You are here: Home Introduction to parallel programming and batch computing with R
Document Actions

Introduction to parallel programming and batch computing with R

(Half day, 22 March, 14:00 – 18:00)

Bernd Bischl and Michel Lang

Faculty of Statistics, TU Dortmund, Germany



We start by introducing the main concepts of high-performance computing and parallel programming. We will then focus on some of the most used R packages for parallel programming:

* multicore

* snow and snowfall

* parallel


During the tutorial we will create insightful examples and applications and discuss some common pitfalls in parallel programming.

The topic of the the second part will be batch computing systems, because most powerful clusters are managed by job schedulers. Resource specifications, job description files and cluster management tools will be explained through brief examples, probably for a PBS/TORQUE based system.

In order to simplify and enhance the working process in such environments, we recommend and will in detail present our own packages BatchJobs and BatchExperiments. BatchJobs offers to manage an arbitrary batch system from within R: Once set up, the package lets you transparently define and submit jobs with only a few lines of code. Powerful tools derived from functional programming assist in the definition of jobs and the collection of results. BatchExperiments is tailored for the everyday task of applying algorithms with varying parameters to problems. Both packages are developed with a strong focus on reproducibility, extensibility and independence of the underlying batch system.

Participants should have at least basic knowledge of R and are encouraged to bring their Wifi-enabled laptops with a recent R installation to take part in examples.


Course fees:

Regular: 75 EUR

Student: 50 EUR


Personal tools