Convergence speed of weighted Bergman kernels towards extremal functions
Guokuan Shao

TL;DR
This paper studies how weighted Bergman kernels, constructed via Bernstein-Markov inequalities on high tensor powers of line bundles, uniformly converge to extremal functions, providing explicit convergence speed estimates.
Contribution
It introduces a method to construct inner products leading to weighted Bergman kernels that converge uniformly to extremal functions with quantifiable speed.
Findings
Established uniform convergence of weighted Bergman kernels to extremal functions.
Derived explicit bounds on the convergence speed.
Applied Bernstein-Markov inequalities to high tensor powers of line bundles.
Abstract
We construct inner products by the Bernstein-Markov inequality on spaces of holomorphic sections of high powers of a line bundle. The corresponding weighted Bergman kernel functions converge to an extremal function. We obtain a uniform convergence speed.
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsGeometry and complex manifolds · Geometric Analysis and Curvature Flows · Holomorphic and Operator Theory
Convergence speed of weighted Bergman kernels towards extremal functions
Guokuan SHAO
Abstract
We construct inner products by the Bernstein-Markov inequality on spaces of holomorphic sections of high powers of a line bundle. The corresponding weighted Bergman kernel functions converge to an extremal function. We obtain a uniform convergence speed.
Classification AMS 2010: 32A36, 32L10, 32Q15, 32U15.
**Keywords: ** Bergman kernel, extremal function, positive line bundle, holomorphic section.
1 Introduction
In this paper we study the convergence speed when the weighted Bergman kernel functions for high powers of a line bundle converge to an extremal function. Let be a positive line bundle over a projective manifold of dimension , where is a smooth Hermitian metric of , is the Kähler form of . Denote by the space of all global holomorphic sections of . A natural inner product on is induced by and . Let be the Bergman kernel for . Let be the Kodaira embedding map when is large. Denote by the Fubini-Study form of complex projective spaces. A famous theorem by Tian-Zelditch [22] states that converges uniformly to in topology. Alternatively we have converges uniformly to [math] on . Since then the convergence of Bergman kernels have been studied actively, see [10, 14, 15, 17] and the references therein. Many works extended the results in more general settings, e.g. the line bundle is big with singular metrics, the base space is normal Kähler complex space etc, see [6, 11]. One of the applications of these results is to explore the zeros of random holomorphic sections of line bundles. For related works, see [5, 7, 12, 18, 19, 21] etc.
Bloom-Shiffman [4] considered a Bergman kernel induced by a new inner product. They used Bernstein-Markov inequalities to construct the new inner products. The limit of is the Siciak’s extremal function [20]. The setting of the work is based on the space of homogeneous polynomials on complex vector spaces. Now we generalize the setting to the case of positive line bundles with a weighted function and obtain a uniform convergence speed.
Some basic settings are the following:
I: Let be a positive line bundle over a projective manifold of dimension . The smooth Hermitian metric of induces the Kähler form of .
II:The Bergman kernel function for is defined by (3). Note that in the Bernstein-Markov inequality, we have the condition (cf. Definition 3.1).
III: The extremal function is defined by (1).
Here is our main theorem.
Theorem 1.1**.**
With the above assumptions, we have uniformly
[TABLE]
Remark 1.2**.**
Note that the Bergman kernel in the theorem is different from the classical one. It depends on the choice of inner products. Different inner products induce different completion spaces of .
The paper is organized as follows. In Section 2 we recall the notion of holomorphic line bundles. In Section 3 we introduce the extremal functions. We consider the Bernstein-Markov inequality of special type. Then we define new inner products which induce the weighted Bergman kernels. The Hörmander’s -estimate for is also mentioned. We conclude Section 4 with the proof of the main theorem by two steps.
2 Holomorphic line bundles
In this section, we introduce basic notions of holomorphic line bundles. Let be a holomorphic line bundle over a complex compact Kähler manifold . The complex dimension of is . Let be the projection map. There exist local trivializations of on an open cover of . The biholomorphisms are , which send isomorphically onto . The transition functions are defined by the formula
[TABLE]
The are non-where vanishing holomorphic functions on . The Cěch cohomology class of defines the first Chern class of .
Denote by the space of all global holomorphic sections of . Let be a Hermitian metric of , the local frame of on . For a section , we write on . Then we can say equally that is a collection of holomorphic functions on which subject to the compatibility condition on . Set , where . Then we can say equally that is a collection of functions with the conditions
[TABLE]
Definition 2.1**.**
A line bundle is said to be Lipschitz if all are Lipschitz functions.
Positive line bundles are always Lipschitz since they admit smooth Hermitian metrics. The curvature current
[TABLE]
represents the first Chern class . Here . The line bundle is called if is strictly positive. In this case we can take to be the Kähler form of . The Kodaira embedding theorem implies that is a projective manifold. Let be the th tensor product of with the natural metric induced by .
3 Extremal functions and weighted Bergman kernels
In this section, we will introduce the extremal functions and define weighted Bergman kernels. From now on, we assume is a positive line bundle over a projective manifold of dimension , where is a smooth Hermitian metric such that is the Kähler form of .
Recall that a (q.p.s.h.) function is an upper semi-continuous (usc) function which is locally the difference of a p.s.h. function and a smooth one. Any q.p.s.h. functions on satisfies
[TABLE]
for some . Denote by the set of all q.p.s.h. functions with . A subset is called if there exists a p.s.h. function such that .
Now we consider a non-pluripolar compact subset and a continuous function . Then the is defined to be the usc regularization of the function
[TABLE]
This generalizes the notion of classical Siciak’s extremal functions [20]. Throughout this paper, we assume that is continuous. The paper [3] gave a local regularity condition to make continuous. Then is equal to its usc regularization. It follows from [13] that . So is a positive closed -current which represents . We write for for simplicity.
Recall that the Bergman kernel for is the Schwartz kernel of the orthogonal projection from the space of global -sections of onto . A natural inner product on is defined by the following formula
[TABLE]
We choose an orthonormal basis with respect to the inner product. Then we obtain the Bergman kernel for . Note that the space of -sections of depends on the definition of inner products on . Hence the Bergman kernel depends also on the definition.
Now we define a new inner product on which yields a weighted Bergman kernel. Let be as before. Consider a positive measure with support on . Note that .
Definition 3.1**.**
The triple satisfies the Bernstein-Markov inequality of special type if we have
[TABLE]
for all , where for some universal constant .
By the above Bernstein-Markov inequality, we have the following well-defined inner product on
[TABLE]
Note that if , then on . The identity theorem implies that is the zero section. Now we choose an orthonormal basis with respect to the above inner product. The weighted Bergman kernel function for is given by
[TABLE]
Here . It is well-known [16] that
[TABLE]
At the end of this section, we recall the Hörmander’s -estimate for [8].
Theorem 3.2**.**
Let be a singular Hermitian holomorphic line bundle over a complete Kähler manifold of dimension . Let be the dual of the canonical line bundle with the natural Hermitian metric . If there exists a continuous function such that , then for any form satisfying
[TABLE]
there exists with and
[TABLE]
If there exists such that
[TABLE]
then for any form with , there exists satisfying
[TABLE]
4 Proof of the main theorem
In this section the proof of the main theorem is divided into two steps. Set
[TABLE]
Recall that the positive measure is in Definition 3.1. First we prove the following
Proposition 4.1**.**
With the above notations and assumptions, we have the uniform estimate on
[TABLE]
In particular,
[TABLE]
Proof.
Recall that is the orthonormal basis of . We write for short. The Bernstein-Markov inequality yields
[TABLE]
Then
[TABLE]
Hence
[TABLE]
For the left inequality, we consider any section satisfying . Let be the Bergman kernel. Then
[TABLE]
So . The proof is completed by using the fact that . ∎
Now we prove the main theorem. Our proof is based on the original idea by Demailly [9].
Proof.
By the above proposition, it suffices to show that
[TABLE]
uniformly on . Recall that is defined in (1).
Let for . Since
[TABLE]
we deduce that
[TABLE]
Locally , then
[TABLE]
So
[TABLE]
By the definition of , .
Fix . Let
[TABLE]
By regularizations and translations, we can assume without loss of generality,
[TABLE]
Denote by the open ball centered at with radius . Since is Lipschitz, we choose for some such that
[TABLE]
and
[TABLE]
Here without confusions, we write locally for .
Let be a cut-off function with support in and on with . Without loss of generality, we assume
[TABLE]
Then is a global section of for all .
Recall that is the canonical line bundle of . There exists an integer such that is positive. Let be the smooth Hermitian metric of . Let be a singular Hermitian metric of for some . Moreover, is smooth in and its Lelong number . That is
[TABLE]
So near we have
[TABLE]
The choice of is possible when is large, since is positive.
The section can be regarded as a -closed -form with values in . Note that the curvature form of is
[TABLE]
which is smooth. Let
[TABLE]
Since
[TABLE]
for some universal constant , so we can apply Hörmander’s -estimate theorem to the case with the metric . Hence there exists a smooth section such that with the following estimate
[TABLE]
Since , is smooth in . We deduce that
[TABLE]
Recall that the Lelong number of at satisfies , i.e. near . But is not integrable near , we must have .
On the open ball , . So
[TABLE]
There exists independent of such that
[TABLE]
Hence
[TABLE]
Set
[TABLE]
Note that and is a non-zero smooth section of . Moreover, we have
[TABLE]
Since on , we apply the mean-value inequality to the subharmonic function , ,
[TABLE]
Note that
[TABLE]
Then we have
[TABLE]
Set
[TABLE]
Then
[TABLE]
[TABLE]
Since is compact, the set of all is bounded. Hence
[TABLE]
for all . Then the proof is completed. ∎
The reference list from the paper itself. Each links out to its DOI / PubMed record.
- 1[1]
- 2[2]
- 3[3] Berman R., Boucksom S., Witt Nyström D., Fekete points and convergence towards equilibrium measures on complex manifolds , Acta Math., 207 (2011), no.1, 1–27.
- 4[4] Bloom T., Shiffman B., Zeros of random polynomials on ℂ m superscript ℂ 𝑚 \mathbb{C}^{m} , Math. Res. Lett., 14 (2007), no.3, 469–479.
- 5[5] Coman D., Marinescu G., Equidistribution results for singular metrics on line bundles , Ann. Sci. École Norm. Supér., 48 (2015), no.3, 497–536.
- 6[6] Coman D., Ma X., Marinescu G., Equidistribution for sequences of line bundles on normal Kähler spaces , available at preprint ar Xiv: 1412.8184.
- 7[7] Coman D., Marinescu G., Nguyên V.-A., Hölder singular metrics on big line bundles and equidistribution , Int. Math. Res. Not., (2016), no.16, 5048–5075.
- 8[8] Demailly J.-P., Complex analytic and differential geometry , available at www.fourier.ujf-grenoble.fr/ ∼ similar-to \sim demailly .
