Parameter Sweep Experiments in Hybrid Computing Systems with R Language
Alexander Rumyantsev1, Oleg Sukhoroslov2, Anna Eparskaya3, Enrico Blanzieri4, Valter Cavecchia5

1Alexander Rumyantsev, Institute of Applied Mathematical Research, Karelian Research Centre of RAS, Petrozavodsk, Russia, Petrozavodsk State University, Petrozavodsk, Russia.

2Oleg Sukhoroslov, Institute for Information Transmission Problems, Kharkevich Institute, RAS, Moscow, Russia.

3Anna Eparskaya, Petrozavodsk State University, Petrozavodsk, Russia.

4Enrico Blanzieri, Petrozavodsk State University, Petrozavodsk, Russia.

5Valter Cavecchia, Petrozavodsk State University, Petrozavodsk, Russia.

Manuscript received on 17 May 2019 | Revised Manuscript received on 24 May 2019 | Manuscript Published on 02 June 2019 | PP: 590-596 | Volume-8 Issue-7S2 May 2019 | Retrieval Number: G11010587S219/19©BEIESP

Open Access | Editorial and Publishing Policies | Cite | Mendeley | Indexing and Abstracting
© The Authors. Blue Eyes Intelligence Engineering and Sciences Publication (BEIESP). This is an open-access article under the CC-BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)

Abstract: High-performance and distributed computing are the backends aimed to equip researchers in addressing computationally intensive modeling and simulation problems. While high-performance systems, such as supercomputers, require specific knowledge and algorithms to be efficiently utilized, the socalled desktop grid systems and hybrid computing systems (as a combination of the former two classes) provide a relatively easy configurable and cheap solution for obtaining necessary computational power. Observing a large number of researchers using R language as a primary solution for applied studies (including modeling, simulation and statistical study in various fields), we introduce the RBOINC package for R language. The package allows to easily perform the so-called parameter sweep type of experiments in R language over BOINC infrastructure, the de-facto standard software solution for desktop grid systems. We discuss the implementation, uncover the possible use-cases and provide some directions for future work.

Keywords: Desktop Grid, Embarrassingly Parallel, BOINC, Parameter Sweep, Everest, Causal Relationship.
Scope of the Article: Autonomic Computing and Agent-Based Systems