FAQ¶

Can I change the directory where .edp scripts are executed ?¶

By default, the interpreted script is saved in a temporary directory which is removed once the FreeFemRunner is destroyed. However, it can also be saved in a custom directory by using the parameter run_dir from the constructor of FreeFemRunner:

python

# Save PyFreeFEM script files in the directory .run/
runner = FreeFemRunner(code,run_dir="run")

An exception is raised if the specified directory does not exist.

Warning

Specifying a running directory is useful to check and possibly to modify the .edp filed generated by PyFreeFEM. However, be warned that temporary directories are preferable to avoid conflicting accesses to stored data: conflicts can occur if several instances of a FreeFemRunner sharing the same running directory are executed (e.g. when running multiple processes of the same python script).

Can I use macro files ?¶

Macro files can be specified with the parameter macro_files of the FreeFemRunner constructor:

python

runner = FreeFemRunner(code,macro_files=["utils.edp","macros.edp"])

These files will be automatically copied to the PyFreeFEM running directory when calling execute(). Macro files also support PyFreeFEM preprocessing instructions.

Alternatively, macro files can be saved in the pyfreefem/edp directory and can be imported using the IMPORT preprocessing instruction:

python

IMPORT "macros.edp"

Can I run scripts with FreeFem++-mpi or ff-mpirun?¶

It is possible to run a script in parallel with the command ff-mpirun instead of FreeFem++ and with a desired number of cpus by using the arguments ncpu or with_mpi of the method ~pyfreefem.freefemrunner.execute:

python

# Run in parallel on 4 cpus with ff-mpirun -np 4
runner.execute(ncpu=4);
# Run on 1 cpu with ff-mpirun -np 1
runner.execute(ncpu=1,with_mpi=1);

In that case, a magic variable WITH_MPI is automatically assigned and set to 1, which can be used to execute instructions depending on whether ff-mpirun is used, e.g.:

freefem

IF WITH_MPI
    /* Instructions to include if the code is run with ff-mpirun */
ELSE
    /* Instructions to include if the code is run with FreeFem++ */
    int mpirank = 0;
    int mpisize = 1;
ENDIF

Can I enable FreeFEM graphics ?¶

FreeFEM graphics are disabled by default. They can be activated by assigning plot=True when calling the method execute():

python script.py¶

from pyfreefem import FreeFemRunner

code="""
mesh=square(10,10);
plot(Th);
"""
FreeFemRunner(code).execute()

Executing the script.py will display the following figure:

_images/Th_freefem.png — FreeFem plot after running script.py¶

Can I change the default FreeFem++ command ?¶

On some configurations, it is useful to change the FreeFem command. This can be achieved by updating the method ~FreeFemRunner.cmd.

For instance, if one wants to prefix the FreeFem++ command when graphics are activated (option -wg), then one can modify ~FreeFemRunner.cmd as follows:

python

from pyfreefem import FreeFemRunner

# Small hack for running FreeFem -wg command with vglrun
def cmd(self, *args, **kwargs):
    cmd = self.old_cmd(*args,**kwargs)
    if self.plot or kwargs.get('plot', False):
        cmd = "vglrun "+cmd
    return cmd
FreeFemRunner.old_cmd = FreeFemRunner.cmd
FreeFemRunner.cmd = cmd

Now, FreeFemRunner.execute(plot=True) will call vglrun FreeFem++ -wg.

See also the example pyfreefem/examples/ex13_change_cmd.py.