The University of Nebraska School of Computing (UNL SoC) is using the JupyterHub https://jupyterhub.org framework for instruction. Our installation of the framework is accessed by directing your browser to https://jhub.unl.edu, and providing your CSE login credentials.

The JupyterHub installation supported by UNL SoC currently provides interactive Python programming capability through a browser interface. Source files created within the interface are saved within the notebook directory under your CSE server home directory, typically /home/ugrad/youruserid/notebook. This allows uploading to your notebook directory from your laptop or lab computer by means of FTP or copying to your mounted CSE home directory. How to transfer via FTP and mount your CSE home directory are topics covered in our FAQ: https://computing.unl.edu/faq in the applicable section for your laptop OS.

As a user of the CSE server, disk space is a shared resource and should be conserved when possible. If you have specific needs for Python modules that would require large amounts of disk space, 20MB or more, it may be beneficial to have these installed by the SoC system administrators instead of installing them locally. Local install of Python module library is typically something you would do for small module libraries or highly specialized ones that others would have no need for. Installing Python module libraries locally also counts against your disk quota, reducing the amount of coursework source and data files you can save on the server.

JupyterHub has it’s own file type associated with Python files, .ipynb so Python source, .py files, copied into your notebook directory can’t be executed directly. The contents of the .py file needs to be included into a .ipynb file before it can be executed (run) within the browser. Your .py files can be directly excuted from the Terminal application within the iPython excution environment provided through JuypterHub.

The JupyterHub terminal interface is sensitive to the type of shell you have assigned. To use the JupyterHub Terminal application, you must use /bin/bash as your default shell. Your shell can be changed using the CSE Account Management utilty provided at https://cse.unl.edu/accounts. Changing your shell, using the Account Settings link, prior to your first JupyterHub login is highly recommended.

Due to the lack of a GPU resource, the JupyterHub hosting server (csce.unl.edu) uses CPU-based TensorFlow libraries. Using CPU-based python libraries for data science research can take enormous memory and CPU resources. Limitations on computating resources are applied to all users on the JupyterHub. These limitations can preclude certain large experiments. If you need greater resources than JupyterHub can provide, you should contact the Holland Computing Center, https://hcc.unl.edu to see if you can obtain an account on their supercomputing servers. You can monitor your experiments resource usage on the hosting server (csce) using the top command to see how much resources your experiment uses while executing with your JupyterHub Notebook. Obviously this requires having an active PuTTY or ssh session into the csce.unl.edu server to execute the top command prior to starting your Notebook ipynb run.