How to convert a Julia 3D array to a Python 3D list

New to Julia
1

I have a Julia 3D array on the format Vector{Vector{Vector{Float64}}} and I am wondering how I can convert it into a 3D list in Python with the same indexing.

Thanks for any help!

2

Is there a reason that your “Julia 3D array” is not an Array{Float64, 3}? Are the internal vectors of different lengths?

By same indexing, do you mean you want 1-based indexing and F-order while in Python?

Why a list rather than a Numpy array?

3

Just to point out some terminology: this is not a 3d array, but a 1d array, or vector, that happens to contain other vectors. In Julia an N-d array is not nested, but ‘flat’, and with special indexing, eg. X[i, j, k] instead of X[i][j][k], which is how you index a nested vector/list.

I don’t think you can have 1-based indexing in Python, but the nested indexing is the similar.

4

This does not answer the indexing part of your question, but it covers the basics of invoking Python, convert to a Python list, and evaluating Python code.

julia> begin
           using Pkg
           Pkg.activate(; temp=true)
           Pkg.add("PythonCall")
       end

julia> using PythonCall

julia> ragged_3d = [
           [
               Float64[1,2,],
               Float64[4,5],
           ],
           [
               Float64[6],
               Float64[7,8],
               Float64[9,10,11],
           ],
       ]
2-element Vector{Vector{Vector{Float64}}}:
 [[1.0, 2.0], [4.0, 5.0]]
 [[6.0], [7.0, 8.0], [9.0, 10.0, 11.0]]

julia> py_ragged_3d = pyconvert(PyList, ragged_3d)
2-element PyList{Any}:
 [[1.0, 2.0], [4.0, 5.0]]
 [[6.0], [7.0, 8.0], [9.0, 10.0, 11.0]]

julia> pyprint(py_ragged_3d)
[[[1.0, 2.0], [4.0, 5.0]], [[6.0], [7.0, 8.0], [9.0, 10.0, 11.0]]]

julia> @pyeval (a=py_ragged_3d) => `print(a[0][0][0])`
1.0
Python: None
1 Like
5

Yes, the internal vectors are of different length. The conversion to python could be to numpy, it is not so important how it is ported over. With same indexing, I mean that the current indexing in julia ( i.e. X[i, j, k]) is maintained into python in the form of X[i][j][k].

I want to save the array as a file and then read it into a Jupyter Notebook.

6

If you’re saving to a file, use some common data format such as JSON. You can use JSON.jl to write it out and Python’s JSON stdlib to read it.