Distinct distance estimates and low degree polynomial partitioning (Q2340417)

If \(\mathcal L\) is a set of lines in \({\mathbf R}^3\), then a point is called \textit{\(r\)-rich} if it lies in at least \(r\) lines of \(\mathcal L\); the set of \(r\)-rich points of \(\mathcal L\) is denoted by \(P_r({\mathcal L})\). The present paper is closely connected with [\textit{L. Guth} and \textit{N. H. Katz}, Ann. Math. (2) 181, No. 1, 155--190 (2015; Zbl 1310.52019)] where different, pretty long arguments are used for the cases with \(r=2\) and \(r>2\). Using only polynomial partitioning the author proves the subsequent theorem: ``For any \(\varepsilon>0\), there are \(D(\varepsilon)\), \(K(\varepsilon)\) so that the following holds. If \(\mathcal L\) is a set of lines in \({\mathbf R}^3\), and there are less than \(L^{(1/2)+\varepsilon}\) lines of \(\mathcal L\) in any irreducible algebraic surface of degree at most \(D\), and if \(2\leq\,r\leq\,2L^{1/2}\), then \[ |P_r({\mathcal L})|\leq\,KL^{(3/2)+\varepsilon}r^{-2}." \] Applying the aforementioned theorem in the arguments of ibidem one gets the following slightly weaker distinct distances estimate: ``For any \(\varepsilon>0\), there is a constant \(c_{\varepsilon}>0\) so that any set of \(N\) points in the plane determines at least \(c_{\varepsilon}N^{1-\varepsilon}\) distinct distances.''