How to coup with matrix dimension?

I have equation e.g a* (b*b') to update Covariance Matrix.

a=(200 * 2) 200 rows and 2 cols

b=(200 * 2)

I need an output of dimension (2 * 2), How to deal with this problem. Is there any mathematical Concept i am Lacking?

Thanks in Advance.

Answer 1

[Wayne King]

a = randn(200,2);
c = a'*a;

You can just use cov()

a = randn(200,2);
y = cov(a);

The equivalence can be made exact by scaling the output of a'*a along with zero-meaning the columns.

So compare:

a = randn(200,2);
ac = bsxfun(@minus,a,sum(a,1)/200);
(ac'*ac)/199

with

cov(a)

[Tallha Akram]

Dear King, b matrix can only be of either dimension (2*2) or (200,200) , as i am taking transpose.

[Wayne King]

I'm not sure what you're saying. If you have a 200x2 matrix, the covariance matrix is 2x2