Asymptotic analysis of simultaneous damages in spatial Boolean models (Q2449392)

The authors consider the planar Boolean model \(\mathcal{D}\) generated by the stationary Poisson point process of germs and a sequence if i.i.d.\ grains being disks with random radius \(R\). It is well known that the distribution of \(\mathcal{D}\) is characterised by its capacity functional \(T(K)=\mathrm{P}\{\mathcal{D}\cap K\neq\emptyset\}\) which can be expressed in terms of the compact set \(K\) and the random variable \(R\). The authors introduce several other functionals that may be used to assess the spatial dependencies: \(C(K_1,K_2)\) being the probability that \(\mathcal{D}\) hits both \(K_1\) and \(K_2\), \(A(K_1,K_2)=C(K_1,K_2)-T(K_1)T(K_2)\), and \(T(K_1,K_2)\) being the probability that a single grain translated by the point of the Poisson process hits both \(K_1\) and \(K_2\). It is shown that \(A(K_1,K_2)\) is nonnegative, meaning that the Boolean model can be considered positively associated. Under regular variation assumptions on \(R\), the authors also derive asymptotic bounds on \(T(K_1,K_2)\) and \(A(K_1,K_2)\) as the minimum distance between the points of \(K_1\) and \(K_2\) grows to infinity.