Stationary random measures on homogeneous spaces

Geometric probability and stochastic geometry (60D05) Point processes (e.g., Poisson, Cox, Hawkes processes) (60G55) Random measures (60G57) Random fields (60G60)

Review text

This paper treats Palm theory for stationary random measures. Especially, the author discusses stationary random measures \(\xi\) on a homogeneous space \(S\) and their Palm measures \({\mathbb P}_{\xi}\). A simple but essential tool is a deterministic transformation of a stationary random measure on \(S\) into a stationary random measure on the group \(G\). Because this transformation preserves the Palm measure, it allows us to extend recent results for groups to general homogeneous spaces in a short way. It starts with such fundamental properties as the refined Campbell theorem. That is to say, let \({\mathbb P}\) be a \(\sigma\)-finite invariant measure on \((\Omega, {\mathcal F})\), \(\xi\) be an invariant random measure on \(S\). Then the invariant random measure \(\xi'\) on a locally compact second countable Hausdorff space \(G\) is defined by \[ \xi' = \int k(x, \cdot) \xi(dx) \] with a kernel \(k\) on \(G\). Let \({\mathbb P}_{\xi'}\) be the Palm measure on \(\Omega\), i.e., \[ {\mathbb P}_{\xi'}(A) = {\mathbb E} \int 1 \{ \theta_g^{-1} \in A \} w'(g) \xi'(dg), \quad \text{for} \quad A \in {\mathcal F}, \] where \(w' : G \to {\mathbb R}_+\) is a measurable function having \[ \int w'(g) \lambda(dg) = 1, \] and the flow \(\theta _g\) is a mapping from \({\mathcal F}\) to \({\mathcal F}\) for \(g \in G\). The refined Campbell theorem connects \({\mathbb P}\) and \({\mathbb P}_{\xi'}\) in the following way: \[ {\mathbb E} \int f( \theta_g^{-1}, g) \xi'(dg) = {\mathbb E}_{ {\mathbb P}_{\xi'} } \int f(\theta_e, g) \lambda(dg) \] for any measurable function \(f : \Omega \times G \to {\mathbb R}_+\), where \(\lambda\) is a left-invariant Haar measure on \(G\). Then the author discusses invariant transports, invariance and transport properties of Palm measures, and stationary partitions. \noindent Theorem 1. Let \(\xi\) and \(\eta\) be two invariant random measures, and \(T\) be an invariant weighted transport-kernel on \(S\). Then \(T\) is \({\mathbb P}\)-a.e. \((\xi, \eta)\)-balancing \[ \int T(\omega, x, \cdot) \xi( \omega, dx) = \eta(\omega, \cdot), \quad \forall \omega \in \Omega \] if and only if \[ {\mathbb E}_{\xi} \iint f(\theta_g^{-1} ) \varDelta^*(x) k(x,dg) T(c, dx) = {\mathbb E}_{\eta} f \] for any measurable function \(f: \Omega \to {\mathbb R}_+\), where the function \(\varDelta^*\) : \(S \to\) \((0, \infty)\) is defined by \(\varDelta^*(x)\) \(=\) \(\varDelta ( g_x^{-1})\), \(x \in S\), \(g_x \in G_{c,x}\) with a modular function \(\varDelta\) : \(G \to\) \((0, \infty)\) and \(G_{c,x}\) \(:=\) \(\{ g \in G\); \(gc = x \}\). \noindent Theorem 2. Let \(\xi\) and \(\eta\) be invariant random measures on \(S\) with positive and finite intensities. Then there exists a \({\mathbb P}\)-a.e. \(( \xi, \eta)\)-balancing invariant weighted transport-kernel \(T\) satisfying \[ \int \varDelta^*(x) T(c, dx) = 1 \] if and only if \({\mathbb E} [ \xi(B) | {\mathcal I}] = {\mathbb E}[ \eta(B) | {\mathcal I} ]\), \({\mathbb P}\)-a.e. for some \(B \in {\mathcal S}\) satisfying \(0 <\) \(\mu(B) <\) \(\infty\). Moreover \noindent Theorem 3. Let \(\xi\) be an invariant random measure on \(S\). The measure \({\mathbb Q}\) on \(\Omega\) is a Palm measure of \(\xi\) at \(c\) with respect to some \(\sigma\)-finite invariant measure if and only if \({\mathbb Q}\) is \(\sigma\)-finite and invariant under \(G_c\), \({\mathbb Q}(\xi(S)\) \(= 0)\) \(=0\), and \[ {\mathbb E}_{ {\mathbb Q} } \iint f(\theta_g^{-1}, g^{-1}c) \varDelta^*(x) k(x, dg) \xi(dx) = {\mathbb E}_{ {\mathbb Q} } \int f(\theta_e, x) \xi(dx) \] for any measurable function \(f : \Omega \times S \to {\mathbb R}_+\). For some basic results on invariance and transport properties of Palm measures, see e.g.\textit{G. Last} and \textit{H. Thorisson} [Ann.\ Probab. 37, No.2, 790--813 (2009; Zbl 1176.60036)]; and \textit{J. Mecke} [Z.\ Wahrsch. Verw.\ Geb.\ 9, 36--58 (1967; Zbl 0164.46601)] for Mecke's intrinsic characterization of Palm measure, and see also \textit{W. Rother} and \textit{M. Zähle} [Math.\ Nachr. 149, 255--263 (1990; Zbl 0723.60054)] for a basic result on Palm distributions in homogeneous spaces.

Recommendations

Cites work

Cited in

(22)

This page was built for publication: Stationary random measures on homogeneous spaces

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q975332)