Numerical Methods for Nonlinear Discrete L1 Approximation Problems
In the problem of curve fitting to discrete data, the L1 norm can play an important role when the data contains large errors, or when some of the data is exceptional. In this paper is considered the problem of providing robust, efficient methods, capable of fast ultimate convergence, for the nonlinear discrete L1 approximation problem. Algorithms based on a Levenberg-like approach are given, together with a convergence analysis, and it is shown how second derivative information may be incorporated. Numerical results for some test problems are included.
Unable to display preview. Download preview PDF.
- 5.Cheney, E W (1966) Introduction to Approximation Theory. McGraw-Hill, New York.Google Scholar
- 7.Hald, J and K Madson (1978) A 2 stage algorithm for minimax optimization. Technical University of Denmark, DK-2800 Lyngby Report No. NI-78–11.Google Scholar
- 9.Jittorntrum, K and M R Osborne (1979) Strong uniqueness and second order convergence in non-linear discrete approximation, Preprint.Google Scholar
- 10.Levenberg, K (1944) A method for the solution of certain non-linear problems in least squares. Quat. Appl. Math. 2, 164–168.Google Scholar
- 11.Madsen, K (1975) An algorithm for minimax solutions of overdetermined systems of non-linear equations. JIMA 16, 321–328.Google Scholar
- 13.Osborne, M R and G A Watson (1978) Nonlinear approximation problems in vector norms. In Numerical Analysis, Dundee (1977) (G A Watson ed.) Springer-Verlag.Google Scholar
- 14.Rockafellar, R T (1970) Convex analysis. Princeton, New Jersey. Princeton Univ. Press.Google Scholar
- 15.Watson, G A (1979) The minimax solution of an overdetermined system of nonlinear equations. JIMA to appear.Google Scholar