Strong stationary duality for diffusion processes (Q501820)

The article extends the circle of ideas surrounding strong stationary duality for Markov chains to a diffusion on a compact interval. Specifically, a strong stationary dual process is defined, its generator is characterized, and relationships between its boundary behavior with that of the original diffusion are enunciated. A sequence of birth and death type Markov chains and their stationary duals are shown to converge to the above pair in law on suitable interpolation and scaling. This allows the authors to extend to the diffusion under consideration the result relating a suitably defined counterpart of separation distance for Markov chains with absorption probability of the dual, and further, its relationship with the spectrum of the generator. Important algorithmic implications of these results are underscored.