On the appearance of Eisenstein series through degeneration (Q951962)

Let \(\Gamma\) be a Fuchsian group acting on the hyperbolic upper half-plane \(\mathbb{H}\). A canonical method of constructing a function which is invariant under \(\Gamma\) is by taking averages of some ``nice'' function \(f\) over a quotient \(\Gamma'\backslash\Gamma\) where \(\Gamma'\) is a subgroup of \(\Gamma\) (of finite or infinite index). By modifying the average (i.e. using the so-called slash action) one can also obtain functions with more general transformation properties (i.e. non-zero weight and/or non-trivial multiplier system). These types of averages are usually called \textit{Poincaré series} and the simplest examples of such functions are the \textit{Eisenstein series} (holomorphic or non-holomorphic). Non-holomorphic (parabolic) Eisenstein series occur naturally in the spectral theory of the hyperbolic Laplacian \(\Delta=-y^{2}\left(\frac{\partial^{2}}{\partial x^{2}}+\frac{\partial^{2}}{\partial y^{2}}\right)\) on the quotient orbifold \(\mathcal{M}=\Gamma\backslash\mathbb{H}\) (recall that an orbifold is essentially a Riemannian manifold where ``corners'' and cusps are also allowed). If \(\Gamma\) has parabolic elements (i.e. \(\mathcal{M}\) has cusps) then it has a continuous spectrum with multiplicity equal to the number of inequivalent cusps. The eigenpacket corresponding to scattering from a cusp \(p\) in \(\Gamma\) is given by the Eisenstein series with respect to \(\Gamma_{p},\) the stabilizer of \(p\) in \(\Gamma.\) For example, if \(p\) is the cusp at \(\infty\), the corresponding Eisenstein series is \[ E_{\infty}\left(z;s\right)=\sum_{\gamma\in\Gamma_{\infty}\backslash\Gamma}\Im\left(\gamma z\right)^{s} \] where \(z\in\mathbb{H}\) and \(\Re s>1\). The function \(E_{\infty}\) has a meromorphic continuation, satisfies a functional equation (in terms of the vector of Eisenstein series with respect to all the cusps of \(\Gamma\)) and satisfies \(\Delta E_{\infty}\left(z;s\right)=s\left(1-s\right)E_{\infty}\left(z;s\right)\). Other than parabolic subgroups of \(\Gamma\) there are two other types of possible (non-trivial) subgroups generated by stabilizers of points in the upper half-plane or the real line: elliptic and hyperbolic. If \(e\in\mathbb{H}\) is the fixed point of an element of \(\Gamma\) then this is said to be an elliptic fixed-point and the stabilizer \(\Gamma_{e}\) is necessarily a finite cyclic group. Similarly, if \(x,y\in\mathbb{R}\) are two different fixed points of some \(\gamma_{h}\in\Gamma\) then the geodesic in \(\mathbb{H}\) with endpoints in \(x\) and \(y\) is fixed under the infinite cyclic subgroup \(\Gamma_{h}\) generated by \(\gamma_{h}\). One can now construct elliptic and hyperbolic Eisenstein series in much the same manner as the parabolic Eisenstein series above. The sum over \(\Gamma_{p}\backslash\Gamma\) will instead be taken over \(\Gamma_{e}\backslash\Gamma\) and \(\Gamma_{h}\backslash\Gamma\). Similarly, the imaginary part, \(z\mapsto\Im z\), which is closely related to the hyperbolic distance function in cartesian coordinates, is replaced by the corresponding function under a suitable change of coordinates. Let \(\left\{ \mathcal{M}_{n}=\Gamma_{n}\backslash\mathbb{H}\right\} \) be an elliptically degenerating family of orbifolds (of finite volume) with limit surface \(\mathcal{M}_{\infty}=\Gamma_{\infty}\backslash\mathbb{H}\). By elliptical degeneration we mean that a ``corner'' is degenerating into a cusp. The main result in the paper under review is that the Eisenstein series are ``well-behaved'' under this kind of degeneration. To be precise: all parabolic, hyperbolic and non-degenerating elliptical Eisenstein series \(E_{*}\) on \(\mathcal{M}_{n}\) converge to the corresponding Eisenstein series on \(\mathcal{M}_{\infty}\). Furthermore, the elliptic Eisenstein series of a degenerating elliptic element converges to the parabolic Eisenstein series on \(\mathcal{M}_{\infty}\) associated to the newly formed parabolic point. In all cases the convergence is uniform on compact subsets bounded away from the degenerating points and half-planes \(\Re s\geq1+\delta\). The method of proof is to first construct certain counting functions and rewrite the Eisenstein series in terms of Stieltjes integrals with respect to the corresponding counting measures. The convergence of the Eisenstein series then essentially follow from convergence of the counting functions. Other than elliptic, one can also consider a family of hyperbolically degenerating orbifolds, i.e. ``pinching'' along a geodesic until a cusp is formed. The authors remark that the corresponding elliptical Eisenstein converge in this situation as well (the convergence of parabolic and hyperbolic Eisenstein series in this setting was proved earlier by \textit{D. Garbin, J. Jorgenson} and \textit{M. Munn} [Comment. Math. Helv. 83, No. 4, 701--721 (2008; Zbl 1154.30032)]).