Hello, I am about to move all of my modelling work into R, and I have been investigating the present state of MCMC and Bayesian methods in R. Following a thread on the mailing list in 2000, I have looked at mcmcpack and Hydra. Three years down the line, is there anything new in this area? I have used both MCSim and WinBUGS in the past. The first one seems promising, but is too focused towards toxicology work, and needs some changes to accomodate sequential Bayesian models. WinBUGS is closed, and runs in windows only. Somebody has posted recently a message in the WinBUGS list asking for feedback on the posibility of developing a Linux version, but that does not seem to be happening anytime soon. What lines of work are being carried out at the moment? Can I esxpect something new in the near future? Regarding the possible use of MCSim (which translates a simple model especification into C code and then compiles it) as a base for a R package, what kind of questions should I look more carefully at? Many thanks, iago MCSim: http://toxi.ineris.fr/activites/toxicologie_quantitative/mcsim/mcsim.php
On Thu, 21 Aug 2003 09:20:50 +0200, "Iago Mosqueira" <imosqueira@suk.azti.es> wrote :> ...Somebody has posted recently a message >in the WinBUGS list asking for feedback on the posibility of developing >a Linux version, but that does not seem to be happening anytime soon. > >What lines of work are being carried out at the moment? Can I esxpect >something new in the near future?Depends what you mean by "near future". The gR project <http://www.r-project.org/gR/> plans to add support for graphical models to R; this will include MCMC methods. There's a workshop next month, at which time some target dates might be set. I'd guess by a year from now there'll be substantial progress, but I'm not sure when the first signs of it will appear. Duncan Murdoch
I also am 'almost' done with an interface between my Hydra MCMC package (see http://hydra-mcmc.sf.net) and R. You write the prior and the likelihood using R functions, and it will run the MCMC sampler in a separate process. Anyone who is interested in helping out with programming or testing, let me know. -G> -----Original Message----- > From: Duncan Murdoch [mailto:dmurdoch@pair.com] > Sent: Thursday, August 21, 2003 10:37 AM > To: Iago Mosqueira > Cc: r-devel@stat.math.ethz.ch > Subject: Re: [Rd] mcmc > > > On Thu, 21 Aug 2003 09:20:50 +0200, "Iago Mosqueira" > <imosqueira@suk.azti.es> wrote : > > > ...Somebody has posted recently a message > >in the WinBUGS list asking for feedback on the posibility of > developing > >a Linux version, but that does not seem to be happening anytime soon. > > > >What lines of work are being carried out at the moment? Can I esxpect > >something new in the near future? > > Depends what you mean by "near future". The gR project > <http://www.r-project.org/gR/> plans to add support for graphical > models to R; this will include MCMC methods. There's a workshop next > month, at which time some target dates might be set. I'd guess by a > year from now there'll be substantial progress, but I'm not sure when > the first signs of it will appear. > > Duncan Murdoch > > ______________________________________________ > R-devel@stat.math.ethz.ch mailing list > https://www.stat.math.ethz.ch/mailman/listinfo/r-devel >LEGAL NOTICE\ Unless expressly stated otherwise, this messag...{{dropped}}