Greetings fellow R entusiasts! We have some problems converting a computer routine written initially for Gauss to estimate a Markov Regime Switching analysis with Time Varying Transition Probability. The source code in Gauss is here: http://www.econ.washington.edu/user/cnelson/markov/programs/hmt_tvp.opt We have converted the code to R, and it's running without errors, but we have some convergence problems. According to the authors of the Gauss code, the initial guess for the Transition Matrix (probability of going from one regime to the other) could be chosen arbitrary, but unfortunately this is not the case for our R code. Also, we do not have Gauss available to test the original source code. A function used in Gauss is called "optmum", while R has a function called "optim". Are these the same? If not, this might be the cause of our convergence problems. I would be glad to share the R program with anyone interested, as well as the panel data used in the analysis. Best, J?rgen Blystad Houge jorgehou at stud.ntnu.no -- View this message in context: http://r.789695.n4.nabble.com/Markov-Switching-with-TVTP-problems-with-convergence-tp3013292p3013292.html Sent from the R help mailing list archive at Nabble.com.
Whit Armstrong
2010-Oct-26 14:42 UTC
[R] Markov Switching with TVTP - problems with convergence
I've looked at the Kim/Nelson gauss code before, and I applaud your effort to convert it to R. I'm happy to have a look at it for you if you are willing to share your example. -Whit On Tue, Oct 26, 2010 at 4:13 AM, Houge <jb.houge at gmail.com> wrote:> > Greetings fellow R entusiasts! > > We have some problems converting a computer routine written initially for > Gauss to estimate a Markov Regime Switching analysis with Time Varying > Transition Probability. The source code in Gauss is here: > http://www.econ.washington.edu/user/cnelson/markov/programs/hmt_tvp.opt > > We have converted the code to R, and it's running without errors, but we > have some convergence problems. According to the authors of the Gauss code, > the initial guess for the Transition Matrix (probability of going from one > regime to the other) could be chosen arbitrary, but unfortunately this is > not the case for our R code. Also, we do not have Gauss available to test > the original source code. > > A function used in Gauss is called "optmum", while R has a function called > "optim". Are these the same? If not, this might be the cause of our > convergence problems. > > I would be glad to share the R program with anyone interested, as well as > the panel data used in the analysis. > > Best, > J?rgen Blystad Houge > jorgehou at stud.ntnu.no > -- > View this message in context: http://r.789695.n4.nabble.com/Markov-Switching-with-TVTP-problems-with-convergence-tp3013292p3013292.html > Sent from the R help mailing list archive at Nabble.com. > > ______________________________________________ > R-help at r-project.org mailing list > https://stat.ethz.ch/mailman/listinfo/r-help > PLEASE do read the posting guide http://www.R-project.org/posting-guide.html > and provide commented, minimal, self-contained, reproducible code. >