[ADMB Users] non-convergence and "inner maxg = 0"

lennart at ii.uib.no lennart at ii.uib.no
Fri Mar 19 05:28:22 PDT 2010


I am an end user of admb, not an expert, so don't necessarily take my 
word for it, but:

I had the same problem. In my case this was (as far as I could tell) 
caused by an infinite loop in the file df1b2-separable/dflocmin.cpp, 
line 74, which reads:

while (counter<20);

This is the end of a do...while loop. The problem is that counter never 
gets updated, so I just replaced it with

while (++counter<20);

..recompiled ADMB, recompiled my program, and everything was fine. By 
fine I mean that the program terminates --- there is probably still 
something wrong with the model, since the termination message now reads:

Error cannot find better value to try and get a positive definite hessian
Can't make a positive definite Hessian

Hope this helps,


On 19/03/10 05:00, Ian Fiske wrote:
> First, I fit a fairly complex model in ADMB and it works great.  Next, 
> I tried to add random effects and use ADMB-RE, but ADMB-RE appears to 
> not converge.  The oddity is that when the program prints out (via 
> cout statements) the parameters, they appear to stabilize to the same 
> values as in the non-RE version.  And (omitting the parms), the 
> program prints

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.admb-project.org/pipermail/users/attachments/20100319/9655ce6f/attachment.html>

More information about the Users mailing list