MATLAB is fastest and easiest when you stick with using numeric arrays and learn how to use the dimensions of the arrays, rather than sticking data in complicated data classes like cell arrays and structures. (most importantly this includes learning how to write vectorized code). If you use the third dimension of an array then your problem is easy using mean:
>> X(:,:,3) = [1,2;3,4];
>> X(:,:,2) = [5,6;7,8];
>> X(:,:,1) = [9,0;1,2];
>> mean(X,3)
ans =
5.0000 2.6667
3.6667 4.6667
The best solution is to create your n matrices in this single numeric array. The second best solution is to create your arrays inside a cell array, which can then be concatenated together like this:
>> Y{3} = [1,2;3,4];
>> Y{2} = [5,6;7,8];
>> Y{1} = [9,0;1,2];
>> X = cat(3,Y{:})
>> mean(X,3)
You will find that the X is the same as in the first example.
The one method you should not use is to dynamically create the n matrices as individual variables (e.g. sequentially numbered) and then attempt to join them together: creating variable names dynamically is buggy, slow and a very poor programming practice.
