with P semiample and R = N(μ*(J + M)). Also, a(T) = 0 forces c₁(M) = 0.
  • The induction on dimension splits on a(T) [2:28]:
    • projective: apply the projective theorem;
    • 0 < a(T) < dim T [2:59]: M must have zero class on a very general fiber of the algebraic reduction h: T → W. Then it descends to a nef line on W, and an ordinary program over W reaches a klt adjoint on a projective base [3:23];
    • a(T) = 0 [4:05]: reduce to tori and to simple symplectic manifolds. Meromorphic nonvanishing for K + M holds because a family of correspondences sweeping T × T would give a forbidden holomorphic one-form.

The rule to use

For a nef sum, R = 0, so P − μ*D is a flat line F from Pic⁰, and L = D + F [4:58]. Faced with a Kähler adjoint problem, read a(T) first, then ask what M must do there: vanish, descend to the base, or meet the projective theorem [5:18].

Generalized abundance for nef adjoint classes on Kähler manifolds

OpenAI Math Preprints

Clase 88 / 100

Ver clases