In this article we show that the proof of the homotopy reconstruction result by Niyogi, Smale, and Weinberger can be streamlined considerably using Federer's work on the reach and several geometric observations. While Niyogi, Smale, and Weinberger restricted themselves to C2 manifolds with positive reach, our proof extends to sets S of positive reach. The sample we consider does not have to lie directly on the set of positive reach. Instead, we assume that the two one-sided Hausdorff distances (delta and epsilon) -- between the sample P to the set S, are bounded. We provide explicit bounds in terms of epsilon and delta, that guarantee that there exists a parameter r such that the union of balls of radii r centered on the points of the sample P deformation retracts to S. We provide even better bounds for the manifold case. In both cases, our bounds improve considerably on the state-of-the-art in almost all settings. In fact the bounds are optimal.
翻译:在文章中,我们展示出尼优吉、斯马利和温伯格的单体重建结果的证据可以使用Federer关于距离和若干几何观测的作品大大简化。尼优吉、斯马利和温伯格将自己限制在C2柱子上,而我们的证据则扩大到S组正接触。我们认为样本不必直接放在正接触的一组上。相反,我们假设Hausdorf的两条单面距离(delta和epsilon) -- -- P到S组的样本之间的两条距离(delta和epsilon) -- -- 被捆绑起来。我们提供了epsilon和delta的清晰界限,保证存在一种参数,即射线球的结合集中在样本的点上。我们为多样的反射线提供了更好的界限。在这两种情况下,我们的界限几乎在所有环境下都大大改进了状态的界限。事实上,界限是最佳的。