Optimal design for the planar Skorokhod embedding problem

Abstract: We revisit the planar Skorokhod embedding problem introduced by Gross and developed further by Boudabra-Markowsky, and we place it in a fully variational framework. For a centered probability measure $μ$ with finite second moment, we show that Gross' $μ$ -domain $U_μ^{G}$ uniquely minimizes the area among all simply connected $μ$ -domains. Equivalently, $U_μ^{G}$ minimizes a natural $H^{\frac{1}{2}}$-type boundary energy, providing an optimal design interpretation of the planar Skorokhod embedding problem. The proof relies on the Fourier characterization of fractional Sobolev spaces on the circle, symmetric decreasing rearrangement of the quantile of $μ$ , and a one-dimensional fractional $Pólya-Szegő$ inequality. Within the Schlicht class, we obtain a sharp model case: under a natural normalization of the quantile, the Gross area is uniquely minimized by the shifted arcsine distribution. This identifies the Gross domain of the arcsine law as the extremal Schlicht solution to the planar Skorokhod embedding problem. In the second part of the paper we return to Brownian symmetrization. Using the optimality of $U_μ^{G}$, we prove that Brownian symmetrization is area-nonincreasing and, in fact, dominates Steiner symmetrization in this regard. We then quantify the ratio $$ρ(U):=\frac{\mathcal{A}(U_μ^{G})}{\mathcal{A}(U)},$$showing that, among simply connected domains of fixed area, every value in $(0,1]$ is attained. A family of thin rectangles provides a striking example where the associated one-dimensional laws collapse to a Dirac mass in $L^{2}$ while the Gross domain does not shrink, illustrating the genuinely fractional nature of the underlying $H^{\frac{1}{2}}$-geometry. Altogether, our results clarify the geometric content of the planar Skorokhod embedding and open a systematic program of "optimal design" via Brownian symmetrization.