DocumentCode :
2663081
Title :
Bezier formulation of multistep algorithms
Author :
Yanilmaz, Mehmet ; Balkir, Sina
Author_Institution :
Dept. of Electr. Eng. & Comput. Sci., Northwestern Univ., Evanston, IL, USA
fYear :
1990
fDate :
1-3 May 1990
Firstpage :
2696
Abstract :
Predictor-based integration formulas are presented by formulating the solution function as a vectored value Bernstein polynomial, namely a Bezier function. The Bezier function subdivision theorem leads to a predictor evaluation algorithm for the next solution point using de Casteljau´s iterative control vertex generation technique. It is shown that the convergence of the algorithms is a consequence of the use of Bernstein uniform approximation basis and that the de Casteljau predictor evaluation scheme leads to a computationally simple estimate of the local truncation error
Keywords :
convergence of numerical methods; integration; iterative methods; polynomials; predictor-corrector methods; roundoff errors; Bernstein uniform approximation; Bezier formulation of multistep algorithms; Bezier function subdivision theorem; computationally simple estimate; convergence; de Casteljau predictor evaluation scheme; de Casteljau´s iterative control vertex generation technique; local truncation error; predictor evaluation algorithm; predictor-based integration formulas; vectored value Bernstein polynomial; Approximation algorithms; Circuit simulation; Computer networks; Differential equations; Iterative algorithms; Lagrangian functions; Least squares approximation; Polynomials; Prediction algorithms; Transient analysis;
fLanguage :
English
Publisher :
ieee
Conference_Titel :
Circuits and Systems, 1990., IEEE International Symposium on
Conference_Location :
New Orleans, LA
Type :
conf
DOI :
10.1109/ISCAS.1990.112565
Filename :
112565
Link To Document :
بازگشت