| Commit message (Collapse) | Author | Age |
|---|
| |
|
|
|
|
|
|
| |
operators, as set by boundary_shiftout* parameters of CoordBase.
Only first-order centered derivative operators are affected.
git-svn-id: https://svn.cct.lsu.edu/repos/numrel/LSUThorns/SummationByParts/trunk@105 f69c4107-0314-4c4f-9ad4-17e986b73f4a
|
| |
|
|
| |
git-svn-id: https://svn.cct.lsu.edu/repos/numrel/LSUThorns/SummationByParts/trunk@97 f69c4107-0314-4c4f-9ad4-17e986b73f4a
|
| |
|
|
|
|
|
|
|
|
| |
makes it compile with certain versions of gfortran.
Change type of argument "part" to set_dmatrix_half from CCTK_INT to
integer since it is called with integer (not CCTK_INT) values.
git-svn-id: https://svn.cct.lsu.edu/repos/numrel/LSUThorns/SummationByParts/trunk@77 f69c4107-0314-4c4f-9ad4-17e986b73f4a
|
| |
|
|
|
|
|
| |
integer constants, whereas real constants are intended.
git-svn-id: https://svn.cct.lsu.edu/repos/numrel/LSUThorns/SummationByParts/trunk@74 f69c4107-0314-4c4f-9ad4-17e986b73f4a
|
| |
|
|
|
|
|
| |
deallocation of the temporary array within the wrong if statement.
git-svn-id: https://svn.cct.lsu.edu/repos/numrel/LSUThorns/SummationByParts/trunk@67 f69c4107-0314-4c4f-9ad4-17e986b73f4a
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
part is an optimisation. Previously the coefficients were constructed every
time the dissipation was applied. This turned out to have a significant
overhead, especially for large domains. Now the coefficients are constructed
only once for each patch on each processor and stored. The thorn now uses
the aliased functions MultiPatch_GetMap and MultiPatch_GetMaps (provided by
thorn MultiPatch) to figure out how many patches are present and which patch
is currently being worked on. These dissipation operators will not work unless
the MultiPatch thorn is used. Right now this should not be a problem (but can
be changed if it is), unless somebody uses dissipation operators outside of
the multipatch infrastructure.
The second part is an update aimed at making the restricted full norm
dissipation operators better behaved on grids with very different number of
grid points in different directions. For that reason the parameter:
diss_fraction
has been made vector valued (of length 3), so that the width of the transition
region from the boundary to the interior operators can be controlled
separately in the 3 dimensions. Unfortunately this means that any parameter
files, that explicitly sets this parameter, will have to be modified. I don't
think too many people are using the restricted full dissipation operators yet,
so it shouldn't cause too many problems. In additiona new parameter:
h_scaling
has been introduced. This parameter is also vector valued of length 3 and
allows the local grid spacing to be rescaled to more closely resemble the
physical grid spacing in any dimension.
Note, these changes only affect the non-diagonal dissipation operators.
git-svn-id: https://svn.cct.lsu.edu/repos/numrel/LSUThorns/SummationByParts/trunk@61 f69c4107-0314-4c4f-9ad4-17e986b73f4a
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
ecplictly setting any derivatives in the y- and/or z-direction to zero
when simulating 1- or 2-D problems with a 3D code. This avoids issues with
errors at roundoff level in the y- and z-directions. Should be used with care
and together with the parameter check_grid_sizes to avoid complaints about
too small grids in the y- and z-directions.
Has only been tested in a few cases. If problems are caused by this commit,
please report them immediately.
git-svn-id: https://svn.cct.lsu.edu/repos/numrel/LSUThorns/SummationByParts/trunk@59 f69c4107-0314-4c4f-9ad4-17e986b73f4a
|
|
|
git-svn-id: https://svn.cct.lsu.edu/repos/numrel/LSUThorns/SummationByParts/trunk@57 f69c4107-0314-4c4f-9ad4-17e986b73f4a
|
