Barry :
>   Ignore this email. I see Hong already did it a different way so you already 
> have access to all this information.

He asks
* Will/does this have a Fortran equivalent?

I'm not sure if the needed Fortran stubs are created automatically or
we must create them manually?
Hong

>
> On May 29, 2014, at 1:43 PM, Barry Smith <[email protected]> wrote:
>
>>
>>  We should add direct support for this. Thus beasties are all stored by 
>> MUMPS in the data structure DMUMPS_STRUC_C which is defined in dmumps_c.h
>> (versions also for single precision and complex).  So what PETSc should 
>> provide in mumps.c is a function something like
>>
>> #undef __FUNCT__
>> #define __FUNCT__ "MatMUMPSGetStruc"
>> PetscErrorCode MatMUMPSGetStruc(Mat A,void **struc)
>> {
>>  Mat_MUMPS      *mumps=(Mat_MUMPS*)A->spptr;
>>
>>  PetscFunctionBegin;
>>  *struc = (void *) mumps->id
>>  PetscFunctionReturn(0);
>> }
>> so stick this function into src/mat/impls/aij/mpi/mumps/mumps.c run make at 
>> the root directory of PETSc to have PETSc libraries recompiled
>> Also add a prototype for this function in petscmat.h
>>
>>
>> Now your code would include dumps_c.h and then call MatMUMPSGetStruc() and 
>> then you can directly access any thing you like.
>>
>> Let us know how it goes and we’ll get this stuff into the development 
>> version of PETSc.
>>
>>  Barry
>>
>>
>>
>>
>> On May 29, 2014, at 10:38 AM, M Asghar <[email protected]> wrote:
>>
>>> Hi,
>>>
>>> Is it possible to access the contents of MUMPS array INFOG (and INFO, 
>>> RINFOG etc) via the PETSc interface?
>>>
>>> I am working with SLEPc and am using MUMPS for the factorisation. I would 
>>> like to access the contents of their INFOG array within our code 
>>> particularly when an error occurs in order to determine whether any 
>>> remedial action can be taken. The error code returned from PETSc is useful; 
>>> any additional information from MUMPS that can be accessed from within ones 
>>> code would be very helpful also.
>>>
>>> Many thanks in advance.
>>>
>>> M Asghar
>>>
>>
>

Reply via email to