Density results for automorphic forms on Hilbert modular groups (Q1413650): Difference between revisions

Let \(F\) be a totally real number field of dimension \(d\), and let \(\mathcal O_F\) be its ring of integers. Let \(\mathbf G:=\text{Res}_{F/\mathbb Q}(\text{SL}_2)\) denote the algebraic group obtained by restriction of scalars applied to \(\text{SL}_2\) over \(F\). Let \(\sigma_1, \dots, \sigma_d\) embeddings of \(F\) into \(\mathbb R\). We have \(G:=\mathbf G_\mathbb R=\text{SL}_2(\mathbb R)^d\) and \(\mathbf G_\mathbb Q\simeq\left\{(x^{\sigma_1},\dots, x^{\sigma_d})\,:\,x\in\text{SL}_2(F)\right\}\). The group \(G\) contains \(K:=\prod_{j=1}^d\text{SO}_2(\mathbb R)\) as a maximal compact subgroup. The image of \(\text{SL}_2(\mathcal O_F)\subset\text{SL}_2(F)\) corresponds to \(\mathbf G_\mathbb Z\). This is a discrete subgroup of \(\mathbf G_\mathbb R\) with finite covolume. It is called the {Hilbert modular group}. Let \(\mathfrak q\) be a non-zero ideal in \(\mathcal O_F\), and let \(\Gamma=\Gamma_0(\mathfrak q)\) denote the congruence subgroup of Hecke type of the Hilbert modular group, that is, \[ \Gamma:=\left\{\left(\begin{matrix} a&b\cr c&d\end{matrix}\right)\in\text{SL}_2(\mathcal O_F)\,:\,c\in\mathfrak q\right\}. \] In the present paper, the authors obtain density results for cuspidal automorphic representations of \(G\) in \(L^2(\Gamma(\mathfrak q)\backslash G)\). The main result, Theorem 3.3, implies that there are infinitely many cuspidal automorphic representations \(\varpi=\bigotimes_{j=1}^d \varpi_j\), even if one restricts some components \(\varpi_j\). The results obtained in the paper imply that there are infinitely many automorphic representations that have a nonzero Fourier coefficient of order \(r\) and components with a prescribed type; that is, for each \(j\), the type of \(\varpi_j\) can be prescribed to be either unitary principal series or discrete series. To prove these density results, the authors use a sum formula of Kuznetsov type for discrete cofinite subgroups \(\Gamma\subset G\), in which all weights contribute. They obtained this sum formula in a previous paper, and applied it there to give estimates for averages of Kloosterman sums for \(F\).