[Trilinos-Users] Epetra_FEVector question

Williams, Alan B william at sandia.gov
Wed Jun 18 19:57:14 MDT 2008


Originally _FEVector was a specialization of _Vector. That turned out to be a limitation for some applications, and so it was altered to inherit the more general _MultiVector. The documentation needs to be updated.
As Mike says, most functions accept a MultiVector so we don't expect this to be a problem. Let us know if it is a problem.
Alan


________________________________
From: trilinos-users-bounces at software.sandia.gov [mailto:trilinos-users-bounces at software.sandia.gov] On Behalf Of Heroux, Michael A
Sent: Monday, June 16, 2008 9:38 PM
To: Kirk, Benjamin (JSC-EG); trilinos-users at software.sandia.gov
Subject: Re: [Trilinos-Users] Epetra_FEVector question

Ben,

All Trilinos solvers accept Epetra_MultiVector objects with Epetra_Vector being a special case, so the fact that Epetra_FEVector derives from Epetra_MultiVector should not be an issue for you.  If you do run into a routine that takes only an Epetra_Vector, you can use the operator() method to get any vector from a  multivector.  For example, if you have a multivector V that contains only a single column, then V(0) is an Epetra_Vector with a view of that column.

Mike


On 6/16/08 7:54 PM, "Kirk, Benjamin (JSC-EG)" <benjamin.kirk-1 at nasa.gov> wrote:

Hi -

I'm migrating a finite element project from PETSc, and this code partitions on elements with shared degrees-of-freedom at nodes/faces on interprocessor boundaries.  Hence the ability to add local contributions to remote data structures is highly desired, which has me leaning to Epetra_FECrsMatrix and Epetra_FEVector...

I was browsing the Epetra documentation (http://trilinos.sandia.gov/packages/docs/r8.0/packages/epetra/doc/html/index.html#Epetra_formulations),
and it states that the Epetra_FEVector is a specialization of Epetra_Vector. However, the inheritance graphs show that Epetra_FEVector is derived from Epetra_MultiVector.

Will this cause problems?  Is there someplace down the road where I will need an Epetra_Vector?  Should the _FEVector be derived from the _Vector?

Thanks,

-Ben


-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://software.sandia.gov/mailman/private/trilinos-users/attachments/20080618/983dcfef/attachment.html 


More information about the Trilinos-Users mailing list