Zair Asrar
08-20-2002, 08:23 PM
Hi
I'm trying to solve an generalized eigenproblem in a form
of
[K]*[eigenvevtor]=[eigenvalue]*[M]*[eigenvector]
to find all the eigenvalue and its corresponding eigenvector.
Matrix K and M are symmetric but have some zero in it like this:
[K] =
2 0 0 3 0 0
0 1 2 0 1 2
0 2 4 0 3 2
3 0 0 5 0 0
0 1 3 0 2 3
0 2 2 0 3 4
matrix [M] also have the same position for the zero value but different nonzero value.
Matrix [K] looks symmetric to me and I thought generalized Jacobi method would work but it didn't. So i just tridiagonolized the [K] matrix using Householder method(to make sure generalized jacobi work) and only after that do the generalized Jacobi method. Is this the right things to do to solve this kind of problem? what should I do for this kind of matrix? Can Cholesky decomposition be used? What method should I follow?:confused:
I'm trying to solve an generalized eigenproblem in a form
of
[K]*[eigenvevtor]=[eigenvalue]*[M]*[eigenvector]
to find all the eigenvalue and its corresponding eigenvector.
Matrix K and M are symmetric but have some zero in it like this:
[K] =
2 0 0 3 0 0
0 1 2 0 1 2
0 2 4 0 3 2
3 0 0 5 0 0
0 1 3 0 2 3
0 2 2 0 3 4
matrix [M] also have the same position for the zero value but different nonzero value.
Matrix [K] looks symmetric to me and I thought generalized Jacobi method would work but it didn't. So i just tridiagonolized the [K] matrix using Householder method(to make sure generalized jacobi work) and only after that do the generalized Jacobi method. Is this the right things to do to solve this kind of problem? what should I do for this kind of matrix? Can Cholesky decomposition be used? What method should I follow?:confused: