From binary Hermitian forms to parabolic cocycles of Euclidean Bianchi groups (Q2116743): Difference between revisions

\textit{D. Zagier} [in: Number theory in progress. Proceedings of the international conference organized by the Stefan Banach International Mathematical Center in honor of the 60th birthday of Andrzej Schinzel, Zakopane, Poland, June 30--July 9, 1997. Volume 2: Elementary and analytic number theory. Berlin: de Gruyter. 1147--1178 (1999; Zbl 0959.11016)] introduced functions \(F_{k,D}:\mathbb{R}\to\mathbb{R}\) defined by \(F_{k,D}(x)=\sum Q(x)^{k-1}\) where \(k>0\) is a positive even integer, \(D>0\) is a non-square integer with \(D\equiv 0,1\bmod 4\), and where \(Q(X)=aX^2+bX+c\in\mathbb{Z}[X]\) ranges over all polynomials with \(a<0<Q(x)\) and \(\operatorname{disc}(Q)=b^2-4ac=D\). It turns out that for \(k=2\) resp. \(4\), these functions are constant with values \(\alpha_D\) resp. \(\beta_D\) related to special values of \(L\)-functions, and they still admit a reasonably explicit description for \(k\geq 6\). For example, it turns out that \(\alpha_D=-5L(\chi_D , -1)\) and \(\beta_D=L(\chi_D , -1)\), where \(\chi_D\) is the character given by the Kronecker symbol \(\bigl(\frac{D}{\cdot}\bigr)\). In the present paper, the author develops what one might call a Hermitian analogue of these functions \(F_{k,D}\). The set-up following the paper's conventions is as follows. Let \(K=\mathbb{Q}(\sqrt{-d})\) be a Euclidean imaginary quadratic number field, so \(d=1, 2, 3, 7, 11\), let \(d_K\) be the discriminant of that field and let \(\mathcal{O}_d\) be its ring of algebraic integers. Let \(h(X,Y)\) be a Hermitian form (with respect to complex conjugation) with Gram matrix \(\begin{pmatrix}a & b\\ \overline{b} & c\end{pmatrix}\), \(a,c\in\mathbb{Z}\), \(b\in\mathcal{O}_d\), and for \(z\in\mathbb{C}\), we put \(h(z):=h(z,1)\). The discriminant of \(h\) is given by \(ac-b\overline{b}=-\Delta\). Finally, let \(k>0\) be an odd integer. The author defines functions \(H_{k,\Delta}:\mathbb{C}\to\mathbb{R}\) by \(H_{k,\Delta}(z)=\sum h(z)^k\), where \(h\) ranges over all Hermitian forms as above of discriminant \(-\Delta\) and such that \(a<0<h(z)\). We assume furthermore in the following statements that \(\Delta\) is a positive integer that is not a norm of an element in \(\mathcal{O}_d\). The first main result (Theorem 1.1) states that \(H_{1,\Delta}(z)\) is constant, and so is \(H_{3,\Delta}(z)\) for \(d=1,3,7\) and \(H_{5,\Delta}(z)\) for \(d=3\). This is done by first showing that it is constant on \(K\) by applying certain transformations to \(z\in K\) that don't change the value of \(H_{k,\Delta}\) (at least in the considered cases) so that because of the Euclidean property, one can then reduce it to the case \(z=0\). Once this is done, one gets that it is constant on all of \(\mathbb{C}\) by showing that the function is continuous (Theorem 1.3). Using the \(\mathcal{O}_d\)-invariance of \(H_{k,\Delta}(z)\), one can define an average value for which one obtains the formula \[\langle H_{k,\Delta}(z)\rangle_{av}=\frac{2\pi\Delta^{k+1}}{(k+1)\sqrt{|d_K|}} \theta(\Delta,k+1)\zeta_{\mathbb{Q}}(k+1)L(\chi_{d_K},k+2)^{-1},\] where \(\theta\) is a certain finite Euler product. This then allows to compute certain special values of \(L\)-functions, for example, \(L(\chi_{d_K},3)=\frac{\pi^3\Delta^2}{6H_{1,\Delta}(0)\sqrt{|d_K|}}\theta(\Delta,2)\) and so, for example in the case \(d=1\) (i.e. \(d_K=-4\)), one gets \(L(\chi_{-4},3)=\frac{\pi^3}{32}\). In the final section of the paper, there is a detailed study of the space of parabolic cocycles on the Euclidean Bianchi group \(\textbf{PSL}(2,\mathcal{O}_d)\) (\(d=1,2,3,7,11\)) where the author obtains a description of this space in terms of a certain space \(W_{k,k}\) of polynomials which in turn contains a polynomial \(P_{k,\Delta}\) that is naturally associated to \(H_{k,\Delta}\), thus giving rise to a nontrivial cocycle on \(\textbf{PSL}(2,\mathcal{O}_d)\). For some odd values of \(k\), the dimensions of \(W_{k,k}\) are also determined assisted by a computer. The section concludes with a general conjecture on these dimensions and an outlook on how these results and conjectures might have an impact on the study of Bianchi cusp forms.