Section 85
Reduction to the non-reduced conics
In this section we write and , in order to match the notation of [43]. Thus is a sextic del Pezzo fibre, is a cubic del Pezzo fibre, and . We shall use the corrected version of [43], namely arXiv:2403.03700v3, dated 27 June 2026. Its Lemma 6.3 gives a non-strict local estimate. We explain below why this estimate still excludes a divisor with non-positive beta-invariant. The generality condition in [43, Theorem 1.1] is not imposed in this section.
Lemma 85.1. One has Moreover, is divisorially stable: for every prime divisor on .
Proof. The effective cone is generated by . The intersection numbers are The divisor is nef for , and is not pseudoeffective for . Its cube is . Hence For the two fibres the positive parts are The corresponding negative parts are and , respectively. These decompositions follow by intersecting with the two rulings of and with a conic fibre; equivalently, they are [43, Propositions 5.1 and 5.2]. Therefore For completeness, the divisorial-stability reduction in [43, Section 5] uses the effective cone to reduce the verification to , by the criterion of Fujita quoted there. The last computation gives , and thus proves the stated divisorial stability. This argument does not require any condition on the singular fibres. ◻
Lemma 85.2. For every point , one has .
Proof. Write , and denote the ruling classes by . Since and , the restriction of is . Choose the ruling through . For , the divisor is nef, and beyond this interval it is not pseudoeffective. There are no fixed components in these restrictions. Thus the two successive expected vanishing orders are Abban–Zhuang adjunction for the flag and Lemma 85.1 now give This also reproduces [43, Lemma 6.1]. ◻
Lemma 85.3. Let be a reduced fibre of the conic bundle . Let be a prime divisor over whose centre is contained in . Then .
Proof. Let be a general point of the centre of . If , apply Lemma 85.2. Hence we may assume that . Let be the fibre of containing ; thus . If is smooth along , then [43, Lemma 6.2] gives .
Suppose that is singular at some point of . The restriction is a projection , so is smooth along . Thus its singular point on is outside . By [43, Lemma 3.5], is a smooth sextic del Pezzo surface. The conic is , where are -curves meeting transversely. Moreover, is a disjoint union of two -curves. We may number them so that and .
Let be the series obtained by refining with respect to . The calculation in the corrected [43, Lemma 6.3] proves We recall the relevant numbers, so that the non-strict inequality is explicit. For , the flag beginning with has first expected vanishing order and point expected vanishing order . The same holds with interchanged. At , blow up on and denote the exceptional curve by . Its log discrepancy is and its first expected vanishing order is . The next expected vanishing order is away from the two points corresponding to , and is at either of these two points. These are all the terms in the surface adjunction estimate. In particular, its lower bound is , including at .
It is essential to retain the positive remainder in the first adjunction step. Applied to a valuation centred in , the last surface estimate gives Indeed, choose a basis simultaneously compatible with the -filtration and , and write its basis divisor as . Then and is the corresponding refined basis divisor. The refined threshold estimate says that is log canonical near for all sufficiently divisible , with an error tending to zero. Inversion of adjunction gives Letting tend to infinity and then tend to zero proves the inequality. This is the strictness argument underlying the equality discussion in [43, Section 8].
Here the centre of is contained in , so . Lemma 85.1 therefore gives the strict inequality Thus the non-strict surface estimate is sufficient. ◻
Proposition 85.4. If a prime divisor over satisfies , then its centre is contained in the reduced support of a non-reduced conic fibre of and is disjoint from .
Proof. The centre is neither a divisor, by Lemma 85.1, nor a subvariety meeting , by Lemma 85.2. Suppose first that the centre is a curve not contracted by . The general cubic fibre is smooth, and only finitely many fibres of are singular. A general point therefore lies on a smooth cubic fibre. Applying [43, Lemma 6.2] at gives .
Suppose next that is contained in a cubic fibre but is not contracted by . The cubic fibre is normal by [43, Lemma 3.2]. The rationality argument in the proof of [43, Lemma 3.4] then shows that its singularities are Du Val. In particular, its singular locus is finite. There are consequently only finitely many conic fibres in meeting this singular locus. A general point avoids their union. The cubic surface is smooth along the conic through this point, so [43, Lemma 6.2] again gives a contradiction.
Thus a curve centre is contained in a conic fibre. A point centre is of course contained in its conic fibre as well. If that fibre were reduced, Lemma 85.3 would give the same contradiction. This proves the assertion. ◻
Remark 85.5. The positivity of in Lemma 85.3 is used only after the centre has been placed in a conic fibre. A horizontal curve need not have positive order along ; it is excluded separately in Proposition 85.4. Thus no strict inequality is inferred from a non-strict lower bound for a local delta-invariant alone.