Regularity of the eta function on manifolds with cusps (Q658332)

If \(A\) is an operator of ``Dirac type'', then the eta function \({\mathrm{Tr}}_{L^2}A\cdot (A^2)^{-(s+1)/2}\) is a measure of the spectral asymmetry of \(A\). If \(M\) is closed, it is well known that the origin is a regular value; this value plays an important role in the geometrical index theorem of Atiyah, Patodi, and Singer. In the present paper, the authors extend this result to a more general context. Let \(M\) be a compact manifold with boundary and let \([0,\epsilon)\times bd(M)\) be a collared neighborhood of the boundary. The interior \(M^o\) of \(M\) is called a conformally cusp manifold if it is endowed with a metric \(g_p\) which near \(x=0\) takes the form \[ g_p=x^{2p}(x^{-4}dx^2+h) \] for some \(p>0\) where \(h\) is a metric on the boundary independent of \(x\). The main examples are complete hyperbolic manifolds of finite volume for which \(p=1\) and \(h\) is flat. Let \(M^o\) be an odd dimensional spin manifold with conformal cusps, \(E\) a twisting bundle of product type on the cusps, and let \(D_p\) be the associated twisted Dirac bundle such that the boundary Dirac operator is invertible. Under these assumptions, the \(L^2\) spectrum of the essentially self-adjoint operator \(D_p\) is discrete and obeys a Weyl-type law; the eta function is holomorphic for \(Re(s)>\dim(M)\) and extends to a meromorphic function with possible \textit{double} poles at certain points. If \(\dim(M)\) is even, the eta function vanishes identically since the spectrum is symmetric. If \(\dim(M)\) is odd, \(s=0\) is a-priori a possible double pole [see \textit{S. Moroianu}, ``Weyl laws on open manifolds'', Math. Ann. 340, 1--21 (2008; Zbl 1131.58019)]. Under these assumptions, authors show: Main Theorem: The eta function is regular for \(Re(s)>-2\) and has at most simple poles at \(s=-2,-4,\dots\). If \(p=1\), if \(M\) is hyperbolic of finite volume, and if \(E\) is a homogeneous vector bundle, then the eta function is entire. Section 1 of the paper presents an introduction. Section 2 shows the eta invariant is a conformal invariant. Section 3 treats the Guillemin-Wodzicki residue density and residue trace. Section 4 contains a new proof that on a spin manifold, the local eta residue of the twisted Dirac operator vanishes. Sections 5 and 6 establish the main theorem using the cusp calculus of Melrose.