multimin question
Brian Gough
bjg@network-theory.co.uk
Tue Apr 11 09:17:00 GMT 2000
I found one problem in the program multimin.c
gsl_multimin_fdf_minimizer_best_step(s);
needs to be
if (status == GSL_SUCCESS)
gsl_multimin_fdf_minimizer_best_step(s);
Without this it goes to the minimum and then heads out to infinity.
Perhaps the best_step should be zero if the function value will not
decrease. The history value for "f" seems to get out of sync in this
case too.
More information about the Gsl-discuss
mailing list