eloping a way of specifying initial values for each
> Markov chain?
>
> Sam
> -Original Message-
> From: Whit Armstrong [mailto:armstrong.w...@gmail.com]
> Sent: 19 December 2011 18:24
> To: Watson, Samuel
> Cc: rcpp-de...@r-forge.wu-wien.ac.at
> Subject: Re: [Rcpp
@gmail.com]
Sent: 19 December 2011 18:24
To: Watson, Samuel
Cc: rcpp-de...@r-forge.wu-wien.ac.at
Subject: Re: [Rcpp-devel] CppBugs vs WinBugs
Oh, that's great. I thought you said the linear model was only showing 2x in
your first email.
Glad you are finding it useful. Eventually, one sh
adapt=2e2L I still get 9.14 seconds. The model is definitely
>> producing the correct answer (as compared to WinBUGS and jags), so I can't
>> work out why I can't an equivalent speed to you.
>> Would you have any other recommendations of things that could be slowing the
>> mod
fference for me for winbugs vs
cppbugs is 40s vs 10s which is a big difference.
I will test some more models and hopefully this model is just an anomaly for me.
Thanks for a great package!
Sam
-Original Message-
From: Whit Armstrong [mailto:armstrong.w...@gmail.com]
Sent: 19 December 201
n
>
> -Original Message-
> From: Whit Armstrong [mailto:armstrong.w...@gmail.com]
> Sent: 19 December 2011 15:23
> To: Watson, Samuel
> Cc: rcpp-de...@r-forge.wu-wien.ac.at
> Subject: Re: [Rcpp-devel] CppBugs vs WinBugs
>
> You have the model specified wrong.
>
Samuel
Cc: rcpp-de...@r-forge.wu-wien.ac.at
Subject: Re: [Rcpp-devel] CppBugs vs WinBugs
You have the model specified wrong.
in your wrapper function you call:
> RadonVaryingInterceptModel m(group,level,basement,N,N_counties);
However, the constructor requires the 'level' variable
meric(county)
> radon.data<-list(log.radon=log.radon,floor=floor,county=county)
> radon.param<-c("a","b")
> radon.inits<-function(){list(a=rnorm(85),b=rnorm(1),sigma.a=runif(1),sigma.y=runif(1),mu.a=rnorm(1)}
>
> #n.burnin = n.iter/2 for bugs
> system.time(ra
me(radon.test<-bugs(radon.data,radon.inits,radon.param,model.file="radon.bug",n.chains=1,n.iter=2,working.directory=getwd(),bugs.directory="D:/R/WinBUGS14",n.thin=5))
---------------
The difference I get is 16.53s
post your code.
The speedup depends on your model. My comparisons were based on
R/JAGS and pymc on linux.
-Whit
On Mon, Dec 19, 2011 at 7:30 AM, Watson, Samuel
wrote:
> I am currently using CppBugs with Rcpp through R. I am very interested to
> use CppBugs as I am finding WinBugs to be prohi
I am currently using CppBugs with Rcpp through R. I am very interested
to use CppBugs as I am finding WinBugs to be prohibitively slow, I use
large amounts of data in large multilevel models, so when I found
cppbugs I was excited. It says on the Github page for cppbugs that I can
achieve speeds of
10 matches
Mail list logo
