A Comparative Study on Nature Inspired Algorithms with Firefly Algorithm

Full Text PDF PDF
Author(s) M. K. A. Ariyaratne | T. G. I. Fernando
Pages 611-617
Volume 4
Issue 10
Date October, 2014
Keywords Firefly Algorithm, Genetic Algorithms, Particle Swarm Optimization Algorithm, Ant Colony Systems, Travelling Salesman Problem, Evolutionary Discrete Firefly Algorithm [EDFA], Nature Inspired Algorithms


Nature inspired algorithms for their powerfulness, acquire a unique place among the algorithms for optimization. This paper intends to provide a comparison of firefly algorithm (FA) with 3 other nature inspired algorithms; genetic algorithms (GA), particle swarm optimization algorithm (PSO) and ant colony systems (ACS). Traveling salesman problem (TSP) has been used as the problem to be solved and hence, discrete versions of firefly algorithm and particle swarm optimization algorithm were used. Four sets of travelling salesman problems with different number of cities (16, 29, 51, and 100) from a popular TSP library and five sets of randomly generated TSP problems with 29 cities were used to obtain conclusions. Accuracy of the final paths given by each algorithm was compared with the so far best path provided by the TSP library. Times required by each algorithm to solve the problems were also taken in to account when making conclusions. To keep the likeness of all algorithms, individuals in a population, number of iterations in a run and number of runs were kept constant. Simulations and results indicate that the firefly algorithm is superior to the other 3 nature inspired algorithms in solving Traveling Salesman Problems. The significant features of this discrete firefly algorithm are the method of calculating distance between two fireflies and their movement strategy. Finally we will discuss some suggestions for further research.

< Back to October Issue