Dynamic Boolean models (Q1275948): Difference between revisions

Let \(X_{\lambda ,\rho }\) be a stationary Boolean model in \({\mathbb R}^d\) with intensity \(\lambda >0\) and primary grain being a ball of radius distribution \(\rho \). The critical intensity \(\lambda _c(\rho)\) is the supremum of such \(\lambda \) for which there is no unbounded connected component in \(X_{\lambda ,\rho }(t)\) almost surely. The authors show first that the property `there is no unbounded connected component' can be replaced e.g.\ by `there is no connected component of infinite volume', or `there is no connected component containing an infinite number of balls', without changing the value of the critical intensity. Then, a dynamics is introduced so that each ball moves in time independently of the others following a given stochastic process in \({\mathbb R}^d\), so that one gets a time-stationary process \(\{X(t):t>0\}\), each \(X_{\lambda ,\rho }(t)\) being a stationary Boolean model of the same distribution as \(X_{\lambda ,\rho }\). The authors show that for a supercritical intensity (\(\lambda >\lambda _c(\rho)\)), an unbounded connected component appears simultaneously in all \(X_{\lambda ,\rho }(t)\), \(t>0\), with probability one. Similarly, for a subcritical intensity (\(\lambda <\lambda _c(\rho)\)), there is no unbounded connected component in any \(X_{\lambda ,\rho }(t)\), \(t>0\), with probability one.