On relative integral monodromy of abelian logarithms and normal functions.

Yves André · 2025

Read On relative integral monodromy of abelian logarithms and normal functions. by Yves André (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-2501-07691.txt.

Introduction

Yves André studies monodromy groups for abelian logarithms in a math paper from 2025. Monodromy tracks how solutions twist as one walks a loop in a parameter space. For abelian schemes, families of abelian varieties, sections add extra structure called normal functions.

An earlier result handled the algebraic version of the problem: the relative group came out maximal. The new work asks the sharper integral question, without smoothing groups into algebraic ones. When the scheme's integral monodromy forms a full lattice, the logarithm's relative group does too. The proof leans on a Hodge theory view of sections. The paper extends the same question to normal functions in general.

Catalog record

CREATORYves André
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE34 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2501-07691

Browse the full library · About the Tower of Babel

Full text

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

On relative integral monodromy of abelian logarithms and normal functions. Yves André Abstract . The relative algebraic monodromy of abelian logarithms (defined as the kernel of a map between algebraic monodromy groups attached to an abelian scheme with and without a section) was computed in [ 1 ] : under natural assumptions, this vector group turns out to be maximal. The relative integral monodromy of abelian logarithms is defined similarly as a kernel of integral monodromy groups, without taking Zariski closures. We show that if the integral monodromy of the abelian scheme is a lattice in the algebraic monodromy (which is not always the case), then the relative integral monodromy of the abelian logarithm is also a lattice in the relative algebraic monodromy. The proof uses a Hodge-theoretic interpretation of sections of abelian schemes. We also consider relative integral monodromy groups in the more general context of normal functions. 1991 Mathematics Subject Classification 11G, 14K 1. Introduction 1.1. The relative monodromy of abelian logarithms of the title refers to the following situation: let f:A\to S be an abelian scheme parametrized by a connected smooth complex variety endowed with a section \sigma . The Betti realization of the family of 1 -motives [\mathbb{Z}_{S}\stackrel{{\scriptstyle 1\mapsto\sigma}}{{\to}}A] lies in an exact sequence of local systems (1.1) 0\to V_{\mathbb{Z}}=R_{1}f^{an}_{\ast}\mathbb{Z}\to{\bf V}_{\mathbb{Z}}\to\mathbb{Z}_{S}\to 0 on S^{an} , where \mathbb{Z}_{S} denotes the constant local system \mathbb{Z} on S^{an} cf. e.g. [ 16 , §10] [ 2 , §1.4] . Let s\in S(\mathbb{C}) be a base point, \Gamma ( resp. \bf\Gamma ) be the image of \pi_{1}(S(\mathbf{C}),s) in GL(V_{s}) ( resp. GL({\bf V}_{s}) ), and H ( resp. \bf H ) be its Zariski-closure. By a theorem of Deligne, the neutral component of H is a semisimple group [ 15 ] . Let \Delta (the relative integral monodromy ) be the kernel of {\bf\Gamma}\to\Gamma and U (the relative algebraic monodromy ) be the kernel of {\bf H}\to H , so that U is a vector group and \Delta is a subgroup of U(\mathbb{Z}) . In [ 1 ] ( cf. also [ 2 , §§1.4, 1.5] ), the following result was proven: 1.1.1 Theorem . Assume that i) A/S has no fixed part, even after pull-back to a finite etale cover of S , ii) \mathbb{Z}.\sigma(S) is Zariski-dense in A (i.e. the image of the section is not contained in any proper S -subgroup of A ). Then there is a natural isomorphism U(\mathbb{Z})\cong H_{1}(A_{s},\mathbb{Z}) . This has been used in a number of contexts, notably in the study of Betti maps in Diophantine Geometry [ 4 ] 1 1 1 the name “Betti map”, due to D. Bertrand, actually refers to this context of Betti realizations. . 1.2. It is tempting to replace the conclusion of Theorem 1.1.1 by 2 2 2 cf. e.g. [ 33 , §3.1] - which however only uses 1.1.1, not its overinterpretation in terms of integral relative monodromy. : \;\;\Delta is of finite index in H_{1}(A_{s},\mathbb{Z})\;\;\;(?) However, as P. Corvaja and U. Zannier pointed out [ 13 ] , it is not clear whether \Delta is Zariski-dense in U : indeed, taking a kernel does not commute with taking a Zariski-closure. The aim of this note is to clarify this point. We prove: 1.2.1 Theorem . Assume moreover that iii) \Gamma is a lattice in H_{\mathbb{R}} . Then \Delta is a lattice in U(\mathbb{Z})\, (equivalently, \bf\Gamma is a lattice in {\bf H}_{\mathbb{R}} ). Assumption iii) is equivalent to: \Gamma is of finite index in H(\mathbb{Z}) [ 8 ] . It holds for instance in the situation of elliptic schemes, which is settled in [ 13 ] 3 3 3 in this particular situation, Theorem 1.1.1 was obtained previously by D. Bertrand, and we reformulate [ 13 ] in Hodge-theoretic terms in §2 below. Theorem 1.2.1 in the case of polarized abelian schemes of relative dimension g over a base which is finite surjective over the moduli space \mathcal{A}_{g} was considered in [ 17 ] ). . However, iii) does not always hold: there are abelian schemes with non-lattice monodromy \Gamma , cf. [ 25 ] . We do not know whether the conclusion holds in such cases. 1.3. One key point in our proof is the Hodge-theoretic interpretation of (holomorphic versus algebraic) sections of abelian schemes, which is of independent interest ( cf. Theorem 3.2.1, which depends on a theorem of Graber-Starr). This allows us to reduce the question to the “modular case” and apply the theorem of Mok-To on sections of Kuga fibrations. 1.4. In most cases, the study of relative integral monodromy can be settled by general results on vanishing of the first cohomology group H^{1}(\Gamma,V) of representations of integral monodromy groups, when the latter are arithmetic. We review these results (Margulis-Starkov, Bass-Milnor-Serre) in §§5.1, 5.2. Such arguments, which do not bear on the geometry of the situation, allow to deal with relative integral monodromy in the more general situation of normal functions, notably in the context of Lefschetz pencils §5.3. When these general results do not apply ( \Gamma is arithmetic but H^{1}(\Gamma,V)\neq 0 - as in the case of elliptic families where \Gamma is an arithmetic subgroup of SL(2) ), our proof rests on the fact that H^{1}(\Gamma,V) carries a mixed Hodge structure with H^{1}(\Gamma,V)^{0,0}=0. 2. The case of families of elliptic curves: Hodge-theoretic interpretation of [ 13 ] 2.1. Replacing S by a finite etale cover, we may assume that for some N>2 , the N -torsion of f is defined over S , whence a (non-constant) map S\to\mathcal{A}_{1,N}=X(N) from S to the modular curve, \Gamma being a subgroup of finite index of \Gamma(N) . Denoting by S^{\prime} the finite etale cover of X(N) with fundamental group \Gamma , there is a cartesian square \begin{matrix}A&\stackrel{{\scriptstyle h}}{{\to}}&A^{\prime}\\ f\downarrow&&\downarrow{f^{\prime}}\\ S&\stackrel{{\scriptstyle g}}{{\to}}&S^{\prime}.\end{matrix} The cocycle e\in H^{1}(S^{an},R_{1}f^{an}_{\ast}\mathbb{Q})=H^{1}(\pi_{1}(S,s),H_{1}(A_{s},\mathbb{Q})) attached to the section \sigma comes (via (h\sigma)^{\ast} ) from a cocycle in H^{1}({\bf\Gamma},H_{1}(A_{s},\mathbb{Q})) . The rank of \Delta is a priori 0,1 or 2 . If it is 2 , \Delta is a lattice in H_{1}(A_{s},\mathbb{R}) . It cannot be 1 , because \mathbb{Q}.\Delta is monodromy-invariant, and \Gamma -invariant subspaces of H_{1}(A_{s},\mathbb{Q}) are 0 or H_{1}(A_{s},\mathbb{Q}) . If it is 0 , i.e. {\bf\Gamma}=\Gamma , then e comes via g^{\ast}=(f^{\prime}h\sigma)^{\ast} from a cocycle in H^{1}(\Gamma,H_{1}(A_{s},\mathbb{Q}))=H^{1}(\pi_{1}(S^{\prime},g(s)),H_{1}(A^{\prime}_{h(s)},\mathbb{Q}))=H^{1}(S^{\prime an},R_{1}{f}^{{}^{\prime}an}_{\ast}\mathbb{Q}). The morphism g^{\ast}:H^{1}(S^{\prime an},R_{1}{f}^{{}^{\prime}an}_{\ast}\mathbb{Q})\to H^{1}(S^{an},R_{1}f^{an}_{\ast}\mathbb{Q}) is a morphism of mixed Hodge structures. The morphism g extends to a morphism \bar{g}:\bar{S}\to\overline{S^{\prime}} between compactifications j:S\hookrightarrow\bar{S},{j^{\prime}}:S^{\prime}\hookrightarrow\overline{S^{\prime}} , and the weight- 0 part of g^{\ast} can be identified with H^{1}(\overline{S^{\prime}},j^{\prime}_{\ast}R^{1}{f^{\prime}}^{an}_{\ast}\mathbb{Q}(1))\to H^{1}(\bar{S},j_{\ast}R^{1}f^{an}_{\ast}\mathbb{Q}(1)). According to [ 34 ] , the (0,0) -component of this map can be identified with the map between sections A^{\prime}(S^{\prime})\otimes\mathbb{Q}\to A(S)\otimes\mathbb{Q} induced by g . Therefore, if \Delta=0 , then \sigma comes from a section of f^{\prime} . But A^{\prime}(S^{\prime}) is torsion according to Shioda’s theorem [ 30 ] . Hence \sigma is torsion and U=0 in that case. 2.1.1 Remark . In the case of Legendre pencil f:A\to S=S^{\prime}=\mathbb{P}^{1}\setminus\{0,1,\infty\} , an example of a non-torsion class e\in H^{1}(S,V_{s}) coming from a holomorphic, non-algebraic, section of f^{an} is given in [ 13 , (2.6)] . 3. Hodge-theoretic interpretation of sections of abelian schemes Let f:A\to S be an abelian scheme without fixed part . Let V_{\mathbb{Z}} be the local system R_{1}f^{an}_{\ast}\mathbb{Z}

Read the complete text