Expand the shape of an array

This is the inverse operation of base::drop(). It is analogous to python's numpy.expand_dims(), but vectorized on which_dim.

expand_dims(x, which_dim = -1L)

Arguments

x	an array. Bare vectors are treated as 1-d arrays.
which_dim	numeric. Desired index position of the new axis or axes in the returned array. Negative numbers count from the back. Can be any length.Throws a warning if any duplicates are provided.

Value

the array x with new dim

Examples

x <- array(1:24, 2:4)
dim(x)
#> [1] 2 3 4
dim(expand_dims(x))
#> [1] 2 3 4 1
dim(expand_dims(x, 2))
#> [1] 2 1 3 4
dim(expand_dims(x, c(1,2)))
#> [1] 1 1 2 3 4
dim(expand_dims(x, c(1,-1)))
#> [1] 1 2 3 4 1
dim(expand_dims(x, 6)) # implicitly also expands dims 4,5
#> [1] 2 3 4 1 1 1
dim(expand_dims(x, 4:6))
#> [1] 2 3 4 1 1 1

# error, implicit expansion with negative indexes not supported
try(expand_dims(x, -6))
#> Error in expand_dims(x, -6) : 
#>   Implicit additional dims for expansion with negative indexes not supported

# supply them explicitly instead
dim(expand_dims(x, -(4:6)))
#> [1] 1 1 1 2 3 4