3.8 Running your simulation

You can run your simulation either in the object directory or in a separate run directory.

Running in the object directory is especially convenient for development and testing. The command for starting a FLASH simulation may be system dependent, but typically you would type something like

mpirun -np N flash4

or

mpiexec -n N flash4

to start a run on $N$ processing entities. On many systems, you can also simply use

./flash4

to start a run on 1 processing entity, i.e., without process-level parallelisation.

If you want to invoke FLASH in a separate run directory, the best way is to copy the flash4 binary into the run directory, and then proceed as above. However, before starting FLASH you also need to do the following:

• Copy a FLASH parfile, normally flash.par, into the run directory.
• If FLASH was configured with Paramesh4.0 in LIBRARY mode (this is not the default): copy the file amr_runtime_parameters, into the run directory. This file should have been generated in the object directory by running ./setup.
• Some code units use additional data or table files, which are also copied into the object directory by ./setup. (These files typically match one of the patterns *.dat or *_table*.) copy or move those files into the run directory, too, if your simulation needs them.