Communications (Jun 1999)
Nonlinear least squares method
Abstract
The paper deals with a comparison of linear and nonlinear least squares approximation. Its aim is to show that the well known transformations of nonlinear dependencies on linear dependencies do not always give exact results. Therefore, the only right possibility of approximation of nonlinear functions is to use one of nonlinear least squares procedures. The Levenberg - Marquardt compromise seems to be by far the best one.
Keywords