Google Groups no longer supports new Usenet posts or subscriptions. Historical content remains viewable.
Dismiss
Eigs with function handle
220 views
Skip to first unread message
Lorenzo
May 19, 2014, 6:02:08 PM5/19/14
to
Dear guys,
I'm having problems using eigs with function handle. Basically when I use the function handle I get completely wrong results. I am trying to find the eigenvector related to the smallest eigenvalue, in modulus, of a large matrix. I know that this smallest eigenvalue is actually zero.
I would like to use eigs with the function handle as I have a more efficient way of computing the matrix vector multiplication (and I don't need to store a huge matrix).
Any idea where the error can come from?
At some point I thought that part of the problem arises because the eigenvalue is ezactly zero and eigs performs a LU decomposition at the beginning.
Thank you,
Best regards
L
I'm having problems using eigs with function handle. Basically when I use the function handle I get completely wrong results. I am trying to find the eigenvector related to the smallest eigenvalue, in modulus, of a large matrix. I know that this smallest eigenvalue is actually zero.
I would like to use eigs with the function handle as I have a more efficient way of computing the matrix vector multiplication (and I don't need to store a huge matrix).
Any idea where the error can come from?
At some point I thought that part of the problem arises because the eigenvalue is ezactly zero and eigs performs a LU decomposition at the beginning.
Thank you,
Best regards
L
dpb
May 19, 2014, 9:52:32 PM5/19/14
to
On 5/19/2014 5:02 PM, Lorenzo wrote:
> Dear guys,
>
> I'm having problems using eigs with function handle. Basically when I
> use the function handle I get completely wrong results. I am trying to
> find the eigenvector related to the smallest eigenvalue, in modulus, of
> a large matrix. I know that this smallest eigenvalue is actually zero.
> I would like to use eigs with the function handle as I have a more
> efficient way of computing the matrix vector multiplication (and I don't
> need to store a huge matrix).
> Any idea where the error can come from?
...
> Dear guys,
>
> I'm having problems using eigs with function handle. Basically when I
> use the function handle I get completely wrong results. I am trying to
> find the eigenvector related to the smallest eigenvalue, in modulus, of
> a large matrix. I know that this smallest eigenvalue is actually zero.
> I would like to use eigs with the function handle as I have a more
> efficient way of computing the matrix vector multiplication (and I don't
> need to store a huge matrix).
> Any idea where the error can come from?
Not w/o anything to look at, no...
--
Steven Lord
May 20, 2014, 11:49:20 AM5/20/14
to
"Lorenzo " <lcam...@usc.edu> wrote in message
news:lldv10$1oh$1...@newscl01ah.mathworks.com...
> Dear guys,
>
> I'm having problems using eigs with function handle. Basically when I use
> the function handle I get completely wrong results. I am trying to find
> the eigenvector related to the smallest eigenvalue, in modulus, of a large
> matrix. I know that this smallest eigenvalue is actually zero.
> I would like to use eigs with the function handle as I have a more
> efficient way of computing the matrix vector multiplication (and I don't
> need to store a huge matrix).
> Any idea where the error can come from?
Not without seeing some code, no.
>
> I'm having problems using eigs with function handle. Basically when I use
> the function handle I get completely wrong results. I am trying to find
> the eigenvector related to the smallest eigenvalue, in modulus, of a large
> matrix. I know that this smallest eigenvalue is actually zero.
> I would like to use eigs with the function handle as I have a more
> efficient way of computing the matrix vector multiplication (and I don't
> need to store a huge matrix).
> Any idea where the error can come from?
My first suspicion is that your function doesn't return what it should. From
the documentation for EIGS:
http://www.mathworks.com/help/matlab/ref/eigs.html?searchHighlight=Afun
y = Afun(x) should return:
A*x, if sigma is not specified, or is a string other than 'sm'
A\x, if sigma is 0 or 'sm'
(A-sigma*I)\x, if sigma is a nonzero scalar (standard eigenvalue problem). I
is an identity matrix of the same size as A.
(A-sigma*B)\x, if sigma is a nonzero scalar (generalized eigenvalue problem)
If you're passing in 0 or 'sm' for sigma, your function should not return
A*x but the solution y to the system of equations A*y = x.
--
Steve Lord
sl...@mathworks.com
To contact Technical Support use the Contact Us link on
http://www.mathworks.com
Andrew Knyazev
May 15, 2015, 6:33:17 PM5/15/15
to
The function handle to the matrix itself is no good. You need a function handle either to the action of the matrix inverse (expensive) on a vector, or of the matrix square (slow convergence of eigs). There is no free lunch either way.
"Lorenzo" wrote in message <lldv10$1oh$1...@newscl01ah.mathworks.com>...
"Lorenzo" wrote in message <lldv10$1oh$1...@newscl01ah.mathworks.com>...
0 new messages