[petsc-dev] MatLUFactor() on MATSEQAIJ with no-inodes

Barry Smith bsmith at mcs.anl.gov
Mon Mar 15 21:35:32 CDT 2010


   Please confirm that this is the absolute latest petsc-dev that you  
get this problem with. I pushed a fix late afternoon for this type of  
problem, that should have fixed the problem you report.

    Barry

On Mar 15, 2010, at 9:28 PM, Lisandro Dalcin wrote:

> A petsc4py testcase (TestMatFactorB2 in tests/test_mat_fact.py) is
> failing with petsc-dev.
>
> Basically, I'm creating a SEQAIJ matrix, assembling a Laplace 1D with
> finite direrences, and calling MatLUFactor() ... As you can see, the
> matrix is not actually using inodes (as expected), but anyway it seems
> that the virtual table is still using MatLUFactorNumeric_SeqAIJ_Inode.
>
> Interestingly a similar test with a diagonal matrix (I mean, SEQAIJ
> where A[i,i] = 1.0/(i+1))
>
>
>
> testFactorILU (__main__.TestMatFactorB2) ... ok
> testFactorLU (__main__.TestMatFactorB2) ...
> Matrix Object:
>  type=seqaij, rows=10, cols=10
>  total: nonzeros=28, allocated nonzeros=50
>    not using I-node routines
> ERROR
>
> ======================================================================
> ERROR: testFactorLU (__main__.TestMatFactorB2)
> ----------------------------------------------------------------------
> Traceback (most recent call last):
>  File "test/test_mat_fact.py", line 76, in testFactorLU
>    self.A.factorLU(r,c,{})
>  File "Mat.pyx", line 838, in petsc4py.PETSc.Mat.factorLU
> (src/petsc4py.PETSc.c:68161)
> Error: error code 77
> [0] MatLUFactor() line 2391 in src/mat/interface/matrix.c
> [0] MatLUFactor_SeqAIJ() line 948 in src/mat/impls/aij/seq/aijfact.c
> [0] MatLUFactorNumeric() line 2561 in src/mat/interface/matrix.c
> [0] MatLUFactorNumeric_SeqAIJ_Inode() line 1238 in src/mat/impls/aij/ 
> seq/inode.c
> [0] Petsc has generated inconsistent data
> [0] Matrix without inode information
>
>
> -- 
> Lisandro Dalcin
> ---------------
> Centro Internacional de Métodos Computacionales en Ingeniería (CIMEC)
> Instituto de Desarrollo Tecnológico para la Industria Química (INTEC)
> Consejo Nacional de Investigaciones Científicas y Técnicas (CONICET)
> PTLC - Güemes 3450, (3000) Santa Fe, Argentina
> Tel/Fax: +54-(0)342-451.1594




More information about the petsc-dev mailing list