An Iterative-Improvement Algorithm for Newton's Forward Difference Interpolation Formula Using Statistical Perspectives of Reduced MSE & Reduced
by Miodrag M Lovric1 & Ashok Sahai.
This paper proposes a computerizable iterative algorithm to intermittently improve the efficiency of the interpolation by the well-known simple-n-popular
“Newton’s Forward Difference Interpolation Polynomial Formula?, using “Statistical” perspective of “Reduced Mean-Square-Error (MSE)” and that of “Reduced-
Bias”. The impugned formula uses the values of the simple forward differences using values of the unknown function “f(x)”at equidistant-points/ knots in the
“Interpolation-Interval”, say [x0, xn]. The basic perspective motivating this iterative algorithm is the fuller use of the “information” available in terms
of these values of the unknown function “f(x)” at the “n+1” equidistant-points/ knots. This information is used to reduce the Interpolation- Error, which is
statistically equivalently measurable in terms of “MSE & BIAS”. The potential of the improvement of the interpolation is tried to be brought forth per an
“empirical study” for which the function “f” is assumed to be known in the sense of simulation. The numerical metric of the improvement uses the sum of
absolute errors, i.e. the differences between the actual (assumed to be known in the sense of the simulating nature of the empirical study) and the
interpolated values at the mid-points of the equidistant-points/ knots in “Interpolation-Interval”, say [0, 1]). This leads to the calibrations of the
respective “Percentage Relative (Relative to actual value of the function at that point) Errors (PREs)”, and hence those of the respective “Percentage
Relative Gains”, in terms of the reduced values of the PREs, compared to that with the use of the Newton’s original forward difference formula.
Interpolation; simulated empirical study; reduced-bias
Ashok Sahai, email@example.com
Miodrag M Lovric, firstname.lastname@example.org
Abdallah Abdelfattahf, email@example.com
READING THE ARTICLE: You can read the article in
portable document (.pdf) format (353430 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 1464 times since APRIL 15, 2011.
Return to the Home Page.