可以并行使用pbapply吗? [英] Is it possible to use pbapply in parallel?

问题描述

我有这段代码想要对矩阵的列求和.

I have this code where I want to sum the columns of a matrix.

我还希望具有几个功能(我的矩阵比下面的示例大得多):

I also want to have a couple of features (my matrix is much bigger than the example below):

• 使用多个内核
• 具有进度栏

我尝试了pbapply包中的pbapply，但是没有运气.我的代码是:

I tried pbapply from the pbapply package but without luck. My code is:

library(pbapply)
library(parallel)

mat <- matrix(ncol=20, nrow = 50, data = runif(1000))

# sum of the columns
matsum <- apply(mat,2,sum)

# now the same in parallel
cl <- makeCluster(2)
pboptions(type = "txt")
parmat <- pbapply(mat,2,sum, cl=cl)
stopCluster(cl)

这会导致错误:

FUN(newX [，i]，...)中的错误:参数的类型"(列表)无效

Error in FUN(newX[, i], ...) : invalid 'type' (list) of argument

我的问题是，根本可以并行运行pbapply吗?

My question is, is it possible at all to run pbapply in parallel?

推荐答案

pbapply不采用cl自变量，因此否.但是，您可以将矩阵转换为data.frame(如果可行)，并使用适用于列的sapply.

pbapply doesn't take the cl argument, so no. You can, however, convert your matrix to a data.frame (if that's feasible), and use sapply which should work on columns.

mat <- as.data.frame(mat)
parmat <- pbsapply(X = mat, FUN = sum, cl = cl)

或者，您可以尝试

colSums(mat)

这是主力军.

这篇关于可以并行使用pbapply吗?的文章就介绍到这了，希望我们推荐的答案对大家有所帮助，也希望大家多多支持IT屋！