[R-sig-hpc] Rmpi matrix manipulation

[Hao Yu]

The first argument of mpi.parLapply is x which is mixed with your x.
Change your x to a different one and your codes should work.

Hao

Xiaochun Sun wrote:
> Hi,
>
> I am using Rmpi and met a problem about using Rmpi to call a function to
> do
> matrix manipulation. I can run the following code using lapply but when I
> use mpi.parLapply the error pop out and saying
> "Error in x %*% b : non-conformable arguments\n"
>
> Could anyone help me handle the problem? Thanks in advance.
>
> Sun
>
>
>
> below is the code:
> n.chains = 2
> ib <- runif(n.chains)
> b <- t(matrix(ib,nrow=n.chains,ncol=500))
> isbeta <- matrix(runif(n.chains,0,0.01),nrow=ncol(x),ncol=n.chains)
> niter=50
> y<- rep(1,500)
> x<- matrix(1,500,500)
> l <- rep(1,500)
>
> Gibbs <- function(y,x,l,isbeta,b,niter){
>    ll <- t(l) %*% l
>    XX <- t(x) %*% x
>    y_star <- y - x %*% b
>    lY <- t(l) %*% y_star
>    XY <- t(x) %*% y
>    ibeta_covar=solve(diag(sbeta))
> }
>
> library(Rmpi)
> mpi.spawn.Rslaves()
>
> bbb_1<- mpi.parLapply(1:2, fun=function(i,Gibbs,y,x,l,isbeta,b,niter){
>
> Gibbs(y,x,l,isbeta[,i],b[,i],niter)},Gibbs=Gibbs,y=y,x=x,l=l,isbeta=isbeta,b=b,niter=niter)
>
> 	[[alternative HTML version deleted]]
>
> _______________________________________________
> R-sig-hpc mailing list
> R-sig-hpc at r-project.org
> https://stat.ethz.ch/mailman/listinfo/r-sig-hpc
>


-- 
Department of Statistics & Actuarial Sciences
Fax Phone#:(519)-661-3813
The University of Western Ontario
Office Phone#:(519)-661-3622
London, Ontario N6A 5B7
http://www.stats.uwo.ca/faculty/yu