View the Project on GitHub

Installing Software on a Mac

Tested on macOS 10, but should work on relatively recent OS X as well.

Compiler code

  1. Install xcode using the App Store
  2. run xcode-select --install from command line
  3. Install homebrew
  4. install gcc (compilers) brew install gcc.
  5. install open-mpi (multi-core processing) brew install open-mpi
  6. install hdf5 with mpi: brew install --build-from-source hdf5 --with-mpi
  7. install netcdf (file storage) brew reinstall --build-from-source netcdf

Note that for the MITgcm as released netcdf does not require hdf5 to be compiled with mpi (though the MITgcm itself requires mpi). However, I am developing a parallel-aware netcdf interface for the MITgcm, so this step is useful.


  1. Install anaconda I use python 2.7 still, but I think just about everything now works w/ python 3.5
  2. Try ipython --pylab. From that command line try something like fig,ax=plt.subplots();ax.plot(np.arange(10)).
  3. Try jupyter-notebook and open the webpage they suggest. See their website for demos and I’ll give some example python notebooks soon.
  4. install xmitgcm
    1. conda install xarray dask
    2. pip install xmitgcm


  1. Download the source code. Its a good idea to put this in a directory with the name the version you are running to avoid compatibility problems. i.e. I keep a version at ~/MITgcmc66b/MITgcm as that is the last version I downloaded. I download the tar balls as they have been tested. I always get errors from the CVS snapshots.
  2. Clone my example project
   git clone

and then follow the instructions at the project README