This is the mail archive of the
gsl-discuss@sources.redhat.com
mailing list for the GSL project.
BFGS minimization comparison between NR and GSL
- From: "Philippe Huber" <huberph at infomaniak dot ch>
- To: <gsl-discuss at sources dot redhat dot com>
- Date: Tue, 20 Aug 2002 18:59:19 +0200
- Subject: BFGS minimization comparison between NR and GSL
Hi all,
I made a program with NR (numerical recipes) routines that minimizes a
function of 35 parameters. I used a BFGS algorithm. I decided a few
weeks ago to make the same program with GSL routines. The result is
somewhat surprising. The GSL BFGS needs about twice as much iterations
as NR's does. Shouldn't it be the same algorithm? I must also say that
sometimes, my program didn't converge with NR and I never had this
problem with GSL.
Your opinion?
Thank you
Philippe