[Trilinos-Users] PETSc interface

Heroux, Mike MHeroux at CSBSJU.EDU
Wed Apr 30 15:57:05 MDT 2014


Lucie, 

We did not update this wrapper interface to include the 64-bit interface
added to Epetra_RowMatrix some time ago.  This is easily fixed.  We can
send you a patch, and get the repository updated.  Sorry for the
inconvenience.  This interface isn't commonly used, but recent events
increase its importance.

Jed,

The situation here is wrapping a PETSc AIJ matrix to implement the
Epetra_RowMatrix interface.  Doing this permits the PETSc matrix to be
used with any Trilinos solver or preconditioner.

Mike

On 4/30/14 4:46 PM, "Jed Brown" <jed at 59A2.org> wrote:

>[resent from subscribed address]
>
>Lucie Freret <lfreret at arrow.utias.utoronto.ca> writes:
>
>> Hello All,
>>
>> I try to apply a Trilinos preconditioner (ML) to a PETSc linear system.
>> As a first test, I consider the example given in EpetraExt package:
>> 
>>http://trilinos.sandia.gov/packages/docs/dev/packages/epetraext/doc/html/
>>epetraext_petsc_cpp.html
>>
>> Unfortunately, I can't compile it and get some errors (see below).
>> I have tried with both petsc-3.4.4 and petsc-3.1.3 (and trilinos 11.4)
>>
>> It is said here 
>> 
>>(http://trilinos.sandia.gov/packages/docs/dev/packages/epetraext/doc/html
>>/epetraext_petsc_interface.html) that the Epetra_PETScAIJMatrix class
>>was first released with Trilinos 9.0 and
>> PETSc-2.3.3
>>
>> Does anyone already tried to use Petsc interface with older versions
>> (of Trilinos and/or PETSc) ?
>
>I'm not sure I understand your situation.  You have an existing PETSc
>Mat that you want to solve using ML?  I suggest configuring PETSc with
>--download-ml and running with -pc_type ml.  This works with the usual
>KSPSolve; you don't need to write any special code.



More information about the Trilinos-Users mailing list