A non-abelian version of Deligne's Fixed Part Theorem

TL;DR

This paper develops a non-abelian analogue of Deligne's Fixed Part theorem, connecting algebraically isomonodromic extensions of local systems with variations of Hodge structures, and demonstrating the constancy of Mumford-Tate groups in this context.

Contribution

It introduces a non-abelian version of Deligne's Fixed Part theorem and explores the relationship between isomonodromic extensions and Hodge structures, extending previous work in the field.

Findings

Established a non-abelian Fixed Part theorem.

Linked algebraically isomonodromic extensions to variations of Hodge structures.

Proved Mumford-Tate group remains constant in certain extensions.

Abstract

We formulate and prove a non-abelian analog of Deligne's Fixed Part theorem on Hodge classes, revisiting previous work of Jost--Zuo, Katzarkov--Pantev and Landesman--Litt. To this aim we study algebraically isomonodromic extensions of local systems and we relate them to variations of Hodge structures, for example we show that the Mumford-Tate group at a generic point stays constant in an algebraically isomonodromic extension of a variation of Hodge structure. v2: a few typos ironed and Thm 1.1 5) completed. v3: there was a Schlamassel leading to a mix-up of files. Apologies. Else identical version (one minor change). v5 final version. Appears in Alg. Geom.