[ADMB Users] ADMB error messages

[Jim Ianelli]

I looked for this in the source code and found:

fmmtr1.cpp:           if (ad_printf) (*ad_printf)("  ic > imax  in fminim is
answer attained ?\n" );
newfmin.cpp:           if (ad_printf) (*ad_printf)("  ic > imax  in fminim
is answer attained ?\n" );

then I looked at imax and it occurs in many places (but difficult to tell
where it's set for the function minimize and I was too lazy to figure that
out).

I suspect that you may have exceeded the number of function evaluations or
that imax is somehow reset within your model (or application to a library).

I'm sure you have, but maybe check if function evaluations is being set as
intended, e.g., 

  maximum_function_evaluations 50, 100, 10000

Cheers,
Jim

| -----Original Message-----
| From: users-bounces at admb-project.org [mailto:users-bounces at admb-
| project.org] On Behalf Of k.curti
| Sent: Wednesday, April 28, 2010 12:04 PM
| To: users at admb-project.org
| Subject: [ADMB Users] ADMB error messages
| 
| Hi-
| 
| I have a general question regarding admb error messages (I'm using the
| most recent version of ADMB and have tried both the Borland and GNU
| compilers).  In several of my model runs, I obtain the following error
| message in the captured tool output, but my other model diagnostics
| appear acceptable.
| 
| "ic > imax  in fminim is answer attained ?"  ...followed by a set of
| derivative/gradient information and then...
| "Function minimizer not making progress ... is minimum attained?
| Minimprove criterion =   0.0000e+00"
| 
| Sometimes I receive this message at the end of an intermediate
| parameter estimation phase, and sometimes I receive it at the end of
| the final estimation phase.  In one of my recent runs I received the
| above error in the last two estimation phases, but the maximum
| gradient component was 0.0020, the Hessian matrix was positive
| definite, and the maximum parameter correlation was -0.7130.  Is this
| error message a show-stopper when it appears at the end of the last
| estimation phase, or can it be ignored since the other diagnostics
| seem acceptable?  In addition, can anyone possibly recommend
| additional diagnostics that can be used to further deduce where the
| problem is in my model?
| 
| Many thanks
| _______________________________________________
| Users mailing list
| Users at admb-project.org
| http://lists.admb-project.org/mailman/listinfo/users