Self-Intersections of Stable Processes in the Plane: Local Times and Limit Theorems

From MaRDI portal





Let \(X_t\) be the symmetric stable process of index \(\beta>1\) in \(R^2\) with transition density \(p_t(x)\) and \(\lambda\)-potential \(G_\lambda(x)\). To study the \(k\)-fold self-intersections of \(X_t\) the author gives meaning to the formal expression \[ \idotsint\limits_{0\leq t_1\leq\ldots\leq t_k\leq t}\delta(X_{t_2}-X_{t_1})\cdots \delta(X_{t_k}-X_{t_{k-1}}). \] By replacing \(\delta\) function with its approximator \(f_ \varepsilon(x)=\varepsilon^{- 2}f(x/\varepsilon)\) the random variable \(\alpha_{k,\varepsilon}(t)\) is obtained. The subtraction from \(\alpha_{k,\varepsilon}(t)\) terms involving lower order intersections leads to the random variable \(\gamma_{k,\varepsilon}(t)\) defined as \[ \gamma_{k,\varepsilon}(t) = \sum_{j=1}^ k(-h_ \varepsilon)^{k- j}\binom{k-1}{j-1}\alpha_{j,\varepsilon}(t), \] where \(h_ \varepsilon=\int f_ \varepsilon(x)G_0(x)\,d^2x\). The limiting behavior of \(\gamma_{k,\varepsilon}(\zeta)\), where \(\zeta\) is an exponential random variable, depends on the value \(\alpha=(2k-1)(2- \beta)-2\). Namely, if \(\alpha<0\), then \(\gamma_{k,\varepsilon}(\zeta)\) converges in \(L^ 2\) to a non-trivial random variable. If \(\alpha=0\), then \(\gamma_{k,\varepsilon}(\zeta)(-\hbox{lg} \varepsilon)^{-1/2}\) converges in distribution to \(C_ 1L\), where \(C_ 1\) is a certain constant, \(L\) is a random variable with density \(\frac12 e^{-| x|}\). At last, if \(\alpha>0\) but \((2(k-1)-1)(2-\beta)<2\), then \(\varepsilon^{\alpha/2}\gamma_{k,\varepsilon}(\zeta)\) converges in distribution to \(C_2L\) for another constant \(C_2\). [For the entire collection see Zbl 0716.00012.]











This page was built for publication: Self-Intersections of Stable Processes in the Plane: Local Times and Limit Theorems

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q3972795)