Burkholder's function via Monge-Ampère equation (Q1928884)

Burkholder's function \[ u_p(x,y)=p\left(1-\frac{1}{p_*}\right)^{p-1}\left(\left| y\right| (p^*-1)\left| x\right|\right)\left(\left| y\right| +\left| x\right|\right)^{p-1}, \] where \(p^*=\max\left(p,\frac{p}{p-1}\right)\) and \(1<p<\infty\), has been shown to play an important role in probability and harmonic analysis. In the setting of probability, it is used in studying the martingale transform and the sharp bounds on \(L^p\). It plays a similar role in harmonic analysis when studying the bounds of the Ahlfors-Beurling transform. The interested reader can consult the numerous references appearing in the paper under review to explore more of these connections. In this paper, the authors use the ``Bellman method'' to show how one can obtain Burkholder's function, by first connecting the function to a solution of a certain Monge-Ampère-type partial differential equation, and then solving the corresponding partial differential equation.