Topic: Is R Slow?

April 19, 2021 from 1pm – 2pm

Leveraging Hardware Resources to Solve Large Scale Problems in the Social Sciences

Join us for a discussion with Shane Sanders and Justin Ehrlich from the Falk College of Sport & Human Dynamics. With expertise in behavioral sports economics, game theory and data computation, visualization and analysis, Sanders and Ehrlich will discuss some of their work in the R programming language that required extra resources beyond desktop machines.

In a computational paper on city quality rankings (Ehrlich, Medcalfe, Sanders; Social Indicators Research, forthcoming), we evaluate the combinatorial and empirical incidence of social choice rank anomalies or aggregation paradoxes in EWB City Ranking Data. Doing so allows us to assess the level of ambiguity in a set of rankings. Given the large sample space and low expected probability of rank anomaly incidence, the problem presents computational challenges that are not easily solved by simulation or brute force method. Instead, a hybrid approach of finding sample space symmetry and utilizing hardware parallelism allows us to search or infer every possible anomaly in the sample space.

While R, utilizing OpenBLAS, is normally slow at such problems, Intel’s Math Kernel Library utilizes parallelism, which greatly speeds up the computation.

h Shane Sanders and Justin Ehrlich from the Falk College of Sport & Human Dynamics

Research Computing Resource Overview

April 28, 2021 from 12:30pm – 1:30pm

Whether you know exactly what you need or you only know that your current resources aren’t enough, join us for an overview of research computing resources available on campus and how to get started.

Register Today!

Spring 2021 Research Computing Series Registration