Scribd
Upload
40 views

Assignment 2b Advanced Programming in R: Ika Pratiwi

This code is running parallel processing on a matrix using the parallel and doSNOW packages in R. It detects the number of CPU cores, creates a cluster across those cores, splits the matrix columns across the cluster, calculates the column means in parallel, and combines the results. As an alternative to clusterApply, it shows using foreach with the %dopar% operator to calculate the column means in parallel and combine the results.

Uploaded by

Ika Pratiwi
Copyright
© © All Rights Reserved
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
40 views

Assignment 2b Advanced Programming in R: Ika Pratiwi

This code is running parallel processing on a matrix using the parallel and doSNOW packages in R. It detects the number of CPU cores, creates a cluster across those cores, splits the matrix columns across the cluster, calculates the column means in parallel, and combines the results. As an alternative to clusterApply, it shows using foreach with the %dopar% operator to calculate the column means in parallel and combine the results.

Uploaded by

Ika Pratiwi
Copyright
© © All Rights Reserved
Available Formats
Download as PPTX, PDF, TXT or read online on Scribd
You are on page 1/ 6

Assignment 2b

Advanced Programming in R
Ika Pratiwi
Run and describe what the code below does.
#include package parallel on library
>library(parallel)

#assign value of cores by get the number of CPU cores
>cores <- detectCores()
>cores
[1] 4

#assign value of cl by create a set of copies of cores
> cl<-makeCluster(cores)
>cl
socket cluster with 4 nodes on host localhost

#assign value of mat which is matrix of value generated by rnorm() function and have 16 column
#rnorm () function is random generation for the normal distribution, with number of observations equal to 160
mat<-matrix(rnorm(160),ncol=16)
>mat

Run and describe what the code below does.
#assign value of splt by splitting the cluster of cl with sequence 1 until the number of dimension of mat
#which is column
splt<-clusterSplit(cl, 1:dim(mat)[2])








#assign value of spltdat which is generated by lapply() function
#lapply() function returns a list with same length as splt by applying function(i)
#function (i) returns the value of all column of mat
spltdat<-lapply(splt, function(i) mat[,i])

Run and describe what the code below does.
#assign value of res by generated function clusterApply()
#function clusterApply() applies function colMeans to component spltdat on every cluster cl
#function colMeans determine the means of each column
res<-clusterApply(cl, spltdat,colMeans)








#assign value of totout by generated function do.call()
#do.call() function execute function call c with argument res to be passed
#c is a function to combine value res into a vector
totout<-do.call(c,res)



#shutdown the cluster cl
stopCluster(cl)
Try to find an alternative function that performs
the same operation
#include package doSNOW and foreach
library(doSNOW)
library(foreach)

#get the number of CPU cores
cores <- detectCores()

#assign value of cl by create a set of copies of cores
cls<-makeCluster(cores)

#register the SNOW(Simple Network Of Workstations) parallel backend with the foreach package
#cls is a cluster object to use for parallel execution
registerDoSNOW(cls)

#assign value of mat is matrix of value generated by rnorm() function and have 16 column
mat<-matrix(rnorm(160),ncol=16)

#foreach object from 1 until number of column of mat
#the result will be generated with c function
#%dopar% evaluates the function mean of each column of mat in parallel
foreach(i=1:ncol(mat), .combine=c) %dopar% (mean(mat[,i]))






Thank you

You might also like