[R] Hessian in constrOptim

Spyridoula Tsonaka spyridoula.tsonaka at med.kuleuven.ac.be
Fri Apr 2 10:16:19 CEST 2004


Dear R-users,

In the function constrOptim there is an option to get an approximation
to the hessian of the surrogate function R at MLE by declaring
hessian=TRUE in the calls to the function optim. I would like to ask
if it is advisable to get an approximate hessian for the funcrion f as
follows:

f''(theta)=R''(theta|theta_k)-B''(theta)

where B''(theta)=mu*sum((g(theta_k)/g(theta)^2)u_i*u_i^T) denotes the
second derivative of the barrier function

(following the notation given by Lange (1999)) where
R''(theta|theta_k) will be replaced by the approximate hessian at MLE
returned by optim.

Thanks in advance,


Spyridoula Tsonaka
Doctoral Student
Biostatistical Centre
Katholieke Universiteit Leuven
Kapucijnenvoer 35
B-3000, Leuven
Belgium

Tel: +32/16/336887
Fax: +32/16/337015




More information about the R-help mailing list