How to convert a cell to matrix?

Example:

A={[1 3 4];
   [2 5 6 4];
   [2 6 8 9 5 6];
   [3 6 5 4 1]}

I know it is not possible by using

B=cell2mat(A);

as it has different array size.

We may use NaN to follow concatenation rules.

How can I do that?

Once converted to matrix, How can I come back to same cell from the matrix?

[the cyclist]

For that input A, what exactly do you want the output to be? For example, do you want

B=[1 3 4 NaN NaN NaN;
   2 5 6  4  NaN NaN;
   2 6 8  9   5   6;
   3 6 5  4   1  NaN]

?

[SM]

Yes, First i want what you have written. Second, I want to come back from B to A.

Answer 1

[Akira Agata]

How about the following way?

% Convert to numeric array
maxLen = max(cellfun(@numel,A));
A = cellfun(@(x)[x, NaN(1,maxLen - numel(x))],A,'UniformOutput',false);
B = cell2mat(A);

% Back to cell array
A2 = mat2cell(B,ones(1,size(B,1)));
A2 = cellfun(@rmmissing,A2,'UniformOutput',false);

Answer 2

[the cyclist]

Here is one straightforward way:

% Original data
A={[1 3 4];
   [2 5 6 4];
   [2 6 8 9 5 6];
   [3 6 5 4 1]};

%%%%%% Convert to B

% Get number of rows of A, and the maximum vector length within A
M = numel(A);
N = max(cellfun(@numel,A));

% Preallocate B to the correct size
B = nan(M,N);

% Fill each row of B with the corresponding row of A
for im = 1:M
    Arow = A{im};
    B(im,1:numel(Arow)) = Arow;
end

%%%%%% Convert B back to A

% Preallocate A2 to the correct number of rows
M2 = size(B,1);
A2 = cell(M2,1);

% Fill A2 with the non-NaN elements of B
for im = 1:M2
    A2{im} = B(im,~isnan(B(im,:)))
end

% Show that they're identical
isequal(A,A2)