The NLopt optimizer returns NLOPT_INVALID_ARGS if you give more than n equality constraints in a problem with n variables. This is totally reasonable in linear settings but not necessarily in nonlinear problems. As far as I can see e.g. ISRES does not depend on this. Could we lift this restriction and maybe just offer a warning with the result?
This feature/restriction was introduced in 2010 as can be seen in the NEWS file.
The problem arises in the functions nlopt_add_equality_mconstraint and nlopt_add_precond_equality_constraint in the file options.c where nlopt_count_constraints is called.
The NLopt optimizer returns NLOPT_INVALID_ARGS if you give more than n equality constraints in a problem with n variables. This is totally reasonable in linear settings but not necessarily in nonlinear problems. As far as I can see e.g. ISRES does not depend on this. Could we lift this restriction and maybe just offer a warning with the result?
This feature/restriction was introduced in 2010 as can be seen in the NEWS file.
The problem arises in the functions
nlopt_add_equality_mconstraint
andnlopt_add_precond_equality_constraint
in the fileoptions.c
wherenlopt_count_constraints
is called.