Mixed Tate motives over \(\mathbb{Z}\) (Q410120)

In this outstanding paper, the author proves a deep conjecture on multiple zeta values. For an \(r\)-tuple of positive integers \(n_1, \dots, n_r\) with \(n_i\geq 1\) and \(n_r\geq 2\), the multiple zeta value \(\zeta(n_1, \dots, n_r)\) is defined as the iterated multiple sum \[ \zeta(n_1, \dots, n_r) = \sum_{0<k_1< \dots < k_r} \frac{1}{k_1^{n_1}\cdots k_r^{n_r}}. \] Multiple zeta values are real numbers that were originally defined by Euler and are also referred to as Euler/Zagier sums. They have recently drawn increasing attention because they appear in many different problems in mathematics and mathematical physics. Let \(n=n_1+\dots+n_r\) be the weight of the multiple zeta value \(\zeta(n_1, \dots, n_r)\) and let \(\mathcal{Z}_n\) denote the \(\mathbb{Q}\)-vector space spanned by all multiple zeta values of weight \(n\). It is a deep conjecture of Zagier's that the dimension of \(\mathcal{Z}_n\) is given by \[ \mathrm{dim}~ \mathcal{Z}_n = d_n \] where \(d_n\) is defined as the coefficient of \(x^n\) in the power series expansion of \(\frac{1}{1-x^2-x^3}\) (see \textit{M. E. Hoffman} [J. Algebra 194, No. 2, 477--495 (1997; Zbl 0881.11067)] and \textit{Y. André} [Une introduction aux motifs. Motifs purs, motifs mixtes, périodes. Paris: Société Mathématique de France (2004; Zbl 1060.14001)]). In particular, one would like to find all \(\mathbb{Q}\)-linear relations between multiple zeta values. A proof of the conjecture \(\mathrm{dim}~ \mathcal{Z}_n = d_n\) seems to be out of reach at present. Nevertheless, it is known that the multiple zeta values in \(\mathbb{R}\) are the images under a \(\mathbb{Q}\)-linear map of certain motivic multiple zeta values \(\zeta^m(n_1, \dots, n_r)\). \textit{T. Terasoma} [Invent. Math. 149, No. 2, 339--369 (2002; Zbl 1042.11043)] and \textit{A. B. Goncharov} [Duke Math. J. 128, No. 2, 209--284 (2005; Zbl 1095.11036)] proved independently that the conjectural dimension formula for the space of the motivic multiple zeta values \(\zeta^m(n_1, \dots, n_r)\) implies the upper bound \[ \mathrm{dim}~ \mathcal{Z}_n \leq d_n. \] In the paper under review, the author proves the motivic dimension formula. In more detail, the author considers the graded \(\mathbb{Q}\)-algebra \(\mathcal{M}\), whose homogeneous elements are the motivic \(\zeta^m(n_1, \dots, n_r)\) in degree \(n=\sum n_i\), and defines a homomorphism \(\mathrm{real}: \mathcal{M} \to \mathbb{R}\) such that \[ \mathrm{real}(\zeta^m(n_1, \dots, n_r))=\zeta(n_1, \dots, n_r). \] The \(\mathbb{Q}\)-linear relations among the \(\zeta^m(n_1, \dots, n_r)\) induce, so called motivic, relations among the \(\zeta(n_1, \dots, n_r)\). The main result of the paper is: The set of elements \(\zeta^m(n_1, \dots, n_r)\) for which each \(n_i\) is either \(2\) or \(3\) form a basis for the \(\mathbb{Q}\)-vector space of motivic multiple zeta values. As a first consequence, one obtains a proof of Hoffman's conjecture that every multiple zeta value is a \(\mathbb{Q}\)-linear combination of values \(\zeta(n_1, \dots, n_r)\) for which each \(n_i\) is either \(2\) or \(3\). The second important consequence is that the category of mixed Tate motives over \(\mathbb{Z}\) is generated by the motivic fundamental group \(\pi\) of \(\mathbb{P}^1-\{0,1,\infty\}\). More precisely, let \(\mathcal{MT}(\mathbb{Z})\) be the category of mixed Tate motives unramified over \(\mathbb{Z}\). It is a Tannakian category let \(\mathcal{G}_{\mathcal{MT}}\) denote its Galois group. Let \(\mathcal{MT}'(\mathbb{Z})\) be the full Tannakian subcategory generated by \(\pi\) and let \(\mathcal{G}_{\mathcal{MT}'}\) be its Galois group. The author proves the deep conjecture: The map \(\mathcal{G}_{\mathcal{MT}} \to \mathcal{G}_{\mathcal{MT}'}\) is an isomorphism. A third important consequence is that all periods of mixed Tate motives over \(\mathbb{Z}\) are \(\mathbb{Q}[\frac{1}{2\pi i}]\)-linear combinations of multiple zeta values. Finally, we remark that the author's proof does not seem to yield an effective algorithm that would allow to determine the linear relations between given multiple zeta values. More about possible and necessary calculations and concrete relations can be found in the paper by \textit{D. Zagier} [Ann. Math. (2) 175, No. 2, 977--1000 (2012; Zbl 1268.11121)]. The reader should also have a look at Deligne's article in the Séminaire Bourbaki, 64ème année, 2011-2012, no. 1048.