Iterates of multidimensional Bernstein-type operators and diffusion processes in population genetics

Takatoshi Hirano, Ryuya Namba

Published 2023-10-21Version 1

The Bernstein operator is known as a typical example of positive linear operators which uniformly approximates continuous functions on $[0, 1]$. In the present paper, we introduce a multidimensional extension of the Bernstein operator which is associated with a transition probability of a certain discrete Markov chain. In particular, we show that the iterate of the multidimensional Bernstein-type operator uniformly converges to the Feller semigroup corresponding to the multidimensional Wright-Fisher diffusion process with mutation arising in the study of population genetics, together with its rate of convergence. The convergence of process-level is obtained as well. Moreover, by taking the limit as both the number of iterate and the dimension of the Bernstein-type operator tend to infinity simultaneously, we prove that the iterate of the multidimensional Bernstein-type operator uniformly converges to the Feller semigroup corresponding to a probability measure-valued Fleming-Viot process with mutation.