There are lot of strategy to wotk around "out of memory" errors in Matlab, the most common way is to split matrices, if possible.

Increase system memory or use systems swap file may help but impact on the performance (in case of using swap file).

You can follow this link:

http://it.mathworks.com/help/matlab/matlab_prog/resolving-out-of-memory-errors.html

In addition to the solutions E. Fiorenza proposed, it might be possible to make the code more efficient in terms of memory use. For example, do you use any for loops when working with the matrix? If so, you should replace these loops with matrix operations (if possible). Maybe you could give some more information on the kind of computations you are performing?

The matrix operations I perform are for loops, and I want to replace each element in a column to its corresponding column number. I'm doing matrix operations, it outputs memory error still. 

If you just want to replace each element with its column number, you could do it without for loops - this is more efficient in terms of memory use. What about this solution? ->

tic

matrixDim = 70000; % matrix dimensions

yourMatrix = rand(matrixDim); % create random matrix

yourMatrix(yourMatrix

But the problem is that the solution proposed is still outside the bounds of memory in my matlab. I'm using Matlab 7.1 by the way.

Hi Jacob, then you will have to follow these instructions, as E. Fiorenza pointed out: http://de.mathworks.com/help/matlab/matlab_prog/resolving-out-of-memory-errors.html

Well, first of all, use an sparse matrix (not a matrix with zeros), get the columns with not null elements using the find command and create a new sparse matrix from these values:

a = sprand(70000,70000,0.1);

[row,col] = find(a > 0);

b = sparse(row,col,col);

Instead 0.1 use any other factor (actually, 0.1 is not very sparse)