Hello Severin,
sorry for the slow response.
for my current output
CarpetIOHDF5::out_vars = "grid::coordinates{out_every=1000000000} HydroBase::rho"
I receive files like
grid-coordinates.file_0.h5 grid-coordinates.file_1.h5
hydrobase-rho.file_0.h5 hydrobase-rho.file_1.h5
Now I have basically two questions left:
- How can I use both of these databases now to generate plots in
visit, i.e., how can I use the coordinates to plot rho?
VisIt is hard-coded to expect files x.file_0.h5, y.file_0.h5, z.file_0.h5 etc so since you had set CarpetIOHDF5::one_file_per_group you will have to extract them afterwards.
This command:
for i in grid-coordinates.file*.h5 ; do for d in x y z ; do h5ls $i | gawk -vd=$d '{sub(" (Dataset|Group).*",""); gsub("\\",""); if(match($0,"GRID::"d) || /Parameters/) { print }}' >dsets.txt hdf5_extract dsets.txt $i $d.${i#*.} done done
will do so when executed in the directory containing the h5 files. This is a bit complex unfortunately. Mostly this is because hdf5_slicer cannot be used (since it changes a hdf5 attribute that then confuses the VisIt Reader).
- If I try to plot only the grid-coordinates in VisIt (e.g. Mesh
plot): - If I plot it for the "Carpet-AMR-grid" I can see the mesh plot for the inner Cartesian grid. - If I try to plot if for the "Carpet Multipatch", the same error as before occcurs:
"The compute engine running on (...) has exited abnormally.
Yes, that is expected. Sorry, the error checking of the VisIt reader is not very good. This happened b/c it could not find the coordinates files and then crashed.
Does this mean that there is something wrong with my grid setup?
There is nothing wrong with your setup. The VisIt reader is to blame. There is a newer version of it which has error checking and some other improvements here:
git clone git@bitbucket.org:rhaas80/carpethdf5.git
though you would have to compile it yourself. I have to go talk to the VisIt devs to have the new version included in VisIt.
Yours, Roland