[Trilinos-Users] [Help] FEM code, Suggestions For new user

[Dharmendar Reddy]

Thanks Kevin.
                         I will follow the procedure you mentioned. With
respect DOLFIN vs Sundance, computation time is not a concern for me since i
will be doing small test problems for now. Given that i have never written a
C++ code but i know the language enough to read and understand a code, can
you please advice me on which of them has easy learning curve.

Also, PETSc has a nice ( atleast i find it very useful) user manual (
http://www.mcs.anl.gov/petsc/petsc-as/snapshots/petsc-current/docs/manual.pdf).
Is there something similar for Epetra  or Sundance ?


I think i am aware of Morfeus packge. I will try to know more about it.


Thanks
Reddy
On Fri, Feb 25, 2011 at 9:37 AM, Kevin Long <kevin.long at ttu.edu> wrote:

>
> Hi Reddy,
>
> If you enable Sundance, all of its other Trilinos dependencies will be
> enabled
> automatically. So basically you would put
>
> -D Trilinos_ENABLE_Sundance:BOOL=ON \
>
> in your cmake command line and all of its dependencies (Epetra, ML, etc)
> will
> be detected and enabled. To use any interesting meshes you'll want exodus,
> which means getting netcdf as a third-party library and then setting
>
> -D Sundance_ENABLE_Trios:BOOL=ON \
>
> If you're using the 10.6 release, I think you also need to set
>
> -D Trilinos_ENABLE_EXPLICIT_INSTANTIATION:BOOL=ON \
>
> to get Sundance to build correctly.
>
> As for Sundance and DOLFIN, they are superficially similar, in that both
> provide a high-level interface for setting up FEM simulations. Sundance is
> pretty tightly integrated with the rest of Trilinos, which is a good thing
> if
> you only want Trilinos solvers, but not good if you want to use Petsc or
> something. DOLFIN isn't tied to any one solver library. Performance of the
> two is similar; some timing comparisons are in a recent SIAM J. Sci. Comp
> article by Long, Kirby, and van Bloemen Waanders.  In those timings,
> Sundance
> is a little faster (a factor of two or so) at building matrices, but then
> in
> any real problem that difference will be dominated by the solve time which
> will be independent of whether you use Sundance or DOLFIN.
>
> With your Fortran background, you might also talk to Damian Rouson about
> his
> Morfeus package (not in Trilinos).
>
> Let me know if you have any questions about Sundance.
>
> Regards,
>
> Kevin
>
>
> --
> ------------------------------------------------------
> Kevin Long
> Associate Professor
> Department of Mathematics and Statistics
> Texas Tech University
> Lubbock, TX
>
> "The end of fear is the beginning of wisdom"
>    -- Bertrand Russell
> ------------------------------------------------------
>



-- 
-----------------------------------------------------
Dharmendar Reddy Palle
Graduate Student
Microelectronics Research center,
University of Texas at Austin,
10100 Burnet Road, Bldg. 160
MER 2.608F, TX 78758-4445
e-mail: dharmareddy84 at gmail.com
Phone: +1-512-350-9082
United States of America.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: https://software.sandia.gov/pipermail/trilinos-users/attachments/20110225/400ccc33/attachment.html