Tips and tricks

What to do if the model fails

There may be some situations in which the model fails. Sometimes this may be a bug with the model itself, but the model can also stop running due to e.g. a computer crash. Another example may be that you are running on a HPC system with a job scheduler, but have not scheduled enough time or compute cores for the model to finish.

It is recommended because of this to run with the dump_data flag in model_setup set to True. This will dump the entire model state into a file (called dump.nc by default, but this is controllable via the dump_filepath model_setup variable).

With this file, it is then possible to load this into Python to look at the model state (e.g. for debugging, see scripts/debug_lateral_flow for an example), or to use this to restart your MONARCHS run (by setting reload_from_dump to True in model_setup.

Data analysis

A few tools exist to aid in looking at MONARCHS output.

get_2d_grid is useful particularly for looking at the progress (i.e. dump) files.