[R] R, Coda, and OpenBUGS

[Uwe Ligges]

Bill Halteman wrote:
> Hi All,
> 
>   I am trying to figure out how to use R-Coda with the output from 
> OpenBugs. I have installed and loaded the packages BRugs and R2WinBUGS.  
> I have successfully run a simple Bayes model in WinBUGS using R2WinBUGS' 
> "bugs" and have used "read.bugs" to build the coda object.  I can 
> successfully switch to OpenBugs and run the same model and get the basic 
> summary back.  However, I cannot build the coda object.  From what I 
> have read "read.openbugs" should do the trick.  However, I have 
> discovered the current version of OpenBUGS (v 3.03) doesn't write any 
> output files in spite of setting codaPkg=TRUE in the "bugs" call.  
> "Str(openbugs.object)" reveals a large collection of elements, one of 
> which is an array that contains the MCMC chains.  Is there a function to 
> build the coda object?


For BRugs: Sure, the function is called buildMCMC(). From the examples 
of ?BRugs:

library(BRugs) # loading BRugs

## Now setting the working directory to the examples' one:
oldwd <- getwd()
setwd(system.file("OpenBUGS", "Examples", package="BRugs"))

## some usual steps (like clicking in WinBUGS):
modelCheck("ratsmodel.txt")          # check model file
modelData("ratsdata.txt")            # read data file
modelCompile(numChains=2)            # compile model with 2 chains
modelInits(rep("ratsinits.txt", 2))  # read init data file
modelUpdate(1000)                    # burn in
samplesSet(c("alpha0", "alpha"))     # alpha0 and alpha should be monitored
modelUpdate(1000)                    # 1000 more iterations ....


##  now let's construct a coda object
## (of class mcmc.list, to be precise)
## for BUGS node "alpha":

alpha <- buildMCMC("alpha")


Best,
Uwe Ligges


> Thanks in advance for any suggestions,
> 
> Bill Halteman
> 
> ______________________________________________
> 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.