Content

This course provides an introduction to successfully running R code on HPC clusters, focusing on language properties that affect performance and offering an overview of different approaches to parallelizing R code. 

In the first part, we establish good practices to avoid some very common performance bottlenecks, and we will also provide hints on how to run R code in the context of a batch scheduling system. 

The second part aims to provide an overview of several commonly used R parallelization packages, their applicability depending on the problem at hand, and minimalist working examples for later reference.

Participation

Intended audience: 
Scientists who work with the R programming language and have recently started (or are about to start) running their R code on an HPC cluster; as well as veteran HPC and R users who want tips on how to improve performance of their R code and on utilizing parallelization packages.

Participation requirements:

• A working knowledge of the R programming language is required. We will focus on specific programming constructs and will not provide a general programming introduction.
• A working R installation (4.1 or newer) is required to run the exercises. Most HPC clusters (see below) should have a fitting R module available for use. The installation of all necessary packages for the exercises will be done during the hands-on part.
• Basic knowledge of working with the Linux command-line interface is strongly recommended.
• Some of the exercises (batch submission, Rmpi) require having access to an HPC cluster that runs the Slurm scheduler.
• The course will be held online via Zoom. Contact details will be made available after successful registration.

Note: The workshop is limited to 50 participants. It is free of charge and will be held in English.

Registration opens in mid Novermber