[petsc-dev] Need mechanism to set LocalToGlobal mapping in a DMShell

Barry Smith bsmith at mcs.anl.gov
Tue Feb 26 20:46:51 CST 2013


On Feb 26, 2013, at 2:24 PM, Richard Tran Mills <rtm at eecs.utk.edu> wrote:

> I think that adding DMLocalToLocalBegin/End makes sense from a consistency standpoint.  Should this be done by adding localtolocalbegin/end pointers to dm->ops?

   Yes, I believe so.

   Barry

> 
> --Richard
> 
> On 2/21/13 4:35 PM, Barry Smith wrote:
>> On Feb 21, 2013, at 3:31 PM, Richard Tran Mills <rtm at eecs.utk.edu> wrote:
>> 
>>> I will add DMShellSetGlobalToLocal (and DMShellSetLocalToGlobal).
>>> 
>>> One other thing: We also need to be do LocalToLocal operations. For the structured grid case, we call DMDALocalToLocalBegin/End, and for the unstructured grid case we construct our own VecScatter context essentially just how it is done in daltol.c for DMDA.  Is there a good reason that there is no DMLocalToLocalBegin/End?
>>   I don't think so. Just laziness in not moving them over when DM was created from DA. You can add them.
>> 
>>    Barry
> 




More information about the petsc-dev mailing list