pipette
ESEspañol

A dual characterization and algorithm of general optimal stopping

Masahiko Egami, Tomohiro Koike

PreprintBold claims, read critically

In the authors' words

We present a novel dual approach to characterizing the solution to optimal stopping problems for strong Feller processes on locally compact, separable metric spaces. Our characterization builds upon reinterpreting a standard result in optimal stopping theory as a dual formulation, which yields computationally tractable algorithms. Our framework is applicable to a broad class of Markov processes. First, in the one-dimensional case, while previous studies have managed to comprehensively solve only diffusion and one-sided jump L\'evy processes over an infinite time horizon, our proposed method successfully extends to both infinite- and finite-horizon scenarios across all these processes, including two-sided jump L\'evy processes. In addition, for multi-dimensional cases, our framework is applicable to diffusions over an infinite horizon with certain structural features such as convex stopping regions. To illustrate this versatility, we explicitly solve several challenging problems including a case where the optimal stopping region consists of two disjoint connected components -- a structural feature that remains unaddressed in the literature.

Main resultLimitation the authors admit

Appeared: Thursday, October 8. arXiv. Preprint, not yet peer-reviewed.