Local and global proportionality (Q1193452)

Let \({\mathcal R}\) be a ring of blue and green vertices (or, in an earlier version of the problem, of black and white balls). For non-negative integers \(l\) and \(r\) and a vertex \(v\) of \({\mathcal R}\), let \(N_{l,r}(v)\) denote the set of \(l+r\) vertices consisting of the \(l\) vertices counterclockwise from \(v\) and the \(r\) vertices clockwise from \(v\). If one knows that each \(N_{l,r}(v)\) with \(v\) green contains at least a certain proportion \(p\) of green vertices, and assuming that at least one vertex of \({\mathcal R}\) is green, what can one deduce about the proportion \(q\) of vertices in \({\mathcal R}\) that are green? In [Discrete Math. 61, 165-179 (1986; Zbl 0619.05045)], \textit{P. C. Fishburn}, \textit{F. K. Hwang} and \textit{H. Lee} raised this problem (in a slightly different form), made some conjectures, and obtained many partial results. In the present paper, the author makes several further conjectures, of which the most interesting is that if \(p>{1\over 2}\) then \(q\geq p\). He proves this conjecture in certain special cases, giving shorter proofs of two results proved by Fishburn, Hwang and Lee (the extreme cases \(l=0\) and \(l=r\)); and he obtains weaker results in other cases. After a brief digression to discuss graphical generalizations, he then proves all his conjectures for the simple case of a periodic ring where the period consists of a block of consecutive green vertices followed by a block of consecutive blue vertices.