On Jul 29, 2011, at 10:28 AM, Robert Matlock wrote:
I'm using Nlopt via R for a likelihood problem. I've run
NLOPT_GN_ORIG_DIRECT_L, NLOPT_GN_ISRES and NLOPT_GD_MLSL
(subsidiary method ) for 1000000 function evaluations each. All
reach the same minimum, but they exit because maxeval is reached not
because xtol_rel is reached.
When I run the local methods on these putative minima they exit
almost instantaneously using xtol_rel or xtol_abs as exit criteria
or both -- i.e. they do converge.
Is this behavior normal for the global methods?
Yes, because they are trying to search the space exhaustively. See:
http://jdj.mit.edu/wiki/index.php/NLopt_Introduction#Termination_tests_for_global_optimization
_______________________________________________
NLopt-discuss mailing list
[email protected]
http://ab-initio.mit.edu/cgi-bin/mailman/listinfo/nlopt-discuss
