Dimensionality Functions

expand_dims(x, ::Val{N})

Expands the dimensions of an array to a given number of dimensions.

Examples The result is a 5D array with singleton dimensions at the end

julia> expand_dims(ones((1,2,3)), Val(5))
1×2×3×1×1 Array{Float64, 5}:
[:, :, 1, 1, 1] =
 1.0  1.0

[:, :, 2, 1, 1] =
 1.0  1.0

[:, :, 3, 1, 1] =
 1.0  1.0

expand_dims(x::AbstractArray{T, N}, dims::Vararg{Int, M})  where {T, N, M}

Insert singleton dimensions at the position of dims into x. Based on a reshape operation.

Examples

julia> expand_dims(zeros((2,2)), 1) |> size
(1, 2, 2)

julia> expand_dims(zeros((2,2)), 2) |> size
(2, 1, 2)

julia> expand_dims(zeros((2,2)), 3) |> size
(2, 2, 1)

julia> expand_dims(zeros((2,2)), 1,3,4) |> size
(1, 2, 1, 1, 2)

expand_size(sz,sz2)

Expands a size tuple sz with the sizes as given in the tuple sz2 for positions which do not exist in sz. Typically one wants to

Example:

julia> expand_size((1,2,3),(4,5,6,7,8,9))
(1, 2, 3, 7, 8, 9)

select_sizes(x::AbstractArray, dim)

Additional size method to access the size at several dimensions in one call. Keep singleton dimensions.

Examples

julia> x = ones((2,4,6,8, 10));

julia> select_sizes(x, (2,3))
(1, 4, 6, 1, 1)

julia> select_sizes(x, 5)
(1, 1, 1, 1, 10)

julia> select_sizes(x, (5,))
(1, 1, 1, 1, 10)

select_sizes_squeeze(x::AbstractArray, dim)

Additional size method to access the size at several dimensions in one call. Remove singleton dimensions.

See also select_sizes which does not remove singleton dimensions.

Examples

julia> select_sizes_squeeze(randn((5,6,7)), (2,3))
(6, 7)

julia> select_sizes_squeeze(randn((5,6,7)), 2)
(6,)

reorient(vec, d::Val{dim}, total_dims=d)

Reorients a 1D vector vec along dimension dim. The total output dimension is total_dims.

Type stable version of reorient!

julia> reorient([1,2,3,4], Val(2))
1×4 Matrix{Int64}:
 1  2  3  4

julia> reorient([1,2,3,4], 2, Val(3))
1×4×1 Array{Int64, 3}:
[:, :, 1] =
 1  2  3  4

julia> x = reshape(1:9, 3, 3);

julia> reorient([1,2,3], 2, Val(ndims(x)))
1×3 Matrix{Int64}:
 1  2  3

idx_to_dim(idx_arr)

Converts an N-dimensional array of NTuple to an N+1 dimensional array by orienting the (inner) tuple along the (outer) trailing+1 dimension. Note that this function is not type stable!

See also: idx_to_view which reinterprets the array to an N+1 dimensional array by unrolling the (inner) tuple. Arguments:

• idx_arr. The array of NTuple to convert

Example:

julia> idx_to_dim([(x,y) for x in 1:3, y in 1:3])
3×3×2 Array{Int64, 3}:
[:, :, 1] =
 1  1  1
 2  2  2
 3  3  3

[:, :, 2] =
 1  2  3
 1  2  3
 1  2  3