从矩阵列表中删除矩阵

Question

I have a list of 12 matrices called M and I'm trying to remove every matrix from the list that has 0 rows.

I know that I can manually remove those matrices with (for example, to remove the second matrix) M[2] <- NULL . I would like to use logic to remove them with something like: M <- M[nrow(M)>0,] (but that obviously didn't work).

Answer 1

Another option that could work is Filter in base R

Filter(nrow, M)

It works because 0 is considered as FALSE and all other values as TRUE

If there are also some attributes, gv from collapse could maintain it

library(collapse)
gv(M, function(x) nrow(x) > 0)

Answer 2

Use sapply() to get a logical vector of non-zero matrices, then use that vector to select/subset:

nzm <- sapply(M, function(m) nrow(m)>0)
M <- M[nzm]

Answer 3

Here is a one-liner.

M <- M[sapply(M, nrow) != 0]

---

Data creation code

M <- lapply(1:5, function(n){
  if(n %% 2 == 0)
    matrix(nrow = 0, ncol = 2)
  else
    matrix(1:4, nrow = 2)
})

Answer 4

I think @Akrun has the cleanest answer. I made this up before I saw his. It is reproducible and also explains a different way of going about it that I used with out thinking about more elegant solutions.

# create list of matrices
matrixlist <- list(matrix(nrow=4,ncol=4,1),
     matrix(nrow=4,ncol=4,1),
     matrix(nrow=0,ncol=4,1),
     matrix(nrow=4,ncol=4,1))
matrixlist

# Identify matrices in my list that have at least one row
idx <- lapply(lapply(matrixlist, `[`), nrow) > 0

# subset the original list with this criteria
my_revised_matrixlist <- matrixlist[idx]

my_revised_matrixlist

All this was accomplished by Akrun's simple Filter(nrow, matrixlist)

Answer 5

Another option worth mentioning with purrr is keep() , discard() , or compact() . The package purrr is my go-to these days with lists.

library(purrr)

# simplest by far
compact(M)

# probably a bit easier to remember what you did later
keep(M, ~ nrow(.) > 0)
discard(M, ~ nrow(.) == 0)