[Limdep Nlogit List] Nlogit result fails to complete with different Halton pts number

William Greene wgreene at stern.nyu.edu
Fri Mar 12 23:54:15 EST 2010


Cathy. The log likelihood differs at high order digits that you are not
seeing. However, I would conclude from your description that the estimation
is OK with PTS=100 (or more). That message means that the log likelihood is
pretty flat near the maximum it has found, and it is not finding that more
iterations are improving logL. This is quite common.
/Bill Greene

----- Original Message -----
From: "Cathy Durham" <cathy.durham at oregonstate.edu>
To: "Limdep and Nlogit Mailing List" <limdep at limdep.itls.usyd.edu.au>
Sent: Thursday, March 11, 2010 12:18:38 PM GMT -05:00 US/Canada Eastern
Subject: [Limdep Nlogit List] Nlogit result fails to complete with different Halton pts number

 I'm running an Nlogit that finishes with normal status if Halton=20, that is 
'Normal exit from iterations. Exit status=0.'
But when I ramp it up to Halton=100
I get 
'Line search does not improve fn. Exit iterations. Status=3'
What does this indicate? The likelihood at exit is the same as the Halton=20 ends normally with.

I'm running a heirarchal form rather than true rpl with parameters on attributes 
That are functions of the rpl variables, that is using (c) in the fcns.
_______________________________________________
Limdep site list
Limdep at limdep.itls.usyd.edu.au
http://limdep.itls.usyd.edu.au


More information about the Limdep mailing list