The Hausdorff measure of the range and level sets of Gaussian random fields with sectorial local nondeterminism (Q2073214): Difference between revisions

Let \(B(x):\mathbb R^N_+\to\mathbb R^d\) be a Brownian sheet with the same Hurst index \(\alpha\) in all directions. Denote by \(\mathcal H_\phi(\cdot)\) the Hausdorff measure with an exact dimension function \(\phi\) (gauge function) and denote by \(|\cdot|\) the \(N\)-dimensional Lebesgue measure. Fix some compact box \(I\subset \mathbb R^N_+\). The author shows that \begin{itemize} \item if \(N<\alpha d\), then for some finite positive \(C_1,C_2\) with probability one, \[ C_1|J|\leq \mathcal H_\phi(B(J))\leq C_2|J|\text{ for all sub-boxes } J\subset I, \] where \(\phi(\varepsilon)=\varepsilon^{N/\alpha} (\log\log(1/\varepsilon))^{N}\); \item if \(N>\alpha d\), then for some finite positive \(C\) and any fixed \(z_0\in \mathbb R^d \) with probability one, \[ CL(z_0,J)\leq \mathcal H_\phi(B^{-1}(z_0)\cap J)<\infty \text{ for all sub-boxes } J\subset I, \] where \(\phi(\varepsilon)=\varepsilon^{N-\alpha d} (\log\log(1/\varepsilon))^{\alpha d}\) and \(L(\cdot,J)\) denotes the local time of \(B\) on \(J\). \end{itemize} Moreover, for some finite positive \(K_1,K_2\) depending on \(I\) there exists a function \(K: I\to [K_1,K_2]\) such that for any fixed \(x_0\in I\) with probability one, \[ \liminf_{\varepsilon\to0}\sup_{|x-x_0|_\infty\leq\varepsilon}\frac{|B(x)-B(x_0)|}{\varepsilon^\alpha(\log\log(1/\varepsilon))^{-\alpha}}=K(x_0). \] In fact, all these results have been obtained for a class of Gaussian fields satisfying sectorial local nondeterminism and some additional assumptions. This class also contains the solution of a system of linear stochastic wave equations in one spatial dimension driven by space-time white noise or coloured noise.