Subject: Re: [ublas] Boundary condition application on matrices
From: Karl Rupp (rupp_at_[hidden])
Date: 2011-04-29 07:54:52
the best approach typically is to not write boundary conditions to the
matrix at all. A little bit of extra logic during the assembly process
can write the values directly to the right hand side vector. This is
typically much faster than manipulating the matrix in a post-assembly step.
Sorry if that's not the let-ublas-do-the-job-answer you may have been
looking for. ;-)
On 04/29/2011 12:40 PM, Umut Tabak wrote:
> Dear all,
> What would be the most efficient way to apply boundary conditions on the
> FE assembled system matrices?
> A couple of days ago, I asked on a post to do that with matrix_indirect,
> using matrix_indirect seems to be fast to extract the free dofs from the
> full system matrices however assigning them to a compressed_matrix for
> further matrix-vector products seems to be very slow, are there any
> workarounds for my problem. What I try to do now, I try get the triplets
> of the compressed_matrix and apply the conditions by using this triplet
> information and write the unconstrained matrix entries into the matrix
> that I would like to return from the function.
> Any other suggestions are very welcome.
> ublas mailing list
> Sent to: rupp_at_[hidden]