Next: Nonlinear Least-Squares Covariance Matrix, Previous: Nonlinear Least-Squares Testing for Convergence, Up: Nonlinear Least-Squares Fitting [Index]
These routines provide a high level wrapper that combines the iteration and convergence testing for easy use.
These functions iterate the nonlinear least squares solver w for a
maximum of maxiter iterations. After each iteration, the system is
tested for convergence with the error tolerances xtol, gtol and ftol.
Additionally, the user may supply a callback function callback
which is called after each iteration, so that the user may save or print
relevant quantities for each iteration. The parameter callback_params
is passed to the callback function. The parameters callback
and callback_params may be set to NULL to disable this feature.
Upon successful convergence, the function returns GSL_SUCCESS
and sets info to the reason for convergence (see
gsl_multifit_nlinear_test
). If the function has not
converged after maxiter iterations, GSL_EMAXITER
is
returned. In rare cases, during an iteration the algorithm may
be unable to find a new acceptable step \delta to take. In
this case, GSL_ENOPROG
is returned indicating no further
progress can be made. If your problem is having difficulty converging,
see Nonlinear Least-Squares Troubleshooting for further guidance.