0
$\begingroup$

Sorry for these questions, I'm trying to understand and cannot seem to figure it out anywhere.

The material I've read so far on Eigenvalues and eigenvectors have showed for 2x2, 3x3.

From what I understand from the Eigenvalues, is that it shows how the data has been transformed (how much by) whereas the Eigenvectors show at what direction.

For a 2x2 matrix, I used the following:

$$\frac{a+b\pm\sqrt{(a+b)^2-4(ab-c^2)}}2.$$

This works fine, however, the problem that I'm facing is that I have a matrix of size: 451x128 which when I compute the Covariance matrix, I get the dimensions of 128x128.

My question is: How do I take the Eigenvalues/eigenvectors of a matrix (128x128) of this magnitude? I was lead to believe that, since, the Eigenvalues and eigenvectors show how to matrix has changed, there should be just a few values?

Hope someone can help

$\endgroup$
0
$\begingroup$

From the definition of an eigenvalue we have:

$$Ax=\lambda x$$ $$Ax-\lambda x =0$$ $$Ax-\lambda I x=0$$

where $I$ is the identity matrix. Then,

$$(A-\lambda I)x=0$$

Since we require that the eigenvector, $x$, be non-zero, the matrix $A-\lambda I$ must be singular. So we can compute the eigenvalues by solving the equation

$$det(A-\lambda I)=0$$

So with this being said, it would be very difficult for you to compute the eigenvalues of a very large matrix by hand. Matlab or Mathematica will have helpful routines for doing so.

$\endgroup$
  • $\begingroup$ Hey thanks for the answer. Im not attempting to calculate the eigenvalues for the 128x128 by hand, but im trying to write an algorithm.. Ive managed to to calculate it for a 2x2, 3x3 using the equation.. But what would i need to do to solve this for a 128x128? I dont understand why there would be 128x128 eogenvalues? Sorry $\endgroup$ – Phorce Jun 5 '14 at 15:19
  • $\begingroup$ A quick google search yielded this: www-users.cs.umn.edu/~saad/eig_book_2ndEd.pdf. This should yield the necessary information. $\endgroup$ – user155385 Jun 5 '14 at 15:22

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.