Additive polylogarithms and their functional equations (Q711568): Difference between revisions

Recall that the classical real analytic \(n\)-polylogarithm \(\mathcal{L}_n\) is a main object in the understanding of weight \(n\) motivic cohomology over a field \(k\) of zero characteristic and is related with the \(Li_{n-j}(z)=\sum_{k=1}^{\infty}\frac{z^{n-j}}{k^{n-j}}\) with \(j\) runs 0 to \(n\), see \textit{D. Zagier} [``Polylogarithms, Dedekind zeta functions, and the algebraic K-theory of fields'', Prog. Math. 89, 391--430 (1991; Zbl 0728.11062)]. Let us denote \(\{b\in R \mid b(1-b)\in R^*\}\) by \(R^{b}\) with \(R\) a ring with unity. One can think that \(\mathcal{L}_n\) is a map from \(\mathbb{Z}[\mathbb{C}^{b}]\) to \(\mathbb R\) obtained by the composition of a map \(l_n\) which maps to the Hopf algebra of Aomoto polynomials over \(k\) with a map usually named volume map \(vol\) from a quotient of Aomoto polynomials to \(\mathbb{R}\) constructed from Hodge mixed theory. The paper under review obtains an analog for the \(n\)-polylogarithm map \(\mathcal{L}_n\) for the infinitesimal case instead of the classical one for \(\mathbb{C}\), which we denote by \(\mathcal{L}_n^{\varepsilon}\) i.e., consider \(k[\varepsilon]_2:=k[\varepsilon]/\varepsilon^2\), \textit{S.Ünver} defines in [``On the additive dilogarithm'', Algebra Number Theory 3, No. 1, 1--34 (2009; Zbl 1194.11073)] a map \(\mathcal{L}_2^{\varepsilon}:\mathbb{Z}[k[\varepsilon]_2^b]\rightarrow k\), and in the present paper under review is defined \(n\)-polylogarithm \(\mathcal{L}_n^{\varepsilon}\), \(k\) has always zero characteristic. The author first construct a family of possible \({\mathcal{L'}}_n^{\varepsilon}:k[\varepsilon]/(\varepsilon^n)^{b}\rightarrow k\) and proves that only one should factor through the natural projection to \(k[\varepsilon]/(\varepsilon^2)\), he call this possible polylogarithm by \(\mathcal{L}_n^{\varepsilon}\). Don Zagier proves the existence of \(\mathcal{L}_n^{\varepsilon}\), result included in this paper under review. In a remark the author includes a referee comment observing that the above construction has similarities with \textit{J. Dupont}'s construction for \(\mathcal{L}_n\) [``On polylogarithms'', Nagoya Math. J. 114, 1--20 (1989; Zbl 0743.33013)]. The author observes that \(\mathcal{L}^{\varepsilon}_n\) is related with the \(n\)-derivative of classical polylogarithm \(Li_1\), and \(\mathcal{L}_n^{\varepsilon}(-1+a\varepsilon)\) is related with \(\zeta(1-n)\) where \(\zeta\) is the classical Riemann zeta function. Finally the author proves in this paper that under some hypothesis the map \(\mathcal{L}_n^{\varepsilon}\) is not zero and it satisfies the expected functional equations for the polylogarithm: inversion formula, distribution formula, 9-term relation for the additive trilogarithm, 9-term relation for the additive tetralogarithm and getting back for \(\mathcal{L}^{\varepsilon}_2(x)\) obtains the Abel and Wojtkowiak function equation. The author also proves that this polylogarithm descend to an analog of the \(n\)-th Bloch group defined by \textit{A. Goncharov} [``Geometry of configurations, polylogarithms and motivic cohomology'', Adv. Math. 114, 197--318 (1995; Zbl 0863.19004)].