2017-04-06 17:53 GMT-03:00 Roy Stogner <[email protected]>:
>
>
> On Thu, 6 Apr 2017, Adriano Côrtes wrote:
>
> Let me see if I got it right? So if I have two system in my
>> equation_system, let me say one called "flow" (with variables
>> {u,v,w,p}) and another called "transport" (with variable {phi}),
>> then if I call
>>
>> --solver_system_names
>> --solver_variable_names
>> --solver_group_flow_u mom
>> --solver_group_flow_v mom
>> --solver_group_flow_w mom
>> --solver_group_flow_p press
>>
>
> This looks correct. IIRC you can use some of the PETSc verbose
> options to check degree of freedom counts to make sure things are
> going where you expect.
I tried the set of command lines above, and I'm getting this error
[0]PETSC ERROR: --------------------- Error Message
--------------------------------------------------------------
[0]PETSC ERROR: Petsc has generated inconsistent data
[0]PETSC ERROR: Unhandled case, must have at least two fields, not 1
It seems that it is not packing {u,v,w} in one group (called mom) and {p}
in another (called press). Any clue?
My system names (created at add_system() time) are those: "flow" and
"transport".
>
> I will be able to pass command lines like
>>
>> -flow_pc_type fieldsplit
>> -flow_pc_fieldsplit_type schur
>> -flow_pc_fieldsplit_schur_fact_type upper
>> -flow_pc_fieldsplit_schur_precondition a11 (question: here a11 will
>> be the press group?)
>>
>> -flow_fieldsplit_mom_ksp_type fgmres
>> -flow_fieldsplit_mom_ksp_max_it 5
>> -flow_fieldsplit_mom_pc_type bjacobi
>>
>> -flow_fieldsplit_press_ksp_type fgmres
>> -flow_fieldsplit_press_ksp_max_it 5
>>
>> If it is not like this, what is wrong and what is the proper way to call,
>> that is the names compositions?
>>
>
> That looks right, but it's been years since I futzed with this. And
> in particular, all I remember about the answer to your question was
> that it was really confusing. We sort our group indices maps by
> string, so the groups end up in alphabetical order as far as PETSc is
> concerned? But there's some PETSc command line options that will let
> you reorder them?
> ---
> Roy
--
Adriano Côrtes
=================================================
Research assistant
High-performance Computing Center (NACAD)
Federal University of Rio de Janeiro (UFRJ)
------------------------------------------------------------------------------
Check out the vibrant tech community on one of the world's most
engaging tech sites, Slashdot.org! http://sdm.link/slashdot
_______________________________________________
Libmesh-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/libmesh-users
