Bill:

All the eigenvectors are for the matrix operating on the left.  We have
found v such that Av=lv, where A is the matrix and l is the eigenvalue. 
You need to premultiply v by A and then divide.

Best wishes,

John

Bill Harris wrote:
> -----BEGIN PGP SIGNED MESSAGE-----
> Hash: SHA1
>
> Just curious: given the principal eigenvector by the power method, would
> it be appropriate to calculate the associated eigenvalue by
> postmultiplying the eigenvector by the original matrix and then dividing
> (say) the first element of that new vector by the first element of the
> principal eigenvector?
>
> I did that with John's approach last night and got a principal
> eigenvalue around 5 for Coyle's 4x4 matrix, about 20-25% high, as I
> recall.  I don't know if that error is due to deviations between his
> eigenvector and the one John's algorithm calculated, to a numerical
> error in my approach, or due to my algorithm being fundamentally flawed.
>
> Thanks,
>
> Bill
> - --
> Bill Harris                      http://facilitatedsystems.com/weblog/
> Facilitated Systems                              Everett, WA 98208 USA
> http://facilitatedsystems.com/                  phone: +1 425 337-5541
> -----BEGIN PGP SIGNATURE-----
> Version: GnuPG v1.4.1 (MingW32)
> Comment: For more information, see http://www.gnupg.org
>
> iD8DBQFEB0EU3J3HaQTDvd8RAssjAJ9b4dHWx1IW4OvTSu7O1ou2HrqJgQCeOTR6
> EhYiKnZ0FsnMSD/lL+rtcKM=
> =PXD/
> -----END PGP SIGNATURE-----
>
> ----------------------------------------------------------------------
> For information about J forums see http://www.jsoftware.com/forums.htm
>


----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to