Adiabatic decomposition of the \(\zeta\)-determinant and scattering theory (Q874376)

The authors are interested in the interplay between the limit behaviour of some ratio, say \(A(R)\), depending on a parameter \(R\), associated to a compatible Dirac operator \({\mathcal D}:C^\infty(M,S)\to C^\infty(M,S)\), acting on sections of a Clifford bundle \(S\to M\), over a closed manifold \(M\) of dimension \(n\), (adiabatic decomposition of the \(\zeta\)-determinant of \({\mathcal D}^2\)), and scattering theory. The \(\zeta\)-determinant of the Dirac Laplacian \({\mathcal D}^2\) is given by the formula \(\det_\zeta{\mathcal D}^2=e^{-\frac{d\zeta_{{\mathcal D}^2}}{ds}}|s=0\), where \[ \zeta_{{\mathcal D}^2}(s)=\sum_{{\lambda_k\not=0}}(\lambda_k^2)^{-s}={{1}\over{\Gamma(s)}}\int_0^\infty t^{s-1}[TR(e^{-t{\mathcal D}^2})-\dim\ker{\mathcal D}]dt. \] This is a holomorphic function of \(s\) from \(\Re(s)\gg 0\) and has the meromorphic extension to the complex plane with \(s=0\) as a regular point. Let us denote by \(M_R\) a manifold \(M\) as before, such that it admits a bicollar neighbourhood \(N_R=[-R,R]_u\times Y\), \(Y\equiv M_{1,R}\bigcap M_{2,R}\), \(M_R=M_{1,R}\bigcup M_{2,R}\), such that the Riemannian structure on \(M_R\) and the Hermitian structure on \(S\), are products of the corresponding structures over \([-R,R]\) and \(Y\) when restricted to \(N_R\), so that \({\mathcal D}\) has the following form \({\mathcal D}_R=G_R(\partial_u+B_R)\) over \(N_R\). Here \(u\) denotes the normal variable, \(G_R:S|_Y\to S|_Y\) is a bundle automorphism and \(B_R\) is a corresponding Dirac operator on \(Y\), and satisfy \(G_R^*=-G_R\), \(G_R^2=-1\), \(B_R=B_R^*\), \(G_RB_R=-B_RG_R\). Then the authors study the following limit (adiabatic decomposition of the \(\zeta\)-determinant of \({\mathcal D}^2\)): \(\lim_{R\to\infty}A(R)\), with \[ A(R)\equiv{{\det_\zeta{\mathcal D}_R^2}\over{\det_\zeta({\mathcal D}_{1,R})^2_{P_1}).\det_\zeta({\mathcal D}_{2,R})^2_{P_2})}} \] where \({\mathcal D}_{i,R}={\mathcal D}_R| _{M_{i,R}}\), \({\mathcal D}_{i,R})^2_{P_i}={\mathcal D}_{i,R}:\text{ dom}({\mathcal D}_{i,R})_{P_i}\to L^2(M_{i,R};S)\), with \[ \text{ dom}({\mathcal D}_{i,R})_{P_i}\equiv\left\{s\in H^1(M_{i,R};S)| P_i(S| _Y)=0\right\},\quad i=1,2. \] \(P_i\) are suitable boundary condition projections, that impose elliptic conditions for \({\mathcal D}_{i,R}\). \(({\mathcal D}_{i,R})^2_{P_i}\) are self-adjoint Fredholm operators with kernels contained in \(C^\infty(M_{i,R},S)\) and having discrete spectra. The authors classify the eigenvalues of \({\mathcal D}_R\) in three different types as \(R\to\infty\): \(l\)-values, \(s\)-values, \(e\)-values. These make their contributions at different time intervals of the integral representation of \(\zeta_{{\mathcal D}^2}(s)\). The main result of the paper is a formula that calculates the following limit: \(\lim_{R\to\infty}R^{-2h}A(R)\), where \(h\) is a constant canonically associated to the problem. The methods adopted partially follow the Duhamel principle and the FPS (finite propagation speed) property, as considered by Cheeger, Gromov and Taylor. The scattering theory, as formulated by \textit{W. Müller} [J. Differ. Geom. 40, No.~2, 311--377 (1994; Zbl 0817.53025)], is used to get a description of the \(s\)-values of \({\mathcal D}_R\) and \(({\mathcal D}_{i,R})_{P_i}\), which allows to make a comparison of \(s\)-values of these operators with the eigenvalues of certain model operators over \(S^1\equiv{\mathbb R}\slash{\mathbb Z}\). Finally, the authors calculate the large time contribution of \(A(R)\) in the adiabatic limit. After a first detailed introduction, the paper splits into four more sections: 2. Small time conditions; 3. Small eigenvalues and scattering matrices; 4. Large time construction. 5. A proof of the decomposition formula of the \(\eta\)-invariant. (Compare with the above quoted paper by W. Müller, and previous results by Wojciechowski and other authors.)