Closed iantaylor-NOAA closed 3 years ago
Okay, try the changes in issue164 branch. The main change is to return from the ::computations1 function instead of exiting. The FINAL_SECTION code will then be called.
Thanks @johnoel! I tested the changes in the issue164 branch with Stock Synthesis and It worked well! Even though I saw the message
Warning -- Hessian does not appear to be positive definite
Hessian does not appear to be positive definite
The FINAL_SECTION code proceeded to run after.
ADMB doesn't enter the FINAL_SECTION when the Hessian is not positive definite.
I confirmed this by adding the following to the end of catage.tpl
which causes the run to end with
but running with
-maxfn 0
ends withThe FINAL_SECTION (which isn't well described in the ADMB manual) is a convenient place to write output that depends on the result of the Hessian inversion (as discussed in #132), but it would be nice to get any output that doesn't depend on the Hessian in cases where the Hessian has failed to invert. Would it be possible to go to the FINAL_SECTION even when the Hessian fails?