[petsc-users] Crash when trying to use FD Jacobian

[Barry Smith]

   If you can provide the non-zero structure of the Jacobian that is all that is needed. Since it is LibMesh I assume you are doing some finite element method. Thus just initially input "a wrong" Jacobian, so long as it has the correct nonzero structure then using the coloring code to compute the Jacobian will work fine.

   Barry

> On Jun 5, 2015, at 12:59 PM, Matthew Knepley <knepley at gmail.com> wrote:
> 
> On Fri, Jun 5, 2015 at 12:32 PM, Harshad Sahasrabudhe <hsahasra at purdue.edu> wrote:
> Hi,
> 
> I'm solving a non-linear equation using NEWTONLS. The SNES is called from a wrapper in the LibMesh library. I'm trying to use the default FD Jacobian by not setting any Mat or callback function for the Jacobian.
> 
> When doing this I get the following error. I'm not able to figure out why I get this error. Can I get some pointers to what I might be doing wrong?
> 
> Ah, this is going to be somewhat harder. Unless PETSc know the connectivity of your Jacobian, which means the influence between
> unknowns, it can only do one vector at a time:
> 
>   -snes_fd
> 
> which is really slow. It is trying to use get coloring for the Jacobian so that it can do many vectors
> at once. Do you have a simplified Jacobian matrix you could use for preconditioner construction?
> Then it could use that.
> 
>   Thanks,
> 
>      Matt
>  
> [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 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: ./nemo on a linux named conte-fe02.rcac.purdue.edu by hsahasra Fri Jun  5 13:25:27 2015
> [0]PETSC ERROR: Libraries linked from /home/hsahasra/NEMO5/libs/petsc/build-real/linux/lib
> [0]PETSC ERROR: Configure run at Fri Mar 20 15:18:25 2015
> [0]PETSC ERROR: Configure options --with-x=0 --download-hdf5=1 --with-scalar-type=real --with-single-library=0 --with-shared-libraries=0 --with-clanguage=C++ --with-fortran=1 --with-cc=mpiicc --with-fc=mpiifort --with-cxx=mpiicpc COPTFLAGS=-O3 CXXOPTFLAGS=-O3 FOPTFLAGS=-O3 --download-metis=1 --download-parmetis=1 --with-valgrind-dir=/apps/rhel6/valgrind/3.8.1/ --download-mumps=1 --with-fortran-kernels=0 --with-blas-lapack-dir=/apps/rhel6/intel/composer_xe_2013.3.163/mkl --download-superlu_dist=1 --with-blas-lapack-dir=/apps/rhel6/intel/composer_xe_2013.3.163/mkl --with-blacs-lib=/apps/rhel6/intel/composer_xe_2013.3.163/mkl/lib/intel64/libmkl_blacs_intelmpi_lp64.so --with-blacs-include=/apps/rhel6/intel/composer_xe_2013.3.163/mkl/include --with-scalapack-lib="-L/apps/rhel6/intel/composer_xe_2013.3.163/mkl/lib/intel64 -lmkl_scalapack_lp64 -lmkl_blacs_intelmpi_lp64" i --with-scalapack-include=/apps/rhel6/intel/composer_xe_2013.3.163/mkl/include --with-pic=1 --with-debugging=1
> [0]PETSC ERROR: ------------------------------------------------------------------------
> [0]PETSC ERROR: MatGetColoring() line 481 in /home/hsahasra/NEMO5/libs/petsc/build-real/src/mat/color/color.c
> [0]PETSC ERROR: SNESComputeJacobianDefaultColor() line 64 in /home/hsahasra/NEMO5/libs/petsc/build-real/src/snes/interface/snesj2.c
> [0]PETSC ERROR: SNESComputeJacobian() line 2152 in /home/hsahasra/NEMO5/libs/petsc/build-real/src/snes/interface/snes.c
> [0]PETSC ERROR: SNESSolve_NEWTONLS() line 218 in /home/hsahasra/NEMO5/libs/petsc/build-real/src/snes/impls/ls/ls.c
> [0]PETSC ERROR: SNESSolve() line 3636 in /home/hsahasra/NEMO5/libs/petsc/build-real/src/snes/interface/snes.c
> [0]PETSC ERROR: solve() line 538 in "unknowndirectory/"src/solvers/petsc_nonlinear_solver.C
> application called MPI_Abort(comm=0x84000000, 73) - process 0
> 
> Thanks,
> Harshad
> 
> 
> 
> -- 
> What most experimenters take for granted before they begin their experiments is infinitely more interesting than any results to which their experiments lead.
> -- Norbert Wiener