Sets of locally maximal visibility and finite unions of starshaped sets (Q1568083)

The visibility set in \(S\) of \(t \in S\) is defined as \(V_t=\{y: t\) sees \(y\) via \(S\}\) (i.e. \( [y,t] \subseteq S\)) similarly, for \(T \subseteq S\), its visibility set in \(S\) is \(V_{T} = \cup\{ V_{t} : t \in T\}\). A subset \(T\) has locally maximal visibility in the set \(S\) iff there is some neighbourhood \(N\) of \(\operatorname {cl} T\) such that for \(s\) in \(N \cap S\), \(V_{s} \subseteq V_{T}\). The convex hull of three non-collinear points is called triangular region and its relative boundary is the union of its edges. A set \(S\) is called starshaped iff there is some point \(p\) in \(S\) such that \(V_{p} = S\); finally the kernel of the set \(S\) is the set \(\text{Ker }S=\{p: V_{p} = S\}\). The author proves that if some finite subset \(T\) of \(S\) has locally maximal visibility in \(S\) then \(V_{T} = S\), where \(S = \text{cl(int } S)\) and \(\operatorname {int} S\) is an open connected subset of some linear topological space such that \(\operatorname {int} S\) contains all the triangular regions whose relative boundaries lie in \(S\). Hence \(S\) is given as the finite union of starshaped sets whose kernels are determined by \(T\). In the second theorem the author, changing the conditions on \(S\) given in the first theorem, proves a similar result as that of the first theorem for an open connected set \(S\) in some linear topological space containing all triangular regions whose relative boundaries lie in \(S\). Finally two counterexamples show that the requirements a) that the triangular regions be contained in \(S\) and b) the finiteness of \(T\) can not be relaxed.