Information Technology Reference
In-Depth Information
We compared our approach in accuracy and performance to the original single-
threaded Discrete Firefly algorithm. Our comparative study between the CPU
and GPU implementations shows that, in general terms, both yield numerical
results closer to the overall and even in some instances reach it. The GPU-
DFA obtains lower times than CPU implementation in the large instances. In
this sense, our GPU implementation produced significantly better optimization
results with significantly less time than CPU model, which in our experiments
yielded a gain time between 1.02 and 21.85.
Besides, we have found that the GPU-DFA model provides a robust parallel
model that would allows to solve instances of different sizes without a great
degradation in the quality of the solutions.
In the future we will explore the expansion by hybridising this technique with
others that can guide the search. Analysing the behaviour of several population
sizes or the use of different natural inspired operators will also be part of future
work. Besides, it would be interesting to evaluate the specific contribution over
other kinds of problem domains or real scenarios to test the feasibility of using
this type of technique.
Acknowledgments.
Authors acknowledge funds from the ANPCyT for
Grant PICT 2011-0639. Dra. Ana C. Olivera gratefully thanks CONICET
References
1. Applegate, D., Bixby, B., Chvatal, V., Cook, B.: The Traveling Salesman Problem:
A Computational Study. Princeton University Press (2007)
2. Baykasoglu, A., Ozsoydan, F.B.: An improved firefly algorithm for solving dynamic
multidimensional knapsack problems. Expert Syst. Appl. 41(8), 3712-3725 (2014)
3. Bojic, I., Podobnik, V., Ljubi, I., Jezic, G., Kusek, M.: A self-optimizing mobile
network: Auto-tuning the network with firefly-synchronized agents. Information
Sciences 182(1), 77-92 (2012)
4. Cano, A., Olmo, J.L., Ventura, S.: Parallel multi-objective ant programming for
classification using GPUs. J. Parallel Distr. Com. 73(6), 713-728 (2013)
5. Chandrasekaran, K., Simon, S.P.: Network and reliability constrained unit com-
mitment problem using binary real coded firefly algorithm. International Journal
of Electrical Power & Energy Systems 43(1), 921-932 (2012)
6. Delevacq, A., Delisle, P., Gravel, M., Krajecki, M.: Parallel Ant Colony Optimiza-
tion on Graphics Processing Units. Journal of Parallel and Distributed Comput-
ing 73(1), 52-61 (2013), metaheuristics on GPUs
7. Donald, D. (ed.): Traveling Salesman Problem, Theory and Applications (2011)
8. Farhoodnea, M., Mohamed, A., Shareef, H., Zayandehroodi, H.: Optimum
placement of active power conditioners by a dynamic discrete firefly algorithm to
mitigate the negative power quality effects of renewable energy-based generators.
International Journal of Electrical Power & Energy Systems 61, 305-317 (2014)
9. Fister, I.: Jr., I.F., Yang, X.S., Brest, J.: A comprehensive review of firefly algo-
rithms. CoRR abs/1312.6609 (2013)
