- FLASH assumes by default that it is being compiled against MPI
version 2 or higher. If you must compile against an MPI-1 library,
define the preprocessor symbol FLASH_MPI1 for your Fortran compiler,
but note that not all code units support this.
- The outflow boundary condition for face-centered variables in the use of
solenoidal magnetic field evolution on AMR fails to ensure the solenoidal constraint
at the physical outflow boundaries. However, numerical solutions with respect to
this error are still physically correct away from the outflow boundaries. This issue
may be resolved in future releases.
- The upwind-biased electric field implementation (i.e., E_upwind=.true.)
for the unsplit staggered mesh solver
in some cases fails to satisfy divergence-free magnetic field evolutions at restart.
If this is the case for your runs, please use E_upwind=.false.
- The new multipole solver is missing the ability to treat
a non-zero minimal radius for spherical geometries, and the
ability to specify a point mass contribution to the potential.
- The “Split” implementation of the diffusion solver is essentially meant for testing purposes.
Moreover, it has not been exercised with PARAMESH.
- Some configurations of hydrodynamic test problems with Chombo grid
show worse than expected mass and total energy conservation. Please
see the Chombo section in the Hydro chapter of this Guide for
details.
- We have experienced the following abort when running
IsentropicVortex problem with Chombo Grid:
"MayDay: TreeIntVectSet.cpp:1995: Assertion `bxNumPts != 0' failed. !!!"
We have been in contact with the Chombo team to resolve this
issue.
- The unsplit MHD solver doesn't support the mode "use_GravPotUpdate=.true."
for 1D when self-gravity is utilized. The solver will still work if it
is set to be .false. In this case the usual reconstruction schemes
will be used in computing the gravitational accelerations at the
predictor step (i.e., at the n+1/2 step) rather than calling the
Poisson solver to compute them.
- Time limiting due to burning, even though has an implementation,
is turned off in most simulations by keeping the value of parameter
enucDtFactor very high. The implementation is therefore not well
tested and should be used with care.
- Mesh replication is only supported for parallel HDF5 and the
experimental derived data type Parallel-NetCDF (+pnetTypeIO)
I/O implementations. Flash will fail at runtime if multiple meshes
are in use without using one of these I/O implementations.
- The unsplit staggered mesh MHD solver shows slight differences
(of relative magnitudes of order
) in restart comparisons (e.g., sfocu comparison),
when there are non-zero values of face-centered magnetic fields.
With PARAMESH4DEV, the current default Grid implementation, we have
observed this problem only with the `force_consistency' flag
(see Runtime Parameter gr_pmrpForceConsistency) turned on.
- In some cases with the default refinement criteria implementation, the
refinement pattern at a given point in time of a PARAMESH AMR
simulation may be slightly different depending on how often plotfiles
and checkpoints are written; with resulting small changes in
simulation results. The effect is expected to also be present in
previous FLASH versions. This is a side effect of
Grid_restrictAllLevels calls that happen in IO to prepare all grid
blocks for being dumped to file. We have determined that this can
only impact how quickly coarser blocks next to a refinement boundary
are allowed to further derefine when their better resolved neighbors
also derefine, in cases where second-derivative criteria applied to the
block itself would allow such derefinement. Users who are concerned
with this effect may want to replace the call to amr_restrict
in Grid_updateRefinement with a call to
Grid_restrictAllLevels, at the cost of a slight increase in
runtime.
- The PG compiler fails to compile source files which contain OpenMP
parallel regions that reference threadprivate data. This h/4.7
appens in
the threaded versions of the Multipole solver and the within block
threaded version of split hydro. A workaround is to remove
“default(none)” from the OpenMP parallel region.
- Absoft compiler (gcc 4.4.4/Absoft Pro fortran 11.1.x86_64 with mpich2 1.4.1p1) generates incorrectly behaving code with some files when used with any optimization. More specifically, we have seen this behavior with gr_markRefineDerefine.F90, but other files may be vulnerable too. An older version (Absoft Fortran 95 9.0 EP/gcc 4.5.1 with mpich-1.2.7p1)
works.
- The -index-reorder setup flag does not work in all the configurations.
If you wish to use it please contact the FLASH team.
- The -noclobber setup option will not force a rebuild of
all necessary source files in a FLASH application with derived data
type I/O (+hdf5TypeIO or +pnetTypeIO). Do not use
-noclobber with derived data type I/O.
