[Trilinos-Users] LOBPCG direct diagonalization

[임재창]

Dear Trilinos developers and users,I'm using LOBPCG method from Anasazi package to diagonalize sparse matrices with Epetra_CrsMatrix.I found that direct solver part becomes extremely slow as the number of block size increases.I am using max iteration =20 and blocksize=1300Direct Diagonalization is done using Teuchos_LAPACK and it supports only openmp.So, my question is that can I use Amesos_Scalapack for direct diagonalization during LOBPCG?or If some use different library for direct diagonalization during LOBPCG,it will be really helpful for me to explain how to modify "AnasaziSolverUtils.hpp" or show me some examplesThank you for reading 
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://trilinos.org/pipermail/trilinos-users/attachments/20160514/37a59fd9/attachment.html>