MATLAB: Allocate a big matrix

allocatearraymatrixzeros

I want to allocate a (10^5,10^4) matrix and after I want to fill the matrix in a for-loop with values. My problem is, that zeros(10^5,10^4) gives the error "matrix exceeds maximum array size preference." and sparse(10^5,10^4) makes the for-loop to slow. Has someone of you a solution of this problem?

Thanks a lot

Best Answer

Use spalloc() specifying the number of expected non-zero values.

Sparse matrices are more efficiently filled working down columns than across rows.

10^5 by 10^4 is 10^9, just under 1 giga-entries. If you were to use the default double() datatype, that would require 8 gigabytes to store. Your preferences have been configured to not allow you to store matrices that large. If you have enough memory for this (and for the other arrays you are using) then you could change your preferences. And if you can get away with using something other than double precision, then create the array that way. For example,

A = zeros(10^5, 10^4, 'int16');

would take only 1/4 of the storage.

Related Solutions

MATLAB: Calculation of a sparse matrix

Sorry if I am saying what you already know, but the way you have defined A, it is decidedly non-sparse. You should not use the pre-allocation to zeros.

You should initialize the matrix A using the "sparse" command, which will define the indexing into A, without allocating memory for the elements. Then you can fill it in with your loops. Hopefully, there are few enough non-zero elements in the matrix to fit into memory.

MATLAB: Assign subscrips to vectors

You probably want something like

 yt = [yt1, yt2, yt3, yt4, yt5] ;

and then use

 yt(:,k)

in the loop, where k is the loop index. If yt's were not all the same dimension, you could use a cell array:

 yt = {y1, y2, y3, y4, y5} ;

and then use

 yt{k}

in the loop.

If you want to avoid copies/aggregation, just define directly the matrix the cell array instead of defining yt1 to yt5, e.g.

 yt = zeros(200,5) ;     % Prealloc.
 yt(:,1) = ... whatever computation you used to define yt1
 yt(:,2) = ... whatever computation you used to define yt2
 ..

 yt{1} = ... whatever computation you used to define yt1
 yt{2} = ... whatever computation you used to define yt2

Best Answer

Related Solutions

MATLAB: Calculation of a sparse matrix

MATLAB: Assign subscrips to vectors

Related Question