如何并行化combn()? [英] How can I parallelize combn()?

问题描述

函数combn()一次生成m个x元素的所有组合.对于nCm小(其中n是x的元素数)，这是非常快速且高效的，但很快就会耗尽内存.例如:

The function combn() generates all combinations of the elements of x taken m at a time. It is very fast and efficient for nCm small (where n is the number of elements of x) but it quickly runs out of memory. For example:

> combn(c(1:50), 12, simplify = TRUE)
Error in matrix(r, nrow = len.r, ncol = count) : 
invalid 'ncol' value (too large or NA)

我想知道是否可以修改功能combn()使其仅生成k个选定的组合.我们将此新函数称为selectedcombn().然后我们将:

I would like to know if the function combn() can be modified such that it generates only k chosen combinations. Let's call this new function chosencombn(). Then we would have:

> combn(c("a", "b", "c", "d"), m=2)
     [,1] [,2] [,3] [,4] [,5] [,6]
 [1,] "a"  "a"  "a"  "b"  "b"  "c" 
 [2,] "b"  "c"  "d"  "c"  "d"  "d" 

>chosencombn(c("a", "b", "c", "d"), m=2, i=c(1,4,6))
     [,1] [,2] [,3]
 [1,] "a"  "b"  "c" 
 [2,] "b"  "c"  "d"

>chosencombn(c("a", "b", "c", "d"), m=2, i=c(4,5))
     [,1] [,2]
 [1,] "b"  "b" 
 [2,] "c"  "d" 

我知道，这种功能需要使用组合的顺序，以便人们可以立即找到给定组合的位置. 是否存在这样的顺序?可以对它进行编码以获得像combn()一样高效的功能吗?

I understand that such a function would require the use of an ordering of the combinations such that one can find the place of a given combination immediately. Does such an ordering exist? Can it be coded to obtain a function as efficient as combn()?

推荐答案

包裹猪蹄" 很有用为此，因为它不会将排列保留在内存中.

Package "trotter" is useful for this as it does not keep the permutations in memory.

library(trotter)

combs = cpv(2, c("a", "b", "c", "d"))
sapply(c(1, 4, 6), function(i) combs[i])
#     [,1] [,2] [,3]
#[1,] "a"  "b"  "c" 
#[2,] "b"  "c"  "d"

这篇关于如何并行化combn()?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！