Modified MKZ Operators Providing Better Estimation on Chosen Part of [1/2, 1)

by Shanaz Ansari Wahid, Ashok Sahai & Vrijesh Tripathi.

Abstract: Recently, özarslan & Duman (2007) introduced a modification of the Meyer-Konig and Zeller (MKZ) operators, (Say MMKZO1k,n (f; x)) which preserves the test-functions f0(x) = 1 and f2(x) = x2, and showed that it provides a better estimation than the classical MKZ-operators on the interval [1/2, 1) with respect to the modulus of continuity and the Lipschitz class functions. In the present paper, we provide a more efficient modification. Our motivation is two-fold. Firstly, both of the aforementioned operators are practically numerically impossible, inasmuch as they require the values of the impugned function at infinitely many knots: namely the values of the function “f” we need are f (k/ (k+n)); k = 0, 1, 2, ..., ?. This impracticality is remedied by our present modification of MKZ-operators (Say FVMMKZO2k,n (f; x)) needing only the values of the function “f” we need are f (k/ (k+n)); k = 0, 1, 2, ..., m for providing an efficient estimation in interval [1/2, m/ (m+n)].In fact, in contrast to the ‘Mathematicians’, who are fond of “abstractions”, the ‘Statisticians-and-Computer-Scientists’ are rather more keen to ensure the “applicational/computational efficiency” of any algorithm. We have, therefore, modified impugned operators: Mk,n (f; x) & its modification MMKZO1k,n (f; x) by their “Finite-Versions” proposed by us, namely by “FVMk,n (f; x)” & “FVMMKZO1k,n (f; x)”, respectively. For illustration purpose, we have chosen the interval to be [1/2, 16/21]. Secondly, we found that our modification would be numerically more efficient than özarslan & Duman (2007)'s. To facilitate an illustration of relative efficiency of approximation of our version of ‘Modified MKZ Operators FVMMKZO2k,n (f; x)’ vs. that of ‘özarslan & Duman (2007)’s Operators FVMMKZO1k,n (f; x)’, we have accommodated a brief numerical study in the last section of this paper. In this 'simulated numerical study' we have chosen four illustrative example-functions: namely exp (x), sin (2+x), ln (2+x), and 10x. For the simplicity of the numerical illustration, we have chosen our illustrative m-n-values as m = 16 & n = 5; m = 32 & n = 10; m = 48 & n = 15, assuming that our chosen interval of interest-of-estimation happens to be [1/2, 16/21]. This illustrative numerical-comparison of “Finite-Versions” has amply supported the fact that our modification performs much better, inasmuch as the absolute approximation error for our version of ‘Modified MKZ Operators FVMMKZO2k,n (f; x))’ is significantly less than that for the classical MKZ operators FVMk,n (f; x)), as also than that for the Ozarslan & Duman (2007)'s modification of the Meyer-Konig and Zeller (MKZ) operators, FVMMKZO1k,n (f; x)).

Key Words: Positive Linear Operators, Rate of Convergence, Modulus of Continuity, Lipschitz Class, Numerical Study

Authors:
Shanaz Ansari Wahid, shanazw@hotmail.com
Ashok Sahai, sahai.ashok@gmail.com
Vrijesh Tripathi, vrijesh.tripathi@sta.uwi.edu

Editor: : Ahmed H. Youssef, ahyoussef@hotmail.com

READING THE ARTICLE: You can read the article in portable document (.pdf) format (266140 bytes.)

NOTE: The content of this article is the intellectual property of the authors, who retains all rights to future publication.

This page has been accessed 4673 times since AUGUST 4, 2009.


Return to the InterStat Home Page.