[ADMB Users] -lprof

John Sibert sibert at hawaii.edu
Mon Jul 21 12:02:17 PDT 2014


Its a tagest issue, so not so simple to send stuff. It could be done or 
I could  set up a run directory and code on katsuo.

The value assigned to fmc.fbest does not seem to change from its default.


John Sibert
Emeritus Researcher, SOEST
University of Hawaii at Manoa
Honolulu HI (GMT-10)
808-294-3842 (mobile)

Visit the ADMB project http://admb-project.org/

On 07/20/2014 03:43 PM, dave fournier wrote:
> On 07/20/2014 06:39 PM, John Sibert wrote:
>
> If you send me the stuff to run it I can take a look.
>
>
>
>> That helped to get a good trace back. Thank.
>>
>> It calls
>>> likeprof_routine(ffbest);
>> with argument 1e+100 near line 263 in modspmin.cpp.
>>
>> There are a lot of complaints "Estimated covariance matrix may not be 
>> positive definite", but it seems to estimate the SDs OK.
>>
>> It may be that the it is not a real minimum, but it would be nice to 
>> have a more graceful exit.
>>
>>
>>
>>
>>
>> John Sibert
>> Emeritus Researcher, SOEST
>> University of Hawaii at Manoa
>> Honolulu HI (GMT-10)
>> 808-294-3842 (mobile)
>>
>> Visit the ADMB project http://admb-project.org/
>>
>> On 07/20/2014 11:25 AM, dave fournier wrote:
>>> feenableexcept(FE_DIVBYZERO | FE_INVALID | FE_OVERFLOW ); 
>>
>>
>
>



More information about the Users mailing list