Version 5 (modified by 12 years ago) ( diff ) | ,
---|
Your First Run — An Example Test Problem
These directions assume that you have already compiled a problem module. If you have not done this yet, see "Setting up and Compiling A Problem" above.
- Move to your system's scratch directory. Most well-managed clusters have a directory like this; it's just a place on a high-volume drive for people to run jobs that produce large amounts of data. The scratch directories for our more heavily-used clusters are listed below:
grass: /grassdata/<username> bluehive: /scratch/<username> bluegene: /scratch/<username> itasca: /scratch1/<username>
- Create a run directory
run_dir
. Then, create a subdirectoryout
withinrun_dir
:mkdir run_dir mkdir run_dir/out
Theout/
directory is where AstroBEAR will store its output files.
- If you are running a problem that uses a source-term lookup table, then copy the
TABLES
directory fromsource
into your problem directory:cp -r <astrobear_path>/source/TABLES run_dir
- Put the problem's data files in
run_dir
. If you are running a sample problem that came with AstroBEAR, such asFieldLoopAdvection
, then the data files can be found in the problem directory insidemodules
.
- Copy the
astrobear
file from your AstroBEAR directory into your run directory:cp astrobear <rundir_path>/run_dir
AstroBEAR is not intended to be run without MPI, so be sure to use thempiexec
command even if you are running a single-processor job. On the plus side, you no longer have to change your data files every time you want to change the number of processors, since AstroBEAR handles decomposition automatically.
Once you have successfully built AstroBEAR and set up a problem directory, you can try running the code.