Stochastic spectral theory for selfadjoint Feller operators: A functional integration approach

Publication date 5 September 2000

Feller operators Feynman-Kac formula Hamiltonian semigroup perturbation theory resolvents Schrödinger semigroup stochastic spectral analysis

Mathematics Subject Classification ID

Applications of functional analysis in quantum physics (46N50) Research exposition (monographs, survey articles) pertaining to operator theory (47-02) Spectrum, resolvent (47A10) Scattering theory of linear operators (47A40) Linear operators defined by compactness properties (47B07) Linear operators on function spaces (general) (47B38) One-parameter semigroups and linear evolution equations (47D06) Integral operators (47G10) Research exposition (monographs, survey articles) pertaining to probability theory (60-02) Continuous-time Markov processes on general state spaces (60J25) Transition functions, generators and resolvents (60J35) Probabilistic potential theory (60J45) Selfadjoint operator theory in quantum theory, including spectral analysis (81Q10)

Review text

The book under review is concerned with the following problems: Let \(E\) denote a locally compact second countable space, \(C_\infty\) the Banach space of continuous functions vanishing at infinity, and let \(m\) denote a non-negative Radon measure on \(E\). Furthermore, it is assumed that there exists a Feller semigroup \((T_t)\) on \(C_\infty\) defining a selfadjoint operator semigroup on \(L^2(E,m)\) with infinitesimal generator \(K_0\). Hence the notation \(T_t=\exp(-t\cdot K_0)\) is justified. The generator \(K_0\), called free Feller operator, defines a Markov process \((X(t): t\geq 0)\) on the state space \(E\). In particular, we have the representation \(\exp(-t\cdot K_0)f(x) =\mathbb{E}_X(f (X(t)))\). And furthermore, by the ``basic assumptions (formulated in Chapter 1) the transition operators have continuous densities w.r.t. \(m\). Throughout the book is concerned with the comparison of properties of semigroups defined by perturbations of the free Feller operator. Regular perturbations are defined by potential functions \(V\): A measurable function \(V\) belongs to \(K(E)\) if \(\|\int^t_0 \exp(-s \cdot K_0) \cdot V ds \|\to 0\) \((t\to 0)\). Define further \(K_{\text{loc}} :=\{V:V \cdot 1_B\in K\) for all compact \(B\}\) and regular Kato-Feller potentials are defined as \(V=V^+ -V^-\) with \(V^+\in K_{\text{loc}}\), \(V^-\in K\). In this case the sum \(K_0+V\) extends to the generator of a Feller semigroup, denoted by \(K\dot+V\). (Under suitable conditions \(K\dot+V\) coincides with the Friedrichs extension, as shown in Chaptfer 2). The perturbated operator generates a Markov process as well and the probabilistic interpretation is given by the Feynman-Kac formula \[ \mathbb{E}_X \left(\exp \left(-\int^t_0 V\bigl( X(u)\bigr) du\right) f\bigl(X(t) \bigr) \right)= \exp\bigl(-t\cdot (K_0\dot+V) \bigr) f(x). \] Singular perturbations are defined by a singularity region \(\Gamma\) (the definition is quite general: it is only assumed that \(\Gamma\) is closed in \(E)\), and \(\Sigma:= E\setminus \Gamma\). The corresponding Markov process and the transition semigroup \((P_\Sigma(t))\) are defined such that \(P_\Sigma(t) f(t) \equiv 0\) \(m\)-a.e. on \(\Gamma\). The corresponding infinitesimal generator on \(L^2(\Sigma, m|_\Sigma)\) is denoted by \((K_0 \dot+V)_\Sigma\). The aim of the book is to compare resolvents and spectral properties, in particular discrete and absolute continuous spectra, of the free and of the (regular and singular) perturbated Feller operators. The first chapter contains basic assumptions for the free operator \(K_0\) which are always assumed to be fulfilled, and which are general enough to cover e.g. general second-order elliptic partial differential operators, Laplace-Beltrami operators and (certain) pseudo-differential operators; in particular, generators of stable processes are included as well. The above-mentioned setup allows to consider the perturbated operators not only as Hilbert-space operators (on \(L^2 (\Gamma,m)\) and \(L^2 (\Gamma,m)\), respectively) but also as transition operators of Markov processes. Hencc not only T. Kato's perturbation theory and Trotter-Kato formulas (addition of unbounded operators) are available, furthermore stochastic models and powerful stochastic methods (in particular, strong Markov property, martingale methods,\dots) may be used for profound investigations. Chapters 2 and 3 are concerned with existence and structure of perturbated Feller operators (operator theory and probabilistic models). In particular, Chapter 3 contains the proofs of the essential features of the perturbated kernels, symmetry and continuity. Chapters 4-6 are concerned with the comparsion of properties of free and perturbated operators. (Throughout, in subsection A regular perturbations and in B singular perturbations.) In particular, operator norms, Hilbert-Schmidt and trace-class norms of differences of semigroups and resolvents are compared. The above-mentioned access to perturbations via stochastic models (``stochastic spectral analysis) allows a quantitative analysis of spectral properties (see in particular Chapter 8) which seem not to be available in the operator-theoretic setup. The book, a bridge between operator theory and stochastic analysis, is organized in a way that is more or less self-contained. Nevertheless, the reader is assumed to have a profound background in functional analysis, operator theory, in probability theory and stochastic processes. The surveys in Chapter 1 (heat kernels, Brownian motion, Brownian bridge, Schrödinger semigroups and Ornstein-Uhlenbeck processes...) and in the Appendix (operator and spectral theory, Feller semigroups and Markov processes, martingales, Dirichlet kernels and probabilistic potential theory) describe the toolbox which is necessary for the investigations. The reader will find precise hints to the background literature and to further investigations as well as a complete index of symbols and of subjects. The history of the main topics is quite long and also the number of authors who have contributed to the rapidly developing theory, too long to be mentioned in details in a review. The results are spread out in the literature (including the authors' own contributions since 1986). The book presents these investigations in a new and unified approach in the interplay of stochastics and analysis.

Recommendations

Cited in

(69)

This page was built for publication: Stochastic spectral theory for selfadjoint Feller operators: A functional integration approach

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4503255)