A sharp Schrödinger maximal estimate in \(\mathbb{R}^2\) (Q2404047)

The authors treat the solution of Schrödinger equation, given by \(e^{it\Delta}f(x)=(2\pi)^{-n}\int e^{i(x\dot \xi+t|\xi|^2)}\hat f(\xi)d\xi\). They consider the problem posed by Carleson: determine the optimal \(s\) for which \(\lim_{t\to 0}e^{it\Delta}f(x)=f(x)\) a.e. whenever \(f\) is in the Sobolev space \(H^s(\mathbb R^n)\). Their main result is: For every \(f\in H^s(\mathbb R^2)\) with \(s>1/3\), \(\lim_{t\to 0}e^{it\Delta}f(x)=f(x)\) almost everywhere. On the other hand, Bourgain showed that the convergence can fail if \(s<n/(2(n+1))\) (in the case \(n=2\): \(s<1/3\)). So, they solve Carleson's problem in the case \(n=2\). For the proof, they use polynomial partitioning and decoupling, introduced by \textit{L. Guth} and \textit{N. H. Katz} [ibid. 181, No. 1, 155--190 (2015; Zbl 1310.52019)]. In the case \(n=1\) the optimal number is \(1/4\), which was proved by Carleson and Dahlberg and Kenig.