Points of middle density in the real line (Q2392509)

Let \(E\subset{\mathbb R}\) be a measurable set for the usual Lebesgue measure \(m\). For \(x\in{\mathbb R}\), let \(d_{E}(x,r):=\frac{1}{2r}m(E\cap\{y:|x-y|<r\})\), \(\underline{d}_{E}(x):=\liminf\limits_{r\downarrow 0}d_{E}(x,r)\) and \(\overline{d}_{E}(x):=\limsup\limits_{r\downarrow 0}d_{E}(x,r)\). A non-trivial Lebesgue measurable set \(E\) is called \(\delta\)-exceptional if, for any \(x\in{\mathbb R}\), we have either \(\underline{d}_{E}(x)<\delta\) or \(\overline{d}_{E}(x)>1-\delta\). Kolyada's problem is to find the infimum \(\delta_0\) of those \(\delta\) for which a \(\delta\)-exceptional set exists. \textit{A. Szenes} [Adv. Math. 226, No. 1, 764--778 (2011; Zbl 1205.28001)] proved that \(\delta_{0}\leq 0.272\), where the upper bound is a positive root of the polynomial \(8x^3 +4x^2 +2x-1\), and conjectures that \(\delta_0\) equals to this root. The authors of the present paper prove that \(\delta_0\) is smaller than the positive root \(0.2710\dots\) of the polynomial \(2x^3 +2x^2 +3x -1\), i.e., the Szenes conjecture is false.