The visible set of a compactum in \(\mathbb{R}^n\) (Q5928495)

Let \(a\) be a point in \(A\subset\mathbb{R}^n\). We say that the point \(a\) can be seen from infinity if there is a ray that meets \(A\) in exactly \(\{a\}\). More generally, \(a\) is \(k\)-visible with respect to \(A\) if there exists a polygonal ray which lies in the union of \(k\) lines and meets \(A\) exactly in \(\{a\}\). The \(k\)-visible set of \(A\) consists of all points in \(A\) that are \(k\)-visible. The main theorem says that we can practically ``hide'' a compactum by arbitrary small ambient isotopy. Precisely, the theorem says the following: Let \(K\) be a compactum and \(f:K\to\mathbb{R}\) an embedding. Given \(\varepsilon>0\) and an integer \(k>0\), there exists an embedding \(g:K\to\mathbb{R}^n\) such that \(f\) and \(g\) are \(\varepsilon\)-ambient isotopic and \(g(K)\) has the property that the \(k\)-visible set lies in a compact 0-dimensional subset of \(g(K)\). Of course, the statement of the theorem is of interest if \(n>1\). For \(n=2\), a simple closed curve \(J\) (named salamander curve) is constructed and a Cantor set \(C\subset J\) so that the 1-visible set of \(J\) lies in \(C\), which illustrates what is happening in higher dimensions.