On 4/22/10 10:36 AM, Jed Brown wrote:
> On Thu, 22 Apr 2010 10:21:26 -0400, Boyce Griffith<[email protected]>
> wrote:
>>
>>
>> On 4/22/10 9:56 AM, Jed Brown wrote:
>>> On Thu, 22 Apr 2010 09:32:12 -0400, Boyce Griffith<[email protected]>
>>> wrote:
>>>> How should one setup the index set? Is it just a mapping from Vec index
>>>> to field index? E.g., each Vec entry is assigned to field 0 or 1 (or 2
>>>> or ...)?
>>>
>>> If your variables are interlaced like
>>>
>>> u0,v0,w0,u1,v1,w1,...
>>
>> In the present context, the order of variables in memory is whatever is
>> determined by libMesh --- and I haven't yet looked to see what that is.
>> It's my impression that variables are not interlaced like this, but I
>> could be mistaken.
>
> Right, but Roy said this:
>
> Block variable ordering with "--node_major_dofs" is perhaps the least
> well-documented option in libMesh, which is saying a lot, but we're
> already using it.
Agh, OK. I should look into this!
>>> split0: 0 3 6 9 | 12 15 18 21
>>> split1: 1 7 | 13 19
>>> split2: 2 4 5 8 10 11 | 14 16 17 20 22 23
>>>
>>> I hope this is clear.
>>
>> Sorry for being dense, but in this case, if I were to create the IS
>> manually using ISCreateGeneral, on process 0, I think that I would use:
>>
>> idx[0] = 0
>> idx[1] = 1
>> idx[2] = 2
>> idx[3] = 0
>> ...
>>
>> and similarly on process 1:
>>
>> ...
>> idx[4] = idx[16-12] = 2
>> idx[5] = idx[17-12] = 2
>> idx[6] = idx[18-12] = 0
>> idx[7] = idx[19-12] = 1
>> ...
>
> To create the index set for split0:
>
> /* This assumes communicator has size 2 */
> PetscInt idx[2][] = {{0,3,6,9},{12,15,18,21}};
> int rank;
> IS is0;
>
> MPI_Comm_rank(comm,rank);
> ISCreateGeneral(comm,4,idx[rank],&is0);
>
> This is a parallel index set containing the global indices of the dofs
> in the desired split.
OK, then I am apparently confused about how PCFieldSplitSetIS is
intended to work. Does one call PCFieldSplitSetIS once for each field?
I was imagining that you call this function only once...
-- Boyce
------------------------------------------------------------------------------
_______________________________________________
Libmesh-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/libmesh-users
