Curvature-direction measures of self-similar sets (Q387587)

The authors obtain a description of the Lipschitz-Killing curvature-direction measures for a wide class of self-similar sets in Euclidean spaces. These measures are the analogues of the high orders mean curvatures of smooth sub-manifolds in the setting of fractal geometry. The authors get a nice decomposition of these measures as products of the Haussdorff measure in the self-similar set and a self-similar fiber measure on the sphere (measuring the distribution of normal vectors). The self-similar sets considered in the paper are given by a Hutchinson's attractor [\textit{J. E. Hutchinson}, Indiana Univ. Math. J. 30, 713--747 (1981; Zbl 0598.28011)] with the strong open set condition, which assumes that the domains used to define the attractor can only overlap in a subset of their boundaries after the application of a contracting similarity. This condition is satisfied for a wide class of fractals as the Sierpinski gasket or the Menger sponge. There are another two assumptions on parallel sets of the self-similar set considered. These conditions are fulfilled in the case of an Euclidean space of dimension \(\leq 3\). A parallel set is an \(r\)-tube neighbouring the fractal, the required conditions states that for almost every \(r > 0\) the complement of this set has positive reach, i.e., there is a notion of normal bundle, and this normal bundle cannot contain simultaneously a vector and an antipodal copy of itself (orientability condition). These conditions are needed to define curvature measures in the parallel sets. The most beautiful part of the paper is a very interesting application of the ergodic theorem from dynamical systems. Almost every point of the self similar set is encoded one-to-one with a sequence of applications of contracting similarities. The shift of this encoding induces an ergodic dynamical system in the Hutchinson's attractor (see [\textit{K. Falconer}, Techniques in fractal geometry. Chichester: John Wiley \& Sons (1997; Zbl 0869.28003)] and [\textit{R. D. Mauldin} and \textit{M. Urbański}, Graph directed Markov systems. Geometry and dynamics of limit sets. Cambridge: Cambridge University Press (2003; Zbl 1033.37025)]) and, hence in the parallel sets. This system preserves covariant neighborhoods. This is the main step in Theorems 3.5, 3.9 and Corollary 3.10. As a consequence curvature-measures coincide, in some sense, almost everywhere and can be computed easier. Remark that some results are achieved with slightly better hypothesis as previously known. The paper is highly technical and it is mainly directed to experts in fractal geometry. Last section, which includes the examples, is just as needed in order to verify the results in the simple case of a ``twisted'' Sierpinski gasket.