On sets of directions determined by subsets of \(\mathbb R^d\) (Q351365): Difference between revisions

Given a subset \(E\) of \(\mathbb R^d\), with \(d\geq 2\), define \[ \mathcal D(E):= \left\{\frac{x-y}{|x-y|}:x,y\in E \text{ and }x\neq y \right\} \] to be the set of directions determined by \(E\). It is proved in Theorem 1.2 that if the Hausdorff dimension of \(E\) is strictly greater than \(d-1\), then \(\sigma(\mathcal D(E))>0\), where \(\sigma\) denotes the Lebesgue measure on the \((d-1)\)-dimensional unit sphere \(S^{d-1}\). The conclusion does not hold in general for \((d-1)\)-dimensional subsets of \(\mathbb R^d\) (as an obvious counter-example, consider a \((d-1)\)-dimensional hyperplane). A less trivial counterexample can be constructed following \textit{K. Simon} and \textit{B. Solomyak} [Real Anal. Exch. 32(2006-2007), No. 1, 67--87 (2007; Zbl 1126.28009)]: there exists a subset \(E\) of \(\mathbb R^d\) with Hausdorff dimension \(d-1\) that is not contained in a hyperplane and for which the Lebesgue measure of \(\mathcal D(E)\) is zero. On the other hand, \textit{T. Orponen} and \textit{T. Sahlsten} [Ann. Acad. Sci. Fenn., Math. 36, No. 2, 677--681 (2011; Zbl 1236.28003)] have recently proved that if \(E\) is a rectifiable subset of \(\mathbb R^d\) that has Hausdorff dimension \(d-1\) and is not contained in any hyperplane, then \(\sigma(\mathcal D(E))>0\). Theorem 1.2 is viewed by the authors as the continuous analog of a result of \textit{J. Pach, R. Pinchasi} and \textit{M. Sharir} [Discrete Comput. Geom. 38, No. 2, 399--441 (2007; Zbl 1146.52011)] on directions determined by finite subsets of \(\mathbb R^d\) (an explicit quantitative connection between the continuous and the discrete counterparts of this problem is given in the third section of the paper). (The authors report at the beginning that \textit{P. Mattila} has recently pointed out to them that their main result can be derived from Theorem 10.11 in his book [Geometry of sets and measures in Euclidean spaces. Fractals and rectifiability. Cambridge: Univ. Press (1995; Zbl 0819.28004)]; however, their method, based on Fourier analysis, allows a more detailed information about the direction set measure and the associated Radon-Nikodym derivative.)