[Trilinos-Users] Epetra_SerialDenseVector vs. Teuchos::SerialDenseVector

Bartlett, Roscoe A rabartl at sandia.gov
Sun Aug 1 18:25:25 MDT 2010


It is very hard to make operator+, operator-, operator* efficient. 

- Ross


> -----Original Message-----
> From: trilinos-users-bounces at software.sandia.gov [mailto:trilinos-
> users-bounces at software.sandia.gov] On Behalf Of Nico Schlömer
> Sent: Sunday, August 01, 2010 6:04 PM
> To: Thornquist, Heidi K
> Cc: Heroux, Michael A; trilinos-users at software.sandia.gov
> Subject: Re: [Trilinos-Users] Epetra_SerialDenseVector vs.
> Teuchos::SerialDenseVector
> 
> > Thus any further development of the SerialDense classes will be done
> in
> Teuchos,
> 
> Suggestion for the further development: Adding
> 
>   operator+
>   operator-
> 
> and scalar multiplication. This is now all possible with operator+= and
> friends, but I just had to evaluate an expression 'a la
> 
>    y = alpha * x1 + beta * x2 + gamma * x3 + delta * x4
> 
> where the x{1,2,3,4} are SerialDenseVectors, alpha,.. are scalars and
> found the only way to do so
> 
> SerialDenseVector y;
> x *= alpha;
> y += x;
> [...]
> 
> The code for the new operators would rather go in SerialDense~Matrix~,
> though, I guess.
> 
> Cheers,
> Nico
> 
> 
> 
> On Tue, 27 Jul 2010 11:20:35 -0600, "Thornquist, Heidi K"
> <hkthorn at sandia.gov> wrote:
> > Hi Nico,
> >
> > There has been some work to add functionality in Teuchos to support
> the
> use
> > of Teuchos::SerialDenseVector with Teuchos::SerialDenseMatrix.  This
> > functionality
> > was in support of Dakota's use of Teuchos::SerialDenseVector.  Thus
> any
> > further
> > development of the SerialDense classes will be done in Teuchos,
> probably
> > not in
> > Epetra.
> >
> > Heidi
> >
> >
> > On 7/27/10 11:10 AM, "Heroux, Michael A" <maherou at sandia.gov> wrote:
> >
> > Nico,
> >
> > My advice is that, if there is functionality overlap between Epetra
> and
> > Teuchos, you should use Teuchos.  Historically, we have migrated a
> lot
> of
> > Epetra functionality into Teuchos, generalizing it as part of the
> > migration.
> > This is what happened in this case.
> >
> > Mike
> >
> >
> > On 7/27/10 12:00 PM, "Christopher Grover Baker" <bakercg at ornl.gov>
> wrote:
> >
> >> Nico,
> >>
> >> Aside from the fact that Teuchos::SerialDenseVector is templated on
> the
> >> underlying scalar type and Epetra_SerialDenseVector is hard-coded to
> >> double,
> >> there should not be much difference between the two. The Epetra
> object
> >> will
> >> see little use outside of Epetra, while the Teuchos object will be
> used
> >> by
> >> templated packages. Although, to be honest, I can't think of any use
> of
> >> Teuchos::SerialDenseVector in any of the templated packages that I
> work
> >> on
> >> (including Tpetra).
> >>
> >> Chris
> >>
> >>
> >> On 7/27/10 12:10 PM, "Nico Schlömer" <nico.schloemer at ua.ac.be>
> wrote:
> >>
> >> Hi all,
> >>
> >> I just saw that for the Epetra_SerialDenseVector class I've always
> used
> >> there seems to be an equivalent in Teuchos::SerialDenseVector.
> >>
> >> Is the use of one of the two classes recommended over the other?
> >>
> >> Cheers,
> >> Nico
> >>
> >>
> >> _______________________________________________
> >> Trilinos-Users mailing list
> >> Trilinos-Users at software.sandia.gov
> >> http://software.sandia.gov/mailman/listinfo/trilinos-users
> >>
> >>
> >>
> >>
> >> _______________________________________________
> >> Trilinos-Users mailing list
> >> Trilinos-Users at software.sandia.gov
> >> http://software.sandia.gov/mailman/listinfo/trilinos-users
> >>
> >
> >
> > _______________________________________________
> > Trilinos-Users mailing list
> > Trilinos-Users at software.sandia.gov
> > http://software.sandia.gov/mailman/listinfo/trilinos-users
> 
> 
> _______________________________________________
> Trilinos-Users mailing list
> Trilinos-Users at software.sandia.gov
> http://software.sandia.gov/mailman/listinfo/trilinos-users



More information about the Trilinos-Users mailing list