extracting only first few eigenvectors

[Mike Miller]

Suppose I have a symmetric matrix of maybe 10000 x 10000 and I want to 
extract only the 10 eigenvectors associated with the 10 largest 
eigenvalues.  Is there a way to do that without doing the usual full 
decomposition like so...

[V, LAMBDA] = eig (A)

...and then pulling out the needed pieces?

This will be a variance-covariance matrix so it should have no negative 
eigenvalues, if that matters.

Mike