THE COMPARISON OF GENETIC ALGORITHMS WHICH SOLVE ORIENTEERING PROBLEM USING COMPLETE AND INCOMPLETE GRAPH

The purpose of this work was to compare two forms of genetic algorithm (complete and incomplete graph version) which solves Orienteering Problem (OP). While in most papers concerning OP graph is complete and satisfies triangle inequality, in our versions such assumptions may not be satisfied. It could be more practical as transport networks are graphs which do not have to satisfy those conditions. In such cases, graphs are usually complemented with fictional edges before they can be used by classic OP solving algorithms which operate on complete graphs. This paper answers the question: Is it better (in terms of results quality and time consumption) to transform graphs to classic OP form before running algorithm (complete graph version) or to solve OP on graphs without any assumptions and changes (incomplete graph version)? The computer experiment was conducted on the real transport network in Poland and its results suggest that it is worth checking both versions of the algorithm on concrete networks.