Multiple points of trajectories of multiparameter fractional Brownian motion (Q1281421)

This paper is concerned with upper bounds for the Hausdorff dimension gauge of the \(k\)-multiple points of multiparameter fractional Brownian motion, which is the \(N\) parameter Gaussian process \(X\) with values in \(d\)-dimensional space and variance \(E(| X(t)-X(s)| ^2)=d| t-s| ^{2\alpha}\). It is shown that if \(kN\leq (k-1)\alpha d\), there are no \(k\)-fold multiple points, if \(kN> (k-1)\alpha d\), the set of \(k\)-fold multiple points has \(\sigma\)-finite Hausdorff measure with respect to the gauge function \(\varphi(t)=t^{kN/\alpha-(k-1)d}(\log \log 1/t)^k\) and in the critical case \(N=\alpha d\) the set of \(k\)-fold multiple points has even \(\sigma\)-finite Hausdorff measure with respect to the gauge function \(\varphi(t)=t^d(\log(1/t)\log \log \log 1/t)^k\). Although the most important case here is the classical Brownian motion, and in this case the results above are known for some time, the Brownian motion case suffers from an overabundance of special properties and passing to a more general class of Gaussian processes forces one to find proofs that rely on more general principles and is thus a worthwile task. The readers of this paper will be rewarded with insights in modern techniques for the study of Gaussian processes and some imaginative proofs. In the classical situation of Brownian motion the upper bounds obtained here are sharp, but the paper leaves the question of lower bounds, which would require different techniques, open.