Discipline: Computer Sciences and Information Management
Subcategory: Computer Science & Information Systems
Levie C. McGee - Philander Smith College
In this research, the performance of a variety of programming languages when considering the task of probability distribution computation is investigated. Computing exact probability distributions is a computational intensive task. Various programming languages such as Matlab and Fortran have been utilized in an attempt to minimize the computational burden of executing this task for certain scenarios; however, to date, there does not appear to be any formal comparison of programming languages for the completion of this task. Researchers tend to rely on the recommendation of other researchers, which is often biased towards the language with which they have the most familiarity or comfort. Recently, methodology has been established to permit the computation of the exact distribution of the multiple window scan statistic for multi-state, higher-order Markovian sequences. This methodology is given in the literature with a complete distribution of the algorithm and pseudo code, which makes it easily accessible for the kind of investigation we are interested in. In the spirit of Aruoba et. al., we conduct a comparison of speed and accessibility to novice programmers for this algorithm, using Fortran and C++. We share our findings about the conditions under which each program performs best. In future work, we will consider a further comparison to computing the distribution using the R program.
Funder Acknowledgement(s): Arkansas Louis Stokes Alliance for Minority Participation Program funded by NSF award number HRD-1304121 and the Arkansas Science and Technology Authority funded by NSF award number HRD-1304121 and the Arkansas Science and Technology Authority.
Faculty Advisor: Deidra A. Coleman, dcoleman@philander.edu
Role: I computed wrote a program in C++ to compare to the Fortran program. In addition, I developed subroutines in both C++ and Fortran to ease development and produced tables and graphs of my findings.