Sára, Antonín; Dolník, Lubomír Measured data nonlinear regression estimates of function dependences of quantities. (Czech. Russian, English translations) Zbl 0556.65100 Sb. Vys. Uč. Tech. Brně 1981, 49-64 (1983). An algorithm for the nonlinear regression is realized in the ALGOL 60 language. Function dependences among variables can be given by a series of functions of arbitrary type, by a system of differential equations (ordinary equations of 1st order with initial conditions) or by both types simultaneously where differentiations are only with respect to one independent variable. The total number of functions describing the process given can be greater than the number of dependences measured. The program applies 3 basic iteration methods for the determination of parameter estimates: the steepest descent method (gradient), the Gauss- Newton method, and the Marquardt method. As optimization criterion the weighted sum of squares of deviations between measured and calculated values is used. MSC: 65C99 Probabilistic methods, stochastic differential equations 62-04 Software, source code, etc. for problems pertaining to statistics 62J02 General nonlinear regression Keywords:least squares method; measured data; iteration methods; steepest descent method; Gauss-Newton method; Marquardt method PDFBibTeX XMLCite \textit{A. Sára} and \textit{L. Dolník}, Sb. Vys. Uč. Tech. Brně 1981, 49--64 (1983; Zbl 0556.65100)