At Thu, 24 Aug 2006 21:24:58 +0200, Marco Maggi wrote: > and then recomputing with GSL step function rk2 > control function 'y' eps_abs = eps_rel = 0.001,
Those are large tolerances -- they are applied to each step, so the global error is O(Nsteps * tolerance). Also rk2 is the least accurate method, try one of the others, e.g. rk4 at least. -- Brian Gough Network Theory Ltd, Publishing the GSL Manual - http://www.network-theory.co.uk/gsl/manual/ _______________________________________________ Help-gsl mailing list Help-gsl@gnu.org http://lists.gnu.org/mailman/listinfo/help-gsl