Kloosterman sums for Clifford algebras and a lower bound for the positive eigenvalues of the Laplacian for congruence subgroups acting on hyperbolic spaces (Q1177114)

For the first positive eigenvalue \(\lambda_1\) of the Laplacian acting in \(L^2(\Gamma\backslash{\mathcal H}^{k+2})\), \(k\geq 2\), the estimate \(\lambda_ 1\geq k/2+1/4\) is proved. \(\Gamma\) is a congruence subgroup acting in hyperbolic \((k+2)\)-space \({\mathcal H}^{k+2}\). This generalizes Selberg's result \(\lambda_ 1\geq 3/16\) for \(k=0\), and Sarnak's result \(\lambda_ 1\geq 3/4\) for \(k=1\). [See \textit{A. Selberg}, Proc. Symp. Pure Math. 8, 1--15 (1965; Zbl 0142.33903), and \textit{P. Sarnak}, Acta Math. 151, 253--295 (1983; Zbl 0527.10022).] The central idea is the computation of the scalar product of two Poincaré series in terms of a Dirichlet series with generalized Kloosterman sums, the Linnik-Selberg series. The presence of a small eigenvalue is reflected in a singularity of this series at a point on the real axis. Suitable estimates for generalized Kloosterman sums give the convergence of the series in a right half- plane. This rules out eigenvalues that are too small. This idea is precisely Selberg's approach. But it is not easy to make it work for the general case \(k\geq 2\). I admire the organization of the difficulties and the description of the discrete groups \(\Gamma\) in such a way that the central idea stays visible. The question `what are the congruence subgroups \(\Gamma\) to work with?' can be answered in more than one way. If one views the hyperbolic space \({\mathcal H}^{k+2}\) as the quotient \(\mathrm{SO}^0(1,k+2)/\mathrm{SO}(k+2)\), it is sensible to use the rational structure of \(\mathrm{SO}_{k+3}(Q)\), with \(Q\) an isotropic quadratic form with rational coefficients and signature \((1,k+2)\). One may also use \(\mathrm{Spin}_{k+3}(Q)\). The image of a congruence subgroup of the spin group is not necessarily a congruence subgroup in \(\mathrm{SO}_{k+3}(Q)\). By working with the spin group the authors have chosen a wider scope. Working with spin groups is implemented as the use of Vahlen groups. Such a group consists of certain \(2\times 2\)-matrices with elements in the Clifford algebra \({\mathcal C}(q)\) associated to a negative definite quadratic form \(q\). Superficially it looks like working in \(\mathrm{SL}_2(\mathbb{R})\) or \(\mathrm{SL}_2(\mathbb{C})\), but here the matrix elements itself are complicated objects. The authors define the generalized Kloosterman sums needed to describe the scalar product of Poincaré series, and provide the estimates that give the convergence of the Linnik-Selberg series on the desired region. To get these estimates, a factorization theorem for generalized Kloosterman sums is given. The factors are estimated separately. The crucial cases are reduced to classical Kloosterman sums, and for these Weil's estimate is used.