Sumsets in difference sets (Q2655770)

In the paper Corollaries 3.1.1 and 3.1.2 by \textit{V. Bergelson} [J. Lond. Math. Soc., II. Ser. 31, 295--304 (1985; Zbl 0579.10029)] are improved and results related to two questions posed in the same paper are proved (one problem is answered in the negative). The improvements contain, inter alia, the following results: (i) Let \(A\subset{\mathbb Z}\) and suppose that the upper Banach density of \(A\) is positive. Then there exists a subset \(B\subset{\mathbb Z}\) of positive asymptotic density such that \(B=-B\), \(0\in B\) and \(A-A\supset B+B\), (ii) Let \(E\subset {\mathbb Z}^2\) have positive upper asymptotic density. Let \(p_i,q_i\in{\mathbb Z}[x]\), \(i=1,\dots,m\), satisfy \(p_i(0)=q_i(0)=0\) for all \(i\). Then there exists \(B\subset {\mathbb Z}\) of positive upper asymptotic density such that \(E-E\supset\cup_{i=1}^m\left(p_i(B)\times q_i(B)\right)\). The above mentioned negative answer says: For every \(0<\alpha<1/2\) there is a set \(E\subset{\mathbb Z}^3\) of asymptotic density \(>\alpha\), such that there is no \(B\subset{\mathbb Z}\) of positive upper asymptotic density satisfying \(E-E\supset B\times B\times B\). One of the results related to the second question says: If \(A\subset {\mathbb Z}\) has positive upper asymptotic density, then for any \(r,s,t\) such that \(r+s+t=0\) the set \(rA+sA+tA\) is a Bohr neighbourhood of \(0\).