This paper was converted on www.awesomepapers.org from LaTeX by an anonymous user.
Want to know more? Visit the Converter page.

Kollár’s Package for Twisted Saito’s S-sheaves

Junchao Shentu [email protected] School of Mathematical Sciences, University of Science and Technology of China, Hefei, 230026, China  and  Chen Zhao [email protected] School of Mathematical Sciences, University of Science and Technology of China, Hefei, 230026, China
Abstract.

We generalize Kollár’s conjecture (including torsion freeness, injectivity theorem, vanishing theorem and decomposition theorem) to Saito’s SS-sheaves twisted by a \mathbb{Q}-divisor. This gives a uniform treatment for various kinds of Kollár’s package in different topics in complex geometry. As a consequence we prove Kollár’s package of pluricanonical bundles twisted by a certain multiplier ideal sheaf. The method of the present paper is L2L^{2}-theoretic.

1. Introduction

Let f:XYf:X\rightarrow Y be a proper morphism between complex spaces111All the complex spaces are assumed to be separated, reduced, paracompact, countable at infinity and of pure dimension throughout the present paper. We would like to point out that the complex spaces are allowed to be reducible. such that YY is irreducible and each irreducible component of XX is mapped onto YY. We say that a coherent sheaf \mathscr{F} on XX satisfies Kollár’s package with respect to ff if the following statements hold.

Torsion Freeness:

Rqf()R^{q}f_{\ast}(\mathscr{F}) is torsion free for every q0q\geq 0 and vanishes if q>dimXdimYq>\dim X-\dim Y.

Injectivity Theorem:

If LL is a semipositive holomorphic line bundle on XX so that LlL^{\otimes l} admits a nonzero holomorphic global section ss for some l>0l>0, then the canonical morphism

Rqf(×s):Rqf(Lk)Rqf(L(k+l))R^{q}f_{\ast}(\times s):R^{q}f_{\ast}(\mathscr{F}\otimes L^{\otimes k})\to R^{q}f_{\ast}(\mathscr{F}\otimes L^{\otimes(k+l)})

is injective for every q0q\geq 0 and every k1k\geq 1.

Vanishing Theorem:

If YY is a projective algebraic variety and LL is an ample line bundle on YY, then

Hq(Y,Rpf()L)=0,q>0,p0.H^{q}(Y,R^{p}f_{\ast}(\mathscr{F})\otimes L)=0,\quad\forall q>0,\forall p\geq 0.
Decomposition Theorem:

Assume moreover that XX is a compact Kähler space. Then Rf()Rf_{\ast}(\mathscr{F}) splits in the derived category D(Y)D(Y) of 𝒪Y\mathscr{O}_{Y}-modules, i.e.

Rf()qRqf()[q]D(Y).Rf_{\ast}(\mathscr{F})\simeq\bigoplus_{q}R^{q}f_{\ast}(\mathscr{F})[-q]\in D(Y).

As a consequence, the spectral sequence

E2pq:Hp(Y,Rqf())Hp+q(X,)E^{pq}_{2}:H^{p}(Y,R^{q}f_{\ast}(\mathscr{F}))\Rightarrow H^{p+q}(X,\mathscr{F})

degenerates at the E2E_{2} page.

These statements date back to J. Kollár [Kollar1986_1, Kollar1986_2], who shows that the dualizing sheaf ωX\omega_{X} satisfies Kollár’s package when XX is smooth projective and YY is projective. Aiming at various geometric applications, Kollár’s results have been further generalized in two directions.

The first direction is to show Kollár’s package for the dualizing sheaf twisted by a \mathbb{Q}-divisor, or a multiplier ideal sheaf more generally. These kinds of (ad-hoc) Kollár’s package have applications in E. Viehweg’s works on the quasi-projective moduli of polarized manifolds [Viehweg1995, Viehweg2010], O. Fujino’s project of minimal model program for log-canonical varieties [Fujino2017] and Kollár-Kovács’ splitting criterion for du Bois singularities [Kollar2010], etc. Besides, K. Takegoshi [Takegoshi1995] proves Kollár’s package for the dualizing sheaf twisted by a Nakano semipositive vector bundle. The injectivity theorem for the dualizing sheaf twisted by a general multiplier ideal sheaf has been investigated by S. Matsumura [Matsumura2018] and Fujino-Matsumura [Matsumura2021]. However the complete proof of the Kollár’s package (listed as above) for the dualizing sheaf twisted by a multiplier ideal sheaf is still missing.

The other direction is to generalize Kollár’s package to certain Hodge theoretic objects such as variations of Hodge structure and Hodge modules. Assume that f:XYf:X\to Y is a morphism between projective varieties. Let 𝕍\mathbb{V} be an \mathbb{R}-polarized variation of Hodge structure on some dense Zariski open subset XoXregX^{o}\subset X_{\rm reg} of the regular loci XregX_{\rm reg}. M. Saito [MSaito1991] constructs a coherent sheaf SX(𝕍)S_{X}(\mathbb{V}) and shows that SX(𝕍)S_{X}(\mathbb{V}) satisfies Kollár’s package with respect to ff. When 𝕍\mathbb{V} is the trivial variation of Hodge structure, SX(𝕍)ωXS_{X}(\mathbb{V})\simeq\omega_{X}. Saito’s work gives an affirmative answer to Kollár’s conjecture [Kollar1986_2, §4]. Together with other deep results of Hodge modules, Kollár’s package for SX(𝕍)S_{X}(\mathbb{V}) plays important roles in the series works of Popa-Schnell [PS2013, PS2014, PS2017]. Recently the authors of the present paper [SC2021_kollar] generalize Saito’s result to the context of non-abelian Hodge theory.

The purpose of the present article is to show that Kollár’s package holds for Saito’s SS-sheaves twisted by a multiplier ideal sheaf associated with a \mathbb{Q}-divisor. This gives a uniform and systematic treatment for various Kollár’s package twisted by a \mathbb{Q}-divisor. This package contains new results even in the case of the dualizing sheaf twisted by a multiplier ideal sheaf. The main tool is the abstract Kollár’s package established in [SC2021_kollar].

1.1. Main result

Before stating the main results let us recall Saito’s construction of SX(𝕍)S_{X}(\mathbb{V}), with two generalizations:

  1. (1)

    We generalize Saito’s construction to complex variations of Hodge structure. In particular we do not make assumptions on the local monodromy. This is interesting in the view of nonabelian Hodge theory because complex variations of Hodge structure are precisely the \mathbb{C}^{\ast} fixed points on the moduli space of certain tame harmonic bundles ([Simpson1990, Theorem 8], [Mochizuki2006, Proposition 1.9]).

  2. (2)

    We generalize Saito’s construction with respect to the Deligne-Manin prolongations of the variation of Hodge structure with indices other than (1,0](-1,0]. This is a combination of Saito’s SX(𝕍)S_{X}(\mathbb{V}) with the multiplier ideal sheaf associated to a boundary \mathbb{Q}-divisor.

Let XX be a complex space. Let 𝕍=(𝒱,,{𝒱p,q},Q)\mathbb{V}=(\mathcal{V},\nabla,\{\mathcal{V}^{p,q}\},Q) be a polarized complex variation of Hodge structure (Definition 3.1) on some dense regular Zariski open subset XoX^{o} of XX. Let AA be an effective \mathbb{Q}-Cartier divisor on XX. We define a coherent sheaf SX(𝕍,A)S_{X}(\mathbb{V},-A) as follows.

Log smooth case:

Assume that XX is smooth, E:=X\XoE:=X\backslash X^{o} is a simple normal crossing divisor and supp(A)E{\rm supp}(A)\subset E. Denote by E=i=1lEiE=\cup_{i=1}^{l}E_{i} the irreducible decomposition and denote A=i=1lriEiA=\sum_{i=1}^{l}{r_{i}}E_{i} with r1,,rl0r_{1},\dots,r_{l}\in\mathbb{Q}_{\geq 0}. Let 𝒓=(r1,,rl)\bm{r}=(r_{1},\dots,r_{l}). Let 𝒱>𝒓1\mathcal{V}_{>\bm{r}-1} be the Deligne-Manin prolongation with indices >𝒓1>\bm{r}-1. It is a locally free 𝒪X\mathscr{O}_{X}-module extending 𝒱\mathcal{V} such that \nabla induces a connection with logarithmic singularities

:𝒱>𝒓1𝒱>𝒓1ΩX(logE)\nabla:\mathcal{V}_{>\bm{r}-1}\to\mathcal{V}_{>\bm{r}-1}\otimes\Omega_{X}(\log E)

where the real part of the eigenvalues of the residue of \nabla along EiE_{i} belongs to (ri1,ri](r_{i}-1,r_{i}] for each ii. Let j:XoXj:X^{o}\to X be the open immersion. Denote S(𝕍):=𝒱pmax,kpmaxS(\mathbb{V}):=\mathcal{V}^{p_{\rm max},k-p_{\rm max}} where pmax=max{p|𝒱p,kp0}p_{\rm max}=\max\{p|\mathcal{V}^{p,k-p}\neq 0\}. Define

SX(𝕍,A):=ωX(jS(𝕍)𝒱>𝒓1).S_{X}(\mathbb{V},-A):=\omega_{X}\otimes\left(j_{\ast}S(\mathbb{V})\cap\mathcal{V}_{>\bm{r}-1}\right).
General case:

Let π:X~X\pi:\widetilde{X}\to X be a proper bimeromorphic morphism such that πo:=π|π1(Xo\supp(A)):π1(Xo\supp(A))Xo\supp(A)\pi^{o}:=\pi|_{\pi^{-1}(X^{o}\backslash{\rm supp}(A))}:\pi^{-1}(X^{o}\backslash{\rm supp}(A))\to X^{o}\backslash{\rm supp}(A) is biholomorphic and the exceptional loci E:=π1((X\Xo)supp(A)))E:=\pi^{-1}((X\backslash X^{o})\cup{\rm supp}(A))) is a simple normal crossing divisor. Then

(1.1) SX(𝕍,A)π(SX~(πo𝕍,πA)).\displaystyle S_{X}(\mathbb{V},-A)\simeq\pi_{\ast}\left(S_{\widetilde{X}}(\pi^{o\ast}\mathbb{V},-\pi^{\ast}A)\right).

When A=A=\emptyset, SX(𝕍,)S_{X}(\mathbb{V},\emptyset) is canonically isomorphic to Saito’s SX(𝕍)S_{X}(\mathbb{V}) (see [MSaito1991], at least when 𝕍\mathbb{V} is \mathbb{R}-polarizable). The main result of the present article is

Theorem 1.1.
  1. (1)

    SX(𝕍,A)S_{X}(\mathbb{V},-A) is a torsion free coherent sheaf on XX which is independent of the choice of the desingularization π:X~X\pi:\widetilde{X}\to X.

  2. (2)

    Let f:XYf:X\to Y be a locally Kähler proper morphism between complex spaces such that YY is irreducible and each irreducible component of XX is mapped onto YY. Let LL be a line bundle on XX such that some multiple mL=B+DmL=B+D where BB is a semipositive line bundle and DD is an effective Cartier divisor on XX. Let FF be an arbitrary Nakano-semipositive vector bundle on XX. Then SX(𝕍,1mD)FLS_{X}(\mathbb{V},-\frac{1}{m}D)\otimes F\otimes L satisfies Kollár’s package with respect to ff.

1.2. Multiplier Grauert-Riemenschneider sheaf

When 𝕍=Xreg\mathbb{V}=\mathbb{C}_{X_{\rm reg}} is the trivial variation of Hodge structure, SX(Xreg,A)S_{X}(\mathbb{C}_{X_{\rm reg}},-A) is exactly the Grauert-Riemenschneider sheaf twisted by the multiplier ideal sheaf associated with AA. This is called the multiplier ideals by Viehweg [Viehweg1995, Viehweg2010] and it also appear in the Nadel vanishing theorem on complex spaces [Demailly2012]. Let us recall its construction for the convenience of readers.

Log smooth case:

Assume that XX is smooth and supp(A){\rm supp}(A) is a simple normal crossing divisor. Then

𝒦X(A):=ωX𝒪X(A).\mathscr{K}_{X}(-A):=\omega_{X}\otimes\mathscr{O}_{X}(-\lfloor A\rfloor).
General case:

Let π:X~X\pi:\widetilde{X}\to X be a proper bimeromorphic morphism such that πo:=π|π1(Xo\supp(A)):π1(Xo\supp(A))Xo\supp(A)\pi^{o}:=\pi|_{\pi^{-1}(X^{o}\backslash{\rm supp}(A))}:\pi^{-1}(X^{o}\backslash{\rm supp}(A))\to X^{o}\backslash{\rm supp}(A) is biholomorphic and the exceptional loci E:=π1((X\Xo)supp(A)))E:=\pi^{-1}((X\backslash X^{o})\cup{\rm supp}(A))) is a simple normal crossing divisor. Then

(1.2) 𝒦X(A):=π(𝒦X~(πA)).\displaystyle\mathscr{K}_{X}(-A):=\pi_{\ast}\left(\mathscr{K}_{\widetilde{X}}(-\pi^{\ast}A)\right).

Certainly 𝒦X(A)SX(Xreg,A)\mathscr{K}_{X}(-A)\simeq S_{X}(\mathbb{C}_{X_{\rm reg}},-A) and one has

𝒦X(A)ωX(A)\mathscr{K}_{X}(-A)\simeq\omega_{X}\otimes\mathscr{I}(-A)

when XX is smooth ((A)\mathscr{I}(-A) is the multiplier ideal sheaf associated with AA). In this case, by Theorem 1.1 one has the following.

Theorem 1.2.

Let f:XYf:X\to Y be a locally Kähler proper morphism between complex spaces, such that YY is irreducible and each irreducible component of XX is mapped onto YY. Let LL be a line bundle such that some multiple mL=B+DmL=B+D where BB is a semipositive line bundle and DD is an effective Cartier divisor on XX. Let FF be an arbitrary Nakano-semipositive vector bundle on XX. Then 𝒦X(1mD)FL\mathscr{K}_{X}(-\frac{1}{m}D)\otimes F\otimes L satisfies Kollár’s package with respect to ff.

Theorem 1.2 has an application to Kollár’s package of pluricanonical bundles.

Corollary 1.3.

Let f:XYf:X\to Y be a morphism from a compact Kähler manifold to an analytic space. Assume that ωXkmA𝒪X(D)\omega_{X}^{\otimes km}\simeq A\otimes\mathscr{O}_{X}(D), k,m>0k,m>0 where AA is a semipositive line bundle (e.g. a semiample line bundle) and DD is an effective Cartier divisor. Let FF be an arbitrary Nakano-semipositive vector bundle on XX. Then 𝒦X(1mD)ωXkF\mathscr{K}_{X}(-\frac{1}{m}D)\otimes\omega^{\otimes k}_{X}\otimes F satisfies Kollár’s package with respect to ff. In particular if ωX\omega_{X} is semipositive, then ωXkF\omega_{X}^{\otimes k}\otimes F satisfies Kollár’s package with respect to ff for every k1k\geq 1.

2. Abstract Kollár’s package

In this section we recall the abstract Kollár’s package established in [SC2021_kollar].

Let XX be a complex space of dimension nn and XoXregX^{o}\subset X_{\rm reg} a dense Zariski open subset. Let (E,h)(E,h) be a hermitian vector bundle on XoX^{o}. Define the 𝒪X\mathscr{O}_{X}-module SX(E,h)S_{X}(E,h) as follows. Let UXU\subset X be an open subset. SX(E,h)(U)S_{X}(E,h)(U) is the space of holomorphic EE-valued (n,0)(n,0)-forms α\alpha on UXoU\cap X^{o} such that for every point xUx\in U, there is a neighborhood VxV_{x} of xx so that

VxXoαα¯<.\int_{V_{x}\cap X^{o}}\alpha\wedge\overline{\alpha}<\infty.
Lemma 2.1.

(Functoriality,[SC2021_kollar, Proposition 2.5]) Let π:XX\pi:X^{\prime}\to X be a proper holomorphic map between complex spaces which is biholomorphic over XoX^{o}. Then

πSX(πE,πh)=SX(E,h).\pi_{\ast}S_{X^{\prime}}(\pi^{\ast}E,\pi^{\ast}h)=S_{X}(E,h).
Lemma 2.2.

([SC2021_kollar, Lemma 2.6]) Let (F,hF)(F,h_{F}) be a hermitian vector bundle on XX (in particular hFh_{F} is smooth on XX). Then

SX(E,h)FSX(EF,hhF).S_{X}(E,h)\otimes F\simeq S_{X}(E\otimes F,h\otimes h_{F}).
Definition 2.3.

(E,h)(E,h) is tame on XX if, for every point xXx\in X, there is an open neighborhood UU containing xx, a proper bimeromorphic morphism π:U~U\pi:\widetilde{U}\to U which is biholomorphic over UXoU\cap X^{o}, and a hermitian vector bundle (Q,hQ)(Q,h_{Q}) on U~\widetilde{U} such that

  1. (1)

    πE|π1(XoU)Q|π1(XoU)\pi^{\ast}E|_{\pi^{-1}(X^{o}\cap U)}\subset Q|_{\pi^{-1}(X^{o}\cap U)} as a subsheaf.

  2. (2)

    There is a hermitian metric hQh^{\prime}_{Q} on Q|π1(XoU)Q|_{\pi^{-1}(X^{o}\cap U)} so that hQ|πEπhh^{\prime}_{Q}|_{\pi^{\ast}E}\sim\pi^{\ast}h on π1(XoU)\pi^{-1}(X^{o}\cap U) and

    (2.1) (i=1rπfi2)chQhQ\displaystyle(\sum_{i=1}^{r}\|\pi^{\ast}f_{i}\|^{2})^{c}h_{Q}\lesssim h^{\prime}_{Q}

    for some cc\in\mathbb{R}. Here {f1,,fr}\{f_{1},\dots,f_{r}\} is an arbitrary set of local generators of the ideal sheaf defining U~\π1(Xo)U~\widetilde{U}\backslash\pi^{-1}(X^{o})\subset\widetilde{U}.

The tameness condition (2.1) is independent of the choice of the set of local generators. In the present paper, a tame hermitian vector bundle (E,h)(E,h) is constructed as a subsheaf of the underlying holomorphic bundle of a variation of Hodge structure. This is a special case of tame harmonic bundles in the sense of Simpson [Simpson1990] and Mochizuki [Mochizuki20072, Mochizuki20071]. In this case, Condition (2.1) comes from the theory of degeneration of variation of Hodge structure [Cattani_Kaplan_Schmid1986].

Theorem 2.4.

([SC2021_kollar, Proposition 2.9 and §4]) Let f:XYf:X\rightarrow Y be a proper locally Kähler morphism from a complex space XX to an irreducible complex space YY. Assume that every irreducible component of XX is mapped onto YY, XoXregX^{o}\subset X_{\rm reg} is a dense Zariski open subset and (E,h)(E,h) is a hermitian vector bundle on XoX^{o} with Nakano semipositive curvature. Assume that (E,h)(E,h) is tame on XX. Then SX(E,h)S_{X}(E,h) is a coherent sheaf which satisfies Kollár’s package with respect to f:XYf:X\to Y.

3. Twisted Saito’s S-sheaf and its Kollár package

3.1. Complex variation of Hodge structure

Definition 3.1.

[Simpson1988, §8] Let XoX^{o} be a complex manifold. Denote by 𝒜Xo0\mathscr{A}^{0}_{X^{o}} the sheaf of CC^{\infty} functions on XoX^{o}. A polarized complex variation of Hodge structure on XoX^{o} of weight kk is a flat holomorphic connection (𝒱,)(\mathcal{V},\nabla) on XoX^{o} together with a decomposition 𝒱𝒪Xo𝒜Xo0=p+q=k𝒱p,q\mathcal{V}\otimes_{\mathscr{O}_{X^{o}}}\mathscr{A}^{0}_{X^{o}}=\bigoplus_{p+q=k}\mathcal{V}^{p,q} of CC^{\infty} bundles and a flat hermitian form QQ on 𝒱\mathcal{V} such that

  1. (1)

    The hermitian form hQh_{Q} which equals (1)pQ(-1)^{p}Q on 𝒱p,q\mathcal{V}^{p,q} is a hermitian metric on the CC^{\infty} complex vector bundle 𝒱𝒪Xo𝒜Xo0\mathcal{V}\otimes_{\mathscr{O}_{X^{o}}}\mathscr{A}^{0}_{X^{o}}.

  2. (2)

    The decomposition 𝒱𝒪Xo𝒜Xo0=p+q=k𝒱p,q\mathcal{V}\otimes_{\mathscr{O}_{X^{o}}}\mathscr{A}^{0}_{X^{o}}=\bigoplus_{p+q=k}\mathcal{V}^{p,q} is orthogonal with respect to hQh_{Q}.

  3. (3)

    The Griffiths transversality condition

    (3.1) (𝒱p,q)𝒜0,1(𝒱p+1,q1)𝒜1,0(𝒱p,q)𝒜0,1(𝒱p,q)𝒜1,0(𝒱p1,q+1)\displaystyle\nabla(\mathcal{V}^{p,q})\subset\mathscr{A}^{0,1}(\mathcal{V}^{p+1,q-1})\oplus\mathscr{A}^{1,0}(\mathcal{V}^{p,q})\oplus\mathscr{A}^{0,1}(\mathcal{V}^{p,q})\oplus\mathscr{A}^{1,0}(\mathcal{V}^{p-1,q+1})

    holds for every pp and qq. Here 𝒜i,j(𝒱p,q)\mathscr{A}^{i,j}(\mathcal{V}^{p,q}) denotes the sheaf of smooth (i,j)(i,j)-forms with values in 𝒱p,q\mathcal{V}^{p,q}.

Denote S(𝕍):=𝒱pmax,kpmaxS(\mathbb{V}):=\mathcal{V}^{p_{\rm max},k-p_{\rm max}} where pmax=max{p|𝒱p,kp0}p_{\rm max}=\max\{p|\mathcal{V}^{p,k-p}\neq 0\}.

Let XX be a complex manifold and i=1lEi=E:=X\XoX\cup_{i=1}^{l}E_{i}=E:=X\backslash X^{o}\subset X a simple normal crossing divisor where E1,,ElE_{1},\dots,E_{l} are irreducible components. Let (𝒱,,{𝒱p,q},Q)(\mathcal{V},\nabla,\{\mathcal{V}^{p,q}\},Q) be a polarized complex variation of Hodge structure on Xo:=X\EX^{o}:=X\backslash E. There is a system of prolongations of 𝒱\mathcal{V}. Let 𝒂=(a1,,al)l\bm{a}=(a_{1},\dots,a_{l})\in\mathbb{R}^{l}. Let 𝒱>𝒂\mathcal{V}_{>\bm{a}} be the Deligne-Manin prolongation with indices >𝒂>\bm{a}. It is a locally free 𝒪X\mathscr{O}_{X}-module extending 𝒱\mathcal{V} such that \nabla induces a connection with logarithmic singularities

:𝒱>𝒂𝒱>𝒂ΩX(logE)\nabla:\mathcal{V}_{>\bm{a}}\to\mathcal{V}_{>\bm{a}}\otimes\Omega_{X}(\log E)

whose real part of the eigenvalues of the residue of \nabla along EiE_{i} belongs to (ai,ai+1](a_{i},a_{i}+1]. Denote

RX(𝕍):=𝒱>𝟏j(S(𝕍))R_{X}(\mathbb{V}):=\mathcal{V}_{>\bm{-1}}\cap j_{\ast}(S(\mathbb{V}))

where j:XoXj:X^{o}\to X is the open immersion and 𝟏=(1,,1)\bm{-1}=(-1,\dots,-1). By the nilpotent orbit theorem [Cattani_Kaplan_Schmid1986] RX(𝕍)R_{X}(\mathbb{V}) is a subbundle of 𝒱>𝟏\mathcal{V}_{>\bm{-1}}, i.e. both RX(𝕍)R_{X}(\mathbb{V}) and 𝒱>𝟏/RX(𝕍)\mathcal{V}_{>\bm{-1}}/R_{X}(\mathbb{V}) are locally free.

3.2. L2L^{2}-adapted local frame on RX(𝕍)R_{X}(\mathbb{V})

Let 𝕍=(𝒱,,{𝒱p,q},Q)\mathbb{V}=(\mathcal{V},\nabla,\{\mathcal{V}^{p,q}\},Q) be a polarized complex variation of Hodge structure over (Δ)n×Δm(\Delta^{\ast})^{n}\times\Delta^{m}. Denote by hQh_{Q} the associated Hodge metric. Let s1,,sns_{1},\dots,s_{n} be holomorphic coordinates of (Δ)n(\Delta^{\ast})^{n} and denote Di:={si=0}Δn+mD_{i}:=\{s_{i}=0\}\subset\Delta^{n+m}. Let NiN_{i} be the unipotent part of ResDi{\rm Res}_{D_{i}}\nabla and let

p:n×Δm(Δ)n×Δm,p:\mathbb{H}^{n}\times\Delta^{m}\to(\Delta^{\ast})^{n}\times\Delta^{m},
(z1,,zn,w1,,wm)(e2π1z1,,e2π1zn,w1,,wm)(z_{1},\dots,z_{n},w_{1},\dots,w_{m})\mapsto(e^{2\pi\sqrt{-1}z_{1}},\dots,e^{2\pi\sqrt{-1}z_{n}},w_{1},\dots,w_{m})

be the universal covering. Let W(1)=W(N1),,W(n)=W(N1++Nn)W^{(1)}=W(N_{1}),\dots,W^{(n)}=W(N_{1}+\cdots+N_{n}) be the monodromy weight filtrations (centered at 0) on V:=Γ(n×Δm,p𝒱)pV:=\Gamma(\mathbb{H}^{n}\times\Delta^{m},p^{\ast}\mathcal{V})^{p^{\ast}\nabla}. The following norm estimate for flat sections is proved by Cattani-Kaplan-Schmid [Cattani_Kaplan_Schmid1986, Theorem 5.21] for the case when 𝕍\mathbb{V} has quasi-unipotent local monodromy and by Mochizuki [Mochizuki20072, Part 3, Chapter 13] for the general case.

Theorem 3.2.

For any 0vGrlnW(n)Grl1W(1)V0\neq v\in{\rm Gr}_{l_{n}}^{W^{(n)}}\cdots{\rm Gr}_{l_{1}}^{W^{(1)}}V, one has

|v|hQ2(log|s1|log|s2|)l1(log|sn|)ln\displaystyle|v|^{2}_{h_{Q}}\sim\left(\frac{\log|s_{1}|}{\log|s_{2}|}\right)^{l_{1}}\cdots\left(-\log|s_{n}|\right)^{l_{n}}

over any region of the form

{(s1,sn,w1,,wm)(Δ)n×Δm|log|s1|log|s2|>ϵ,,log|sn|>ϵ,(w1,,wm)K}\left\{(s_{1},\dots s_{n},w_{1},\dots,w_{m})\in(\Delta^{\ast})^{n}\times\Delta^{m}\bigg{|}\frac{\log|s_{1}|}{\log|s_{2}|}>\epsilon,\dots,-\log|s_{n}|>\epsilon,(w_{1},\dots,w_{m})\in K\right\}

for any ϵ>0\epsilon>0 and an arbitrary compact subset KΔmK\subset\Delta^{m} .

The rest of this part is devoted to the norm estimate of the singular hermitian metric hQh_{Q} on RX(𝕍)R_{X}(\mathbb{V}).

Lemma 3.3.

Assume that n=1n=1. Then W1(N1)RX(𝕍)𝟎=0W_{-1}(N_{1})\cap R_{X}(\mathbb{V})_{\bf 0}=0.

Proof.

Assume that W1(N1)RX(𝕍)𝟎0W_{-1}(N_{1})\cap R_{X}(\mathbb{V})_{\bf 0}\neq 0. Let kk be the weight of 𝕍\mathbb{V}. Let l=max{l|Wl(N1)RX(𝕍)𝟎0}l=\max\{l|W_{-l}(N_{1})\cap R_{X}(\mathbb{V})_{\bf 0}\neq 0\}. Then l1l\geq 1. By [Schmid1973, 6.16], the decomposition 𝒱>𝟏p+q=kj𝒱p,q𝒱>𝟏\mathcal{V}_{>\bm{-1}}\simeq\bigoplus_{p+q=k}j_{\ast}\mathcal{V}^{p,q}\cap\mathcal{V}_{>\bm{-1}} induces a pure Hodge structure of weight m+km+k on Wm(N1)/Wm1(N1)W_{m}(N_{1})/W_{m-1}(N_{1}). Moreover

(3.2) Nl:Wl(N1)/Wl1(N1)Wl(N1)/Wl1(N1)\displaystyle N^{l}:W_{l}(N_{1})/W_{l-1}(N_{1})\to W_{-l}(N_{1})/W_{-l-1}(N_{1})

is an isomorphism of type (l,l)(-l,-l). Denote S(𝕍)=𝒱p,kpS(\mathbb{V})=\mathcal{V}^{p,k-p}. By the definition of ll, any nonzero element αWl(N1)RX(𝕍)𝟎\alpha\in W_{-l}(N_{1})\cap R_{X}(\mathbb{V})_{\bf 0} induces a nonzero [α]Wl(N1)/Wl1(N1)[\alpha]\in W_{-l}(N_{1})/W_{-l-1}(N_{1}) of Hodge type (p,klp)(p,k-l-p). Since (3.2) is an isomorphism, there is βWl(N1)/Wl1(N1)\beta\in W_{l}(N_{1})/W_{l-1}(N_{1}) of Hodge type (p+l,kp)(p+l,k-p) such that Nl(β)=[α]N^{l}(\beta)=[\alpha]. However, β=0\beta=0 since p+l=0\mathcal{F}^{p+l}=0. This contradicts to the fact that [α]0[\alpha]\neq 0. W1(N1)RX(𝕍)𝟎W_{-1}(N_{1})\cap R_{X}(\mathbb{V})_{\bf 0} therefore has to be zero. ∎

Denote by TiT_{i} the local monodromy operator of 𝕍\mathbb{V} around DiD_{i}. Since T1,,TnT_{1},\dots,T_{n} are pairwise commutative, there is a finite decomposition

𝒱>𝟏|𝟎=1<α1,,αn0𝕍α1,,αn\mathcal{V}_{>\bm{-1}}|_{\bf 0}=\bigoplus_{-1<\alpha_{1},\dots,\alpha_{n}\leq 0}\mathbb{V}_{\alpha_{1},\dots,\alpha_{n}}

such that (Tie2π1αiId)(T_{i}-e^{2\pi\sqrt{-1}\alpha_{i}}{\rm Id}) is unipotent on 𝕍α1,,αn\mathbb{V}_{\alpha_{1},\dots,\alpha_{n}} for each i=1,,ni=1,\dots,n.

Let

v1,,vNRX(𝕍)|𝟎1<α1,,αn0𝕍α1,,αnv_{1},\dots,v_{N}\in R_{X}(\mathbb{V})|_{\bf 0}\cap\bigcup_{-1<\alpha_{1},\dots,\alpha_{n}\leq 0}\mathbb{V}_{\alpha_{1},\dots,\alpha_{n}}

be an orthogonal basis of RX(𝕍)|𝟎Γ(n,pS(𝕍))pR_{X}(\mathbb{V})|_{\bf 0}\simeq\Gamma(\mathbb{H}^{n},p^{\ast}S(\mathbb{V}))^{p^{\ast}\nabla}. Then v1~,,vN~\widetilde{v_{1}},\dots,\widetilde{v_{N}} that are determined by

(3.3) vj~:=exp(i=1nlogzi(αiId+Ni))vj if vj𝕍α1,,αn,j=1,,N\displaystyle\widetilde{v_{j}}:={\rm exp}\left(\sum_{i=1}^{n}\log z_{i}(\alpha_{i}{\rm Id}+N_{i})\right)v_{j}\textrm{ if }v_{j}\in\mathbb{V}_{\alpha_{1},\dots,\alpha_{n}},\quad\forall j=1,\dots,N

form a frame of 𝒱>𝟏jS(𝕍)\mathcal{V}_{>\bm{-1}}\cap j_{\ast}S(\mathbb{V}). To be precise, we always use the notation αEi(vj~)\alpha_{E_{i}}(\widetilde{v_{j}}) instead of αi\alpha_{i} in (3.3). By (3.3) we acquire that

|vj~|hQ2\displaystyle|\widetilde{v_{j}}|^{2}_{h_{Q}} |i=1nziαEi(vj~)exp(i=1nNilogzi)vj|hQ2\displaystyle\sim\left|\prod_{i=1}^{n}z_{i}^{\alpha_{E_{i}}(\widetilde{v_{j}})}{\rm exp}\left(\sum_{i=1}^{n}N_{i}\log z_{i}\right)v_{j}\right|^{2}_{h_{Q}}
|vj|hQ2i=1n|zi|2αEi(vj~),j=1,,N\displaystyle\sim|v_{j}|^{2}_{h_{Q}}\prod_{i=1}^{n}|z_{i}|^{2\alpha_{E_{i}}(\widetilde{v_{j}})},\quad j=1,\dots,N

where αEi(vj~)(1,0]\alpha_{E_{i}}(\widetilde{v_{j}})\in(-1,0], i=1,n\forall i=1,\dots n.

By Theorem 3.2 and Lemma 3.3 one has

|vj|hQ2(log|s1|log|s2|)l1(log|sn|)ln,l1l2ln1,\displaystyle|v_{j}|^{2}_{h_{Q}}\sim\left(\frac{\log|s_{1}|}{\log|s_{2}|}\right)^{l_{1}}\cdots\left(-\log|s_{n}|\right)^{l_{n}},\quad l_{1}\leq l_{2}\leq\dots\leq l_{n-1},

over any region of the form

{(s1,sn,w1,,wm)(Δ)n×Δm|log|s1|log|s2|>ϵ,,log|sn|>ϵ,(w1,,wm)K}\left\{(s_{1},\dots s_{n},w_{1},\dots,w_{m})\in(\Delta^{\ast})^{n}\times\Delta^{m}\bigg{|}\frac{\log|s_{1}|}{\log|s_{2}|}>\epsilon,\dots,-\log|s_{n}|>\epsilon,(w_{1},\dots,w_{m})\in K\right\}

for any ϵ>0\epsilon>0 and an arbitrary compact subset KΔmK\subset\Delta^{m}. Therefore we obtain that

1|vj||z1zr|ϵ,ϵ>0.\displaystyle 1\lesssim|v_{j}|\lesssim|z_{1}\cdots z_{r}|^{-\epsilon},\quad\forall\epsilon>0.

The local frame (v1~,,vN~)(\widetilde{v_{1}},\dots,\widetilde{v_{N}}) is L2L^{2}-adapted in the sense of S. Zucker [Zucker1979, page 433].

Definition 3.4.

Let (E,h)(E,h) be a vector bundle with a possibly singular hermitian metric hh on a hermitian manifold (X,ds02)(X,ds^{2}_{0}). A holomorphic local frame (v1,,vN)(v_{1},\dots,v_{N}) of EE is called L2L^{2}-adapted if, for every set of measurable functions {f1,,fN}\{f_{1},\dots,f_{N}\}, i=1Nfivi\sum_{i=1}^{N}f_{i}v_{i} is locally square integrable if and only if fivif_{i}v_{i} is locally square integrable for each i=1,,Ni=1,\dots,N.

To see that (v1~,,vN~)(\widetilde{v_{1}},\dots,\widetilde{v_{N}}) is L2L^{2}-adapted, let us consider the measurable functions f1,,fNf_{1},\dots,f_{N}. If

j=1Nfjvj~=exp(i=1nNilogzi)(j=1Nfji=1n|zi|αEi(vj~)vj)\sum_{j=1}^{N}f_{j}\widetilde{v_{j}}={\rm exp}\left(\sum_{i=1}^{n}N_{i}\log z_{i}\right)\left(\sum_{j=1}^{N}f_{j}\prod_{i=1}^{n}|z_{i}|^{\alpha_{E_{i}}(\widetilde{v_{j}})}v_{j}\right)

is locally square integrable, then

j=1Nfji=1n|zi|αEi(vj~)vj\sum_{j=1}^{N}f_{j}\prod_{i=1}^{n}|z_{i}|^{\alpha_{E_{i}}(\widetilde{v_{j}})}v_{j}

is locally square integrable because the entries of the matrix exp(i=1nNilogzi){\rm exp}\left(-\sum_{i=1}^{n}N_{i}\log z_{i}\right) are LL^{\infty}-bounded. Since (v1,,vN)(v_{1},\dots,v_{N}) is an orthogonal basis, |fjvj~|hQi=1n|zi|αEi(vj~)|fjvj|hQ|f_{j}\widetilde{v_{j}}|_{h_{Q}}\sim\prod_{i=1}^{n}|z_{i}|^{\alpha_{E_{i}}(\widetilde{v_{j}})}|f_{j}v_{j}|_{h_{Q}} is locally square integrable for each j=1,,Nj=1,\dots,N.

In conclusion, we obtain the following proposition.

Proposition 3.5.

Let (X,ds02)(X,ds^{2}_{0}) be a hermitian manifold and EE a normal crossing divisor on XX. Let 𝕍\mathbb{V} be a polarized complex variation of Hodge structure on Xo:=X\EX^{o}:=X\backslash E. Then there is an L2L^{2}-adapted holomorphic local frame (v1~,,vN~)(\widetilde{v_{1}},\dots,\widetilde{v_{N}}) of RX(𝕍)R_{X}(\mathbb{V}) at every point xEx\in E. There are moreover αEi(vj~)(1,0]\alpha_{E_{i}}(\widetilde{v_{j}})\in(-1,0], i=1,,ri=1,\dots,r, j=1,,Nj=1,\dots,N and positive real functions λjC(X\E)\lambda_{j}\in C^{\infty}(X\backslash E), j=1,,Nj=1,\dots,N such that

(3.4) |vj~|2λji=1r|zi|2αEi(vj~),j=1,,N\displaystyle|\widetilde{v_{j}}|^{2}\sim\lambda_{j}\prod_{i=1}^{r}|z_{i}|^{2\alpha_{E_{i}}(\widetilde{v_{j}})},\quad\forall j=1,\dots,N

and

1λj|z1zr|ϵ,ϵ>01\lesssim\lambda_{j}\lesssim|z_{1}\cdots z_{r}|^{-\epsilon},\quad\forall\epsilon>0

for each j=1,,Nj=1,\dots,N. Here z1,,znz_{1},\cdots,z_{n} are holomorphic local coordinates on XX so that Ei={zi=0}E_{i}=\{z_{i}=0\}, i=1,ri=1,\cdots r and E={z1zr=0}E=\{z_{1}\cdots z_{r}=0\}.

3.3. Twisted Saito’s S-sheaf

Let XX be a complex space and XoXregX^{o}\subset X_{\rm reg} a dense Zariski open subset. Let 𝕍=(𝒱,,{𝒱p,q},Q)\mathbb{V}=(\mathcal{V},\nabla,\{\mathcal{V}^{p,q}\},Q) be a polarized complex variation of Hodge structure on XoX^{o}. Let AA be an effective \mathbb{Q}-Cartier divisor on XX. We define a coherent sheaf SX(𝕍,A)S_{X}(\mathbb{V},-A) as follows.

Log smooth case:

Assume that XX is smooth, E:=X\XoE:=X\backslash X^{o} is a simple normal crossing divisor and supp(A)E{\rm supp}(A)\subset E. Denote by E=i=1lEiE=\cup_{i=1}^{l}E_{i} the irreducible decomposition and denote A=i=1lriEiA=\sum_{i=1}^{l}{r_{i}}E_{i} with r1,,rl0r_{1},\dots,r_{l}\in\mathbb{Q}_{\geq 0}. Let 𝒓=(r1,,rl)\bm{r}=(r_{1},\dots,r_{l}). Let 𝒱>𝒓1\mathcal{V}_{>\bm{r}-1} be the Deligne-Manin prolongation with indices >𝒓1>\bm{r}-1. It is a locally free 𝒪X\mathscr{O}_{X}-module extending 𝒱\mathcal{V} such that \nabla induces a connection with logarithmic singularities

:𝒱>𝒓1𝒱>𝒓1ΩX(logE)\nabla:\mathcal{V}_{>\bm{r}-1}\to\mathcal{V}_{>\bm{r}-1}\otimes\Omega_{X}(\log E)

where the real part of the eigenvalues of the residue of \nabla along EiE_{i} belongs to (ri1,ri](r_{i}-1,r_{i}] for each ii. Let j:XoXj:X^{o}\to X be the open immersion. Denote S(𝕍):=𝒱pmax,kpmaxS(\mathbb{V}):=\mathcal{V}^{p_{\rm max},k-p_{\rm max}} where pmax=max{p|𝒱p,kp0}p_{\rm max}=\max\{p|\mathcal{V}^{p,k-p}\neq 0\}. Define

SX(𝕍,A):=ωX(jS(𝕍)𝒱>𝒓1).S_{X}(\mathbb{V},-A):=\omega_{X}\otimes\left(j_{\ast}S(\mathbb{V})\cap\mathcal{V}_{>\bm{r}-1}\right).
General case:

Let π:X~X\pi:\widetilde{X}\to X be a proper bimeromorphic morphism such that πo:=π|π1(Xo\supp(A)):π1(Xo\supp(A))Xo\supp(A)\pi^{o}:=\pi|_{\pi^{-1}(X^{o}\backslash{\rm supp}(A))}:\pi^{-1}(X^{o}\backslash{\rm supp}(A))\to X^{o}\backslash{\rm supp}(A) is biholomorphic and the exceptional loci E:=π1((X\Xo)supp(A)))E:=\pi^{-1}((X\backslash X^{o})\cup{\rm supp}(A))) is a simple normal crossing divisor. Then

(3.5) SX(𝕍,A)π(SX~(πo𝕍,πA)).\displaystyle S_{X}(\mathbb{V},-A)\simeq\pi_{\ast}\left(S_{\widetilde{X}}(\pi^{o\ast}\mathbb{V},-\pi^{\ast}A)\right).

Let LL be a line bundle such that some multiple mL=B+DmL=B+D where BB is a semipositive line bundle and DD is an effective Cartier divisor on XX. Let hBh_{B} be a hermitian metric on BB with semipositive curvature and hDh_{D} the unique singular hermitian metric on 𝒪X(D)\mathscr{O}_{X}(D) determined by the effective divisor DD. hDh_{D} is a singular hermitian metric, smooth over X\DX\backslash D, defined as follows. Let sH0(X,𝒪X(D))s\in H^{0}(X,\mathscr{O}_{X}(D)) be the defining section of DD and let h0h_{0} be an arbitrary smooth hermitian metric on 𝒪X(D)\mathscr{O}_{X}(D). Then hDh_{D} is defined by |ξ|hD=|ξ|h0/|s|h0|\xi|_{h_{D}}=|\xi|_{h_{0}}/|s|_{h_{0}} which is independent of the choice of h0h_{0}. Denote hL:=(hBhD)1mh_{L}:=(h_{B}h_{D})^{\frac{1}{m}}. The main result of this section is

Theorem 3.6.

SX(𝕍,1mD)LSX(S(𝕍)L,hQhL)S_{X}(\mathbb{V},-\frac{1}{m}D)\otimes L\simeq S_{X}(S(\mathbb{V})\otimes L,h_{Q}h_{L}). In particular SX(𝕍,1mD)S_{X}(\mathbb{V},-\frac{1}{m}D) is independent of the choice of the desingularization π:X~X\pi:\widetilde{X}\to X.

Proof.

By Lemma 2.1, the proof can be reduced to the log smooth case, that is, XX is smooth, E:=X\XoE:=X\backslash X^{o} is a simple normal crossing divisor and supp(D)E{\rm supp}(D)\subset E. Denote j:Xo:=X\EXj:X^{o}:=X\backslash E\to X to be the inclusion. We are going to show that

SX(𝕍,1mD)L=SX(S(𝕍)L,hQhL)S_{X}(\mathbb{V},-\frac{1}{m}D)\otimes L=S_{X}(S(\mathbb{V})\otimes L,h_{Q}h_{L})

as subsheaves of ωXj(S(𝕍))L\omega_{X}\otimes j_{\ast}(S(\mathbb{V}))\otimes L. Since the problem is local, we assume that X=ΔnX=\Delta^{n} is the polydisc. Denote E:={z1zl=0}E:=\{z_{1}\cdots z_{l}=0\} where Ei:={zi=0}E_{i}:=\{z_{i}=0\} for each i=1,,li=1,\dots,l. Let 𝕍=(𝒱,,{𝒱p,q},hQ)\mathbb{V}=(\mathcal{V},\nabla,\{\mathcal{V}^{p,q}\},h_{Q}) be a polarized complex variation of Hodge structure on XoX^{o}. Let 𝟎=(0,,0)X{\bf 0}=(0,\dots,0)\in X and let (v1~,,vN~)(\widetilde{v_{1}},\dots,\widetilde{v_{N}}) be an L2L^{2}-adapted local frame of RX(𝕍)R_{X}(\mathbb{V}) at 𝟎{\bf 0} as in Proposition 3.5. Let f1,,fN(j𝒪Xo)𝟎f_{1},\dots,f_{N}\in(j_{\ast}\mathscr{O}_{X^{o}})_{\bf 0} and let ee be the local frame of LL at 𝟎\bm{0}. We are going to prove that

i=1Nfi[vi~dz1dzne]𝟎SX(S(𝕍)L,hQhL)𝟎\sum_{i=1}^{N}f_{i}[\widetilde{v_{i}}dz_{1}\wedge\cdots\wedge dz_{n}\otimes e]_{\bf 0}\in S_{X}(S(\mathbb{V})\otimes L,h_{Q}h_{L})_{\bf 0}

if and only if

fi𝒪X(j=1lrimαEj(vi~)Ej)𝟎f_{i}\in\mathscr{O}_{X}\big{(}-\sum_{j=1}^{l}\lfloor\frac{r_{i}}{m}-\alpha_{E_{j}}(\widetilde{v_{i}})\rfloor E_{j}\big{)}_{\bf 0}

for every i=1,,Ni=1,\dots,N.

Denote ds02=i=1ndzidz¯ids^{2}_{0}=\sum_{i=1}^{n}dz_{i}d\bar{z}_{i}. Since (v1~,,vN~)(\widetilde{v_{1}},\dots,\widetilde{v_{N}}) is an L2L^{2}-adapted frame as in Proposition 3.5, the integral

|i=1Nfivi~dz1dzn|2|e|hL2volds02=|i=1Nfivi~|2|e|hL2volds02\int|\sum_{i=1}^{N}f_{i}\widetilde{v_{i}}dz_{1}\wedge\cdots\wedge dz_{n}|^{2}|e|^{2}_{h_{L}}{\rm vol}_{ds^{2}_{0}}=\int|\sum_{i=1}^{N}f_{i}\widetilde{v_{i}}|^{2}|e|^{2}_{h_{L}}{\rm vol}_{ds^{2}_{0}}

is finite near 𝟎{\bf 0} if and only if

(3.6) |fivi~|2|e|hL2volds02|fi|2j=1r|zj|2αEj(vi~)2rimλivolds02\displaystyle\int|f_{i}\widetilde{v_{i}}|^{2}|e|^{2}_{h_{L}}{\rm}{\rm vol}_{ds^{2}_{0}}\sim\int|f_{i}|^{2}{\rm}\prod_{j=1}^{r}|z_{j}|^{2\alpha_{E_{j}}(\widetilde{v_{i}})-\frac{2r_{i}}{m}}\lambda_{i}{\rm vol}_{ds^{2}_{0}}

is finite near 𝟎{\bf 0} for every i=1,,Ni=1,\dots,N. Here λi\lambda_{i} is a positive real function so that

(3.7) 1λi|z1zr|ϵ,ϵ>0.\displaystyle 1\lesssim\lambda_{i}\lesssim|z_{1}\cdots z_{r}|^{-\epsilon},\quad\forall\epsilon>0.

Denote

vj(f):=min{l|fl0 in the Laurant expansion f=ifizji}.v_{j}(f):=\min\{l|f_{l}\neq 0\textrm{ in the Laurant expansion }f=\sum_{i\in\mathbb{Z}}f_{i}z_{j}^{i}\}.

By Lemma 3.7, the local integrability of (3.6) is equivalent to that

(3.8) vj(fi)+αEi(vj~)rim>1,j=1,,l.\displaystyle v_{j}(f_{i})+\alpha_{E_{i}}(\widetilde{v_{j}})-\frac{r_{i}}{m}>-1,\quad\forall j=1,\dots,l.

This is equivalent to

(3.9) vj(fi)αEi(vj~)+rim,j=1,,l.\displaystyle v_{j}(f_{i})\geq\lfloor-\alpha_{E_{i}}(\widetilde{v_{j}})+\frac{r_{i}}{m}\rfloor,\quad\forall j=1,\dots,l.

As a consequence, SX(S(𝕍)L,hQhL)𝟎S_{X}(S(\mathbb{V})\otimes L,h_{Q}h_{L})_{\bf 0} is generated by

dz1dzneexp(i=1nlogzi(αEi(vj~)+rimId+Ni))vj~,j=1,,N.dz_{1}\wedge\cdots\wedge dz_{n}\otimes e\otimes{\rm exp}\left(\sum_{i=1}^{n}\log z_{i}(\lfloor-\alpha_{E_{i}}(\widetilde{v_{j}})+\frac{r_{i}}{m}\rfloor{\rm Id}+N_{i})\right)\widetilde{v_{j}},\quad\forall j=1,\dots,N.

These are exactly the generators of ωXL(j(S(𝕍))𝒱>𝒓m1)\omega_{X}\otimes L\otimes(j_{\ast}(S(\mathbb{V}))\cap\mathcal{V}_{>\frac{\bm{r}}{m}-1}) at 𝟎{\bf 0}. The proof is finished. ∎

The proof of the following lemma is omitted.

Lemma 3.7.

Let ff be a holomorphic function on Δ:={z|0<|z|<1}\Delta^{\ast}:=\{z\in\mathbb{C}|0<|z|<1\} and aa\in\mathbb{R}. Then

|z|<12|f|2|z|2a𝑑z𝑑z¯<\int_{|z|<\frac{1}{2}}|f|^{2}|z|^{2a}dzd\bar{z}<\infty

if and only if v(f)+a>1v(f)+a>-1. Here

v(f):=min{l|fl0 in the Laurant expansion f=ifizi}.v(f):=\min\{l|f_{l}\neq 0\textrm{ in the Laurant expansion }f=\sum_{i\in\mathbb{Z}}f_{i}z^{i}\}.

3.4. Kollár package

In this section we prove the main theorem (Theorem 1.1) of the present paper. Let XX be a complex space and XoXregX^{o}\subset X_{\rm reg} a dense Zariski open subset. Let 𝕍:=(𝒱,,{𝒱p,q},Q)\mathbb{V}:=(\mathcal{V},\nabla,\{\mathcal{V}^{p,q}\},Q) be a polarized complex variation of Hodge structure of weight kk on XoX^{o}. Let

=θ¯++¯+θ\nabla=\overline{\theta}+\partial+\bar{\partial}+\theta

be the decomposition according to (3.1). For the reason of degrees, S(𝕍)S(\mathbb{V}) is a holomorphic subbundle of 𝒱\mathcal{V} and θ¯(S(𝕍))=0\overline{\theta}(S(\mathbb{V}))=0.

Lemma 3.8.

(S(𝕍),hQ)(S(\mathbb{V}),h_{Q}) is a Nakano semipositive vector bundle which is tame on XX.

Proof.

To see that (S(𝕍),hQ)(S(\mathbb{V}),h_{Q}) is Nakano semipositive, we take the decomposition

=θ¯++¯+θ\nabla=\overline{\theta}+\partial+\bar{\partial}+\theta

according to (3.1). Since θ¯(S(𝕍))=0\overline{\theta}(S(\mathbb{V}))=0, it follows from Griffiths’ curvature formula

Θh(S(𝕍))+θθ¯+θ¯θ=0\Theta_{h}(S(\mathbb{V}))+\theta\wedge\overline{\theta}+\overline{\theta}\wedge\theta=0

that

1Θh(S(𝕍))=1θ¯θNak0.\sqrt{-1}\Theta_{h}(S(\mathbb{V}))=-\sqrt{-1}\overline{\theta}\wedge\theta\geq_{\rm Nak}0.

To prove the tameness we use Deligne’s extension. Since the problem is local, we assume that there is a desingularization π:X~X\pi:\widetilde{X}\to X such that π\pi is biholomorphic over XoX^{o} and D:=π1(X\Xo)D:=\pi^{-1}(X\backslash X^{o}) is a simple normal crossing divisor. By abuse of notations we identify XoX^{o} and π1(Xo)\pi^{-1}(X^{o}). There is an inclusion S(𝕍)𝒱>𝟏|XoS(\mathbb{V})\subset\mathcal{V}_{>\bm{-1}}|_{X^{o}}. Let z1,,znz_{1},\dots,z_{n} be holomorphic local coordinates such that Di={zi=0}D_{i}=\{z_{i}=0\}, i=1,,ki=1,\dots,k and D={z1zk=0}D=\{z_{1}\cdots z_{k}=0\}. By Theorem 3.2, one has the norm estimate

(3.10) |z1zk||s|h0|s|h\displaystyle|z_{1}\cdots z_{k}||s|_{h_{0}}\lesssim|s|_{h}

for any holomorphic local section ss of 𝒱>𝟏\mathcal{V}_{>\bm{-1}}. Here h0h_{0} is an arbitrary (smooth) hermitian metric on 𝒱>𝟏\mathcal{V}_{>\bm{-1}}. This shows that (S(𝕍),hQ)(S(\mathbb{V}),h_{Q}) is tame. ∎

Theorem 3.9.

Let XX be a complex space and XoXregX^{o}\subset X_{\rm reg} a dense Zariski open subset. Let 𝕍:=(𝒱,,{𝒱p,q},Q)\mathbb{V}:=(\mathcal{V},\nabla,\{\mathcal{V}^{p,q}\},Q) be a polarized complex variation of Hodge structure of weight kk on XoX^{o}. Let LL be a line bundle such that some multiple mL=A+DmL=A+D where AA is a semipositive line bundle and DD is an effective Cartier divisor on XX. Let FF be an arbitrary Nakano-semipositive vector bundle on XX. Then SX(𝕍,1mD)FLS_{X}(\mathbb{V},-\frac{1}{m}D)\otimes F\otimes L satisfies Kollár’s package with respect to any locally Kähler proper morphism f:XYf:X\to Y such that YY is irreducible and each irreducible component of XX is mapped onto YY.

Proof.

Let hAh_{A} be a hermitian metric on AA with semipositive curvature and hDh_{D} the singular hermitian metric on 𝒪X(D)\mathscr{O}_{X}(D) determined by the effective divisor DD. Denote hL:=(hAhD)1mh_{L}:=(h_{A}h_{D})^{\frac{1}{m}}. Then

1ΘhL(L)|X\D=1mΘhA(A)|X\D0.\sqrt{-1}\Theta_{h_{L}}(L)|_{X\backslash D}=\frac{\sqrt{-1}}{m}\Theta_{h_{A}}(A)|_{X\backslash D}\geq 0.

Hence (L|U,hL|U)(L|_{U},h_{L}|_{U}) has semipositive curvature and is tame on XX. Therefore by Lemma 3.8 (S(𝕍)LF|U,hQhLhF|U)(S(\mathbb{V})\otimes L\otimes F|_{U},h_{Q}h_{L}h_{F}|_{U}) has semipositive curvature on U=Xo\supp(D)U=X^{o}\backslash{\rm supp}(D) and is tame on XX. By Lemma 2.2, Theorem 2.4 and Theorem 3.6 we obtain that SX(𝕍,1mD)FLSX(S(𝕍)LF|U,hQhLhF|U)S_{X}(\mathbb{V},-\frac{1}{m}D)\otimes F\otimes L\simeq S_{X}(S(\mathbb{V})\otimes L\otimes F|_{U},h_{Q}h_{L}h_{F}|_{U}) satisfies Kollár’s package with respect to f:XYf:X\to Y. ∎

References