Intrinsic ultracontractivity for stable semigroups on unbounded open sets (Q2391050)

From MaRDI portal
scientific article
Language Label Description Also known as
English
Intrinsic ultracontractivity for stable semigroups on unbounded open sets
scientific article

    Statements

    Intrinsic ultracontractivity for stable semigroups on unbounded open sets (English)
    0 references
    0 references
    24 July 2009
    0 references
    Let \((X_t)\) denote the isotropic stable process of index \(\alpha<2\), i.e., a subordination of a standard Brownian motion by an \(\alpha/2\)-stable subordinator on \(\mathbb R_+\). Let \(D\subseteq \mathbb R^d\) be an open -- here unbounded -- domain. Let, furthermore, \(\tau_D\) denote the first exit time from \(D\) and define \((P_t^D)\) to be the transition semigroup of the process killed on exiting \(D\), \(P_t^Df(x)=E_x(\tau_D>t; f(X_t))= \int p_t^D(x,y)f(y)\,dy\). Here \(p_t^D\) denotes the kernel of \(P_t^D\). Throughout it is assumed that the operators \(P_t^D\) are compact operators on \(L^2\) and \((\varphi_n = \varphi_n^D)_{n\geq 1}\) is a complete ONS of eigenfunctions with corresponding eigenvalues \(\lambda_n = \lambda_n^D\). \((P_t^D)\) is called intrinsically ultracontractive (IU) if for all \(t>0\) we have \(p_t^D(x,y)\leq C_{D,t}\varphi_1(x)\varphi_1(y)\). Previous investigations were often restricted to the case of bounded domains or \(\alpha =2\), i.e., to Brownian motions \((X_t)\). Since for \(\alpha < 2\) the process is a jump-process, the results differ essentially. The first results are obtained for general open domains, in particular (Theorem 1), \(\varphi_1(x) \approx \left(1+|x|\right)^{-d-\alpha} E_x(\tau_D)\) and (Theorem 2) property (IU) is equivalent to an upper estimate \(p_t^D(x,y)\leq C_{D,t}\left(1+|x|\right)^{-d-\alpha}\left(1+|y|\right)^{-d-\alpha}\). For horn-shaped domains \(D=D_f= \{x:x_1>0\), \(|(x_2, \dots x_d)|\leq f(x_1)\}\) which are non-degenerate, property (IU) is characterized by the behaviour of \((f(u))^\alpha\log u\) for \(u\to\infty\) (Theorem~5). The proofs rely on boundary Harnack inequalities, in fact, on a uniform version proved by \textit{K.\,Bogdan, T.\,Kulczycki} and \textit{M.\,Kwaśnicki} [Probab.\ Theory Relat.\ Fields 140, No.\,3--4, 345--381 (2008; Zbl 1146.31004)]; see also the literature there. For the case of Brownian motions and unbounded domains, see, e.g., \textit{B.\,Davies} [J.~Funct.\ Anal.\ 100, No.\,1, 162--180 (1991; Zbl 0766.47026)], \textit{P.\,J.\,Méndes-Hernándes} [Mich.\ Math.\ J.\ 47, No.\,1, 79--99 (2000; Zbl 0978.47032)] and the references therein.
    0 references
    0 references
    0 references
    0 references
    0 references
    intrinsic ultracontractivity
    0 references
    isotropic stable process
    0 references
    first eigenfunction
    0 references
    first exit time
    0 references
    horn shaped domain
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references
    0 references