ARPN Journal of Science and Technology Logo

ARPN Journal of Science and Technology >> Volume 7, Issue 1, January 2017

ARPN Journal of Science and Technology


Matrix Multiplication by (Map Reduce) Segmentation

Full Text Pdf Pdf
Author Egbenimi Beredugo Eskca, Eltaeib Tarik
ISSN 2225-7217
On Pages 785-788
Volume No. 4
Issue No. 12
Issue Date January 01, 2015
Publishing Date January 01, 2015
Keywords Segmentation, Matrix, Map reduce, Second square, Computational-mapping, Hama



Abstract

This paper was produced as a result of my interest in metric multiplication of high ordered matrix. Following several attempts to simplify the task of matrix multiplication, there abound a plethora of information pertaining to various methods of achieving, varying degree of simplicity and accuracy in the multiplication of matrices of high order. My research primarily focused on the need to split large matrices into smaller segment, multiply the smaller segments in a systematic manner and reassembling the results to form the product of what would be the result of multiplying the large matrices. I have envisioned my approach as a new approach to the common map Reduce approach for metric multiplication. The result of my research and analysis, and the consequent methodology, is a fast, and practical matrix multiplication system that is applicable to parallel and distributed computing environment. It is my hope that the methods included in this article will greatly improve the speed of multiplying large matrices, as well as reduce the accompanying computational complexity required to implement the matrix multiplication system. Initial experimentation with a simple C++ simulation produced very impressive results. This article is mainly meant to expose the logic, algorithm and the mathematically deduction that serve as the foundation of the proposed system without diminishing the broader applicability of the techniques outlined in the paper.


Back
Seperator
    Journal of Computing | Journal of Systems and Software     
Copyrights
2014 ARPN Publishers