The Heckman-Opdam Markov processes (Q880941): Difference between revisions

Let \({\mathcal A}\) be an \(n\)-dimensional Euclidean vector space. A finite family \({\mathcal R}\subset {\mathcal A}\) is called an integral root system if it satisfies \(r_\alpha({\mathcal R})={\mathcal R}\) and \(\alpha\check{}({\mathcal R}) \subset Z\), where \(r_\alpha(x)=x-(\alpha\check{},x)\alpha\) and \(\alpha\check{}=(2/| \alpha| ^2)\alpha\). For a vector \(u\) such that \((\alpha,u)\neq 0\) for all \(\alpha\in {\mathcal R}\), the positive root system \({\mathcal R}^+\) is a subset of roots \(\alpha \in {\mathcal R}\) such that \((\alpha,u)>0\). The Weyl chambers are the cones delimitted by the hyperplanes orthogonal to the vectors of \({\mathcal R}\). Also the positive Weyl chamber \({\mathcal A}_+\) is defined by \({\mathcal A}_+=\{x\in {\mathcal A}:(\alpha,x)>0,\forall \alpha \in {\mathcal R}^+\}\). For a fixed vector \(\xi\), let \(T_\xi\) be the Dunkl-Cheredrick operator and \({\mathcal L}\) the Laplacian defined by \({\mathcal L}=\sum_{i=1}^n T^2_{\xi_i}\) for an orthonormal basis \(\{\xi_1,\dots,\xi_n\}\). The Hechman-Opdam process (HO-process) is defined as the process with generator \(D={1\over 2}({\mathcal L}-| \rho| ^2)\) for \(\rho=(1/2)\sum_{\alpha\in {\mathcal R}^+}k(\alpha)\alpha\). The radial process \(X^W_t\) on \(\bar{\mathcal A}^+\) is defined as the process corresponding to the differential part of \(D\). In this paper, using the stochastic calculus related to the Dirichlet forms, the SDE characterizing \(X^W_t\) is given. The law of large number, the central limit theorem and some characterizations of the jumping part of \(X_t\) are given. Let \(X^T_t=\sqrt{T}X_{t/T}\) be the normalized HO-process. Then the convergence of \(X^T_t\) to the Dunkl process as \(T\to \infty\) is shown. As the main result, it is shown that the bridge of the radial HO-process of length \(T\) converges in law to some process. Furthermore, after a normalization, this process converges to the intrinsic Brownian motion.