| bf9b5d84 | 09-Mar-2016 |
Toby Isaac <tisaac@uchicago.edu> |
DMFOrest: added DMForestGetAdaptivitySF()
It makes most sense to compute transfer SFs when the new forest is being
setup, but we may not need them if we're doing some initiali a prior
refinement wit
DMFOrest: added DMForestGetAdaptivitySF()
It makes most sense to compute transfer SFs when the new forest is being
setup, but we may not need them if we're doing some initiali a prior
refinement with nothing to transfer.
show more ...
|
| 26d9498a | 08-Mar-2016 |
Toby Isaac <tisaac@uchicago.edu> |
DMForest: added DMForestSetAdaptivityPurpose()
DMDestroy() can't do cycle checks with forest->adapt. I could try to
introduce a robust, extensible cycle check into DM, but that's a bigger
project.
DMForest: added DMForestSetAdaptivityPurpose()
DMDestroy() can't do cycle checks with forest->adapt. I could try to
introduce a robust, extensible cycle check into DM, but that's a bigger
project. This lets the user move forest->adapt to dm->coarseDM or
dm->fineDM, where it can be picked up in cycle detection.
show more ...
|
| 20e8089b | 24-Feb-2016 |
Toby Isaac <tisaac@uchicago.edu> |
DMP4est: simplify DMForestTemplate() interface
Now when you want to create dmAdapted from dm,
just call
DMForestTemplate(dm,MPI_COMM_NULL,&dmAdapted).
If you called DMSetUp(dmAdapted) right af
DMP4est: simplify DMForestTemplate() interface
Now when you want to create dmAdapted from dm,
just call
DMForestTemplate(dm,MPI_COMM_NULL,&dmAdapted).
If you called DMSetUp(dmAdapted) right after this, the topology of
dmAdapted would essentially be a copy of dm, but you can do things like
set adaptivity labels, change the overlap, and change the partitioning
before calling set up, and then all of those changes will be executed in
DMSetUp().
show more ...
|