Eigen values/vectors large symetric matrix

[Gaso]

I need to calculate eigen values and eigen vectors of a large symmetric real matrix, but all eigen vectors have to be exact not just 2/3 off all eigenvectors.
I tried with CPPLAPACK and dgeev routine:
http://cpplapack.sourceforge.net/
but only about 2/3 of eigen values and vectors are exact, for my project I need all vectory exactly calculated with the norm of unity, for the time expansion of my system in eigen space. I use MKL and ICC compiler, how can I calculate numerically all eigen vectors exactly in c++, which method would be exact.

 

[Born2bwire]

If you have MKL then you already have a C++ library wrapper for Blas and Lapack. The MKL manuals will explain how to use them, but as far as I know dgeev is a direct routine, you should get the exact results (or as close as you can get with finite precision). However, since you have a real symmetric matrix, there are more efficient methods, like dsyev.

 

[Gaso]

I have contacted Yuki Onishi from CPPLAPACK project.
dgeev had a bug and I reported it on 24.09.2009, the bug was solved, I have also changed my code and started to use dsyev, I didn't know that MKL have c++ library wrapper, thanks for the suggestion.

 

[Born2bwire]

Oh wow... whoops, I didn't notice how old this thread was when I responded first. Well at least you got it figured out then.