[petsc-users] [petsc-maint] VecGetSubVector causes NaN

[ehsan sadrfaridpour]

Thanks a lot for your prompt reply and useful advice.

I have another question. In my current case, I don't change any value in
the sub-vector and in another words I used it as read-only.
What should I do after I don't need it anymore?
Should I call VecRestoreSubVector method or it is faster if I call the
Destroy method on the sub-vector ?

If I understand you correctly, the subvector doesn't have any space for
itself, in that case Destroy method does not release the memory.

And finally, I destroy the original vector after I call either Restore or
Destroy on the sub-vector.

Best,
Ehsan

On Wed, Jun 22, 2016 at 12:44 PM, Matthew Knepley <knepley at gmail.com> wrote:

> On Wed, Jun 22, 2016 at 9:21 AM, ehsan sadrfaridpour <it.sadr at gmail.com>
> wrote:
>
>> Hi,
>>
>> If I use VecGetSubVector to get a subvector it works fine as far as I
>> don't destroy the original vector.
>> But, if I destroy the original vector, I get NaN values. So, it is my
>> misunderstanding of what the function suppose to do, or is it a problem?
>>
>
> You cannot destroy the original vector since you still need to call
>
>
> http://www.mcs.anl.gov/petsc/petsc-current/docs/manualpages/Vec/VecRestoreSubVector.html#VecRestoreSubVector
>
>
>> I do the same thing with MatGetSubMatrix
>> <http://www.mcs.anl.gov/petsc/petsc-current/docs/manualpages/Mat/MatGetSubMatrix.html#MatGetSubMatrix>
>> and I haven't had any problem with it.
>>
>
> MatGetSubMatrix() is a hybrid since for INITIAL_MATRIX it creates a new
> matrix, whereas for REUSE_MATRIX it refills and existing
> one. However there is not Restore().
>
>
> http://www.mcs.anl.gov/petsc/petsc-current/docs/manualpages/Mat/MatGetSubMatrix.html
>
>    Matt
>
> Best,
>> Ehsan
>>
>
>
>
> --
> 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
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/petsc-users/attachments/20160622/03d7e875/attachment.html>