Make structures from a cell and store them in a new cell

I have a function yal which gives a square cell of size nxn, each cell containing a vector of two columns and multiple rows. I want to make structure for each cell, the structures will have:

 a.fc=[:,1];  % First column of structure a will store first column of vector
 a.sc=[:,2]; % Second column of structure a will store second column of vector
 a.addition=[:,1]+[:,2]; % Third column is the sum of both columns of vector
 (There can be more fields added to the structure a)

At the end these structures need to be stored in a new cell.

Can anyone please help me out..

[Stephen23]

First column of structure a will store first column of vector

What you show is one field of a structure storing the first column of the matrix. Your structure appears to be scalar.

"At the end these structures need to be stored in a new cell."

Why? Why not simply use a scalar or non-scalar structure ?

[Asim Ismail]

There will be n structures in case of cell yal, because it has nxn vectors and I need to arrange them. I think storing them in a cell will be more convenient.

[Stephen23]

" I think storing them in a cell will be more convenient."

Splitting up a perfectly good non-scalar structure into the cells of a cell array will be inefficient, very inconvenient, and most likely a total waste of time.

[Asim Ismail]

Ok lets ignore the final part (storing them in a cell) and just get structures for each vector of the cell.

Answer 1

[Guillaume]

As per Stephen's comment using a non-scalar structure as the final result would make a lot more sense (and be a lot easier).

Anyway, to do your conversion the easiest way would be to rearrange your cell array so that the columns of the matrices are actually split into cells of the cell array in the 3rd dimension. A call to cell2struct then do the conversion to structure:

C = arrayfun(@(x) randi([1 25], x, 2), magic(5), 'UniformOutput', false);  %generated demo data that fits the bill (nx2 matrix in each cell)

temp = cellfun(@(c) cat(3, {c(:, 1)}, {c(:, 2)}, {c(:, 1) + c(:, 2)}), C, 'UniformOutput', false);
temp = reshape([temp{:}], [size(C), 3]);   %the 3 is the number of field
resultstruct = cell2struct(temp, {'fc', 'sc', 'addition'}, 3);  %the 3 means use 3rd dimension for fields

If you do really want a cell array of scalar structure then you can do:

resultcell = num2cell(resultstruct);

As said, this would be harder to use.

[Asim Ismail]

Thanks

The structures will be non-scalar, thats right as Stephen said. But the final result should be a cell storing all these non-scalar structures.

[Asim Ismail]

Cant we use loop to perform all these tasks?

[Stephen23]

"But the final result should be a cell storing all these structures."

Why should the output be a cell array? Using a non-scalar structure is much simpler.

[Guillaume]

In your example:

a.fc = ...
a.sc = ...
a.addition = ...

is a scalar structure. The fields may be non-scalar but the structure is.

a(3,2).fc = ...
a(3,2).sc = ...
a(3,2).addition = ...

would be a non-scalar structure.

In any case, my answer provides you with either a non-scalar structure (of the same size as the cell array) of non-scalar fields, or a cell array of scalar structures of non-scalar fields.

You can use either. In my opinion, splitting the non-scalar structure into cells of a cell array adds an unnecessary level of indirection.

[Asim Ismail]

Ok then lets just get the results in non-scalar structures.

[Guillaume]

Ok then lets just get the results in non-scalar structures.

As said, this is what my answer does.

[Asim Ismail]

Can it be done in a loop because there are many fields I have to add in the structure?