I’ve been tinkering with it for a while, but I am finally sitting down to write about QuTiP, a great toolbox for doing quantum mechanics in the Python computing language. A couple things came together recently to re-ignite my interest and get me to think about QuTiP again. First was an email and later blog post by Markus Baden about how to set up QuTiP on picloud. The second was the discovery of picloud itself. Last week I was finally starting to dive in to a project to wire up five servers into a mini-cluster and then this week I learn about massively scalable computing in the cloud for mere pennies (picloud.com).
I haven’t done a lot with it yet, but I know enough to realize that I’ll have to find another role for these servers. I estimate that keeping them on all year will cost about $800 in electricity (not including the AC to cool them). For that I can have 16,000 core hours of computing power per year. My cluster would only have 20 cores total so I would have to run a full month of computing jobs (800 hours = 33 days). Picloud is such an obvious win that I don’t even have to factor in the time I would spend maintaining my own cluster.