Edge Coloring of Set of Graphs with The Use of Data Decomposition and Clustering

This paper presents solution to a problem of parallel edge coloring of sizable sets of cubic graphs. Since edge coloring of the graph is considered to be non-dividable operation which can be, in some cases, computed several hundreds to thousands of times, use of the parallel computing is necessary. The paper presents design and implementation of methodology based on clustering of graphs and data decomposition model. Methodology aims to satisfy time criterium of effective decomposition of problem to set of subproblems. Proposed methodology was experimentally tested on all available datasets related to group of graphs called snarks.