Parallel Programming in Python

A two-day, intermediate course on making scientific Python run efficiently on many cores: recognising which problems can be parallelised, measuring performance, then working hands-on with Dask arrays, Numba, delayed evaluation, parallel design patterns and asyncio. Run by the Netherlands eScience Center's Digital Skills Programme in the Carpentries style, with learners coding along as instructors write on screen. The emphasis is on how the methods compare and when to use each. 9:30 to 17:00 on 13 and 14 October; lunch and closing drinks included. €50 for people affiliated with Dutch research institutions, €200 for industry; master's students are not accepted. You need basic Python, NumPy and Jupyter. Late cancellations incur a no-show fee. At Science Park 402. Why go: two focused days from the people who write research software for a living, at a fraction of commercial training prices. For researchers, research software engineers and data scientists whose code has outgrown one core.