This collection of functions and jupyter notebooks aims at making
the analysis of PISM ensembles more convenient. Main functions are
in pism_ens_analysis/pism_ens_analysis.py.
Have a look to the example folder, try out the jupyter notebooks.
Start with get_ensemble_indicators.ipynb.
For more analyses, look at indicator_dependencies.ipynb.
You would like to fire up a notebook on the machine that hosts the PISM files. On your local machine, this would just be:
jupyter notebook
For running jupyter on the cluster, you need an ssh tunnel to connect your browser with jupyter running on the cluster:
from your local shell:
ssh -L localhost:choose_a_port1:localhost:choose_a_port2 username@cluster2015.pik-potsdam.de
And then, on the cluster:
jupyter notebook --no-browser --port=choose_a_port2
Open your browser and go to
localhost:choose_a_port1
Choose_a_port2 should be chosen randomly for different users, so that they do not interfere. There are probably restrictions to the ports that can be used, staying between 7000 and 9000 is probably safe. For a long list, see https://en.wikipedia.org/wiki/List_of_TCP_and_UDP_port_numbers
This code is licensed under GPLv3, see the LICENSE.txt. See the commit history for authors.