0% found this document useful (0 votes)
22 views11 pages

Effective Methods in Complex Geometry

The document discusses methods for computing effective bounds for the degree of embeddings in projective spaces of algebraic varieties, particularly focusing on the case of adjoint line bundles. It outlines the use of various analytic techniques, such as the theory of positive currents and plurisubharmonic functions, to derive these bounds. Additionally, it covers fundamental concepts in hermitian differential geometry and the positivity of holomorphic vector bundles, along with significant results like the Kodaira embedding theorem.

Uploaded by

Chor Yin Ho
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
22 views11 pages

Effective Methods in Complex Geometry

The document discusses methods for computing effective bounds for the degree of embeddings in projective spaces of algebraic varieties, particularly focusing on the case of adjoint line bundles. It outlines the use of various analytic techniques, such as the theory of positive currents and plurisubharmonic functions, to derive these bounds. Additionally, it covers fundamental concepts in hermitian differential geometry and the positivity of holomorphic vector bundles, along with significant results like the Kodaira embedding theorem.

Uploaded by

Chor Yin Ho
Copyright
© All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd

L2 -Methods and Effective

Results in Algebraic Geometry

Jean-Pierre Demailly

Université de Grenoble I, Institut Fourier


URA 188 du CNRS, BP74, F-38402 Saint-Martin d’Hères, France

Abstract. One important problem arising in algebraic geometry is the computation of effective bounds
for the degree of embeddings in a projective space of given algebraic varieties. This problem is intimately
related to the following question: Given a positive (or ample) line bundle L on a projective manifold X,
can one compute explicitly an integer m0 such that mL is very ample for m > m0 ? It turns out that
the answer is much easier to obtain in the case of adjoint line bundles 2(KX + mL), for which universal
values of m0 exist. We indicate here how such bounds can be derived by a combination of powerful
analytic methods: theory of positive currents and plurisubharmonic functions (Lelong), L2 estimates
for ∂ (Andreotti-Vesentini, Hörmander, Bombieri, Skoda), Nadel vanishing theorem, Aubin-Calabi-Yau
theorem, and holomorphic Morse inequalities.

1. Basic concepts of hermitian differential geometry


Let X be a complex manifold of dimension n and let F be a C ∞ complex vector
bundle of rank r over X. A connection D on F is a linear differential operator D acting
on spaces C ∞ (X, Λp,q TX

⊗ F ) of F -valued differential forms, increasing the degree by 1
and satisfying Leibnitz’ rule
D(f ∧ u) = df ∧ u + (−1)deg f f ∧ Du
for all forms f ∈ C ∞ (X, Λa,b TX

), u ∈ C ∞ (X, Λp,q TX

⊗ F ). As usual, we split
′ ′′
D = D + D into its (1, 0) and (0, 1) parts, where
D′ + D′′ : C ∞ (X, Λp,q TX

⊗ F ) −→ C ∞ (X, Λp+1,q TX

⊗ F ) ⊕ C ∞ (X, Λp,q+1 TX

⊗ F ).

With respect to a trivialization τ : F↾Ω −→ Ω × Cr , a connection D can be written
Du ≃τ du + Γ ∧ u, where Γ = Γ′ + Γ′′ is an arbitrary (r × r)-matrix of 1-forms and
d acts componentwise. A standard computation shows that D2 u ≃τ Θ(D) ∧ u, where
Θ(D) = dΓ + Γ ∧ Γ is a global 2-form on X with values in Hom(F, F ). This form is called
the curvature tensor of F . In the important case of rank 1 bundles, Θ(F ) = dΓ is a
d-closed form with complex values; it is well known that the De Rham cohomology class
i i
of θ(F ) := 2π Θ(F ) = 2π D2 is the image in De Rham cohomology of the first Chern class
c1 (F ) ∈ H 2 (M, Z). For any line bundles F1 , . . . , Fp on X and any compact p-dimensional
analytic set Y in X, we set
Z
F1 · . . . · Fp · Y = c1 (F1 ) ∧ . . . ∧ c1 (Fp ).
Y
2 L2 -Methods and Effective Results in Algebraic Geometry

If F is equipped with a C ∞ hermitian metric h, the connection D is said to be compatible


with h if
dhu, vih = hDu, vih + hu, Dvih

for all smooth sections u, v of F . This is equivalent to the antisymmetry condition


Γ⋆ = −Γ (in a unitary frame), i.e. Γ′′ = −Γ′⋆ . In particular, a compatible connection
D is uniquely determined by its (0, 1)-component D′′ . If F has a holomorphic structure,
we precisely have a canonical (0, 1)-connection D′′ = ∂ obtained by letting ∂ act
componentwise. Hence, there exists a unique (1, 0)-connection D′ that makes D = D′ + ∂
compatible with the hermitian metric. This connection is called the Chern connection.
Let (eλ ) be a local holomorphic frame of F↾Ω and let H = (hλµ ), hλµ = heλ , eµ i be the
hermitian matrix representing the metric. Standard computations show that the Chern
connection and curvature are given by

−1 −1 −1
D′ ≃τ ∂ + H ∂H ∧ • = H ∂(H •), Θ(F ) = ∂(H ∂H).

In the special case where F has rank 1, it is convenient to write the unique coefficient
H = h11 of the hermitian metric in the form H = e−2ϕ . The function ϕ is called the
weight of the metric in the local coordinate patch Ω. We then find Θ(F ) = 2∂∂ϕ. It is
important to observe that this formula still makes sense in the context of distribution
theory if ϕ is just an arbitrary L1loc function. As we shall see later, the case of logarithmic
poles is very important for the applications.

(1.1) Definition. A singular hermitian metric on a line bundle F is a metric given in



any trivialization τ : F↾Ω −→ Ω × C by

kξk = |τ (ξ)| e−ϕ(x) , x ∈ Ω, ξ ∈ Fx ,

where ϕ ∈ L1loc (Ω) is an arbitrary function. The associated curvature current is

i
θ(F ) = ∂∂ϕ.
π

The Lelong-Poincaré equation states that πi ∂∂ log |f | = [Df ], where f is a holomorphic


or meromorphic function and [Df ] is the current of integration over the divisor of f .
More generally, we have
i
∂∂ log kσk = [Dσ ] − θ(F )
π
for every section σ ∈ H 0 (X, F ), as follows from the equality kσk = |f |e−ϕ , if f = τ (σ).
As a consequence, the De Rham cohomology class of [Dσ ] coincides with the first Chern
2
class c1 (F )R ∈ HDR (X, R).

2. Positivity and ampleness


Let (z1 , . . . , zn ) be holomorphic coordinates on X and let (eλ )16λ6r be an orthonor-
mal frame of F . Let the curvature tensor of F be
X
Θ(F ) = cjkλµ dzj ∧ dzk ⊗ e⋆λ ⊗ eµ .
16j,k6n, 16λ,µ6r
Positivity and ampleness 3

Clearly, this tensor can be identified with a hermitian form on TX ⊗ F , namely


X X ∂
e )(t) =
Θ(F cjkλµ tjλ tkµ , t= tjλ ⊗ eλ ∈ TX ⊗ F.
∂zj
(2.1) Definition (Kodaira, Nakano, Griffiths). A holomorphic vector bundle F is
e )(t) > 0 for all nonzero tensors t ∈ TX ⊗ F ;
• positive in the sense of Nakano if Θ(F
e )(ξ ⊗ v) > 0 for all nonzero decomposable
• positive in the sense of Griffiths if Θ(F
tensors ξ ⊗ v ∈ TX ⊗ F .

In particular, a holomorphic line bundle F is positive if and only if its weights ϕ are
strictly plurisubharmonic (psh), i.e. if (∂ 2 ϕ/∂zj ∂zk ) is positive definite.

P
(2.2) Example. Let D = αj Dj be a divisor with coefficients αj ∈ Z and let
F = O(D) be the associated invertible sheaf of meromorphic functions u such that
div(u) + D > 0 ; the corresponding line bundle can be equipped with the singular metric
defined by kukQ= |u|. If gj is a generator of the ideal of Dj on an open set Ω ⊂ X,
α
then τ (u) = u gj j defines a trivialization
P of O(D) over Ω; thus, our singular metric is
associated with the weight ϕ = αj log |gj |. By the Lelong-Poincaré equation, we find

i  i
Θ O(D) = ∂∂ϕ = [D] > 0,
π π
P
where [D] = αj [Dj ] denotes the current of integration over D.

(2.3) Example. Assume that σ1 , . . ., σN are nonzero holomorphic sections of F . Then


we can define a natural (possibly singular) hermitian metric on F by

|τ (ξ)|2
kξk2 = P 2
16j6N |τ (σj (x))|

with Prespect to any  trivialization τ . The associated weight function is ϕ(x) =


log |τ (σj (x))|2 1/2 . In this case ϕ is a psh function; thus, iΘ(F ) is a closed pos-
itive current.
T Let us denote by Σ the linear system defined by σ1 , . . . , σN and by
BΣ = σj−1 (0) its base locus. Let

ΦΣ : X \ BΣ → PN−1 , x 7→ (σ1 (x) : σ2 (x) : . . . : σN (x))

i
be the associated map. Then θ(F ) = 2π log(|σ1 |2 + · · · + |σN |2 ) is the pullback over
X \ BΣ of the Fubini-Study metric ωFS on PN−1 .

(2.4) Definition. A holomorphic line bundle F over a compact complex manifold X is


• very ample, if the map Φ|F | : X → PN−1 defined by the complete linear system
|F | = P (H 0 (X, F )) is a regular embedding (this means in particular that B|F | = ∅);
• ample, if mF is very ample for some positive integer m.

Here we used an additive notation for Pic(X) = H 1 (X, O⋆ ), i.e. mF = F ⊗m . By


Example (2.2), every ample line bundle F has a smooth hermitian metric with positive
definite curvature form; indeed, if Φ|mF | is an embedding, then we get a positive definite
4 L2 -Methods and Effective Results in Algebraic Geometry

curvature form θ(F ⊗m ) = Φ⋆|mF | (ωFS ) and we need only extract the mth root of this
metric to get the desired smooth metric on F . The converse is also true:
(2.5) Kodaira embedding theorem (1954). A line bundle F is ample if and only if
F can be equipped with a smooth hermitian metric of positive curvature.
In this context, Fujita [Fuj87] has raised the following important conjecture.
(2.6) Conjecture (Fujita, 1987). If L is an ample line bundle on a projective n-fold X,
then KX + (n + 1)L is globally generated and KX + (n + 2)L is very ample.
Here KX = Λn TX ⋆
is the canonical bundle. The example of curves shows that KX is
needed to get a uniform answer (if L is a bundle of degree 1 on a curve, then in general
mL does not have any nonzero section unless m > g = genus). Also, the example of
projective spaces show that Fujita’s bounds would be optimal, because KPn = O(−n−1).
Such questions have attracted a lot of attention in recent years. First, the case
of surfaces has been completely settled by Reider; in [Rei88] he obtains a very sharp
criterion for global generation and very ampleness of line bundles in dimension 2. In
higher dimensions, let us mention [Dem90, 93, 94, 95] and the works of Fujita [Fuj87, 94],
Kollár [Kol93], Ein-Lazarsfeld [EL92, 93], Lazarsfeld [Laz93], and Siu [Siu93, 94]. Our
goal is to describe a few powerful analytic methods that are useful in this context.

3. Bochner technique and vanishing theorems


Let X be a compact
P complex n-fold equipped with a Kähler metric, namely a positive
(1, 1)-form ω = i ωjk dzj ∧ dz k with dω = 0. Let F be a holomorphic vector bundle on
X equipped with a hermitian metric, and let

∆′ = D′ D′⋆ + D′⋆ D′ , ∆′′ = D′′ D′′⋆ + D′′⋆ D′′ ,

be the complex Laplace operators associated with the Chern connection D. Here
the adjoints
R D′⋆ , D′′⋆ are the formal adjoints computed with respect to the L2 norm
kuk2 = X |u(x)|2 dVω (x), where |u| is the pointwise hermitian norm and dVω = ω n /n!
is the volume form. The fundamental results of Hodge theory imply isomorphisms

H q (X, ΩpX ⊗ F ) = H p,q (X, F ) ≃ Hp,q (X, F )


between sheaf cohomology groups, Dolbeault ∂-cohomology groups, and the space Hp,q
of harmonic (p, q)-forms ∆′′ u = 0. The next fundamental fact is an identity originally
used by Bochner to prove vanishing results for Betti numbers. Slightly later, the identity
was extended to the complex situation by Kodaira and Nakano.
P
(3.1) Bochner-Kodaira-Nakano formula (1954). For all u = uJ,K,λ dzI ∧dz J ⊗eλ
of class C ∞ and type (p, q), we have

∆′′ u = ∆′ u + Ap,q
F,ω u,

where Ap,q p,q


F,ω is the hermitian endomorphism such that hAF,ω u, ui =
X X X
cjkλµ uJ,jS,λ uJ,kS,µ + cjkλµ ukR,K,λ ujR,K,µ − cjjλµ uJ,K,λ uJ,K,µ ,

and the summations are extended to all relevant indices 1 6 j, k 6 n, 1 6 λ, µ 6 r, and


all relevant multiindices |J| = p, |K| = q, |R| = p − 1, |S| = q − 1.
Hörmander’s L2 estimates and existence theorems 5

As h∆′ u, ui = kD′ uk2 + kD′⋆ uk2 > 0 the Bochner-Kodaira-Nakano formula implies
Z
′′
h∆ u, ui > hAp,q
F,ω u, ui dVω .
X

If Ap,q
F,ω is positive definite, every (p, q)-harmonic form has to vanish and we conclude
that H q (X, ΩpX ⊗ F ) = 0. In the special case of rank 1 bundles, we can take at each
point x ∈ X simultaneous diagonalizations
X X
ω(x) = i dzj ∧ dz j , Θ(F )(x) = i γj (x)dzj ∧ dz j ,

where γ1 (x) 6 · · · 6 γn (x) are the curvature eigenvalues. Then cjjλµ = γj and
X X X 
hAp,q
F,ω u, ui = γj − γj |uJK |2 > (γ1 + · · · + γq − γn−p+1 − · · · − γn )|u|2 .
J,K j∈K j ∈J
/

Assume now that iΘ(F ) is positive. The choice ω = iΘ(F ) yields γj = 1 for j = 1, 2, . . . , n
and hAp,q 2
F,ω u, ui = (p + q − n)|u| . From this, we immediately infer:

(3.2) Akizuki-Kodaira-Nakano vanishing theorem (1954). If F is a positive line


bundle on a compact complex manifold X, then

H p,q (X, F ) = H q (X, ΩpX ⊗ F ) = 0 for p + q > n + 1.

The above vanishing result is optimal. Unfortunately, it cannot be extended to semipos-


itive or numerically effective line bundles of bidegrees (p, q) with p < n, as shown by a
counterexample of Ramanujam [Ram74].

4. Hörmander’s L2 estimates and existence theorems


The basic existence theorem is the following result, which is essentially due to
Hörmander [Hö65] and, in a more geometric setting, to Andreotti-Vesentini [AV65].
(4.1) Theorem. Let (X, ω) be a complete Kähler manifold. Let F be a hermitian vector
bundle of rank r over X, and assume that A = Ap,q
F,ω is positive definite everywhere on
p,q ⋆
Λ TX ⊗ F , q > 1. Then for any form g ∈ L (X, Λp,q TX
2 ⋆
⊗ F ) with
Z
′′
D g=0 and h(Ap,q
F,ω )
−1
g, gi dVω < +∞,
X

there exists a (p, q − 1)-form f such that D′′ f = g and


Z Z
2
|f | dVω 6 h(Ap,q
F,ω )
−1
g, gi dVω .
X X

The proof can be ultimately reduced to a simple duality argument for unbounded
operators on a Hilbert space, based on the a priori inequality
Z
′′ 2 ′′⋆
kD uk + kD uk > 2
hAp,q
F,ω u, ui dVω .
X
6 L2 -Methods and Effective Results in Algebraic Geometry

The above L2 existence theorem can be applied in the fairly general context of weakly
pseudoconvex manifolds (i.e. manifolds possessing a weakly psh exhaustion function),
thanks to the fact that every weakly pseudoconvex Kähler manifold (X, ω) carries a
complete Kähler metric. In particular, the existence theorem can be applied on compact
manifolds, pseudoconvex open sets in Cn , Stein manifolds, etc. By regularization
arguments, the existence theorem also applies when F is a line bundle and the hermitian
metric is a singular metric with positive curvature in the sense of currents. In fact, the
solutions obtained with the regularized metrics have weak L2 limits satisfying the desired
estimates. Especially, we get the following more tractable version in the case p = n.
(4.2) Corollary. Let (X, ω) be a Kähler weakly pseudoconvex complex manifold of
dimension n. Let F be a holomorphic line bundle on X, equipped with a singular metric
whose local weights ϕ ∈ L1loc satisfy iΘ(F ) = 2i∂∂ϕ > εω for some ε > 0. For every
g ∈ L2 (X, Λn,q TX

⊗ F ) with D′′ g = 0, there exists f ∈ L2 (X, Λp,q−1 TX

⊗ F ) such that
′′
D f = g and Z Z
2 −2ϕ 1
|f | e dVω 6 |g|2 e−2ϕ dVω .
X qε X

This result leads in a natural way to the concept of multiplier ideal sheaves, according
to Nadel [Nad89]. The basic idea was already implicit in the work of Bombieri [Bom70]
and Skoda [Sk72].

(4.3) Multiplier ideal sheaves. Let ϕ be a psh function on an open subset Ω ⊂ X.


We define I(ϕ) ⊂ OX to be the sheaf of germs f ∈ OΩ,x such that |f |2 e−2ϕ is integrable
on a small neighborhood V of x with respect to the Lebesgue measure.
(4.4) Main property ([Nad89], [Dem93]). The ideal sheaf I(ϕ) ⊂ OX is a coherent
analytic sheaf. Its zero variety V (I(ϕ)) is the set of points in a neighborhood of which
e−2ϕ is nonintegrable.
A basic observation is that the zero variety V (I(ϕ)) is closed related to the sublevel sets
of Lelong numbers of ϕ.
(4.5) Definition. The Lelong number of a psh function ϕ at a point x ∈ X is the limit
ν(ϕ, x) := lim inf z→x ϕ(z)/ log |z − x|. The function ϕ is said to have a logarithmic pole
of coefficient γ if γ = ν(ϕ, x) > 0.
(4.6) Lemma ([Sk72]). Let ϕ be psh on Ω and let x ∈ Ω.
• If ν(ϕ, x) < 1, then e−2ϕ is integrable near x ⇒ I(ϕ)x = OΩ,x .
s+1
• If ν(ϕ, x) > n + s, s ∈ N, then e−2ϕ > C|z − x|−2n−2s near x and I(ϕ)x ⊂ mΩ,x .
P
(4.7) Simple algebraic [Link] ϕ = αj log |gj |, αj ∈ Q+ , be associated with a
normal crossing Q-divisor D = αj Dj , Dj = gj−1 (0). An easy computation gives
X
I(ϕ) = O(− ⌊αj ⌋Dj ) = O(−⌊D⌋),
where ⌊αj ⌋ = the integral part of αj . If the assumption on normal crossings is omitted,
a desingularization of D has to be used in combination with the following fonctoriality
property for direct images.
(4.8) Basic fonctoriality property. Let µ : X ′ → X be a modification (i.e. a proper
generically 1 : 1 holomorphic map), and let ϕ be a psh function on X. Then

µ⋆ O(KX ′ ) ⊗ I(ϕ ◦ µ) = O(KX ) ⊗ I(ϕ).
Numerical criteria for very ample line bundles 7

Let us now consider the case of general algebraic singularities


α 
ϕ∼ log |f1 |2 + · · · + |fN |2
2

with α ∈ Q+ and fj holomorphic on an open set Ω ⊂ X. By Hironaka’s theorem, there


e → X of X such that µ⋆ (f1 , . . . , fN ) is an invertible
exists a smooth modification µ : X P
sheaf O(−D) associated with a normal crossing divisor D = λj Dj . Then
X 
I(ϕ) = µ⋆ OX
e (ρj − ⌊αλj ⌋)Dj ,
P
where R = ρj Dj is the zero divisor of the jacobian Jµ of the blow-up map. In
this context, we get the following important vanishing theorem, which can be seen
as a generalization of the Kawamata-Viehweg vanishing theorem (see [Kaw82], [Vie82],
[EV86]).

(4.9) Nadel vanishing theorem ([Nad89], [Dem93]). Let (X, ω) be a Kähler weakly
pseudoconvex manifold, and let F be a holomorphic line bundle over X equipped with a
singular hermitian metric of weight ϕ. Assume that iΘ(F ) > εω for some continuous
positive function ε on X. Then

H q X, O(KX + F ) ⊗ I(ϕ) = 0 for all q > 1.

Proof. In virtue of Hörmander’s L2 estimates applied on small balls, the ∂-complex of


L2loc (n, q)-forms is a (fine) resolution of the sheaf O(KX + F ) ⊗ I(ϕ). The global L2
cohomology is also zero by the L2 estimates applied globally on X.
(4.10) Corollary. Let x1 , . . . , xN be isolated points in the zero variety V (I(ϕ)). Then
there is a surjective map
M 
H 0 (X, KX + F ) −→
−→ O(KX + L) ⊗ OX /I(ϕ) xj
.
16j6N

In particular, if the weight function ϕ is such that ν(ϕ, x) > n + s at some point x ∈ X
and ν(ϕ, y) < 1 at nearby points, then H 0 (X, KX + F ) generates all s-jets at x.

(4.11) Remark. It is an easy exercise (left to the reader!) to show that Corollary (4.10)
implies the Kodaira embedding theorem.

5. Numerical criteria for very ample line bundles


The simplest approach to this problem is a recent technique due to [Siu94], which
rests merely on Nadel’s vanishing theorem and the Riemann-Roch formula. We formulate
here a slightly improved version (see also [Dem94, 95]).

(5.1) Theorem. Let L be an ample line bundle on a projective n-fold X. Let xj ∈ X


and sj ∈ N be given, 1 6 j 6 N . For

X 3n + 2sj − 1
m > m0 = 2 +
n
16j6N
8 L2 -Methods and Effective Results in Algebraic Geometry

H 0 (X, 2KX +mL) generates simultaneously jets


 of order sj at all points xj . In particular,
3n+1
2KX + mL is very ample for m > 2 + n .

Proof. By a result of Fujita, KX + mL is ample for m > m0 (in fact Fujita has shown
that KX + mL is nef for m > m + 1 and ample for m > n + 2). The idea is to use a
recursion procedure for the construction of psh weights (ϕν )ν>1 on KX + m0 L such that
(α) the curvature of KX + m0 L is positive definite: i∂∂ϕν > εν ω for some εν > 0, where
ω is the Kähler metric;
(β) ν(ϕν , xj ) > n + sj for all j;
(γ) I(ϕν+1 ) ) I(ϕν ) whenever dim V (I(ϕν )) > 0.

Indeed, Nadel’s vanishing theorem implies

H q (X, O(2KX + mL) ⊗ O/I(ϕν )) = 0 for m > m0 and q > 1.

Hence, h0 = χ is large for some m ∈ [m0 , 2m0 − 1], and the existence of a section σ
vanishing at order 2(n + sj ) at all points xj follows by the Riemann-Roch formula and
an elementary count of dimensions. We then set inductively

ϕν+1 = log(eϕν + e(1−m/2m0 )ψ |σ|1/2 ),

where ψ is a weight for a smooth metric of positive definite curvature on L. Condition


(γ) guarantees that the process stops after a finite number of steps.

One weak point of the above result is that large multiples of L are required. Instead,
we would like to find conditions on L implying that 2(KX + L) is very ample. For this,
we need a convenient measurement of how large L is.

(5.2) Definition. Let L be a numerically effective line bundle, i.e. a line bundle such
that Lp · Y > 0 for all p-dimensional subvarieties Y . For every S ⊂ X, we set

µ(L, S) = min (Lp · Y )1/p ,


Y ∩S6=∅

where Y runs over all p-dimensional subvarieties intersecting S. The main properties of
this invariant are:
• Linearity: ∀k > 0, µ(kL, S) = k µ(L, S) ;
• Nakai-Moı̆shezon criterion: L is ample if and only if µ(L, X) > 0.
(5.3) Theorem ([Dem93]). Let s, m ∈ N, s > 1, m > 2. If L is ample and satisfies

(m − 1) µ(L, X) > 6(n + s)n − s,

then 2KX + mL generates s-jets. Moreover, the result still holds with 6(n + s)n replaced
by 12nn if s = 1; in particular, 2KX + 12nn L is always very ample.
Proof. By Corollary (4.10), the main point is to construct psh weights ϕ that achieve
the desired ideals I(ϕ)xj for the jets. This is done by solving a complex Monge-Ampère
equation  
i
ω + ∂∂ϕ = f, ω = θ(L),
π
Holomorphic Morse inequalities 9

where f is a linear combination of Dirac measures δxj and of a uniform density with
respect to ω n . The solution ϕ does exist by the Aubin-Calabi-Yau theorem, but in
general, the poles of ϕ are not isolated. Hence, the Lelong numbers have to be estimated
precisely: this is indeed possible by means of intersection inequalities for positive currents.
We refer to [Dem93, 94] for details.

6. Holomorphic Morse inequalities


The starting point is the following differential geometric asymptotic inequality, in
which X(6q, L) denotes the set of points x ∈ X at which θh (L)(x) has at most q negative
eigenvalues. The proof is obtained by a careful study of the spectrum of the complex
Laplace operator ∆′′ . See [Dem85, 91] for details.

(6.1) Strong Morse inequalities ([Dem85]). Let X be a compact complex n-fold and
(L, h) a hermitian line bundle. Then, as k → +∞,
X Z
q−j j kn n
(−1) h X, kL) 6 (−1)q θh (L) + o(k n ).
n! X(6q,L)
06j6q

(6.2) Special case (algebraic version). Let L = F −G, where F and G are numerically
effective. Then for all q = 0, 1, . . . , n,
X  
q−j j kn X q−j n
(−1) h (X, kL) 6 (−1) F n−j · Gj + o(k n ).
n! j
06j6q 06j6q

In particular, for q = 1 we get

kn n
h0 (X, kL) − h1 (X, kL) > (F − nF n−1 · G) − o(k n ).
n!

(6.3) Corollary. If F, G are nef and F n > 0, then k(mF − G) has sections as soon as
m > nF n−1 · G/F n and k ≫ 0.
(6.4) Corollary. If F, G are nef and F n > 0, then H 0 (X, KX + mF − G) 6= 0 for some
m 6 nF n−1 · G/F n + n + 1.
Proof. Set m0 := ⌊nF n−1 · G/F n ⌋ + 1. By Corollary (6.3), m0 F − G has a psh weight ϕ
with i∂∂ϕ ≫ 0; thus, H q (X, O(KX + mF − G) ⊗ I(ϕ)) = 0 for q > 1 and m > m0 . The
Hilbert polynomial is thus equal to

h0 (X, O(KX + mF − G) ⊗ I(ϕ)) > 0,

and it must be nonzero for some m ∈ [m0 , m0 + n] because there are at most n roots.

A similar proof yields

(6.5) Corollary. If F, G are nef with F n > 0, and Y is a p-dimensional subvariety, then
H 0 (Y, ωY ⊗ OY (mF − G)) 6= 0 for some m 6 pF p−1 · G · Y /F p · Y + p + 1, where ωY is
the L2 dualizing sheaf of Y .
10 L2 -Methods and Effective Results in Algebraic Geometry

A proof by backward induction on dim Y then yields the following effective version of
the big Matsusaka theorem ([Mat72], [KoM83]), improving Siu’s result [Siu93].

(6.6) Theorem ([Siu93], [Dem94, 95]) Let F and G be nef line bundles on a projective
n-fold X. Assume
 that F is ample and set H = λn (KX + (n + 2)F ) with λ2 = 1 and
λn = 3n+1
n − 2n for n > 3. Then mF − G is very ample for

n−1 n−2
n−1
(3n−1 −1)/2 (F · (G + H))(3 +1)/2 (F n−1 · H)3 (n/2−3/4)−1/4
m > (2n) .
(F n )3n−2 (n/2−1/4)+1/4

In particular mF is very ample for

 3n−2 (n/2+3/4)+1/4
n−2 F n−1 · KX
m > Cn (F n )3 n+2+
Fn
n−1 n−2
−1)/2
with Cn = (2n)(3 (λn )3 (n/2+3/4)+1/4
.

References

[AV65] A. Andreotti and E. Vesentini, Carleman estimates for the Laplace-Beltrami


equation in complex manifolds, Publ. Math. I.H.E.S. 25 (1965) 81–130.
[Bom70] E. Bombieri, Algebraic values of meromorphic maps, Invent. Math. 10 (1970),
267–287; Addendum, Invent. Math. 11 (1970), 163–166.
[Dem85] J.-P. Demailly, Champs magnétiques et inégalités de Morse pour la d′′ -coho-
mologie, Ann. Inst. Fourier (Grenoble) 35 (1985) 189–229.
[Dem89] J.-P. Demailly, Une généralisation du théorème d’annulation de Kawamata-
Viehweg, C. R. Acad. Sci. Paris Série I Math. 309 (1989), 123–126.
[Dem90] J.-P. Demailly, Singular hermitian metrics on positive line bundles, in: Proc.
Complex Algebraic Varieties, Bayreuth, April 1990 (K. Hulek, Th. Peternell,
M. Schneider and F. Schreyer, eds.), Lecture Notes in Math. 1507, Springer-
Verlag, Berlin and New York, 1992.
[Dem91] J.-P. Demailly, Holomorphic Morse inequalities, Proc. Sympos. Pure Math.
52 (1991), 93–114.
[Dem93] J.-P. Demailly, A numerical criterion for very ample line bundles, J. Differen-
tial Geom. 37 (1993), 323–374.
[Dem94] J.-P. Demailly, L2 vanishing theorems for positive line bundles and adjunction
theory, in: Lecture notes of a CIME course, Transcendental Methods in
Algebraic Geometry, Cetraro, Italy, July 1994.
[Dem95] J.-P. Demailly, Effective bounds for very ample line bundles, to appear in the
special volume of Invent. Math. in honor of R. Remmert, 1995.
[EL92] L. Ein and R. Lazarsfeld, Seshadri constants on smooth surfaces, Journées de
Géométrie Algébrique d’Orsay, July 1992, Astérisque 282 (1993), 177–186.
[EL93] L. Ein and R. Lazarsfeld, Global generation of pluricanonical and adjoint linear
series on smooth projective threefolds, J. Amer. Math. Soc. 6 (1993), 875–903.
Holomorphic Morse inequalities 11

[EV86] H. Esnault and E. Viehweg, Logarithmic De Rham complexes and vanishing


theorems, Invent. Math. 86 (1986) 161–194.
[Fuj87] T. Fujita, On polarized manifolds whose adjoint bundles are not semipositive,
in: Algebraic Geometry, Sendai, 1985 (T. Oda, ed.), Adv. Stud. Pure Math.
10, North-Holland, Amsterdam and New York, 1987, 167–178.
[Fuj94] T. Fujita, Remarks on Ein-Lazarsfeld criterion of spannedness of adjoint
bundles of polarized threefold, preprint, Tokyo Inst. of Technology at Oh-
okayama, 1994.
[Hö65] L. Hörmander, L2 -estimates and existence theorems for the ∂ operator, Acta
Math. 113 (1965), 89–152.
[Kaw82] Y. Kawamata, A generalization of Kodaira-Ramanujam’s vanishing theorem,
Math. Ann. 261 (1982), 43–46.
[Kol93] J. Kollár, Effective base point freeness, Math. Ann. 296 (1993), 595–605.
[KoM83] J. Kollár and T. Matsusaka, Riemann-Roch type inequalities, Amer. J. Math.
105 (1983), 229–252.
[Laz93] R. Lazarsfeld, with the assistance of G. Fernández del Busto, Lectures on
linear series, Park City, IAS Math. Series, vol. 3, (1993).
[Mat72] T. Matsusaka, Polarized varieties with a given Hilbert polynomial, Amer. J.
Math. 94 (1972), 1027–1077.
[Nad89] A.M. Nadel, Multiplier ideal sheaves and Kähler-Einstein metrics of positive
scalar curvature, Proc. Nat. Acad. Sci. U.S.A. 86 (1989), 7299–7300, and Ann.
of Math. (2), 132 (1990), 549–596.
[Ram74] C.P. Ramanujam, Remarks on the Kodaira vanishing theorem, J. Indian Math.
Soc. (New Ser.) 36 (1972), 41–50; 38 (1974), 121–124.
[Rei88] I. Reider, Vector bundles of rank 2 and linear systems on algebraic surfaces,
Ann. of Math. 127 (1988), 309–316.
[Siu93] Y.T. Siu, An effective Matsusaka big theorem, Ann. Inst. Fourier (Grenoble)
43 (1993), 1387–1405.
[Siu94] Y.T. Siu, Effective very ampleness, preprint (1994), to appear in Invent. Math.
1995.
[Sk72] H. Skoda, Sous-ensembles analytiques d’ordre fini ou infini dans Cn , Bull. Soc.
Math. France 100 (1972), 353–408.
[Vie82] E. Viehweg, Vanishing theorems, J. Reine Angew. Math. 335 (1982), 1–8.

Common questions

Powered by AI

Very ample line bundles are significant because they allow for an embedding of a complex manifold into projective space, expressing the manifold as an algebraic variety. Numerically, very ample line bundles can be detected using the Riemann-Roch formula and Morse inequalities in combination with the Nadel vanishing theorem. Specifically, very ampleness is tied to the ability of the line bundle to separate points and tangents on the manifold, which is assured if it has sufficient sections. This can be numerically estimated through their intersection numbers and positivity properties .

A multiplier ideal sheaf, denoted as I(ϕ), consists of holomorphic functions whose local integrals against the weight defined by a plurisubharmonic function ϕ are finite. This sheaf is coherent and analytically captures the non-integrability locus of the plurisubharmonic weights, directly relating to the sublevel sets of the Lelong numbers of the function ϕ. Multiplier ideal sheaves are thus pivotal in understanding the measure concentration behavior in complex geometry .

The desingularization process replaces a singular locus with a simpler one, allowing the study of divisors in a 'normal crossings' situation, which simplifies the handling of singularities. Specifically, applying desingularization to a divisor D allows for the expression of the associated multiplier ideal sheaf I(ϕ) in terms of an invertible sheaf and normal crossing divisors. This transformation is critical for performing vanishing theorems and calculations involving the multiplier ideal, as it ensures analytically manageable singularities and thus aids in achieving coherent and finer estimates .

Pseudoconvexity is vital in applying L2-methods because it ensures the existence of a plurisubharmonic exhaustion function necessary for complex analysis on manifolds. In particular, weakly pseudoconvex manifolds admit a complete Kähler metric, making it feasible to use L2 existence theorems for establishing the existence of certain forms and cohomology elements, pivotal for extending L2-methods beyond compact settings to more general pseudoconvex open sets and Stein manifolds .

The Kodaira embedding theorem states that a holomorphic line bundle over a compact complex manifold is very ample if it can be used to embed the manifold into projective space. This embedding requires that the line bundle has a smooth Hermitian metric with a positive definite curvature form. In essence, the theorem links the positivity property of line bundles to their ability to provide projective embeddings, thus allowing the manifold to be represented as an algebraic variety .

Lelong numbers quantify the amount of mass a plurisubharmonic function has concentrated at a singular point. They are defined as the limit of the ratio of the function to the logarithm of the distance to the singularity. A plurisubharmonic function is considered to have a logarithmic pole of coefficient γ at a point x if its Lelong number at that point equals γ. This indicates that the function behaves essentially like a logarithmic potential near x .

The Nadel vanishing theorem leverages multiplier ideal sheaves to establish vanishing results for higher cohomology groups of twisted canonical bundles. Mulitplier ideal sheaves act as an intermediary to handle singularities by ensuring that the sheaf O(KX + F) ⊗ I(ϕ), where ϕ is the weight function of a singular Hermitian metric, vanishes for all q ≥ 1 if the curvature iΘ(F) is greater than or equal to εω for some positive ε .

Singular metrics in the context of line bundles on Kähler manifolds allow for the extension of L2 methods to broader settings, including weakly pseudoconvex manifolds. These metrics, when equipped on line bundles, ensure that the curvature form is positive in the sense of currents. This is crucial for the application of the L2 existence theorem, which provides weak L2 limits of solutions that satisfy the desired estimates on compact manifolds and pseudoconvex sets .

The construction of plurisubharmonic weights is crucial for applying holomorphic Morse inequalities, which are used to count sections of line bundles on complex manifolds. These inequalities depend on the eigenvalues of curvature forms of line bundles, which are directly influenced by the choice of plurisubharmonic weights. Effective construction can yield precise estimates for the number of sections, as these weights control the asymptotic growth properties of line bundle cohomology, thus refining the counts predicted by Morse inequalities .

Nef line bundles ensure that the intersection numbers with curves on the manifold are non-negative, playing a crucial role in vanishing theorems and the numerical study of divisors. They are instrumental in the construction and analysis of more complex bundles, as exemplified in the proofs of vanishing theorems and very ampleness criteria in projective geometry, where they help in establishing the existence of sections under various conditions .

You might also like