Running thermodynamics only with MPI

[einola]

Running something simple and quick but with MPI is essential for those working on coupling (@andreapiacentini, @monsieuralok). Here is how I do this (even though it crashes at the end until this issue is resolved).

Get nextsim running in your favourite environment; just be sure to run cmake with -DENABLE_MPI. I used the Docker images in the Docker/ directory.

The setup consists of the single column config_column.cfg configuration run on a 30x30 grid. This should result in a spatially uniform solution that is time-varying.

I created the init file using make_init_column.py, changing nfirst and nsecond to 30 (lines 3 and 4 in the file). I also changed the name of the output file to init_column_30x30.nc for clarity).

Then, from the run/ directory do:

root@5323bdc8d33f:/nextsim/nextsimdg/run# mpirun -n 2 --allow-run-as-root ../build/nextsim --config-file=config_column.cfg --model.partition_file=partition_metadata_2.nc --model.init_file=init_column_30x30.nc 

You can also run it with just one MPI partition by using --model.partition_file=partition.nc in the above line. If you want to change the number of rows and columns, you must edit the partition.cdl file, replacing NX, NY, domain_extent_x, domain_extent_y with the desired dimensions. Then recreate partition.nc with

ncgen -b partition.cdl

The result of the 30x30 run is spatially uniform, but ice thickness and temperatures vary with time, e.g.

[andreapiacentini]

Thanks, I can perfectly reproduce your run.
It's a good starting point.
It crashes on a

HDF5-DIAG: Error detected in HDF5 (1.10.4) thread 0:
  #000: H5F.c line 549 in H5Fflush(): invalid file identifier
    major: Invalid arguments to routine
    minor: Inappropriate type
terminate called after throwing an instance of 'netCDF::exceptions::NcHdfErr'

and totalview places it in

Talk to you soon