On the lifetime of a conditioned Brownian motion on a fish bowl (Q2475572)

Let \(\Omega\) be a bounded domain with nonempty boundary, and consider a Brownian motion \(X\) in \(\Omega\), starting from \(x\) and killed at \(y\). It has been conjectured that for planar domains, the expected lifetime of \(X\) is maximal for boundary points \(x\) and \(y\). This article does not solve this conjecture (it does not consider a planar domain), but it verifies that this conjecture is false on a subdomain of the two-dimensional sphere.