Scattering by obstacles of finite capacity (Q1328298)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Scattering by obstacles of finite capacity
scientific article

    Statements

    Scattering by obstacles of finite capacity (English)
    0 references
    0 references
    3 January 1995
    0 references
    Assume that \(X\) is a locally compact, second countable space equipped with a Radon measure \(m\) of full support. Let \(G\subset X\) be an open set and \(B= X\backslash G\) considered as an obstacle. Assume further that a self-adjoint non-negative operator \(H\) on \(L_ 2(X)\) is given such that: (1) the form \(\mathfrak h\) associated to \(H\) is a regular Dirichlet form, (2) for any \(t>0\), \(\|\exp(- tH): L_ 1\to L_ 1\|< \infty\). Another operator \(H_ G\) on \(L_ 2(G)\) is associated to the closure of the form \({\mathfrak h}\mid D({\mathfrak h})\cap C_ c(G)\). Let \(J^*: L_ 2(G)\to L_ 2(X)\) be the embedding and \(J: L_ 2(X)\to L_ 2(G)\) be the orthogonal projection. The paper develops a method to study the difference \[ D(t):= \exp(- tH)- J^*\exp(- tH_ G)J. \] Sufficient conditions are stated for the operator \(D(t)\) to be Hilbert-Schmidt respectively trace class. As a consequence, one can show that the essential respectively absolutely continuous spectra of \(H\) and \(H_ G\) coincide. The method is based on two theoretical results. The first one is the existence of a Markov process \((\Omega,(\mathbb{P}^ x; x\in X)\), \(({\mathcal X}_ t; t>0))\), i.e. for each \(x\in X\), \((\Omega,\mathbb{P}^ x,({\mathcal X}_ t)_{t>0})\) is a stochastic process with the state space \(X\cup\{\infty\}\). Moreover, it holds \[ \exp(- tH)f(x)= \mathbb{E}^ x(f\circ{\mathcal X}_ t) \] for \(t\geq 0\), \(f\in L_ 2\) and a.e. \(x\in X\) [\textit{M. Fukushima}, `Dirichlet forms and Markov processes', Amsterdam (1980; Zbl 0422.31007)]. The second source is a factorization technique based on the `little Grothendieck theorem'' which yields a remarkable criterion for an operator to be Hilbert-Schmidt respectively trace class. The starting point of the method is then an application of the Feynman- Kac formula giving \[ \exp(- tH_ G)f(x)= \mathbb{E}^ x(\mathbf{1}_{\{\tau_ b>t\}}\cdot f\circ{\mathcal X}_ t). \] Here \(\tau_ B= \inf\{s\geq 0;\;{\mathcal X}_ t\in B\}\) is the hitting time of \(B\).
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    locally compact, second countable space equipped with a Radon measure of full support
    0 references
    essential spectra
    0 references
    little Grothendieck theorem
    0 references
    obstacle
    0 references
    regular Dirichlet form
    0 references
    orthogonal projection
    0 references
    Hilbert-Schmidt
    0 references
    trace class
    0 references
    absolutely continuous spectra
    0 references
    Markov process
    0 references
    stochastic process
    0 references
    state space
    0 references
    factorization technique
    0 references
    Feynman-Kac formula
    0 references
    0 references