the optimization algorithm BHHH
Michael Creel
michael.creel at uab.es
Thu Jul 23 04:35:33 CDT 2009
george216 wrote:
>
> Dear friends,
> I am very interested with the optimization algorithm of Berndt, B. Hall,
> R.
> Hall, and Jerry Hausman
> <http://en.wikipedia.org/wiki/Jerry_Hausman>(Berndt, E., B. Hall, R.
> Hall, and J. Hausman, (1974), “Estimation and
> Inference in Nonlinear Structural Models”, Annals of Social Measurement,
> Vol. 3, 653-665.). Such algorithm is advised for the estimation of smooth
> transition autoregression models. Is there any implementation of this
> algorithm in Octave. many thanks in advance.
> George.
>
> _______________________________________________
> Help-octave mailing list
> Help-octave at octave.org
> https://www-old.cae.wisc.edu/mailman/listinfo/help-octave
>
>
If I recall correctly, that algorithm approximates the Hessian using the
outer product of the gradient. I also seem to recall that the algorithm
often has trouble converging due to the OPG becoming rank deficient during
the course of intermediate iterations. Consulting the foggy annals of my
experience, I believe that for most problems where BHHH would be a
possibility, BFGS or LBFGS will work better. Both of those are available in
the optim package.
Michael
Michael
--
View this message in context: http://www.nabble.com/the-optimization-algorithm-BHHH-tp24616173p24622378.html
Sent from the Octave - General mailing list archive at Nabble.com.
More information about the Help-octave
mailing list