[petsc-users] memory access out of range with MUMPS

[Jose E. Roman]

If you are going to run in parallel then you should not provide the communicator comm=PETSc.COMM_SELF in any object.
Jose


> El 16 abr 2018, a las 17:42, Yongxiang Wu <yongxiang27 at gmail.com> escribió:
> 
> Hello everyone,
> I am using slepc4py solving a generalized eigenvalue problem, with a large spase matrix.  The following is my python code. The matrix is first created using PETSc.Mat().createAIJ() and saved to file matrix-A.dat . 
> 
> 
>    opts = PETSc.Options()
>    opts.setValue('-st_pc_factor_mat_solver_package','mumps')
> 
>     viewer = PETSc.Viewer().createBinary('matrix-A.dat', 'r',comm=PETSc.COMM_SELF)
>     A = PETSc.Mat().create(comm=PETSc.COMM_SELF)
>     A.load(viewer)
>     A.setFromOptions()
>     A.setUp()
>        
>     Print('load petsc binary finished')
>     
>     t0 = time()
>     
>     E = SLEPc.EPS().create(comm=PETSc.COMM_SELF)
>     E.setOperators(A,None)
>     E.setDimensions(noEigs2solv,PETSc.DECIDE)           # set number of  eigenvalues to compute
>     E.setWhichEigenpairs(E.Which.TARGET_MAGNITUDE)
>     E.setProblemType(SLEPc.EPS.ProblemType.GNHEP)    # generalized non-Hermitian eigenvalue problem
>     
>     E.setTolerances(arnoldiTol,1000)     # set tolerance 
>     if SIGMA is not None:
>         E.setTarget(SIGMA)               # set the desired eigenvalue 
>     
>     if options_v0 is not None:
>         E.setInitialSpace(options_v0)    # set the initial vector
>        
>     st = E.getST()
>     st.setType('sinvert')
>            
>     E.setFromOptions()
>     Print('Start solving ...')
>     E.solve()
> 
> My problem here is, when I run this code on cluster, I would end up with the following error. Strange thing is that I used the save library on my local PC, and it works without any error.  If any one of you could give me hint what is wrong?  Thanks
> 
> 
>  ** ERROR RETURN ** FROM ZMUMPS INFO(1)=   -1
>  ** INFO(2)=               0
> [0]PETSC ERROR: ------------------------------------------------------------------------
> [0]PETSC ERROR: Caught signal number 11 SEGV: Segmentation Violation, probably memory access out of range
> [0]PETSC ERROR: Try option -start_in_debugger or -on_error_attach_debugger
> [0]PETSC ERROR: or see http://www.mcs.anl.gov/petsc/documentation/faq.html#valgrind
> [0]PETSC ERROR: or try http://valgrind.org on GNU/linux and Apple Mac OS X to find memory corruption errors
> [0]PETSC ERROR: likely location of problem given in stack below
> [0]PETSC ERROR: ---------------------  Stack Frames ------------------------------------
> [0]PETSC ERROR: Note: The EXACT line numbers in the stack are not available,
> [0]PETSC ERROR:       INSTEAD the line number of the start of the function
> [0]PETSC ERROR:       is given.
> [0]PETSC ERROR: [0] PetscInitializeMUMPS line 1188 /zhome/academic/HLRS/iag/iagyonwu/.local/lib/petsc/src/mat/impls/aij/mpi/mumps/mumps.c
> [0]PETSC ERROR: [0] MatGetFactor_aij_mumps line 2125 /zhome/academic/HLRS/iag/iagyonwu/.local/lib/petsc/src/mat/impls/aij/mpi/mumps/mumps.c
> [0]PETSC ERROR: [0] MatGetFactor line 4341 /zhome/academic/HLRS/iag/iagyonwu/.local/lib/petsc/src/mat/interface/matrix.c
> [0]PETSC ERROR: [0] PCSetUp_LU line 59 /zhome/academic/HLRS/iag/iagyonwu/.local/lib/petsc/src/ksp/pc/impls/factor/lu/lu.c
> [0]PETSC ERROR: [0] PCSetUp line 886 /zhome/academic/HLRS/iag/iagyonwu/.local/lib/petsc/src/ksp/pc/interface/precon.c
> [0]PETSC ERROR: [0] KSPSetUp line 294 /zhome/academic/HLRS/iag/iagyonwu/.local/lib/petsc/src/ksp/ksp/interface/itfunc.c
> [0]PETSC ERROR: [0] STSetUp_Sinvert line 96 /zhome/academic/HLRS/iag/iagyonwu/.local/lib/slepc-3.8.2/src/sys/classes/st/impls/sinvert/sinvert.c
> [0]PETSC ERROR: [0] STSetUp line 233 /zhome/academic/HLRS/iag/iagyonwu/.local/lib/slepc-3.8.2/src/sys/classes/st/interface/stsolve.c
> [0]PETSC ERROR: [0] EPSSetUp line 104 /zhome/academic/HLRS/iag/iagyonwu/.local/lib/slepc-3.8.2/src/eps/interface/epssetup.c
> [0]PETSC ERROR: [0] EPSSolve line 129 /zhome/academic/HLRS/iag/iagyonwu/.local/lib/slepc-3.8.2/src/eps/interface/epssolve.c
> application called MPI_Abort(MPI_COMM_WORLD, 59) - process 0
> [unset]: write_line error; fd=-1 buf=:cmd=abort exitcode=59
> 
> with regards
> Yong