Solve generalized eigenvalue problem or reduce it to single

General CULA Dense (LAPACK & BLAS) support and troubleshooting. Use this forum if you are having a general problem or have encountered a bug.

Solve generalized eigenvalue problem or reduce it to single

Postby horacioemilio » Thu Mar 22, 2012 4:14 am

We want to accelerate on GPUs the generalized symmetric eigenvalue problem, i.e., Ax=lambda*Bx, where A and B are real matrices. We wonder whether CULA has support for this problem, or if using other CULA functions, we could reduce this problem to the simple eigenvalue problem, A'x=lambda*x
horacioemilio
 
Posts: 2
Joined: Thu Mar 22, 2012 4:11 am

Re: Solve generalized eigenvalue problem or reduce it to sin

Postby kyle » Thu Mar 22, 2012 10:37 am

We have plans to add generalized eigen-solvers in the next CULA Dense release.

Are you interested in general or symmetric systems?
kyle
Administrator
 
Posts: 301
Joined: Fri Jun 12, 2009 7:47 pm

Re: Solve generalized eigenvalue problem or reduce it to sin

Postby horacioemilio » Thu Mar 22, 2012 12:15 pm

only in symmetric systems

from the other side, would it be possible, using CULA tools, to reduce the generalized problem to the single one? (Ax=lambda*x)
horacioemilio
 
Posts: 2
Joined: Thu Mar 22, 2012 4:11 am

Re: Solve generalized eigenvalue problem or reduce it to sin

Postby kyle » Thu Mar 22, 2012 1:07 pm

horacioemilio wrote:... would it be possible, using CULA tools, to reduce the generalized problem to the single one? (Ax=lambda*x)

No. There are core components that need to be implemented first.
kyle
Administrator
 
Posts: 301
Joined: Fri Jun 12, 2009 7:47 pm


Return to CULA Dense Support

Who is online

Users browsing this forum: No registered users and 2 guests