Using Matlab and Octave with Condor: a case study

The goal of this project is to document and improve computational statistics capabilities and resource utilization within the Department of Statistics. Specifically, we look at the commonly used software Matlab [13] and alternative approaches to its use with GNU Octave [2] and Condor [9, 12]. By providing concrete examples of the process of modifying long-running simulation code from Matlab to Octave and Condor, we hope that more people will take advantage of these resources and improve the efficiency of their research. However, the intention of this project is not to promote the virtues of Matlab or Octave, one over the other. In fact, R [10] is more suitable for general statistical computing and is similarly workable with Condor1. Rather, we want to show those bound to Matlab for what ever reason that they may use their code with Octave, after slight modification, and compute in parallel with Condor in ways that does not violate departmental policy.