Re: [petsc-users] KSPSetDM problem

Michele Rosso Fri, 15 Nov 2013 15:35:37 -0800

Barry,

thank you.

I have another question: how does petsc compute the restriction(assuming the user does not supply his own restriction)?Does it use a simple transpose of the interpolation operator? Could youpoint me to the source code where this is done?


Thank you,
Michele

On 11/14/2013 06:51 PM, Barry Smith wrote:

https://bitbucket.org/petsc/petsc/wiki/Home#markdown-header-contributing-a-small-patch

On Nov 14, 2013, at 8:29 PM, Michele Rosso <[email protected]> wrote:

Barry,

thank you. I will have to go with option 1) since with 2) the number of 
iterations is almost the double (I tried with ex34).
I am testing the piecewise constant interpolation for periodic BCs we discussed 
about in a previous conversation: in case I am successful, how do I submit the 
fix?

Michele

On 11/14/2013 06:11 PM, Barry Smith wrote:

    Since you set the DM inactive   the PCSetUp_MG() has no code to compute the 
coarser grid operators. You need to either

1) provide a DMKSPSetComputeOperators () function that will compute the 
operator for all levels or

2) run with -pc_mg_galerkin to have the coarse grid operators computed by 
multiplying the fine grid operator by the restriction/interpolation.

    Barry

On Nov 14, 2013, at 7:49 PM, Michele Rosso
<[email protected]>
  wrote:

Hi,

I am  solving a Poisson equation (cell-centered discretization) with multigrid.
I am using a 2D version of ksp/ksp/examples/tutorials/ex34.c as a reference.
The following code performs the same calculations as ex34:

   call PetscInitialize(PETSC_NULL_CHARACTER,ierr)

     call DMDACreate2d(PETSC_COMM_WORLD, DMDA_BOUNDARY_NONE, 
DMDA_BOUNDARY_NONE,&
          & DMDA_STENCIL_STAR, 16, 16, PETSC_DECIDE, PETSC_DECIDE, &
          & 1, 1, PETSC_NULL_INTEGER, PETSC_NULL_INTEGER,da,ierr)
     call DMDASetInterpolationType(da, DMDA_Q0, ierr);
     call DMCreateGlobalVector(da,b,ierr)
     call create_rhs(da,b)
     call DMCreateMatrix(da,'aij',A,ierr)
     call create_matrix(da,A)
     call VecDuplicate(b,x,ierr)

! Create solver

     call KSPCreate(PETSC_COMM_WORLD,solver,ierr)
     call KSPSetDM(solver,da,ierr)
     call KSPSetDMActive(solver,PETSC_FALSE,ierr)
     call KSPSetOperators(solver,A,A,SAME_NONZERO_PATTERN,ierr)
     call KSPSetType(solver,'cg',ierr)
     call KSPSetNormType(solver,KSP_NORM_UNPRECONDITIONED,ierr)
     call KSPSetFromOptions(solver,ierr)
     call KSPSolve(solver,b,x,ierr)

rerr = error(da,x)call PetscFinalize(ierr)



Both ex34 and my code produce exactly the same results except when I run with

-ksp_view -ksp_monitor_true_residual -ksp_converged_reason -ksp_type cg 
-pc_type mg -pc_mg_levels 4

In this case my code results in the following error, despite KSPSetDMActive is 
used after KSPSetDM:

[0]PETSC ERROR: --------------------- Error Message 
------------------------------------
[0]PETSC ERROR: Object is in wrong state!
[0]PETSC ERROR: You called KSPSetDM() but did not use 
DMKSPSetComputeOperators() or KSPSetDMActive(dm,PETSC_FALSE);!
[0]PETSC ERROR: 
------------------------------------------------------------------------
[0]PETSC ERROR: Petsc Release Version 3.4.3, Oct, 15, 2013
[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: ./test on a linux-gnu-dbg named enterprise-A by mic Thu Nov 14 
17:38:23 2013
[0]PETSC ERROR: Libraries linked from /opt/petsc/petsc-3.4.3/linux-gnu-dbg/lib
[0]PETSC ERROR: Configure run at Tue Oct 29 22:43:45 2013
[0]PETSC ERROR: Configure options --known-mpi-shared="0 " --known-memcmp-ok  --with-debugging="1 " --with-fortran-datatypes 
--with-shared-libraries=0 --with-dynamic-loading=0 --with-mpi-compilers="1 " --download-blacs="1 " 
--download-superlu_dist="1 " --download-metis="1 " --download-parmetis="1 " --download-ml=1 PETSC_ARCH=linux-gnu-dbg
[0]PETSC ERROR: 
------------------------------------------------------------------------
[0]PETSC ERROR: KSPSetUp() line 230 in 
/opt/petsc/petsc-3.4.3/src/ksp/ksp/interface/itfunc.c
[0]PETSC ERROR: PCSetUp_MG() line 730 in 
/opt/petsc/petsc-3.4.3/src/ksp/pc/impls/mg/mg.c
[0]PETSC ERROR: PCSetUp() line 890 in 
/opt/petsc/petsc-3.4.3/src/ksp/pc/interface/precon.c
[0]PETSC ERROR: KSPSetUp() line 278 in 
/opt/petsc/petsc-3.4.3/src/ksp/ksp/interface/itfunc.c
[0]PETSC ERROR: KSPSolve() line 399 in 
/opt/petsc/petsc-3.4.3/src/ksp/ksp/interface/itfunc.c


Am I missing anything?
Thank you,

Michele

Re: [petsc-users] KSPSetDM problem

Reply via email to