Tannakian duality for Anderson-Drinfeld motives and algebraic independence of Carlitz logarithms (Q2471788): Difference between revisions

This paper is a milestone in transcendence theory in characteristic \(p\). The author shows that every algebraic relation between periods of \(t\)-motives is ``motivic'', that is, it arises from a relation between \(t\)-motives. This is a characteristic \(p\) version of Grothendieck's period conjecture. Previously, \textit{D. Brownawell, G. Anderson} and the author had shown that any \textit{linear} relation between periods of a \(t\)-motive is motivic [Ann. Math. (2) 160, No. 1, 237--313 (2004; Zbl 1064.11055)]. Since the category of \(t\)-motives is closed under tensor product, one may expect to deduce results on \textit{algebraic} independence from this. To do so, the author develops a Tannakian machinery for \(t\)-motives, loosely based on similar approaches for differential and difference Galois theory. As a consequence he associates to every \(t\)-motive an algebraic group: its Tannakian Galois group (to be compared with the motivic fundamental group or the Mumford-Tate group of a motive). He shows that the transcendence degree of the set of periods of a \(t\)-motive is the dimension of its Tannakian Galois group. To illustrate the spectacular power of his theorem, the author determines all algebraic relations between Carlitz logarithms of algebraic points.