I've never used the coda file for this, but I continue with more iterations all the time when the convergence isn't satisfactory.  If I've used Update to create 20,000 iterations and the convergence doesn't look good enough, I go back into the Update window and tell it to do some more.  Of course I don't exit from the BUGS program between the two updates.  It never occurred to me that there was something wrong with this.  Am I missing something?

Cory

[log in to unmask]" type="cite">

Hi. Mick,

 

Thanks much for your reply. Then, are the hidden parameters started with a random value and adjusted continuously as each iteration goes up? If so, due to the difference of the hidden parameters, each iteration is updated differently? Please correct me if I misunderstood.

 

Also, if you don’t mind, would it be fine to ask more questions? Related to my question, I’m still wondering if there is any way to restart the iteration once it is finished? Or if there is any way to speed up updates such as using multi-core (cpu)? I’m using only a single chain, not multi chains, so I doubt if multi-core would be applicable.

 

Thanks again and have a great weekend,

Ashley

 

 

 

 

On Sat, Feb 4, 2017 at 5:29 PM, Michael Mccarthy <[log in to unmask]> wrote:

Most MCMC sampling methods have additional parameters in them, other than the ones being estimated. These are hidden from user of OpenBUGS, and are used to tune the particular sampler being used to make it more efficient. So I don’t think what you are proposing will work, because those hidden parameters will need to re-tuned.

 

Regards,

 

Mick

-----------------

Michael McCarthy

School of BioSciences

The University of Melbourne

Parkville VIC 3010 Australia

 

+61 3 8344 6856

mickresearch.wordpress.com

qaeco.com

@mickresearch

 

 

 

From: (The BUGS software mailing list) [mailto:[log in to unmask]] On Behalf Of Ashley
Sent: Saturday, 4 February 2017 9:42 PM
To: [log in to unmask]
Subject: [BUGS] Setting up initial values with the last value in coda file

 

Hello. All in List,

 

My Openbugs code ran successfully with 20,000 updates, but results of convergence test showed that more updates are required. I’m thinking to extend the updates up to 50,000 (i.e., additional 30,000 updates). At this moment, because I already did 20,000 updates and don’t want to waste  time, I wondered what if I use the last values in coda files (which are saved in ‘State’ file) as initial values. If do so, can the first update from the second run be treated as 20,001th update? Thinking of how MCMC works, because the current estimates are obtained from the previous estimates and the first estimates are obtained from the initial values, I thought it might be ok. But, because I haven’t seen anything like that, I’d like to listen to your thoughts. Please let me know what I am missed.

 

Take care,

Ashley

------------------------------------------------------------------- This list is for discussion of modelling issues and the BUGS software. For help with crashes and error messages, first mail [log in to unmask] To mail the BUGS list, mail to [log in to unmask] Before mailing, please check the archive at www.jiscmail.ac.uk/lists/bugs.html Please do not mail attachments to the list. To leave the BUGS list, send LEAVE BUGS to [log in to unmask] If this fails, mail [log in to unmask], NOT the whole list

 

------------------------------------------------------------------- This list is for discussion of modelling issues and the BUGS software. For help with crashes and error messages, first mail [log in to unmask] To mail the BUGS list, mail to [log in to unmask] Before mailing, please check the archive at www.jiscmail.ac.uk/lists/bugs.html Please do not mail attachments to the list. To leave the BUGS list, send LEAVE BUGS to [log in to unmask] If this fails, mail [log in to unmask], NOT the whole list

-- 
Cory Atwood
Statwood Consulting
2905 Covington Rd
Silver Spring, MD 20910
301-589-7158