[petsc-users] using petsc tools to solve isolated irregular domains with finite difference

Bishesh Khanal bisheshkh at gmail.com
Mon Oct 28 01:30:39 CDT 2013 

On Sun, Oct 27, 2013 at 6:47 PM, Matthew Knepley <knepley at gmail.com> wrote:

> On Sun, Oct 27, 2013 at 10:04 AM, Bishesh Khanal <bisheshkh at gmail.com>wrote:
>
>>
>>
>>
>> On Sun, Oct 27, 2013 at 2:53 PM, Matthew Knepley <knepley at gmail.com>wrote:
>>
>>> On Sat, Oct 26, 2013 at 8:27 AM, Bishesh Khanal <bisheshkh at gmail.com>wrote:
>>>
>>>>
>>>>
>>>>
>>>> On Sat, Oct 26, 2013 at 2:57 PM, Matthew Knepley <knepley at gmail.com>wrote:
>>>>
>>>>> On Sat, Oct 26, 2013 at 3:12 AM, Bishesh Khanal <bisheshkh at gmail.com>wrote:
>>>>>
>>>>>>
>>>>>>
>>>>>>
>>>>>> On Fri, Oct 25, 2013 at 10:21 PM, Matthew Knepley <knepley at gmail.com>wrote:
>>>>>>
>>>>>>> On Fri, Oct 25, 2013 at 2:55 PM, Bishesh Khanal <bisheshkh at gmail.com
>>>>>>> > wrote:
>>>>>>>
>>>>>>>> On Fri, Oct 25, 2013 at 8:18 PM, Matthew Knepley <knepley at gmail.com
>>>>>>>> > wrote:
>>>>>>>>
>>>>>>>>> On Fri, Oct 25, 2013 at 12:09 PM, Bishesh Khanal <
>>>>>>>>> bisheshkh at gmail.com> wrote:
>>>>>>>>>
>>>>>>>>>> Dear all,
>>>>>>>>>> I would like to know if some of the petsc objects that I have not
>>>>>>>>>> used so far (IS, DMPlex, PetscSection) could be useful in the following
>>>>>>>>>> case (of irregular domains):
>>>>>>>>>>
>>>>>>>>>> Let's say that I have a 3D binary image (a cube).
>>>>>>>>>> The binary information of the image partitions the cube into a
>>>>>>>>>> computational domain and non-computational domain.
>>>>>>>>>> I must solve a pde (say a Poisson equation) only on the
>>>>>>>>>> computational domains (e.g: two isolated spheres within the cube). I'm
>>>>>>>>>> using finite difference and say a dirichlet boundary condition
>>>>>>>>>>
>>>>>>>>>> I know that I can create a dmda that will let me access the
>>>>>>>>>> information from this 3D binary image, get all the coefficients, rhs values
>>>>>>>>>> etc using the natural indexing (i,j,k).
>>>>>>>>>>
>>>>>>>>>> Now, I would like to create a matrix corresponding to the laplace
>>>>>>>>>> operator (e.g. with standard 7 pt. stencil), and the corresponding RHS that
>>>>>>>>>> takes care of the dirchlet values too.
>>>>>>>>>> But in this matrix it should have the rows corresponding to the
>>>>>>>>>> nodes only on the computational domain. It would be nice if I can easily
>>>>>>>>>> (using (i,j,k) indexing) put on the rhs dirichlet values corresponding to
>>>>>>>>>> the boundary points.
>>>>>>>>>> Then, once the system is solved, put the values of the solution
>>>>>>>>>> back to the corresponding positions in the binary image.
>>>>>>>>>> Later, I might have to extend this for the staggered grid case
>>>>>>>>>> too.
>>>>>>>>>> So is petscsection or dmplex suitable for this so that I can set
>>>>>>>>>> up the matrix with something like DMCreateMatrix ? Or what would you
>>>>>>>>>> suggest as a suitable approach to this problem ?
>>>>>>>>>>
>>>>>>>>>> I have looked at the manual and that led me to search for a
>>>>>>>>>> simpler examples in petsc src directories. But most of the ones I
>>>>>>>>>> encountered are with FEM (and I'm not familiar at all with FEM, so these
>>>>>>>>>> examples serve more as a distraction with FEM jargon!)
>>>>>>>>>>
>>>>>>>>>
>>>>>>>>> It sounds like the right solution for this is to use PetscSection
>>>>>>>>> on top of DMDA. I am working on this, but it is really
>>>>>>>>> alpha code. If you feel comfortable with that level of
>>>>>>>>> development, we can help you.
>>>>>>>>>
>>>>>>>>
>>>>>>>> Thanks, with the (short) experience of using Petsc so far and being
>>>>>>>> familiar with the awesomeness (quick and helpful replies) of this mailing
>>>>>>>> list, I would like to give it a try. Please give me some pointers to get
>>>>>>>> going for the example case I mentioned above. A simple example of using
>>>>>>>> PetscSection along with DMDA for finite volume (No FEM) would be great I
>>>>>>>> think.
>>>>>>>> Just a note: I'm currently using the petsc3.4.3 and have not used
>>>>>>>> the development version before.
>>>>>>>>
>>>>>>>
>>>>>>> Okay,
>>>>>>>
>>>>>>> 1)  clone the repository using Git and build the 'next' branch.
>>>>>>>
>>>>>>
>>>>>> I encountered errors when doing make on the 'next' branch. The errors
>>>>>> are as follows (I tried attached the configure.log file but the email
>>>>>> bounced back saying it awaits moderator approval for having too big an
>>>>>> attachment, so I'm sending this one with only make.log attached. ) :
>>>>>>
>>>>>
>>>>> They are fixed. Pull again and rebuild.
>>>>>
>>>>
>>>> doing git pull in the next branch says "Already up-to-date.", I'm not
>>>> sure if it should have said that then. Still tried ./configure and make
>>>> again but returns the same error. Does it take sometime to get updated in
>>>> the servers or do I need to do anything special again other than the
>>>> following ? :
>>>> git checkout next
>>>> git pull
>>>>
>>>
>>> Okay, it should be this
>>>
>>> git checkout next
>>> git pull
>>> make allfortranstubs
>>> make
>>>
>>
>> Thanks, but now it reports another error:
>>
>>          CXX
>> arch-linux2-cxx-debug/obj/src/dm/impls/redundant/dmredundant.o
>>          CXX
>> arch-linux2-cxx-debug/obj/src/dm/impls/redundant/ftn-auto/dmredundantf.o
>>          CXX arch-linux2-cxx-debug/obj/src/dm/impls/plex/plexcreate.o
>> src/dm/impls/da/hypre/mhyp.c: In function ‘PetscErrorCode
>> MatSetupDM_HYPREStruct(Mat, DM)’:
>> src/dm/impls/da/hypre/mhyp.c:485:58: error: invalid conversion from
>> ‘PetscInt** {aka int**}’ to ‘const PetscInt** {aka const int**}’
>> [-fpermissive]
>> /home/bkhanal/Documents/softwares/petsc/include/petscdmda.h:68:27:
>> error:   initializing argument 3 of ‘PetscErrorCode
>> DMDAGetGlobalIndices(DM, PetscInt*, const PetscInt**)’ [-fpermissive]
>> src/dm/impls/da/hypre/mhyp.c: In function ‘PetscErrorCode
>> MatSetupDM_HYPRESStruct(Mat, DM)’:
>> src/dm/impls/da/hypre/mhyp.c:963:58: error: invalid conversion from
>> ‘PetscInt** {aka int**}’ to ‘const PetscInt** {aka const int**}’
>> [-fpermissive]
>> /home/bkhanal/Documents/softwares/petsc/include/petscdmda.h:68:27:
>> error:   initializing argument 3 of ‘PetscErrorCode
>> DMDAGetGlobalIndices(DM, PetscInt*, const PetscInt**)’ [-fpermissive]
>>
>
> Yes, one of the test branched had a problem with Hypre. I pushed a fix.
> Please try again.
>

Thanks, it works now. So (1) is done, I'm almost done with (3), will get
back soon.


>
>   Thanks,
>
>      Matt
>
>
>> gmake[2]: *** [arch-linux2-cxx-debug/obj/src/dm/impls/da/hypre/mhyp.o]
>> Error 1
>> gmake[2]: *** Waiting for unfinished jobs....
>> gmake[2]: Leaving directory `/home/bkhanal/Documents/softwares/petsc'
>> gmake[1]: *** [gnumake] Error 2
>> gmake[1]: Leaving directory `/home/bkhanal/Documents/softwares/petsc'
>> **************************ERROR*************************************
>>   Error during compile, check arch-linux2-cxx-debug/conf/make.log
>>   Send it and arch-linux2-cxx-debug/conf/configure.log to
>> petsc-maint at mcs.anl.gov
>> ********************************************************************
>> make: *** [all] Error 1
>>
>>
>>>
>>>   Thanks,
>>>
>>>      Matt
>>>
>>>
>>>>
>>>>>   Matt
>>>>>
>>>>>
>>>>>>           CXX arch-linux2-cxx-debug/obj/src/
>>>>>> mat/order/ftn-auto/spectralf.o
>>>>>>          CXX
>>>>>> arch-linux2-cxx-debug/obj/src/mat/order/ftn-custom/zsorderf.o
>>>>>> src/mat/order/wbm.c: In function ‘PetscErrorCode
>>>>>> MatGetOrdering_WBM(Mat, MatOrderingType, _p_IS**, _p_IS**)’:
>>>>>> src/mat/order/wbm.c:12:24: warning: variable ‘cntl’ set but not used
>>>>>> [-Wunused-but-set-variable]
>>>>>> src/mat/order/wbm.c:15:36: warning: unused variable ‘num’
>>>>>> [-Wunused-variable]
>>>>>> src/mat/order/wbm.c:15:56: warning: variable ‘icntl’ set but not used
>>>>>> [-Wunused-but-set-variable]
>>>>>> src/mat/order/wbm.c:15:66: warning: unused variable ‘info’
>>>>>> [-Wunused-variable]
>>>>>>          CXX arch-linux2-cxx-debug/obj/src/mat/matfd/fdmatrix.o
>>>>>> src/mat/order/ftn-auto/spectralf.c: In function ‘void
>>>>>> matcreatelaplacian_(Mat, PetscReal*, PetscBool*, _p_Mat**, int*)’:
>>>>>> src/mat/order/ftn-auto/spectralf.c:44:40: error: ‘MatCreateLaplacian’
>>>>>> was not declared in this scope
>>>>>> gmake[2]: ***
>>>>>> [arch-linux2-cxx-debug/obj/src/mat/order/ftn-auto/spectralf.o] Error 1
>>>>>> gmake[2]: *** Waiting for unfinished jobs....
>>>>>> src/mat/order/hslmc64.c: In function ‘PetscErrorCode HSLmc64AD(const
>>>>>> PetscInt*, PetscInt*, PetscInt*, PetscInt*, const PetscInt*, const
>>>>>> PetscInt*, PetscScalar*, PetscInt*, PetscInt*, PetscInt*, PetscInt*,
>>>>>> PetscInt*, PetscScalar*, PetscInt*, PetscScalar*, PetscInt*)’:
>>>>>> src/mat/order/hslmc64.c:332:21: warning: variable ‘warn1’ set but not
>>>>>> used [-Wunused-but-set-variable]
>>>>>> src/mat/order/hslmc64.c:332:28: warning: variable ‘warn2’ set but not
>>>>>> used [-Wunused-but-set-variable]
>>>>>> src/mat/order/hslmc64.c:332:35: warning: variable ‘warn4’ set but not
>>>>>> used [-Wunused-but-set-variable]
>>>>>> gmake[2]: Leaving directory `/home/bkhanal/Documents/softwares/petsc'
>>>>>> gmake[1]: *** [gnumake] Error 2
>>>>>> gmake[1]: Leaving directory `/home/bkhanal/Documents/softwares/petsc'
>>>>>> **************************ERROR*************************************
>>>>>>   Error during compile, check arch-linux2-cxx-debug/conf/make.log
>>>>>>   Send it and arch-linux2-cxx-debug/conf/configure.log to
>>>>>> petsc-maint at mcs.anl.gov
>>>>>> ********************************************************************
>>>>>> make: *** [all] Error 1
>>>>>>
>>>>>>
>>>>>>
>>>>>>>
>>>>>>> 2) then we will need to create a PetscSection that puts unknowns
>>>>>>> where you want them
>>>>>>>
>>>>>>> 3) Setup the solver as usual
>>>>>>>
>>>>>>> You can do 1) an 3) before we do 2).
>>>>>>>
>>>>>>>
>>>>>>>  If not, just put the identity into
>>>>>>>>> the rows you do not use on the full cube. It will not hurt
>>>>>>>>> scalability or convergence.
>>>>>>>>>
>>>>>>>>
>>>>>>>> In the case of Poisson with Dirichlet condition this might be the
>>>>>>>> case. But is it always true that having identity rows in the system matrix
>>>>>>>> will not hurt convergence ? I thought otherwise for the following reasons:
>>>>>>>> 1)  Having read Jed's answer here :
>>>>>>>> http://scicomp.stackexchange.com/questions/3426/why-is-pinning-a-point-to-remove-a-null-space-bad/3427#3427
>>>>>>>>
>>>>>>>
>>>>>>> Jed is talking about a constraint on a the pressure at a point. This
>>>>>>> is just decoupling these unknowns from the rest
>>>>>>> of the problem.
>>>>>>>
>>>>>>>
>>>>>>>> 2) Some observation I am getting (but I am still doing more
>>>>>>>> experiments to confirm) while solving my staggered-grid 3D stokes flow with
>>>>>>>> schur complement and using -pc_type gamg for A00 matrix. Putting the
>>>>>>>> identity rows for dirichlet boundaries and for ghost cells seemed to have
>>>>>>>> effects on its convergence. I'm hoping once I know how to use PetscSection,
>>>>>>>> I can get rid of using ghost cells method for the staggered grid and get
>>>>>>>> rid of the identity rows too.
>>>>>>>>
>>>>>>>
>>>>>>> It can change the exact iteration, but it does not make the matrix
>>>>>>> conditioning worse.
>>>>>>>
>>>>>>>    Matt
>>>>>>>
>>>>>>>
>>>>>>>>  Anyway please provide me with some pointers so that I can start
>>>>>>>> trying with petscsection on top of a dmda, in the beginning for
>>>>>>>> non-staggered case.
>>>>>>>>
>>>>>>>> Thanks,
>>>>>>>> Bishesh
>>>>>>>>
>>>>>>>>>
>>>>>>>>>   Matt
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>> Thanks,
>>>>>>>>>> Bishesh
>>>>>>>>>>
>>>>>>>>>
>>>>>>>>>
>>>>>>>>>
>>>>>>>>> --
>>>>>>>>> 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
>>>>>>>>>
>>>>>>>>
>>>>>>>>
>>>>>>>
>>>>>>>
>>>>>>> --
>>>>>>> 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
>>>>>>>
>>>>>>
>>>>>>
>>>>>
>>>>>
>>>>> --
>>>>> 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
>>>>>
>>>>
>>>>
>>>
>>>
>>> --
>>> 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
>>>
>>
>>
>
>
> --
> 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/20131028/f9edac80/attachment-0001.html>

More information about the petsc-users mailing list