[ADMB Users] Opposite of mfexp()

Mark Payne mpa at aqua.dtu.dk
Mon Jun 25 11:03:10 PDT 2012


I seem to have struck a problem with some code that I've narrowed down to the folllowing.The problem is essentially that

nll -= tmpl;


nll -= log(mfexp(tmpl));

(where nll is my objective function and tmpl is a temporary dvariable)

lead to different results in the optimisation - the first one sensible, and the second one not. Am I doing something outstandingly silly here? Or is there something that I haven't understood?


More information about the Users mailing list