Converting irregular nested python list to matlab array

I am trying to convert an irregular python list ( by irregular I mean size of elements within each sub-list may not be same) to matlab array. The structure of the list is as follows :-

%Python list

nested_list = [
        [
            [[1,1], [2,2]],
            [[3,3], [4,4], [5,5]]]
        ],
        [
          [[1,1], [2,2]],
          [[3,3], [4,4], [5,5]]],
          [[6,6]]
        ]
    ]
    

As you can see the array is 4D and size of each sub list is not constant. I don't wish to use scipy.io.savemat function. I have tried matlab's cell function but coudn't find a way to call it recursively. Is there a way to convert this list to a matlab array/cell array?

Thank You.

Answer 1

[Bjorn Gustavsson]

Wouldn't that become a 3-D cell-array of 1-by-2 arrays? Maybe something like this:

nested_list{1,1,1} = [1,1];
nested_list{1,1,2} = [2,2];
nested_list{1,2,1} = [3,3];
nested_list{1,2,2} = [4,4];
nested_list{1,2,3} = [5,5];
nested_list{2,1,1} = [1,1];
nested_list{2,1,2} = [2,2];
nested_list{2,2,1} = [3,3];
nested_list{2,2,2} = [4,4];
nested_list{2,2,3} = [5,5];
nested_list{3,2,3} = [6,6];

This would be very clean to index into after checking the nonemptiness of the cell-elements.

Or do you want a cell-array of cell-arrays of cell-arrays?

%Python list

row11 = {[1,1],[2,2]};
row12 = {[3,3], [4,4], [5,5]};
layer1 = {row11,row12};
row21 = {[1,1],[2,2]};
row22 = {[3,3], [4,4], [5,5]};
row22 = {[6,6]};
layer2 = {row21,row22};

nested_list = {layer1,layer2};

This becomes a mess to index into:

nested_list{1}{2}{3} % etc...

Then you'll have to check the type of each cell-element along the way, I'd try to avoid that hassle and try to stick with the first data-structure - since cell-elements can be whatever types you need I don't easily see the benefits of having cell-arrays nested too deep.

HTH

[Soumil Parnami]

Thank You for the reply! I think the first data structure will work for my case. Is there a simple way to convert the above list into the mentioned datastructure. The code I am using to do this is taking too long to do this compared to scipy. I am using nested for loops. Is there a faster way to do this?

Thank You

[Bjorn Gustavsson]

My python-skills are "very limited". The first naive approach would be to build a one-D list with elements of this type (possibly in python?):

[id1, id2, id2, cell_content_vector]

While at the same time keeping track of the largest indices in each dimension. After generating that it should be rather straightforward to loop through that array:

testCell = cell(max_id1,max_id2,max_id3);
for i0 = 1:numel(intermediate_cellarray,1)
  i1 = intermediate_cellarray(i0,1);
  i2 = intermediate_cellarray(i0,2);
  i3 = intermediate_cellarray(i0,3);
  vec = intermediate_cellarray(i0,4:end);
  final_3D_cell{i1,i3,i3} = vec;
end