Minimax approximations subject to a constraint
HTML articles powered by AMS MathViewer
- by C. T. Fike and P. H. Sterbenz PDF
- Math. Comp. 25 (1971), 295-298 Request permission
Abstract:
A class of approximation problems is considered in which a continuous, positive function $\varphi (x)$ is approximated by a rational function satisfying some identity. It is proved under certain hypotheses that there is a unique rational approximation satisfying the constraint and yielding minimax relative error and that the corresponding relative-error function does not have an equal-ripple graph. This approximation is, moreover, just the rational approximation to $\varphi (x)$ yielding minimax logarithmic error. This approximation, in turn, is just a constant multiple of the rational approximation to $\varphi (x)$ yielding minimax relative error but not necessarily satisfying the constraint.References
-
W. J. Cody & Anthony Ralston, "A note on computing approximations to the exponential function," Comm. ACM, v. 10, 1967, pp. 53-55.
- I. F. Ganžela and C. T. Fike, Sterbenz, P. H, Math. Comp. 23 (1969), 313–318. MR 245199, DOI 10.1090/S0025-5718-1969-0245199-6 W. Kahan, "Library tape functions EXP, TWOXP, and .XPXP.," Programmers’ Reference Manual, University of Toronto, 1966. (Mimeographed.) W. J. Cody, "Double-precision square root for the CDC-3600," Comm. ACM, v. 7, 1964, pp. 715-718.
Additional Information
- © Copyright 1971 American Mathematical Society
- Journal: Math. Comp. 25 (1971), 295-298
- MSC: Primary 41A20
- DOI: https://doi.org/10.1090/S0025-5718-1971-0303176-X
- MathSciNet review: 0303176