Eigenvectors seem wrong



On 03/20/2013 04:10 PM, Raymond Toy wrote:
>>>>>> "Ben" == Ben Blomberg <bblomberg at mail.bradley.edu> writes:
>      Ben> I recently posted about trouble finding eigenvalues with a
>      Ben> large Matrix. IT was suggested I try loading lapack and using
>      Ben> dgeev. This worked like a charm. However if I try and find
>      Ben> the eigenvectors with something like this
>
>      Ben>  [eigenvals, R_eigenvec,
>      Ben> L_eigenvec]:dgeev(aa1, True, True);
>
>
>      Ben> I get eigenvectors that differ from similar programs. The
>      Ben> first two seem correct but after that the values are much
>      Ben> different from that in say mathmatica.
>
>      Ben> Does anyone have any thought?
>
> A small example would be helpful.  But remember that eigenvectors are
> not unique.  If x is an eigenvector, so is a*x.  Perhaps Mathematica
> uses a different normalization of the eigenvectors?
>
and when there is an eigenvalue with multiplicity n>1, if x1,...,xn are 
n eigenvectors corresponding to that eigenvalue, any linear combination 
of them will also be an eigenvector.

Jaime