On the Wolff circular maximal function (Q2452322)

The main result is the sharp \(L^3\) bound \(\|M_\Phi^\delta f\|_{L^3([\frac12,1])}\leq C_\varepsilon \delta^{-\varepsilon}\|f\|_{L^3(\mathbb{R}^2)}\), where \(M_\Phi^\delta\) is a ``variable coefficient generalization'' of the Wolff circular maximal function, introduced by \textit{L. Kolasa} and \textit{T. Wolff} [Pac. J. Math. 190, No. 1, 111--154 (1999; Zbl 1019.42013)]. This in particular recovers the original theorem from [\textit{T. Wolff}, Am. J. Math. 119, No. 5, 985--1026 (1997; Zbl 0892.52003)] for the circular case \(\Phi(x,y)=|x-y|\) in a way that ``avoids some technical lemmas from Wolff's proof''. Now the function \(\Phi\) is only required to be smooth and to satisfy Sogge's ``cinematic curvature condition'', but it is immediately replaced by a polynomial approximation \(\Psi\) of controlled degree \(k\), so that the actual analysis, involving incidences of (\(\delta\)-neighbourhoods of arcs of) \(\Psi\)-circles \(\Gamma(x,r)=\{y:\Psi(x,y)=r\}\), can be carried out with tools of real algebraic geometry. After several ``standard reductions'', the heart of the matter is an incidence counting problem for certain bipartite collections \((\mathcal{W},\mathcal{B})\) of \(\Psi\)-circles, and the required bound is established by induction on the quantity \((\#\mathcal{W})(\#\mathcal{B})\). This induction exploits the ``polynomial ham sandwich theorem'', which is inventive in that ``to the best of the author's knowledge, the work of \textit{L. Guth} [Acta Math. 205, No. 2, 263--286 (2010; Zbl 1210.52004)] and \textit{J. Bourgain} and \textit{L. Guth} [Geom. Funct. Anal. 21, No. 6, 1239--1295 (2011; Zbl 1237.42010)] are the only other examples where [it is used] to solve a non-discrete problem in harmonic analysis.''