Nonlinear least squares via automatic derivative evaluation (Q789873)

In a former paper the authors have described a method (they call the table method) to automatically evaluate for a certain class of functions \(f:{\mathbb{R}}^ n\to {\mathbb{R}}\) all partial derivatives up to a given order. The method uses in a straightforward way common rules of calculus such as the chain rule. It does not provide for any optimization of the evaluation scheme with respect to storage or speed of computation. Now, the authors have written a Fortran program to solve a two-dimensional nonlinear least squares problem by means of the Newton-Raphson method. The table algorithm is used to supply the derivatives needed.