dolfin team mailing list archive

[Anders Logg <logg@xxxxxxxxx>]

On Wed, Nov 28, 2007 at 11:22:36PM +0100, Mads Hoel wrote:
> Sorry, this was meant to be sent to the Dolfin mailing list.
> 
> ------- Forwarded message -------
> From: "Mads Hoel" <mfhoel@xxxxxxxxxx>
> To: "Matthew Knepley" <knepley@xxxxxxxxx>
> Cc:
> Subject: Re: [DOLFIN-dev] Solving eigenvalue problems with dolfin
> Date: Wed, 28 Nov 2007 23:21:02 +0100
> 
> On Tue, 27 Nov 2007 16:35:52 +0100, Matthew Knepley <knepley@xxxxxxxxx>
> wrote:
> 
> > On Nov 27, 2007 9:33 AM, Anders Logg <logg@xxxxxxxxx> wrote:
> >> On Tue, Nov 27, 2007 at 04:30:43PM +0100, Thomas Witkowski wrote:
> >> > Hi,
> >> >
> >> > I want to use dolfin for solving eigenvalue problems. I've installed
> >> > dolfin 0.7.1 with petsc and slepc. The matrix, for which the  
> >> eigenvalues
> >> > should be computed, is defined in the following way:
> >> >
> >> > PETScMatrix C(TS, TS);
> >> > C.set(.....);
> >> > C.set(.....);
> >>
> >> Try calling
> >>
> >>   C.apply();
> >
> > Why change the name from assemble()? You even used it in your  
> > description,
> > and apply() is easily confused with operator application.
> 
> Perhaps they wanted to avoid confusion with assembling contributions from
> the element matrices, atleast I found that confusing when I started out
> with PETSc. The PETSc manual often refer to assembling the matrices,
> refering to MatAssemblyBegin/MatAssemblyEnd(), that has to be done once
> the precaching of values are done. But I agree, apply might be equally
> confusing. What about the name applyChanges? Or perhaps processValues?
> 
> Then again, not using the name assemble might be confusing if one knows
> that it is a PETSc matrix one is working with.

It's not necessarily a PETSc matrix. It can also be a uBLAS matrix or an
Epetra matrix or a CRSMatrix and we want to use the same interface for
all.

-- 
Anders


> >
> >   Matt
> >
> >> after the set functions.
> >>
> >> PETSc needs to "assemble" the changes you made when calling set()
> >> before you can do anything with the matrix.
> >>
> >>
> >>
> >>
> >> > Now, when I'm trying to solve the problem with:
> >> >
> >> >  SLEPcEigenvalueSolver esolver;
> >> >  esolver.solve(C);
> >> >
> >> > I get the following error message:
> >> >
> >> >
> >> > [0]PETSC ERROR: --------------------- Error Message
> >> > ------------------------------------
> >> > [0]PETSC ERROR: Object is in wrong state!
> >> > [0]PETSC ERROR: Not for unassembled matrix!
> >> > [0]PETSC ERROR:
> >> >  
> >> ------------------------------------------------------------------------
> >> > [0]PETSC ERROR: Petsc Release Version 2.3.3, Patch 8, Fri Nov 16
> >> > 17:03:40 CST 2007 HG revision:  
> >> 414581156e67e55c761739b0deb119f7590d0f4b
> >> > [0]PETSC ERROR: See docs/changes/index.html for recent updates.
> >> > [0]PETSC ERROR: See docs/faq.html for hints about trouble shooting.
> >> > [0]PETSC ERROR: See docs/index.html for manual pages.
> >> > [0]PETSC ERROR:
> >> >  
> >> ------------------------------------------------------------------------
> >> > [0]PETSC ERROR: ./mass on a linux-gnu named ... by witkowsk Tue Nov 27
> >> > 16:17:51 2007
> >> > [0]PETSC ERROR: Libraries linked from
> >> > /u/witkowski/local/petsc-2.3.3-p8/lib/linux-gnu-cxx
> >> > [0]PETSC ERROR: Configure run at Tue Nov 27 11:24:27 2007
> >> > [0]PETSC ERROR: Configure options --with-mpi-dir=/u/witkowski/local
> >> > --with-clanguage=cxx --download-hypre=yes --with-hypre=1
> >> > --download-umfpack=yes --with-umfpack=1 --with-shared=0
> >> > [0]PETSC ERROR:
> >> >  
> >> ------------------------------------------------------------------------
> >> > [0]PETSC ERROR: MatMult() line 1616 in src/mat/interface/matrix.c
> >> > [0]PETSC ERROR: STApply_Shift() line 30 in src/st/impls/shift/shift.c
> >> > [0]PETSC ERROR: STApply() line 49 in src/st/interface/stsolve.c
> >> > [0]PETSC ERROR: EPSBasicArnoldi() line 86 in  
> >> src/eps/impls/arnoldi/arnoldi.c
> >> > [0]PETSC ERROR: EPSSolve_KRYLOVSCHUR() line 91 in
> >> > src/eps/impls/krylovschur/krylovschur.c
> >> > [0]PETSC ERROR: EPSSolve() line 71 in src/eps/interface/solve.c
> >> > Eigenvalue solver (krylovschur) converged in 1 iterations.
> >> >
> >> >
> >> > What does it meen? And what have I made wrong?
> >> >
> >> > Regards,
> >> >
> >> > Thomas
> >> > _______________________________________________
> >> > DOLFIN-dev mailing list
> >> > DOLFIN-dev@xxxxxxxxxx
> >> > http://www.fenics.org/mailman/listinfo/dolfin-dev
> >> _______________________________________________
> >> DOLFIN-dev mailing list
> >> DOLFIN-dev@xxxxxxxxxx
> >> http://www.fenics.org/mailman/listinfo/dolfin-dev
> >>
> >
> >
> >
> 
> 
>