Grain rotations and distortions in the asymptotic variance of vacancy of the Boolean model (Q719600)

Consider a Boolean model \(\Xi = \bigcup_i(\xi_i + S_i)\subset {\mathbb{R}}^k\), \(k\geq 2\), where \(\{\xi_i\}\) is a homogeneous Poisson process with intensity \(\lambda>0\). Denote \(y+ B=\{y+z: z\in B\}\) for \(y\in\mathbb{R}^k\) and \(B\subset {\mathbb{R}}^k\). Random grains \(S_i\) are independent and identically distributed (i.i.d.) as the random closed set \(S\), and \(\{S_i\}\) is independent of \(\{\xi_i\}\). Besides, \(\operatorname{E}(|S|^2) <\infty\), where \(|\cdot|\) denotes \(k\)-dimensional volume. The vacancy of \(\Xi\) in a fixed Riemann measurable set \({\mathcal R}\subset \mathbb{R}^k\) of finite positive content is defined as \(V=|{\mathcal R}\setminus \Xi|\). As in [\textit{P. Hall}, Introduction to the theory of coverage processes. New York etc.: Wiley (1988; Zbl 0659.60024)] a Riemann measurable set is a set whose indicator function is directly Riemann integrable. As the distribution of \(V\) is complicated, it is of interest to study the limiting behavior under either a ``high intensity'' or ``moderate intensity'' regime. The authors consider the limit of \(\lambda\, \operatorname{Var}(V)\), that is the asymptotic variance of vacancy as \(\lambda \to \infty\) and the grains are distributed (and scaled) as \(\delta S\), where \(\delta=\delta(\lambda)\downarrow 0\) such that \(\delta^k\,\lambda\to \rho\in [0, \infty]\). Section 2.3 displays the limiting asymptotic variance of vacancy for the cases \(\rho<\infty\) and \(\rho=\infty\), as derived in [loc. cit.]. It is also known [loc. cit., 144--145 and 189--190] that in the both cases the asymptotic variance of vacancy does not increase when the grains are subjected to i.i.d. random uniformly distributed rotations. More precisely, in the setup of [loc. cit.] the grains are i.i.d. and distributed as \(S^* = {\mathbf T} S = \{\mathbf {T s}:\, {\mathbf s}\in S\}\), where \({\mathbf T}\) is distributed as the unique probability measure on \(SO(k)\) such that \(Q{\mathbf T} = {\mathbf T}\) in distribution. In the present paper, the authors investigate continuity properties of the asymptotic variance of vacancy when \({\mathbf T}\) is a rotation with arbitrary distribution. The paper is organized as follows. Section 2 contains notations and basic concepts, as well as the results on the asymptotic variance of vacancy obtained in [loc. cit.]. Sections 3 and 4 discuss, separately for the cases \(\rho<\infty\) and \(\rho=\infty\), optimality and continuity properties for distortions and continuity of rotations under symmetry assumptions. Section 5 titled ``concluding remarks'', completes the paper.