On a shared, remote resource like the Trillium supercomputer, running Python computation with specific packages requires some care.  The "virtual environment" approach is the most suitable for a supercomputer, but there are a few tricks to get it to work.  We will explain how these virtual environments work, and why they are better suited than other options like conda.   We will also show how to incorporate these virtual environments into the Open Ondemand web interface that is attached to the Trillium system.  Format: Virtual