[Developers] FW: sd_report error

Mark Maunder mmaunder at iattc.org
Fri Mar 9 14:24:03 PST 2012


Thanks Johnoel,

How did you fix this? Did you make it so that an error message was created or did you make it so it ignored the error if the product of the diagonal values got too large and just continued?

Mark



From: Johnoel Ancheta [mailto:johnoel at hawaii.edu]
Sent: Friday, March 09, 2012 2:09 PM
To: Mark Maunder
Subject: Re: FW: sd_report error

Hi Mark,  fixed the problem see

http://www.admb-project.org/redmine/issues/30

You will need to svn update to the latest trunk.

Johnoel
On Wed, Feb 29, 2012 at 6:46 AM, Mark Maunder <mmaunder at iattc.org<mailto:mmaunder at iattc.org>> wrote:
Johnoel,

See the attached. Please don't distribute the files (particularly the data file). Some details are below.

Thanks,

Mark

-----Original Message-----
From: Mark Maunder
Sent: Tuesday, February 28, 2012 2:12 PM
To: 'Teresa A'mar'
Subject: RE: sd_report error

Teresa,

I have attached the files I used for the model that does not produce an error in the hessian, but does not produce a std file. The pin file is from a par file so it does not have to estimate the parameters. I am using Borlands 5.5 with ADMB 10.1 downloaded from the website (I have to use 10.1 since I am using the negative binomial, which has a bug in 10.0).

I use the following to run it

smod -nox -maxfn 10000 -mno 15000 -ams 2000000 -gbs 100000000

It would be interesting to see what happens when it is compiled and run using a different compiler.

Please do not distribute the data file.

Thanks,

Mark

-----Original Message-----
From: Teresa A'mar [mailto:zta at uw.edu<mailto:zta at uw.edu>]
Sent: Tuesday, February 28, 2012 1:39 PM
To: Mark Maunder
Subject: sd_report error

Hi Mark,

Could you send some files which produce the error you have been referring to?  Also, which version or Subversion revision of ADMB are you using?

Thanks.

-Teresa

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.admb-project.org/pipermail/developers/attachments/20120309/36946056/attachment.html>


More information about the Developers mailing list