Re: [deal.II] MUMPS with PETScWrappers::MPI::BlockSparseMatrix

[Richard Schussnig]

Hi Konrad!

You can use the parallel direct solver in the schur complement, for 
orientation, take a look at step-57 (should be Navier-Stokes with direct 
solver for the A-block, if im not mistaken).
However, my inferior C++ knowledge did not allow me to do the factorization 
in the constructor of the 'preconditioner', but you can initialize it 
before creating the preconditioner and pass it as an argument!
This is considerably faster than doing the factorization inside the 
preconditioner of course, since you would re-do it every iteration. 
Nevertheless, this is perfectly fine for testing, to see if it is really 
doing the job!

Kind regards, Richard

Am Sonntag, 8. September 2019 15:20:20 UTC+2 schrieb Konrad:
>
> Thank you, David, so I will see if I can maybe still use it in the Schur 
> complement somehow.
>
> Best,
> Konrad
>
> On Friday, September 6, 2019 at 7:46:23 PM UTC+2, David Wells wrote:
>>
>> Hi Konrad,
>>
>> I don't think that it is possible to use MUMPS with a block matrix for 
>> exactly this reason. I think that if you want to use MUMPS you will need to 
>> copy the block matrix into a non-block matrix.
>>
>> Thanks,
>> David
>>
>> On Fri, Sep 6, 2019 at 12:25 PM Konrad  wrote:
>>
>>> Dear deal.ii community,
>>>
>>> is it possible to use MUMPS with 
>>> a PETScWrappers::MPI::BlockSparseMatrix? Don't find anything but I see 
>>> that PETScWrappers::MPI::BlockSparseMatrix does not inherit from 
>>> PETScWrappers::MatrixBase.
>>>
>>> Best,
>>> Konrad
>>>
>>> -- 
>>> The deal.II project is located at http://www.dealii.org/
>>> For mailing list/forum options, see 
>>> https://groups.google.com/d/forum/dealii?hl=en
>>> --- 
>>> You received this message because you are subscribed to the Google 
>>> Groups "deal.II User Group" group.
>>> To unsubscribe from this group and stop receiving emails from it, send 
>>> an email to dea...@googlegroups.com.
>>> To view this discussion on the web visit 
>>> https://groups.google.com/d/msgid/dealii/290c3820-4c09-4515-a2f9-847691627427%40googlegroups.com
>>>  
>>> 
>>> .
>>>
>>

-- 
The deal.II project is located at http://www.dealii.org/
For mailing list/forum options, see 
https://groups.google.com/d/forum/dealii?hl=en
--- 
You received this message because you are subscribed to the Google Groups 
"deal.II User Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to dealii+unsubscr...@googlegroups.com.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/dealii/0c5e62ad-fe63-49bf-9344-7af0652ad6bb%40googlegroups.com.

Re: [deal.II] MUMPS with PETScWrappers::MPI::BlockSparseMatrix

[Konrad]

Thank you, David, so I will see if I can maybe still use it in the Schur 
complement somehow.

Best,
Konrad

On Friday, September 6, 2019 at 7:46:23 PM UTC+2, David Wells wrote:
>
> Hi Konrad,
>
> I don't think that it is possible to use MUMPS with a block matrix for 
> exactly this reason. I think that if you want to use MUMPS you will need to 
> copy the block matrix into a non-block matrix.
>
> Thanks,
> David
>
> On Fri, Sep 6, 2019 at 12:25 PM Konrad > 
> wrote:
>
>> Dear deal.ii community,
>>
>> is it possible to use MUMPS with a PETScWrappers::MPI::BlockSparseMatrix? 
>> Don't find anything but I see that PETScWrappers::MPI::BlockSparseMatrix 
>> does not inherit from PETScWrappers::MatrixBase.
>>
>> Best,
>> Konrad
>>
>> -- 
>> The deal.II project is located at http://www.dealii.org/
>> For mailing list/forum options, see 
>> https://groups.google.com/d/forum/dealii?hl=en
>> --- 
>> You received this message because you are subscribed to the Google Groups 
>> "deal.II User Group" group.
>> To unsubscribe from this group and stop receiving emails from it, send an 
>> email to dea...@googlegroups.com .
>> To view this discussion on the web visit 
>> https://groups.google.com/d/msgid/dealii/290c3820-4c09-4515-a2f9-847691627427%40googlegroups.com
>>  
>> 
>> .
>>
>

-- 
The deal.II project is located at http://www.dealii.org/
For mailing list/forum options, see 
https://groups.google.com/d/forum/dealii?hl=en
--- 
You received this message because you are subscribed to the Google Groups 
"deal.II User Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to dealii+unsubscr...@googlegroups.com.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/dealii/da7bcf06-9daa-4b52-9230-956ef11e68d3%40googlegroups.com.

Re: [deal.II] MUMPS with PETScWrappers::MPI::BlockSparseMatrix

[David Wells]

Hi Konrad,

I don't think that it is possible to use MUMPS with a block matrix for
exactly this reason. I think that if you want to use MUMPS you will need to
copy the block matrix into a non-block matrix.

Thanks,
David

On Fri, Sep 6, 2019 at 12:25 PM Konrad  wrote:

> Dear deal.ii community,
>
> is it possible to use MUMPS with a PETScWrappers::MPI::BlockSparseMatrix?
> Don't find anything but I see that PETScWrappers::MPI::BlockSparseMatrix
> does not inherit from PETScWrappers::MatrixBase.
>
> Best,
> Konrad
>
> --
> The deal.II project is located at http://www.dealii.org/
> For mailing list/forum options, see
> https://groups.google.com/d/forum/dealii?hl=en
> ---
> You received this message because you are subscribed to the Google Groups
> "deal.II User Group" group.
> To unsubscribe from this group and stop receiving emails from it, send an
> email to dealii+unsubscr...@googlegroups.com.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/dealii/290c3820-4c09-4515-a2f9-847691627427%40googlegroups.com
> 
> .
>

-- 
The deal.II project is located at http://www.dealii.org/
For mailing list/forum options, see 
https://groups.google.com/d/forum/dealii?hl=en
--- 
You received this message because you are subscribed to the Google Groups 
"deal.II User Group" group.
To unsubscribe from this group and stop receiving emails from it, send an email 
to dealii+unsubscr...@googlegroups.com.
To view this discussion on the web visit 
https://groups.google.com/d/msgid/dealii/CABrTbYSsPwK7WUcUZroMygKuoqszL03X2L53fGXjWVwJs0BWig%40mail.gmail.com.