Quantized Volume Comparison for Fano Manifolds, II

Abstract: In this note, the second author's quantized volume comparison conjecture is solved: If $X$ is a smooth complex $K$-semistable Fano variety of dimension $n$, then for every integer $m\geq1$, [ h^0(X,-mK_X)\leq h^0(\mathbb P^n,-mK_{\mathbb P^n}) =\binom{n+m(n+1)}{n}, ] and equality for one $m$ characterizes projective space. Somewhat surprisingly, the same statement actually holds whenever $T_X$ is slope semistable with respect to $-K_X$. The idea is to apply a jet-dimension counting trick to a filtration of subsheaves induced by $H<sup>0(X,-mK_X)$. Then the slope semistability condition yields the desired dimension bound.

• The paper proves that every smooth Fano n-fold with slope-semistable tangent bundle satisfies h⁰(X,−mK_X) ≤ binom(n+m(n+1), n) for all m≥1, with equality at one level forcing X ≅ ℙⁿ.
• The authors combine a jet-filtration slope bound with sharp weighted jet counting, converting semistability of symmetric powers of the cotangent bundle into finite-level Hilbert function estimates.
• The result implies the sharp volume bound (−K_X)ⁿ ≤ (n+1)ⁿ under tangent-bundle slope semistability and extends to polarized varieties, pseudoeffective line bundles, and ℚ-Fano varieties.

Overview and main results

This paper by Lyu and Zhang resolves the quantized volume comparison conjecture formulated by the second author in earlier work. The conjecture asks for a finite-level refinement of Fujita's volume comparison: if $X$ is a smooth complex $K$-semistable Fano $n$-fold, Fujita proved $(-K_X)^n \le (n+1)^n$ with equality only for $\mathbb P^n$, and since $h^0(X,-mK_X) = \frac{(-K_X)^n}{n!}m^n + O(m^{n-1})$, it is natural to ask whether the comparison holds level by level. The paper proves that for every $m \ge 1$,

$$h^0(X,-mK_X) \le h^0(\mathbb P^n,-mK_{\mathbb P^n}) = \binom{n+m(n+1)}{n},$$

with equality at a single value of $m$ forcing $X \simeq \mathbb P^n$. The notable strengthening is that the same statement holds under the strictly weaker hypothesis that the tangent bundle $K$0 is slope semistable with respect to $K$1; $K$2-semistability enters only via Li's extension of Tian's theorem, which produces this semistability. The result also yields, as a corollary, the Fujita–Liu volume bound $K$3 under slope semistability alone — a sharp estimate that, according to the authors, had not previously been recorded under this hypothesis. Ran had earlier obtained only the non-sharp bound $K$4 under the same semistability assumption plus additional hypotheses.

Context: Yau's conjecture and prior work

The conjecture is the compact analogue of Yau's conjecture for complete noncompact Kähler manifolds $K$5 with nonnegative holomorphic bisectional curvature: $K$6, with equality only for $K$7. That statement was proved by Ni under maximal volume growth, in full generality with rigidity by Chen–Fu–Yin–Zhu, and strengthened by Liu, who replaced bisectional curvature with the weaker nonnegative holomorphic sectional curvature using his three-circle theorem. The paper notes a precise structural parallel: with $K$8, the Euclidean model dimension $K$9 coincides with the Fano comparison term, so both results are sharp jet-counting statements with model-space rigidity, differing in the nature of the hypotheses (pointwise curvature versus algebraic slope semistability).

In the predecessor paper, the second author proved the comparison for all sufficiently large $n$0, and conditionally in dimension four. The present work removes both restrictions.

The proof mechanism

The argument combines two ingredients, neither of which uses the valuative criterion for K-stability directly.

Jet-filtration slope bound. For $n$1 with $n$2, $n$3, the kernels of the evaluation maps $n$4 into sheaves of principal parts form a decreasing filtration $n$5. The $n$6-th graded image $n$7 is a torsion-free subsheaf of $n$8. Since $n$9 is $(-K_X)^n \le (n+1)^n$0-semistable, so is $(-K_X)^n \le (n+1)^n$1, and hence each $(-K_X)^n \le (n+1)^n$2 is semistable with slope $(-K_X)^n \le (n+1)^n$3 — this uses the Mehta–Ramanathan restriction to general complete-intersection curves, the characteristic-zero tensor product theorem, and the symmetrizing idempotent. Summing the slope inequalities $(-K_X)^n \le (n+1)^n$4 over the filtration and using the telescoping identity $(-K_X)^n \le (n+1)^n$5 (which follows from termination of the filtration via Krull intersection and finite-dimensionality of $(-K_X)^n \le (n+1)^n$6) yields, for general $(-K_X)^n \le (n+1)^n$7,

$(-K_X)^n \le (n+1)^n$8

Sharp jet counting. Conversely, a Taylor-expansion (hockey-stick) estimate shows that if $(-K_X)^n \le (n+1)^n$9 with $\mathbb P^n$0, then the weighted jet sum $\mathbb P^n$1 strictly exceeds $\mathbb P^n$2 — a contradiction. Hence $\mathbb P^n$3.

Equality case. If $\mathbb P^n$4 attains the bound, equality forces the order-$\mathbb P^n$5 jet evaluation map $\mathbb P^n$6 to be an isomorphism, so $\mathbb P^n$7 generates $\mathbb P^n$8 jets at $\mathbb P^n$9, whence $h^0(X,-mK_X) = \frac{(-K_X)^n}{n!}m^n + O(m^{n-1})$0. The Bauer–Szemberg characterization of projective space by anticanonical Seshadri constants then gives $h^0(X,-mK_X) = \frac{(-K_X)^n}{n!}m^n + O(m^{n-1})$1. This is a rigidity statement at a single finite level $h^0(X,-mK_X) = \frac{(-K_X)^n}{n!}m^n + O(m^{n-1})$2, which is stronger than what asymptotic methods provide.

Generalizations

The core theorem is stated for an arbitrary polarization $h^0(X,-mK_X) = \frac{(-K_X)^n}{n!}m^n + O(m^{n-1})$3 and line bundle $h^0(X,-mK_X) = \frac{(-K_X)^n}{n!}m^n + O(m^{n-1})$4: if $h^0(X,-mK_X) = \frac{(-K_X)^n}{n!}m^n + O(m^{n-1})$5 is $h^0(X,-mK_X) = \frac{(-K_X)^n}{n!}m^n + O(m^{n-1})$6-semistable and $h^0(X,-mK_X) = \frac{(-K_X)^n}{n!}m^n + O(m^{n-1})$7, then

$h^0(X,-mK_X) = \frac{(-K_X)^n}{n!}m^n + O(m^{n-1})$8

and, for pseudoeffective $h^0(X,-mK_X) = \frac{(-K_X)^n}{n!}m^n + O(m^{n-1})$9 with $m \ge 1$0, one obtains $m \ge 1$1. The anticanonical result follows by taking $m \ge 1$2. A final remark extends the argument to $m \ge 1$3-semistable $m \ge 1$4-Fano varieties via reflexive sheaves on the regular locus; codimension-one regularity ensures that reflexive extension does not affect slopes, and the singular Seshadri characterization of Liu–Zhuang handles equality. The extension to $m \ge 1$5-Fano varieties includes levels $m \ge 1$6 for which $m \ge 1$7 is not Cartier.

Limitations and open questions

The paper is explicit about what its method does not deliver. First, it does not recover the earlier fixed-level result conditional on $m \ge 1$8: the present filtration controls the ordinary vanishing-order filtration at a general point, whereas $m \ge 1$9 is an infimum over all valuations, and the argument does not deduce the finite-level delta invariant from $$h^0(X,-mK_X) \le h^0(\mathbb P^n,-mK_{\mathbb P^n}) = \binom{n+m(n+1)}{n},$$0-semistability. Second, and more significantly, finite-level rigidity does not imply volume rigidity: equality of volumes fixes only the leading coefficient of the anticanonical Hilbert polynomial and is compatible with a strict $$h^0(X,-mK_X) \le h^0(\mathbb P^n,-mK_{\mathbb P^n}) = \binom{n+m(n+1)}{n},$$1 deficit at every finite level. The authors accordingly leave open whether a smooth Fano $$h^0(X,-mK_X) \le h^0(\mathbb P^n,-mK_{\mathbb P^n}) = \binom{n+m(n+1)}{n},$$2-fold with $$h^0(X,-mK_X) \le h^0(\mathbb P^n,-mK_{\mathbb P^n}) = \binom{n+m(n+1)}{n},$$3-semistable $$h^0(X,-mK_X) \le h^0(\mathbb P^n,-mK_{\mathbb P^n}) = \binom{n+m(n+1)}{n},$$4 and $$h^0(X,-mK_X) \le h^0(\mathbb P^n,-mK_{\mathbb P^n}) = \binom{n+m(n+1)}{n},$$5 must be $$h^0(X,-mK_X) \le h^0(\mathbb P^n,-mK_{\mathbb P^n}) = \binom{n+m(n+1)}{n},$$6; under $$h^0(X,-mK_X) \le h^0(\mathbb P^n,-mK_{\mathbb P^n}) = \binom{n+m(n+1)}{n},$$7-semistability the answer is affirmative by Fujita's equality theorem.

A further point of context is the declaration on the use of AI: the jet-filtration slope bound, one of the two central ingredients, is credited to ChatGPT 5.6, likely building on Ran's work on sheaves of differential operators; the authors disclaim credit for that lemma while taking responsibility for the content.

Conclusion

The paper settles the quantized volume comparison conjecture in full, and in a stronger form than conjectured: sharp finite-level Hilbert function bounds and single-level rigidity hold for Fano manifolds whose tangent bundle is merely slope semistable, with $$h^0(X,-mK_X) \le h^0(\mathbb P^n,-mK_{\mathbb P^n}) = \binom{n+m(n+1)}{n},$$8-semistability used only through Li's theorem. The method — a jet filtration graded into semistable symmetric powers of the cotangent sheaf, combined with sharp jet counting at a general point — is purely algebraic and adapts to polarized pairs, general line bundles, and $$h^0(X,-mK_X) \le h^0(\mathbb P^n,-mK_{\mathbb P^n}) = \binom{n+m(n+1)}{n},$$9-Fano varieties. The main open problem left by the work is whether volume equality $m$0 alone forces projective space under tangent-bundle semistability.