Re: generalized eigenvalue problem
- From: "Greg von Winckel" <see@xxxxxxxxxxxxx>
- Date: Sun, 20 May 2007 16:13:48 -0400
You mentioned earlier that the matrix A is singular. This does not
agree with your subsequent assertion that it is SPD. I'm not sure
what the inequality is that you put. This looks like something to do
with diagonal dominance.
Are your matrices full? If they are than you can try converting it to
a regular EVP. This is not a good idea if they are sparse, since the
resulting EVP will likely be full.
Infinite eigenvalues can happen:
A=[1,1;1,2]
B=[1,-1;-1,1]
eig(A,B) returns the eigenvalues 1/5 and Inf. I do know that it is
possible to get complex eigenvalues from a symmetric pencil though.
If you say something more about A,B and where they come from, I might
be able to add something more useful.
Edwin Ramayya wrote:
.
Dear Greg,
Thankyou so much for ur quick response. I did try what you
suggested!
but still I get some junk eigenvalues which are complex.
I checked for (SPD) Aii*Ajj > |Aij|, i ~= j, and checked for
Hermitian condition too.
A, B satisfy the condition!
anyother idea?
thanks,
Edwin.
- Follow-Ups:
- Re: generalized eigenvalue problem
- From: Edwin Ramayya
- Re: generalized eigenvalue problem
- References:
- generalized eigenvalue problem
- From: Edwin Ramayya
- Re: generalized eigenvalue problem
- From: Greg von Winckel
- Re: generalized eigenvalue problem
- From: Edwin Ramayya
- generalized eigenvalue problem
- Prev by Date: Re: MATLAB Programming Contest: May 9 - May 16, 20
- Next by Date: Re: problem with function
- Previous by thread: Re: generalized eigenvalue problem
- Next by thread: Re: generalized eigenvalue problem
- Index(es):
Relevant Pages
|
