The exceptional locus of a motivic local system

Luca Terenzi · 2026

Read The exceptional locus of a motivic local system by Luca Terenzi (2026) 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-2603-22171.txt.

Introduction

Terenzi studies the exceptional locus of a Nori motivic local system. The system lives over a smooth, connected complex algebraic variety. The locus tracks where stalk complexity varies. Terenzi defines the locus through motivic Galois groups and Artin motives. He proves a motivic version of the Cattani-Deligne-Kaplan theorem. The locus is a countable union of closed algebraic subvarieties. He also shows it is defined over any field where the system has a model. A similar account covers the splitting locus of the weight filtration.

Catalog record

CREATORLuca Terenzi
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE159741
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2603.22171

Browse the full library · About the Tower of Babel

Full text

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

The exceptional locus of a motivic local system Luca Terenzi Abstract . Given a Nori motivic local system over a smooth, connected complex algebraic variety, we define its exceptional locus as a way to measure the variation in the motivic complexity of its stalks. The definition is given explicitly in terms of motivic Galois groups and Artin motives. Our main result is a motivic analogue of the Cattani–Deligne–Kaplan Theorem, asserting that the exceptional locus is a countable union of closed algebraic subvarieties. Moreover, we show that the maximal such subvarieties are defined over any algebraically closed subfield over which the ambient variety and the motivic local system admit models, and that they are stable under Galois conjugation when these models descend to a further subfield. This strengthens and extends previous results by André in the pure case. We obtain a similar geometric description for the splitting locus of the motivic weight filtration. In the case of 1 1 -motivic local systems, the above properties pass to the underlying variations of mixed Hodge structure via André’s fullness theorem. † † 1991 Mathematics Subject Classification : 14F42, 14C15, 14D07, 18M25. † † Key words and phrases : Nori motives, motivic Galois groups, Hodge locus, weight filtration. Introduction Motivation and goal of the paper Let X be a smooth, connected complex algebraic variety, and let V be an admissible, graded-polarizable variation of mixed Hodge structure over X . One can think of V as a local system of mixed Hodge structures, and it is natural to ask how the complexity of the stalks V_{x} varies with respect to x\in X(\mathbb{C}) . This is measured by the (tensorial) Hodge locus of V , defined as the set of points x\in X(\mathbb{C}) such that the mixed Hodge structure V_{x} admits more Hodge tensors than the very general stalk. A fundamental result describes such a locus as a countable union of strict closed algebraic subvarieties of X : in the case of pure variations, this was proved in Cattani–Deligne–Kaplan’s celebrated paper [ CDK95 ] ; it was later extended to mixed variations by Brosnan–Pearlstein–Schnell in [ BPS10 ] , where a similar geometric description was provided for the locus where the weight filtration splits. The result is particularly interesting when the variation V is of geometric origin – that is, when it arises from the relative cohomology of X -schemes. In this case, it can be promoted to a Nori motivic local system, the stalks of which are Nori motives rather than Hodge structures. The main goal of the present paper is to define a natural motivic version of the (tensorial) Hodge locus, for which we establish the analogue of the Cattani–Deligne–Kaplan Theorem; similarly for the splitting locus of the motivic weight filtration. We are also able to describe the fields of definition of these loci and to prove their stability under Galois conjugation; these arithmetic aspects are only partially understood in the Hodge-theoretic setting. In the case of pure motivic local systems, similar results were obtained by André in [ And96 ] , using the language of motivated cycles. Following André’s Tannakian approach, our results are based on the theory of motivic Galois groups. Our treatment relies systematically on the six-functor formalism for Nori motivic sheaves, initiated by Ivorra–Morel in [ IM24 ] and completed by the author in [ Ter26 ] . Of particular importance are the structural properties of motivic Galois groups of smooth algebraic varieties, established by Jacobsen in [ Jac26 ] . History and previous work The study of the Hodge locus began with Weil’s famous observation about Hodge classes in the relative cohomology of a smooth projective family over a smooth complex algebraic variety, from [ Wei79 ] . To fix notation, let p\colon Y\to X be such a family, and let n\geq 0 . As x\in X(\mathbb{C}) varies, the rational Hodge structures V_{x}:=H^{2n}(Y_{x};\mathbb{Q})(n) assemble into a polarizable pure variation of Hodge structure V of weight 0 over X . The Hodge locus of V is defined as the subset of points x\in X(\mathbb{C}) such that the stalk V_{x} contains more classes of type (0,0) compared to the very general stalk. Since the weight and Hodge filtrations on the stalks of V vary holomorphically, this locus is a countable union of strict closed analytic subvarieties of X . However, the Hodge Conjecture predicts that all rational (0,0) -classes on V_{x} arise from algebraic cycles on the fibre Y_{x} . This would imply that the analytic subvarieties forming the Hodge locus are in fact algebraic. This expectation was established, unconditionally on the Hodge Conjecture and for arbitrary polarizable pure variations, by Cattani–Deligne–Kaplan in [ CDK95 , Thm. 1.1] . As later proved by Brosnan–Pearlstein–Schnell in [ BPS10 , Cor. 2] , the result remains true for arbitrary admissible, graded-polarizable variations of mixed Hodge structure: this is deduced from the pure case via the theory of higher normal functions. Using the same method, the authors obtained a similar geometric description for the splitting locus of the weight filtration. All these results, and particularly the proof of [ CDK95 , Thm. 1.5] , are based on substantial input from Complex Analysis. As explained by André in [ And92 , § 2] , it is natural to consider Hodge classes not only on V but on all of its tensor combinations at once, also known as Hodge tensors. It is thus natural to replace the usual Hodge locus of V by the tensorial Hodge locus , defined as the set of points x\in X(\mathbb{C}) such that V_{x} contains more Hodge tensors than the very general stalk – or, equivalently, such that the Mumford–Tate group of V_{x} is smaller than that of the very general stalk. The main advantage of the tensorial Hodge locus over the usual one is its amenability to group-theoretic techniques. If the variation V arises from a smooth projective family p\colon Y\to X , the Hodge locus (both in the usual and in the tensorial version) is expected to be defined over any algebraically closed subfield k\subset\mathbb{C} over which the family admits a model: again, this would follow from the validity of the Hodge Conjecture. So far, such property has not been established unconditionally in general; see [ KOU23 ] , [ SS16 ] , [ Sim90 ] and [ Voi07 ] for partial results in this direction. Variations arising from smooth projective families are basic examples of variations of geometric origin . All such variations are realizations of finer cohomological objects known as Nori motivic local systems , defined by the author in [ Ter26 , § 6] : these can be thought of as local systems with values in the Tannakian category \mathcal{M}(\mathbb{C}) of Nori motives over \mathbb{C} . There is a canonical monoidal Hodge realization functor \iota_{\mathbb{C}}^{\textup{Hdg}}\colon\mathcal{M}(\mathbb{C})\to\mathsf{MHS}, which is conjectured to be fully faithful, with essential image stable under subquotients. Thus, given a motivic local system M over X , with underlying variation V , the motivic Galois group of M_{x} is expected to coincide with the Mumford–Tate group of V_{x} for every x\in X(\mathbb{C}) . While the fullness conjecture remains widely open in general, its validity is known for the Tannakian subcategory of \mathcal{M}(\mathbb{C}) generated by the

Read the complete text