On 19/06/16 18:53, John Peterson wrote: > >> On Jun 19, 2016, at 3:50 AM, Torquil Macdonald Sørensen <torq...@gmail.com> >> wrote: >> >> Hi! >> >> I have solved a PDE generalized eigenproblem (A*v = lambda*B*v) in >> libmesh, but I would like to do some linear algebra computations that >> involves the A and B matrices, some of the eigenvectors v, as well as >> A2/B2-matrices that arise from a different type of discretization of the >> same PDE. >> >> I guess I will probably have to create a second EigenSystem > Or derive from EigenSystem and implement something more specialized to your > specific app? > >> in order to >> create A2 and B2, but how can I perform linear algebra computations >> involving all of these matrices and vectors? > Look at the SparseMatrix and NumericVector interfaces. Everything you need > should be available, if not we can certainly add something. > > >> Are there any known libmesh examples where computations such as v^T*A2*v >> are done, where v is an eigenvector from the A/B generalized >> eigenproblem, and A2 comes from a different eigenproblem? > Sparse matrix-vector and vector-vector products are available in the > interfaces I mentioned. > > >> Is it possible to get access to A/B/v/A2/B2 as PETSc-objects that can be >> manipulated freely using PETSc functions? > Yes, but I'd just stick to the generic SparseMatrix interface if possible. >
Thanks! Your suggestions pointed me in the right direction. Best regards, Torquil Macdonald Sørensen ------------------------------------------------------------------------------ Attend Shape: An AT&T Tech Expo July 15-16. Meet us at AT&T Park in San Francisco, CA to explore cutting-edge tech and listen to tech luminaries present their vision of the future. This family event has something for everyone, including kids. Get more information and register today. http://sdm.link/attshape _______________________________________________ Libmesh-users mailing list Libmesh-users@lists.sourceforge.net https://lists.sourceforge.net/lists/listinfo/libmesh-users
