C-version of the .rgigjd and .rgigjd1 functions in fBasics
Yohan Chalabi
chalabi at phys.ethz.ch
Tue Mar 30 17:44:23 CEST 2010
>>>> "EP" == Ester Pantaleo <esterpantaleo at gmail.com>
>>>> on Tue, 23 Mar 2010 14:27:41 +0100
EP> To the Maintainers of the package fBasics,
EP>
EP>
EP>
EP> I thought you might be interested in the attached code
EP> written by Robert B. Gramacy and me which. It is inspired by
EP> the rgig function in ghyp which is essentially identical to
EP> the .rgigjd and .rgigjd1 functions in fBasics. The attached
EP> code is a merging of a C-version of the code in rgig and some
EP> code from monomvn (a package for estimation of multivariate
EP> normal and student-t data of arbitrary dimension where the
EP> pattern of missing data is monotone).
EP>
EP>
EP> I am sending this in the spirit of the open source community
EP> because I believe that the C-version, being much faster,
EP> could contribute to your package (see file time.R for a
EP> comparison between the R and C codes). Some details about the
EP> code follow. The R function uniroot has been replaced by a
EP> modified version of ZEROIN from the C math library (algorithm
EP> from G.Forsythe, M.Malcolm, C.Moler). Inverse gaussian variates
EP> (case lambda=-0.5) are generated by the function rinvgauss
EP> (algorithm from Michael,Schucany,Haas as given in Gentle's
EP> book on page 193) essentially in the same way as in ghyp.
EP>
EP>
EP> Should you have any questions, please do not hesitate to ask.
EP>
EP>
EP> Best regards,
EP>
EP>
EP> Ester Pantaleo
EP>
EP>
EP> Dipartimento di Fisica
EP>
EP> Universita' di Bari
EP>
EP> Italy
Dear Ester,
Thank you very much for your contribution.
I have forwarded your message to David Luethi, the maintainer of the
ghyp package.
Kind regards,
Yohan
--
PhD candidate
Swiss Federal Institute of Technology
Zurich
www.ethz.ch
More information about the Rmetrics-core
mailing list