No label defined (Q3972585)

[For the entire collection see Zbl 0685.00014.] Consider the selfadjoint operator \(H_{ren}\) defined by the formal expression \({1\over 2}\sum_{k\in\mathbb{Z}^ d}p^ 2_ k+{1\over 2}\sum_{i,k\in\mathbb{Z}^ d}d_{ik}x_ ix_ k\) in \(L_ 2(\mathbb{R}^{\mathbb{Z}^ d},\gamma_ s)\) where \(\gamma_ s\) is the Gaussian measure with the correlation function \(S=D^{-1/2}\) where \(D\) is the linear operator in \(\ell^ 2(\mathbb{Z}^ d)\) defined by the matrix \(d_{ik}\). It is assumed that \(\sigma(D)=\sigma_{ac}(D)=[m^ 2,M^ 2]\), \(0<m^ 2<M^ 2<\infty\). In fact \(H_{ren}\) is the Dirichlet operator of the measure \(\gamma_ s\) [see, e.g., \textit{Yu. M. Berezanskij} and \textit{Yu. G. Kondrat'ev}, Spectral methods in infinite dimensional analysis, Kiev, Naukova Dumka (1988; Zbl 0707.47001) (in Russian)]. Consider also the operator \(H_ 1=P^ 2/2+\omega_ 0^ 2 Q^ 2/2\) acting in \(L_ 2(\mathbb{R},\gamma_ 0)\), \(d\gamma_ 0(\kappa)=\sqrt{{w_ 0 \over\pi}}e^{-\omega_ 0\kappa^ 2}d\kappa\) and the operator \(V_ \lambda=\sum_{k\in\mathbb{Z}^ d}\lambda_ k x_ k Q\), \(\sum_{k\in\mathbb{Z}^ d}\lambda_ k^ 2<\infty\). Then the operator \(H_{\lambda,ren}=\mathbf{1}\otimes H_{ren}+H_{ren}\otimes\mathbf{1}+V_ \lambda\) describes the interaction of an infinite system with a single oscillator. Let \(\varphi(t)=(Q(t)1,1)L_ 2(\mathbb{R}^{\mathbb{Z}^ d},\gamma_ s)\), \({\mathcal D}_ \lambda= \begin{pmatrix} \omega^ 2_ 0 & \ell_ \lambda\\ \ell^*_ \lambda & D \end{pmatrix}\) where \(\ell_ \lambda x=\sum_{k\in\mathbb{Z}^ d}\lambda_ k x_ k\), \((\ell_ \lambda^*\kappa)_ k=\kappa\lambda_ k\). Theorem 1. If \(\| S\lambda\|\leq\omega_ 0\) and \(\sigma({\mathcal D}_ \lambda)=\sigma_{ac}({\mathcal D}_ \lambda)\) then \(\lim_{t\to\infty}\varphi(t)=0\). Theorem 2. Let \(\Delta(s)=\omega_ 0^ 2-s+((s-D)^{- 1}\lambda,\lambda)\). Suppose that the domain of definition \({\mathcal D}(\Delta)\) of \(\Delta(s)\) contains \([m^ 2,M^ 2]\) and \(\Delta(s)\neq0\), \(s\in{\mathcal D}(\Delta)\cap[m^ 2,M^ 2]\). Then \(\sigma({\mathcal D}_ \lambda)=\sigma_{ac}({\mathcal D}_ \lambda)\cap\Sigma\), where \(\Sigma\) is the point spectrum of \({\mathcal D}_ \lambda\) specified by the table \[ \begin{array}{|l|l|l|} \hline & \Delta(M^ 2)<0 & \Delta(M^ 2)\geq0 \\ \hline \Delta(m^ 2)>0 & \Sigma=\emptyset & \Sigma=\{\omega_ M\} \\ \hline \Delta(m^ 2)\leq0 & \Sigma=\{\omega_ m\} & \Sigma=\{\omega_ m,\omega_ M\} \\ \hline \end{array} \] where \(\omega_ m<m^ 2\), \(\omega_ M>M^ 2\).