Does using cells cause dramatically slow down the code?

Does using cells instead of matrixes make big difference?

Answer 1

[Guillaume]

A cell array of scalar numbers will always be slower to process that a matrix with the same scalar numbers, yes. There's also no point in using a cell array for that since you can't perform mathematical operations directly on a cell array.

What is the context behind your question?

[IBM watson]

In my code there are many matrices coming out of loops. But all of them just for one iteration and I should store them for all iterations. (I havent add the outer loop which is for iterations) I think I have two options:

1) Rearranging the code to add every matrices to first one in every iteration:

iter1

a=[1 2 3

3 4 5]

iter2

a=[1 2 3

3 4 5

4 5 6

5 6 7]

iter3

a=[1 2 3

3 4 5

4 5 6

5 6 7

6 7 8

1 5 7 ]

2) Storing them in a cell.

cell{1,1}= a=[1 2 3

3 4 5]

cell{1,2}= a= [4 5 6

5 6 7]

I have two many matrices like 'a' and they more complex.

I hope I can clarify.

[Guillaume]

With the two schemes you describe, using cell arrays and concatenating the whole lot at the end would be faster than growing a matrix.

Probably, the best scheme, assuming it's always the same size of matrix returned and that you know the number of loops beforehand, is to create a matrix the right size to start with and fill it in the loop:

numiterations = 12;  %for example
outputsize = [2, 3]; %for example

result = zeros(outputsize .* [numiterations, 1]); %create a matrix the correct size
for iteration = 1:numiteration
    result((1:outputsize(1)) + (numiteration-1)*outputsize(1), :) = yourfunction();
end

[IBM watson]

Thank you!

[Walter Roberson]

Looks like you might not be preallocate the cell array