JAXA Repository / AIREX 未来へ続く、宙(そら)への英知

このアイテムに関連するファイルはありません。

タイトルNLINEAR - NONLINEAR CURVE FITTING PROGRAM
著者(英)Everhart, J. L.
発行日1994-01-01
言語eng
内容記述A common method for fitting data is a least-squares fit. In the least-squares method, a user-specified fitting function is utilized in such a way as to minimize the sum of the squares of distances between the data points and the fitting curve. The Nonlinear Curve Fitting Program, NLINEAR, is an interactive curve fitting routine based on a description of the quadratic expansion of the chi-squared statistic. NLINEAR utilizes a nonlinear optimization algorithm that calculates the best statistically weighted values of the parameters of the fitting function and the chi-square that is to be minimized. The inputs to the program are the mathematical form of the fitting function and the initial values of the parameters to be estimated. This approach provides the user with statistical information such as goodness of fit and estimated values of parameters that produce the highest degree of correlation between the experimental data and the mathematical model. In the mathematical formulation of the algorithm, the Taylor expansion of chi-square is first introduced, and justification for retaining only the first term are presented. From the expansion, a set of n simultaneous linear equations are derived, which are solved by matrix algebra. To achieve convergence, the algorithm requires meaningful initial estimates for the parameters of the fitting function. NLINEAR is written in Fortran 77 for execution on a CDC Cyber 750 under NOS 2.3. It has a central memory requirement of 5K 60 bit words. Optionally, graphical output of the fitting function can be plotted. Tektronix PLOT-10 routines are required for graphics. NLINEAR was developed in 1987.
NASA分類STATISTICS AND PROBABILITY
レポートNO94M10318
LAR-13934
権利No Copyright
URIhttps://repository.exst.jaxa.jp/dspace/handle/a-is/312696


このリポジトリに保管されているアイテムは、他に指定されている場合を除き、著作権により保護されています。