[petsc-users] Errors with KSP solver

Hien HN- hoangnhanhien at gmail.com
Mon Nov 2 20:38:13 CST 2020 

Hi Matt,

1) As I understood, the VecScatterCreateToAll function will create the
vector sol_to_all, and it does not need to create this vector with, for
example, VecCreate or VecCreateMPI function. Is this correct?
                  call VecScatterCreateToAll(petsc_
solution,ctx,sol_to_all,petsc_err)

I have used the VecView to print the vector sol_to_all to a file. It seems
no problem with sol_to_all.

I installed PETSc with Intel MKL and MS-MPI libraries. So should I enable
the option of Intel MKL that performs vector operations in my Fortran
project?


2) This system is not the Laplacian.

Thank you very much,
Hoang

On Mon, 2 Nov 2020 at 20:15, Matthew Knepley <knepley at gmail.com> wrote:

> On Mon, Nov 2, 2020 at 12:31 AM Hien HN- <hoangnhanhien at gmail.com> wrote:
>
>> Dear PETSc developers,
>>
>>
>>
>> *I have used KSP solver to solve a linear equation system in parallel
>> with 5 processors.*
>>
>>
>>
>> *The linear system consists of a pressure matrix (pressure_coeff_mat) and
>> its RHS (p_coeff_mat) whose values are set every time step using
>> MatSetValue and VecSetValues functions and then assembled using:*
>>
>>    call MatAssemblyBegin(pressure_coeff_mat,mat_final_assembly,petsc_err)
>>
>>    call MatAssemblyEnd(pressure_coeff_mat,mat_final_assembly,petsc_err)
>>
>>    call VecAssemblyBegin(p_coeff_rhs,petsc_err)
>>
>>    call VecAssemblyEnd(p_coeff_rhs,petsc_err)
>>
>>
>>
>>
>>
>> *One processor solves this linear system using KSBCGS Krylov solver with
>> ILU preconditioner, and the result is scattered to all processors*
>>
>>   call KSPGetPC(petsc_ksp,petsc_pc,petsc_err)
>>
>>    if (preconditioner==0) then
>>
>>       call PCSetType(petsc_pc,PCJACOBI,petsc_err
>>
>>       call
>> PetscOptionsSetValue(PETSC_NULL_OPTIONS,"sub_ksp_type","preonly",petsc_err)
>>
>>       call
>> PetscOptionsSetValue(PETSC_NULL_OPTIONS,"sub_pc_type","ilu",petsc_err)
>>
>>       call PCSetFromOptions(petsc_pc,petsc_err)
>>
>>    else
>>
>>       call PCSetType(petsc_pc,PCBJACOBI,petsc_err)
>>
>>    end if
>>
>>    call KSPSetFromOptions(petsc_ksp,petsc_err)
>>
>>    call KSPSetUp(petsc_ksp,petsc_err)
>>
>>    call KSPSolve(petsc_ksp,p_coeff_rhs,petsc_solution,petsc_err)
>>
>>    call VecScatterCreateToAll(petsc_solution,ctx,sol_to_all,petsc_err)
>>
>>    call
>> VecScatterBegin(ctx,petsc_solution,sol_to_all,INSERT_VALUES,SCATTER_FORWARD,petsc_err)
>>
>>    call VecScatterEnd(ctx,petsc_solution,sol_to_all,
>> INSERT_VALUES,SCATTER_FORWARD,petsc_err)
>>
>>    call VecGetValues(sol_to_all,ncells, loc_id_to_global-1,sol, petsc_err)
>>
>>    call VecScatterDestroy(ctx,petsc_err)
>>
>>    call VecDestroy(sol_to_all,petsc_err)
>>
>>
>>
>>
>>
>> *However, I got the following errors with vector scattering, i.e.,
>> VecScatterBegin, VecScatterEnd, and also VecGetValues. I have checked the
>> pressure matrix, its RHS, and the solution for the first call of KSPSolve.
>> Their values seem to be okay (see attached file). So, could you please tell
>> me what does this error mean and how can I fix it.*
>>
>
> 1) The error message says that 'sol_to_all'  has not been created
>
> 2) Is this system the Laplacian? If so, ILU is a bad preconditioner. You
> should try gamg.
>
>   Thanks,
>
>      Matt
>
>
>>
>>
>> *Thank you,*
>>
>>
>>
>> *Error message:*
>>
>>
>> -----------------------------------------------------------------------------
>>
>> Configure options --with-cc="win32fe cl" --with-fc="win32fe ifort"
>> --with-cxx="win32fe cl"
>> --with-blaslapack-lib="[/cygdrive/c/PROGRA~2/Intel/COMPOS~1/mkl/lib/ia32/mkl_intel_c_dll.lib,/cygdrive/c/PROGRA~2/Intel/COMPOS~1/mkl/lib/ia32/mkl_sequential_dll.lib,/cygdrive/c/PROGRA~2/Intel/COMPOS~1/mkl/lib/ia32/mkl_core_dll.lib]"
>> --with-mpi-lib="[/cygdrive/c/PROGRA~2/MICROS~1/MPI/Lib/x86/msmpifec.lib,/cygdrive/c/PROGRA~2/MICROS~1/MPI/Lib/x86/msmpi.lib]"
>> --with-mpi-include="[/cygdrive/c/PROGRA~2/MICROS~1/MPI/Include,/cygdrive/c/PROGRA~2/MICROS~1/MPI/Include/x86]"
>> --with-mpiexec=/cygdrive/c/PROGRA~1/MI564C~1/bin/mpiexec.exe
>> --with-debugging=1 --with-shared-libraries=0 --CFLAGS=-Od --FFLAGS=-Od
>> CXXFLAGS=-Od --with-clean=1
>>
>> #4930 VecScatterBegin() line 83 in
>> C:\petsc\PE4D24~1.13-\src\vec\vscat\INTERF~1\vscatfce.c
>>
>> [1]PETSC ERROR:
>>
>> [2]PETSC ERROR:
>>
>> [3]PETSC ERROR:
>>
>> [0]PETSC ERROR:
>>
>> [4]PETSC ERROR:
>>
>> #4930 VecScatterBegin() line 83 in
>> C:\petsc\PE4D24~1.13-\src\vec\vscat\INTERF~1\vscatfce.c
>>
>> #4930 VecScatterBegin() line 83 in
>> C:\petsc\PE4D24~1.13-\src\vec\vscat\INTERF~1\vscatfce.c
>>
>> #4930 VecScatterBegin() line 83 in
>> C:\petsc\PE4D24~1.13-\src\vec\vscat\INTERF~1\vscatfce.c
>>
>> #4930 VecScatterBegin() line 83 in
>> C:\petsc\PE4D24~1.13-\src\vec\vscat\INTERF~1\vscatfce.c
>>
>> --------------------- Error Message
>> --------------------------------------------------------------
>>
>> [1]PETSC ERROR:
>>
>> [2]PETSC ERROR:
>>
>> [3]PETSC ERROR:
>>
>> [0]PETSC ERROR:
>>
>> [4]PETSC ERROR:
>>
>> --------------------- Error Message
>> --------------------------------------------------------------
>>
>> --------------------- Error Message
>> --------------------------------------------------------------
>>
>> --------------------- Error Message
>> --------------------------------------------------------------
>>
>> --------------------- Error Message
>> --------------------------------------------------------------
>>
>> Null argument, when expecting valid pointer
>>
>> [1]PETSC ERROR:
>>
>> [2]PETSC ERROR:
>>
>> [3]PETSC ERROR:
>>
>> [0]PETSC ERROR:
>>
>> [4]PETSC ERROR:
>>
>> Null argument, when expecting valid pointer
>>
>> Null argument, when expecting valid pointer
>>
>> Null argument, when expecting valid pointer
>>
>> Null argument, when expecting valid pointer
>>
>> Null Object: Parameter # 3
>>
>> [1]PETSC ERROR:
>>
>> [2]PETSC ERROR:
>>
>> [3]PETSC ERROR:
>>
>> [0]PETSC ERROR:
>>
>> [4]PETSC ERROR:
>>
>> Null Object: Parameter # 3
>>
>> Null Object: Parameter # 3
>>
>> Null Object: Parameter # 3
>>
>> Null Object: Parameter # 3
>>
>> --
>> Hoang Nhan Hien, Ph.D.
>>
>> 1) Nuclear Safety Department, Institute for Nuclear Science and
>> Technology, Vietnam Atomic Energy Insitute
>>
>> 2) Nuclear Safety Analysis Method Development, Enviroment & Energy
>> Technology, Inc.
>>
>> Email: hoangnhanhien at gmail.com; h <hnhien at kaeri.re.kr>ien at en2t.com
>> Phone: +81-10-2635-2411
>>
>>
>
> --
> 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
>
> https://www.cse.buffalo.edu/~knepley/
> <http://www.cse.buffalo.edu/~knepley/>
>


-- 
Hoang Nhan Hien, Ph.D.

1) Nuclear Safety Department, Institute for Nuclear Science and Technology,
Vietnam Atomic Energy Insitute

2) Nuclear Safety Analysis Method Development, Enviroment & Energy
Technology, Inc.

Email: hoangnhanhien at gmail.com; h <hnhien at kaeri.re.kr>ien at en2t.com
Phone: +81-10-2635-2411
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20201103/6eb88389/attachment.html>

More information about the petsc-users mailing list