Stopping distributions for right processes (Q806175)

The stopping distribution problem with non-randomized stopping times has been studied by several authors, among them Dubins, Root, Baxter, Chocon, Rost, and Falkner. In the paper under review the authors prove the existence of the desired non-randomized stopping time T under hypotheses more general than those previously considered. Let X be a right process and let \({\mathcal E}^ e\) be the \(\sigma\)-algebra on the state space E generated by q-excessive functions, then \[ {\mathcal E}\subset {\mathcal E}^ e\subset {\mathcal E}^* \] where \({\mathcal E}\) is the Borel \(\sigma\)-algebra and \({\mathcal E}^*\) is the universal completion of \({\mathcal E}\). The main theorem 1.2 is stated by the authors as follows: ``Let m be an excessive measure for X and assume that each \({\mathcal E}^ e\)-measurable semipolar set is m-polar. Let \(\mu\) be a measure on E such that its X-potential \(\mu\) U is \(\sigma\)-finite and \(\mu U\ll m\). Let \(\nu\) be another measure on E. If \(\mu\) \(U\geq \nu U\) and there exists a set \(C\in {\mathcal E}^*\) such that for each m-polar set \(Z\in {\mathcal E}^ e\) we have \(\nu (Z)=\mu (Z\cap C)\), then there is a non-randomized stopping time T such that \(\mu P_ T=\nu\) (and conversely).