Re: [eigen] Student contribution

[ Thread Index | Date Index | More Archives ]

Hi Gael,

Thank you for your answer.
We have decided to improve the current SVD. We will have to present our work between the 17th and the 20th of June, and we would expect a feedback from the community before the presentation. We have never implemented any SVD, nonetheless we have already used it in Statistics (PCA). Also doing research on the algorithm is a part of our project.
We are familiar with devide-and-conquer method.

Best regards,

Gauthier Brun, Nicolas Carre, Jean Ceccato and Pierre Zoppitelli.

Le 2013-05-21 12:03, Gael Guennebaud a écrit :
Hi and welcome!

Indeed, a divide-and-conquer SVD would still be very useful. There is
already a bidiagonalization class so that you can focus on the divide
and conquer part, even though it might certainly be accelerated for
large matrices to exploit block operations. How much time to have for
this project? How familiar are you with the computation of the SVD in
general and with this divide-and-conquer method in particular?


On Tue, May 21, 2013 at 10:08 AM, BRUN Gauthier <brunga@xxxxxxxxxxxxxx> wrote:

We are four students in an engineering school of computer science and applied mathematics, and we would like to contribute to eigen for a project.
We would like to know if there is a specific modul you would to see
implemented, or improved?
Our teacher for this project already send you a mail, and it appear you are
interessted in a better SVD modul, using "Bidiagonalization +
divide-and-conquer bidiagonal SVD "

Best regards,

Gauthier Brun, Nicolas Carre, Jean Ceccatoet Pierre Zoppitelli.

Mail converted by MHonArc 2.6.19+