Optimally stopping multidimensional Brownian motion

Abstract: We solve optimal stopping for multidimensional Brownian motion in a bounded domain, a question raised in Dynkin and Yushkevich (1967), where the one-dimensional case was presented. Taking a geometric approach, under regularity conditions we construct the optimal stopping free boundary in the multidimensional case. We characterise the value function as the pointwise infimum of potentials with recursive extensions dominating the gain function, and obtain its continuity. Explicit examples illustrate the result.