Bounds on arithmetic projections, and applications to the Kakeya conjecture (Q1574747)

Let \(A, B, C\) be sets of integers of cardinality at most \(N\), and let \(G\) be a subset of \(A\times B\) such that the differences \(\{ a-b: (a,b) \in G \}\) are all distinct, and the sums \(\{ a+b: (a,b) \in G \}\) all lie in \(C\). The cardinality of \(G\) is trivially bounded by \(N^2\), however this was improved by Gowers to \(N^{2-\varepsilon}\) for some \(\varepsilon > 0\). Bourgain obtained the quantitative value of \(\varepsilon = 1/13\); this is improved to \(1/6\) in this paper. If one makes the additional assumption that the weighted sums \(\{ a+2b: (a,b) \in G \}\) lie in another set \(D\) of cardinality at most \(N\), then one can improve this further to \(1/4\). As an application the authors show that Besicovitch sets (sets in \(R^n\) which contain a line segment in every dimension) have Minkowski dimension at least \((4n+3)/7\) and Hausdorff dimension at least \((6n+5)/11\).