Text size

Section 116

Proof for every smooth member

Proof of Theorem 108.1. Let QQ be a prime divisor over XX. Suppose that βX(Q)≤0\beta_X(Q)\le0. Proposition 110.6 puts its centre on one of the two lines through the singular point of an A1A_1 or A2A_2 fibre. Denote this line by Γ\Gamma and choose a closed point pp of the centre.

If pp is the singular point of the fibre, Proposition 112.1 and Corollary 113.6 give δp(X)>1\delta_p(X)>1, a contradiction. If p∈Ep\in E, use Lemma 114.3 or Proposition 114.6. Their bounds are again strictly greater than one.

Assume that pp is outside these two points. Interchange the plane projections so that the first contracts Γ\Gamma. If p∉D2p\notin D_2, Lemma 111.10 supplies a smooth quintic del Pezzo section through pp on which pp lies on none of the ten lines. Lemmas 111.5, 111.7 and 111.8, with the two projections interchanged, give δp(X)≥1152/1127>1\delta_p(X)\ge1152/1127>1. If p∈D2p\in D_2, Lemma 111.10 says that the fibre has type A1A_1 and pp is its remaining smooth corner. Proposition 115.1 gives δp(X)>1\delta_p(X)>1 there. This is the last possibility.

In every case AX(Q)>SX(Q)A_X(Q)>S_X(Q), contrary to our assumption. Thus every prime divisor over XX has positive beta invariant. The valuative criterion proves K-stability. ◻

Example 116.1. The pencil in Example 109.5 has three A2A_2 fibres. At each singular point the weighted extraction gives 64/6364/63. Each of the two lines has one point on EE; its ordinary point extraction gives 48/4148/41. Every other point on these lines is treated by an opposite quintic section. Thus even this pencil, whose six ramification contributions are all concentrated at three points, is covered by the proof.

The extraction arguments do not infer a refined series from a volume alone. In each case, actual anticanonical sections supply independent initial forms, and the exact Hilbert count proves that they exhaust the associated graded space. The barycentres or binary-space sums then compute SX(G)S_X(G) with its original normalization. This is why the Abban–Zhuang inequalities apply without a Zariski decomposition on the extracted threefold.