Approximation by Chebyshevian Bernstein Operators versus Convergence of Dimension Elevation

Rachid Ait-Haddou, Marie-Laurence Mazure

Research output: Contribution to journalArticlepeer-review

8 Scopus citations

Abstract

On a closed bounded interval, consider a nested sequence of Extended Chebyshev spaces possessing Bernstein bases. This situation automatically generates an infinite dimension elevation algorithm transforming control polygons of any given level into control polygons of the next level. The convergence of these infinite sequences of polygons towards the corresponding curves is a classical issue in computer-aided geometric design. Moreover, according to recent work proving the existence of Bernstein-type operators in such Extended Chebyshev spaces, this nested sequence is automatically associated with an infinite sequence of Bernstein operators which all reproduce the same two-dimensional space. Whether or not this sequence of operators converges towards the identity on the space of all continuous functions is a natural issue in approximation theory. In the present article, we prove that the two issues are actually equivalent. Not only is this result interesting on the theoretical side, but it also has practical implications. For instance, it provides us with a Korovkin-type theorem of convergence of any infinite dimension elevation algorithm. It also enables us to tackle the question of convergence of the dimension elevation algorithm for any nested sequence obtained by repeated integration of the kernel of a given linear differential operator with constant coefficients. © 2016 Springer Science+Business Media New York
Original languageEnglish (US)
Pages (from-to)425-461
Number of pages37
JournalConstructive Approximation
Volume43
Issue number3
DOIs
StatePublished - Mar 17 2016

Bibliographical note

KAUST Repository Item: Exported on 2020-10-01

Fingerprint

Dive into the research topics of 'Approximation by Chebyshevian Bernstein Operators versus Convergence of Dimension Elevation'. Together they form a unique fingerprint.

Cite this