[ADMB Users] "Hessian does not appear to be positive definite"

[dave fournier]

Your problme is that the data are too good so the model fits perfectly
with an estimated sigma of almost 0. I added a tiny bit of noise to the
data and it converged fine.

  Kd=column(Data,1);
 random_number_generator rng(301);
 dvector tmp(1,Nobs_tot);
 Kd=elem_prod(Kd,exp(.01*tmp));   // add some lognormal noise to Kd

However with real data you may need to civilize the code a bit.

-- 
David A. Fournier
P.O. Box 2040,
Sidney, B.C. V8l 3S3
Canada
Phone/FAX 250-655-3364
http://otter-rsch.com