[R-pkgs] New version: dclone 1.7-1 and dcmle 0.1-4 rolled out

Peter Solymos solymos at ualberta.ca
Tue Feb 14 18:09:20 CET 2012

Dear R Community,

I am happy to introduce the next release of the *dclone* package (see
R Journal, 2(2): 29-37, 2010) which now includes a suite of functions
for parallel computations:

- see functions parJagsModel, parUpdate, parCodaSamples using 'snow'
type clusters; and the
- jags.parfit function that can be used with 'snow' type clusters or
multiple cores.
- iterative model fitting with the data cloning algorithm can exploit
various types of parallelism (parallel chains, 'size' balancing or the
two combined) through the dc.parfit function using JAGS, WinBUGS or
- JAGS related parallel functions can safely handle RNGs even for
large numbers of parallel chains using the "lecuyer" JAGS module.

The *dcmle* package defines S4 classes around infrastructure provided
by the dclone package to make package development with data cloning
for hierarchical models easy as a breeze.

The testing suite of the two packages now includes 30 examples (see
listDcExamples() ).

cl <- makePSOCKcluster(3)
m <- dcmle(seeds, n.clones=c(1,2,4), cl=cl, partype="parchains")

I am grateful for the feedback and support provided by a growing user
base. Special thanks to Martyn Plummer for exposing RNG handling in
rjags for better integration, and to Khurram Nadeem for catching the
most annoying bugs on the test run.



Peter Solymos
Alberta Biodiversity Monitoring Institute
and Boreal Avian Modelling project
Department of Biological Sciences
CW 405, Biological Sciences Bldg
University of Alberta
Edmonton, Alberta, T6G 2E9, Canada
Phone: 780.492.8534
Fax: 780.492.7635
email <- paste("solymos", "ualberta.ca", sep = "@")

More information about the R-packages mailing list