I actually am going to sum a bunch of rank 1 matrices before I find the eigen vectors, so there will be a full rank matrix before the eigenvecs/values are found.
Thanks for the suggestions though! Dave On Thu, Mar 6, 2008 at 7:00 PM, Warren Weckesser < [EMAIL PROTECTED]> wrote: > Don't use a gsl_eigen_* function to find the eigenvalues and eigenvectors. > > The matrix a*b^T has rank 1. The only nonzero eigenvalue is b^T*a (i.e. > the dot product of b and a), and the corresponding eigenvector is a. > > The eigenspace of the zero eigenvalue is the set of all vectors normal to > b, i.e. x such that b^T*x = 0, so just find a basis for this space to get > the eigenvectors of the zero eigenvalue. > > --Warren > > ________________________________________ > From: [EMAIL PROTECTED] > [EMAIL PROTECTED] On Behalf Of David Doria [EMAIL PROTECTED] > Sent: Thursday, March 06, 2008 5:08 PM > To: [email protected] > Subject: [Help-gsl] eigenvectors of non symmetric matrix? > > I am taking an outer product: > > a b^T > where a and b are column vectors. Then I want the eigen values and > vectors > of the resulting matrix (called mat3). > > I tried to use: > gsl_eigen_symmv_workspace * EigenWorkspace = gsl_eigen_symmv_alloc (2); > gsl_eigen_symmv (mat3, EigenValues, EigenVectors, EigenWorkspace); > > but it gave the wrong results. I guess this is because it was expecting a > symmetric matrix? Is the only other choice to use: > gsl_eigen_hermv_workspace * EigenWorkspace = gsl_eigen_hermv_alloc (2); > gsl_eigen_hermv (mat3, EigenValues, EigenVectors, EigenWorkspace); > > but for that, I'd have to first make mat3 a complex matrix (or so says the > error haha)? > > Please let me know. > > -- > Thanks, > > David > _______________________________________________ > Help-gsl mailing list > [email protected] > http://lists.gnu.org/mailman/listinfo/help-gsl > -- Thanks, David _______________________________________________ Help-gsl mailing list [email protected] http://lists.gnu.org/mailman/listinfo/help-gsl
