Sets of integers with missing differences (Q5918027)

Given a set of positive integers \(M\), a set \(S\) of positive integers is said to be an \(M\)-set if \(a,b\in S\) implies \(a-b\notin M\). If \(\mu(M)\) denotes the supremum of the upper (asymptotic) densities of the all \(M\)-sets, and \(d(M)=\sup_{x\in(0,1)} \min_{m\in M}\|xm\|\), then \textit{D. G. Cantor} and \textit{B. Gordon} [J. Comb. Theory, Ser. A 14, 281-287 (1973; Zbl 0277.10043)] proved that \(\mu(M)\geq d(M)\). The author proves (1) the exact value for \(\mu(M)\) when \(M\) is a finite arithmetical sequence with a coprime difference and offset, or a four element set of the type \(M=\{i,j,2i,i+j\}\) with \((i,j)=1\) and \(j\equiv i+1\pmod{3}\), or \(M=\{i,j,i+j,2j\}\) with \((i,j)=1\) and \(j\equiv i+2\pmod{3}\), (2) lower bounds for \(\mu(M)\) and \(d(M)\) for wide classes of three element sets \(M\), where in some cases the bounds are exact (some of them are conjectured to be exact).