Eigenvalues and transformations
- From: Sensei <senseiwa@xxxxxxx>
- Date: Thu, 20 Apr 2006 19:47:42 +0200
Back again on matrices! I'll be probably be damned for my posts :)
I read a book on linear algebra, and I was thinking back on my post... so, let me formulate my question.
We have a matrix A over a field K, NxN, with KerA = { 0_K }, so I can say that A has full rank. We have a matrix P over K, MxN, with kerP = { 0_K }, so again P has full rank. Now, we can split A with something similar to a congruence with the difference that P is not square:
A = P^T X P
The matrix X as you pointed out can be obtained by the left inverse of P: QP=I, and X = Q^T A Q. X is a matrix over K, MxM. Let's have M > N.
Is there any connection between the eigenvalues/eigenvectors of X and A?
I read about the singular value decomposition, and since P is real (I'm not interested in somehting too advanced, it's too early!) and A is real too, it can be decomposed into three matrices, the same can be done on Q (QP=I) and I can say this, if I'm not mistaken:
Q = U S V^T
V is such that V^T V = V V^T = I, the same is for U (if I understand, since they are real matrices). So, I can write:
X = Q^T A Q = (U S V^T)^T A (U S V^T) = V S (U^T A U) S V^T
Between the parenthesis, U^T A U, is equal to U^-1 A U. Should this preserve the egienvalues of A? And what about the other matrices? Can I preserve the eigenvalues by restricting P (or X)?
I have an book on this, but it is only on square matrices... I cannot find anywhere something about non-square ones...
Thanks for any hint! :)
--
Sensei <senseiwa@xxxxxxx>
The optimist thinks this is the best of all possible worlds.
The pessimist fears it is true. [J. Robert Oppenheimer]
.
- Follow-Ups:
- Re: Eigenvalues and transformations
- From: Sensei
- Re: Eigenvalues and transformations
- Prev by Date: Re: Should be easy
- Next by Date: Chess boards & connections.
- Previous by thread: the mathematics behind Mp3 files, and more
- Next by thread: Re: Eigenvalues and transformations
- Index(es):
Relevant Pages
|
Loading