On the property of monotonic convergence for multivariate Bernstein-type operators (Q1344176)

The authors mention first that many sequences \((L_ n)\) of one- dimensional linear positive operators satisfy the monotonicity property: \(L_ nf \geq L_{n + 1}f\), if \(f\) is a convex function. For example, in the case of Bernstein polynomials \(B_ nf\), by using expressions, in terms of second-order divided differences, for the remainder term and for the difference of two consecutive terms of the sequence \((B_ n)\) (see the reviewer's papers [J. Soc. Ind. Appl. Math. Ser. B, 1, 137-163 (1964; Zbl 0143.079)], respectively [Math. Z. 98, 46-51 (1967; Zbl 0169.074)]), one can obtain at once the relations: \(B_ nf \geq B_{n + 1} f \geq f\) on \(I = [0,1]\), if \(f\) is convex on this interval. Such a result was obtained by W. B. Temple in 1954 and in 1959 by B. Averbach, but without using the divided differences. The aim of this paper is to apply probabilistic methods for finding similar results for the multivariate Bernstein-type operators. But in the multivariate case the situation is not so simple. So, we want to mention that \textit{Th. Sauer} [Comput. Aided Geom. 8, 465-478 (1991; Zbl 0756.41029)] succeeded to establish such a result assuming that the function \(f\) is axially convex, a notion introduced by H. J. Schmid.