How is it possible to make a matrix of array in R? - r

Question

What i want to have is a matrix in which each element is an array itself.
This array is taken subsetting a dataframe, but the example can be generalized for any array.
I tried with:
My_matrix <- matrix(array(), nrow = NROW, ncol = NCOL)
for (i in 1:NROW){
for(j in 1:NCOL){
My_matrix[i,j] <- df[df$var1 == j & df$var2== i,]$var3
}
}
but I got this message error:
Error in My_matrix[i,j] <- df[df$var1== j & df$var2== i,]$var3 :
number of items to replace is not a multiple of replacement length
How should I define and access each element of the matrix and each element of the contained array?

Answer 1

I think I understand that: (1) the base array is 45x3; (2) each cell has a differently sized matrix; and (3) this is not known apriori. Gotcha. Not possible. An array (matrix) is always perfectly dimensioned, and while you can dynamically change one or more of the dimensions, you change for all cells.
Alternative: list-columns.
dat <- data.frame(x=1:3, y=11:13)
dat$z <- lapply(3:5, function(i) matrix(seq_len(i^2), nr=i))
dat
# x y
# 1 1 11
# 2 2 12
# 3 3 13
# z
# 1 1, 2, 3, 4, 5, 6, 7, 8, 9
# 2 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16
# 3 1, 2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25
That doesn't look very appealing, but if you want a different presentation, you might consider assigning it as a tibble::tbl_df (available whenever dplyr is loaded as well). (Note that presentation is distinct from storage and accessibility.)
library(tibble)
as_tibble(dat)
# # A tibble: 3 x 3
# x y z
# <int> <int> <list>
# 1 1 11 <int[,3] [3 x 3]>
# 2 2 12 <int[,4] [4 x 4]>
# 3 3 13 <int[,5] [5 x 5]>
Subsetting is consistent:
dat$z[ dat$x == 2 & dat$y == 12 ]
# [[1]]
# [,1] [,2] [,3] [,4]
# [1,] 1 5 9 13
# [2,] 2 6 10 14
# [3,] 3 7 11 15
# [4,] 4 8 12 16
### note that you need an extra [[1]] to get to the real data
m <- dat$z[ dat$x == 2 & dat$y == 12 ][[1]]
m
# [,1] [,2] [,3] [,4]
# [1,] 1 5 9 13
# [2,] 2 6 10 14
# [3,] 3 7 11 15
# [4,] 4 8 12 16
m[3,4]
# [1] 15