[ADMB Users] problems running mcmc simulations

[Felice Griffiths]

I was multiplying the log-likelihood by 15, so that explains why I was seeing differences in my results with different weights.  Turns out the real problem was that I didn't initialize the penalty vectors I have in the model. Putting the weight on the log-likelihood was seemingly solving the problem by getting the mcmc simulations to run, but was not correcting the real issue.  Thanks for the insight!


----- Original Message -----
From: "dave fournier" <otter at otter-rsch.com>
To: users at admb-project.org
Sent: Sunday, July 25, 2010 2:31:28 AM GMT -08:00 US/Canada Pacific
Subject: Re: [ADMB Users] problems running mcmc simulations

Are you multiplying the likelihood or the log-likelihood by 15?
Multiplying the log-likelihood by 15 sharpens the modes (figuratively
speaking) and is not equivalent to the original problem.

Otherwise without looking at your code and data it is hard to say
much about what is going on.
_______________________________________________
Users mailing list
Users at admb-project.org
http://lists.admb-project.org/mailman/listinfo/users