On exact sequences of Hodge theoretic fundamental groups

Simon Shuofeng Xu · 2025

Read On exact sequences of Hodge theoretic fundamental groups by Simon Shuofeng Xu (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2503-13307.txt.

Introduction

Xu builds a Hodge-theoretic version of the fundamental group in this 2025 mathematics paper. The fundamental group encodes the loop structure of a space. Hodge theory adds rich linear-algebra data from complex geometry.

Xu defines the group for smooth complex algebraic varieties. He then studies families of varieties over a base, proving a right exact sequence linking the groups of fiber, total space, and base. He asks when the sequence is fully exact and connects that to known results in non-abelian Hodge theory. An obstruction to splitting the sequence comes from étale fundamental groups. His main examples are the universal curve and a bundle of line bundles over it.

Catalog record

CREATORSimon Shuofeng Xu
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE70 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2503-13307

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

On exact sequences of Hodge theoretic fundamental groups Simon Shuofeng Xu Abstract . The goal of this paper is to first define a Hodge theoretic fundamental group for smooth connected complex algebraic varieties and then prove and study a right exact sequence of Hodge theoretic fundamental groups associated to a smooth projective family of algebraic varieties f : X → B f\colon X\to B . In particular, we study when this right exact sequence is exact, relate this question to some prior results in non-abelian Hodge theory, and give an obstruction to splitting in terms of étale fundamental groups. The main examples we consider in this note is the universal curve f : 𝒞 g → ℳ g f\colon\mathcal{C}_{g}\to\mathcal{M}_{g} and the moduli space of degree 1 1 line bundles on the universal curves p : Pic 𝒞 g / ℳ g 1 → ℳ g p\colon\Pic^{1}_{\mathcal{C}_{g}/\mathcal{M}_{g}}\to\mathcal{M}_{g} . 1. Introduction In this note, we define and study a notion of a Hodge theoretic fundamental group. Let X be a smooth connected complex algebraic variety. Consider the category of \VMHS(X) whose objects are admissible, graded-polarizable \mathbb{Q} -variations of mixed Hodge structure on X whose underlying local systems are defined over \mathbb{Z} . Any choice of a base point x\in X(\mathbb{C}) defines a fiber functor F_{x} from this category to the category of \mathbb{Q} -vector spaces, and the pair (\VMHS(X),F_{x}) is a \mathbb{Q} -linear neutral Tannakian category. Definition 1.1 . We define the Hodge theoretic fundamental group of X with respect to some base point x\in X(\mathbb{C}) to be the Tannakian fundamental group \pi_{1}(\VMHS(X),F_{x}) . Remark 1.2 . The term “Hodge theoretic fundamental group” has appeared in the literature before (see e.g. [ Ara10 ] and [ Jac22 ] ) and it was used to refer to a \pi_{1}(\LS^{\text{hdg}}(X)) , an object which we are about to introduce, except the local systems considered by Arapura does not necessarily have integral structures. As we will see, our Hodge theoretic fundamental is bigger and contains \pi_{1}(\LS^{\text{hdg}}(X)) as a normal subgroup. 1 1 1 We thank the anonymous reviewer for alerting us to the paper of Arapura [ Ara10 ] . Let \LS(X) be the category of \mathbb{Q} -local systems on X with an underlying integral structure. There’s an obvious forgetful functor of Tannakian categories from \VMHS(X)\to\LS(X) , and the Tannakian subcategory generated by the essential image of this functor is the subcategory of \LS(X) whose objects are local systems that are subquotients of local systems underlying variations of mixed Hodge structure in \VMHS(X) . Following the notation in [ DE22 ] , we denote this subcategory \LS^{\text{Hdg}}(X) , and via Tannakian duality, we have a closed embedding \pi_{1}(\LS^{\text{Hdg}}(X),F_{x})\to\pi_{1}(\VMHS(X),F_{x}) . On the other hand, the constant variations of mixed Hodge structure form a sub-Tannakian category of \VMHS(X) which is isomorphic to the category of graded-polarizable \mathbb{Q} -mixed Hodge structures \MHS , and so Tannakian duality gives us a map \pi_{1}(\VMHS(X),F_{x})\to\pi_{1}(\MHS,F) , where F is the forgetful functor from \MHS to the category of \mathbb{Q} -vector spaces. From now on, we will omit the fiber functor when the context is clear. André [ And21 , Theorem C. 11] showed that the following sequence is a (split) short exact sequence 1\to\pi_{1}(\LS^{\text{Hdg}}(X))\to\pi_{1}(\VMHS(X))\to\pi_{1}(\MHS)\to 1. See also the work of D’Addezio and Esnault [ DE22 , Corollary 4.7] who worked with \mathbb{Q} -variations of mixed Hodge structure with no underlying integral structure. 2 2 2 Arapura also attempted to prove the exactness of this sequence [ Ara10 , Theorem 2.6] , but a gap was found in his work by D’Addezio and Esnault [ DE22 , Remark 4.8] . Remark 1.3 . We have chosen to work with variation of mixed Hodge structures with an underlying integral structure, mainly because that assumption is needed when we discuss non-abelian Hodge locus in section 4 and it also makes the computation we are going to do in section 5 more tractable. In section 2 , we first generalize this short exact sequence to the relative setting. Throughout this paper, by a smooth projective family of algebraic varieties , we mean a smooth projective morpshim f\colon X\to B between smooth, connected complex algebraic varieties with connected fibers. Fix some base point b\in B and x\in X_{b}:=f^{-1}(b) . We have the following proposition: Theorem 1.4 . The following natural sequence of Hodge theoretic fundamental groups induced by f\colon X\to B is an exact sequence of affine group schemes: (1.1) \pi_{1}(\LS^{\text{Hdg}}(X_{b}))\to\pi_{1}(\VMHS(X))\to\pi_{1}(\VMHS(B))\to 1 For the rest of the paper, we further study this sequence (\ref{Hodge SES}) . In particular, we focus on the questions of when the first map \pi_{1}(\LS^{\text{Hdg}}(X_{b}))\to\pi_{1}(\VMHS(X)) is a closed embedding, and when the sequence has no section. The main examples we will be working with is the universal curve f\colon\mathcal{C}_{g}\to\mathcal{M}_{g} of genus g and the moduli space of degree 1 line bundles on the universal curve p\colon\Pic^{1}_{\mathcal{C}_{g}/\mathcal{M}_{g}}\to\mathcal{M}_{g} . Here we view these Deligne-Mumford stacks as orbifolds so we may talk about their (orbifold) fundamental group, and local systems as well as variations of (mixed) Hodge structure on them (for an exposition of the orbifold approach to \mathcal{M}_{g} , see [ ACG11 , Chapter XII, section 4] ). In particular, we prove the following theorem Theorem 1.5 . (1) For any g\geq 2 , there exists some curve C of genus g such that the sequence 1\to\pi_{1}(\LS^{\text{Hdg}}(C))\to\pi_{1}(\VMHS(\mathcal{C}_{g}))\to\pi_{1}(\VMHS(\mathcal{M}_{g}))\to 1 is not left exact (see Theorem 3.2 ); (2) When g\geq 3 , the map \pi_{1}(\VMHS(\mathcal{C}_{g}))\to\pi_{1}(\VMHS(\mathcal{M}_{g})) induced by f\colon\mathcal{C}_{g}\to\mathcal{M}_{g} and the map \pi_{1}(\VMHS(\Pic^{1}_{\mathcal{C}_{g}/\mathcal{M}_{g}}))\to\pi_{1}(\VMHS(\mathcal{M}_{g})) induced by p\colon\Pic^{1}_{\mathcal{C}_{g}/\mathcal{M}_{g}}\to\mathcal{M}_{g} do not splits (see Corollary 6.6 ). Now part (1) of Theorem 1.5 says that in general the sequence ( 1.1 ) is not short exact. However, one can also show that at least after passing to the pro-reductive quotient, under some assumptions, this sequence is exact for an analytically very general choice of base point b\in B . More precisely, we prove the following Theorem 1.6 . Suppose f\colon X\to B is a smooth projective family of algebraic varieties such that \pi_{1}^{\text{\'{e}t}}(X_{b}) injects into \pi_{1}^{\text{\'{e}t}}(X) . Let \mathcal{G}^{\text{red}} be the pro-reductive quotient of an affine group scheme \mathcal{G} . Define NG(X/B)^{\text{red}}:=\left\{b\in B:(\pi_{1}(\LS^{\text{Hdg}}(X_{b})))^{\text{red}}\to(\pi_{1}(\VMHS(X)))^{\text{red}}\text{ is not injective}\right\}. Then NG(X/B)^{\text{red}} is a countable union of closed analytic subsets of B . Assuming some standard conjectures in non-abelian Hodge theory, this set NG(X/B)^{\text{red}} is in fact a countable union of closed algebraic subsets of B (see section 3 ). Therefore, it’s natural to ask if this is the case without having to pass to pro-reductive quotients: we can define NG(X/B):=\left\{b\in B:\pi_{1}(\LS^{\text{Hdg}}(X_{b}))\to\pi_{1}(\VMHS(X))\text{ is not injective}\right\}. Question 1.7 . Is the locus NG(X/B) is always a countable union of closed algebraic subsets of B ? In fact, we show that this locus can be empty even in non-isotrivial families: Theorem 1.8 . For any g\geq 2 and any curve C of genus g , \pi_{1}(\LS^{\text{Hdg}}(\Pic^{1}(C)) may be identified with \mathbb{G}_{a}^{2g}\times\widehat{\mathbb{Z}}^{2g} and the sequence 1\to\mathbb{G}_{a}^{2g}\times\widehat{\mathbb{Z}}^{2g}\to\pi_{1}(\VMHS(\Pic^{1}_{\mathcal{C}_{g}/\mathcal{M}_{g}}))\to\pi_{1}(\VMHS(\mathcal{M}_{g}))\to 1 is always short exact. Part (2) of Theorem 1.5 is interesting as it provides some examples for which a “Hodge theoretic question” has an affirmative answer (see also [ Xu24 ] for a slightly different formulation of the Hodge theoretic section question): Let f\colon X\to B be a smooth projective family of algebraic varieties. Any algebraic section s\colon B\to X

Read the complete text