| Commit message (Collapse) | Author | Age |
... | |
| |
|
| |
|
|
|
|
|
|
| |
Removed definition of OD
Added Tensor syntax checking
|
|
|
|
|
|
|
|
|
| |
Added Errors package
Removed SpaceMask from CreateMoLThorn; Kranc shouldn't know anything
about CactusEinstein!
Added some error checking to CreateBaseThorn
|
| |
|
|
|
|
| |
Removed a large chunk of redundant code
|
|
|
|
|
|
|
| |
Added error checking to SpaceSeparated
Fixed CommaInitSeparated as it was buggy; it didn't initialise the last
variable!
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Added query about excessive sorting
Renamed "derivativesUsed" to "oldDerivativesUsed" as we now have
two types of derivative.
Added error checking to make sure that only recognized functions are
used in a calculation and that shorthand names do not conflict with
gridfunction names. Ensure that shorthands are assigned before they
are used.
Fixed typo; a semicolon was missing in UncommentSourceSync. This code
could not possibly have worked for C thorns with more than one loop in
a calculation.
Ensure that the grid loop range is correct for *both* types of
derivatives, not just the old ones.
Removed check on whether or not derivatives are used when deciding
whether to synchronize. It makes no sense; we want to know if
*subsequent* loops will use derivatives, and that is too hard to
tell. So we now synchronize any variables which are set (either in the
source file or in the schedule file).
|
| |
|
|
|
|
| |
allow for cleaner scheduling; schedule Translator now at INITIAL!
|
| |
|
|
|
|
| |
CodeGen`SOURCELANGUAGE)
|
|
|
|
| |
Mma 5.1.
|
|
|
|
| |
(because no definition in that loop contains derivatives)
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
| |
options verbatim, default empty (ICN)
Index: Parfiles.m
|
|
|
|
|
| |
Added IOBasic::out_format = .16e,
Added 2D robust stability tests
|
| |
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
post-process the code block structure returned by CreateCalculationFunction
and extract the information about what to SYNC from comment lines
via a function GrepSyncGroups[code_struct, function_name].
If a function contains more than one loop, the calls to CCTK_SyncGroup
will be uncommented and no SYNCs will be written into the schedule file
for that function.
Typically, then a schedule structure will contain entries like:
SynchronizedGroups -> GrepSyncGroups[setrhs, lookup[calc, Name]]
|
| |
|
| |
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
done for Setter thorns:
If the Calculation structure has only one loop then SYNCS are done
in schedule.ccl, otherwise they are done in the source code as before,
and a warning message is printed.
This is supposed to be the minimal change necessary to adapt to the
Multipatch infrastructure. As long as Multipatch is not used, the user
does not need to care.
|
| |
|
| |
|
|
|
|
| |
Helpers.m : EnsureDirectory
|
|
|
|
| |
several notebooks
|
|
|
|
|
| |
explicit initialisation of variables in Fortran to avoid implicit assignment
of SAVE attribute.
|
|
|
|
| |
continuation lines, staying within compiler limits for some thorns.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
| |
pointed out by Erik:
if ( (excision).AND.(find_excision_boundary) ) then
is illegal in F90, because the Cactus-BOOLEAN-variables become integers
in Fortran. Now we use:
if ( (excision.ne.0).AND.(find_excision_boundary.ne.0) ) then
|
|
|
|
| |
would otherwise be duplicated in local code
|
|
|
|
| |
was set twice, and in the diagonal case first to no and then to yes.
|
| |
|
|
|
|
|
|
|
|
| |
specify which partial derivative to use when they define a connection
(this will break existing code). This is not a final solution, and
will probably be reworked, so unless you desparately need TensorTools
to work with an existing thorn urgently, don't worry about this until
later.
|
| |
|
|
|
|
|
|
|
| |
thorn (e.g. for #include "gsl.h" with the GNU Scientific Library).
This should probably be extended to all the other thorn types, but
unless needed, I will wait until KrancThorns is refactored, as then it
will only have to be done once, not five times....
|
|
|
|
| |
Removed several unused functions.
|
|
|
|
|
| |
GNU Scientific Library. Is not necessarily the way to go about it,
but will not break existing code (i hope).
|
|
|
|
| |
cluttered the display.
|
|
|
|
|
|
|
|
| |
False (the default), the result is that no tensor type tags are
guessed. When True, an attempt is made to guess tensor type tags
based on the names of the members of the groups. However, this can
fail if the user is not using single groups for each tensor. The
default is now "False", which should maintain compatibility.
|
|
|
|
| |
evaluator thorns (since they only have one timelevel)
|
|
|
|
| |
Carpet warning messages
|
|
|
|
|
| |
experimental, but should not have any effect unless one activates
Erik's rotating symmetry or tensor thorns, which are still undocumented.
|