Professional Documents
Culture Documents
Comparativestudyofbio-Inspiredoptimizationalgorithms Andtheirapplicationtodielectricfunction Tting
Comparativestudyofbio-Inspiredoptimizationalgorithms Andtheirapplicationtodielectricfunction Tting
To cite this article: D. Liu & K. A. Michalski (2016) Comparative study of bio-inspired optimization
algorithms and their application to dielectric function fitting, Journal of Electromagnetic Waves and
Applications, 30:14, 1885-1894, DOI: 10.1080/09205071.2016.1219277
Article views: 93
1. Introduction
Bio-inspired algorithms are widely used for solving optimization problems, including non-
deterministic polynomial problems, which require exponential computation time to solve
in the worst case scenario. The earliest and most well-known bio-inspired optimization
(BIO) algorithm is the genetic algorithm (GA),[1] which generates solutions to optimization
problems using techniques inspired by natural evolution, such as inheritance, mutation,
selection, and crossover. Other well-established BIO algorithms include differential evolu-
tion (DE), which is a variant of GA,[2] particle swarm optimization (PSO) [3] and ant colony
optimization (ACO).[4] These algorithms are still widely researched and have been applied
in electronmagnetics [5–11] and other areas. [12–15] Since the beginning of this decade,
new BIO algorithms have emerged, viz. bacterial foraging (BFO),[16–27] firefly (FFA),[28–35]
cuckoo search (CSA),[36–39] krill herd (KHA),[40–47], and cuttlefish (CFA) [48–50] with the
names clearly identifying the species whose behavior inspired them. The algorithm names
and their abbreviation are listed in Table 1. A comparison of some of these algorithms was
recently published by Singh and Singh [51].
The purpose of the present paper is to present a comprehensive study of the per-
formance of both the latest and the more established BIO algorithms, when applied to
a suite of unimodal and multimodal test functions, and to examine the performance of
these algorithms when applied to a “real-life” problem of fitting a model of the dielectric
function of metals at optical wavelengths to tabulated experimental data. This work should
be a useful guide for researchers interested in applying BIO algorithms in various areas of
engineering and science.
Table 6. Average result rankings of different optimization algorithms for the benchmark functions.
f1 f2 f3 f4 f5 f6 f7 f8 Average rank
PSO 3 3 4 5 3 5 3 2 3.5
GA 5 5 5 4 2 2 4 6 4.125
BFO 7 6 3 2 7 1 1 7 4.25
FFA 4 4 1 3 5 4 5 4 3.75
CSA 1 1 2 1 1 3 2 1 1.5
KHA 6 2 6 6 4 7 7 3 5.125
CFA 2 7 7 7 6 6 6 5 5.75
are listed in Table 4 and these parameter values can be found in [40,55–58], the meaning of
each parameters in Table 4 can be found in [1,3,16,28,40,48].
1888 D. LIU AND K. A. MICHALSKI
Table 8. Cost function values achieved by the tested optimization algorithms in the dielectric function
fitting.a
Au Ag Cu
2.99E−02 7.46E−02 2.55E−02
PSO
9.41E−02±3.89E−02 1.71E−01±1.18E−01 8.44E−02±3.14E−02
7.20E−01 1.37E−01 7.11E−02
GA
3.05E+00±4.33E+00 1.62E+01±5.26E+01 2.79E+00±3.92E+00
3.39-02 6.70E−02 3.15E−02
BFO
1.24E−01±6.82E−02 2.43E−01±5.49E−02 9.37E−02±5.34E−02
3.78E−02 1.13E−01 3.98E−02
FFA
1.61E−01±7.55E−02 2.77E−01±1.84E−01 1.46E−01±7.26E−02
2.82E−02 8.09E−02 2.59E−02
CSA
4.59E−02±1.22E−02 1.09E−02±1.32E−02 4.02E−02±9.74E−02
3.75E−02 1.09E−01 3.50E−02
KHA
1.21E−01±4.02E−02 2.35E−01±9.58E−02 1.06E−01±3.27E−02
7.79E−02 1.17E−01 5.91E−02
CFA
1.97E−01±5.87E−02 3.12E−01±1.05E−01 1.77E−01±6.27E−02
a In each cell, the top line is the best result and the bottom line gives the mean ± standard deviation of all results.
Table 9. Performance ranking of the tested optimization algorithms based on the average optimization
results for the dielectric function fitting.
PSO GA BFO FFA CSA KHA CFA
Au 2 7 4 5 1 3 6
Ag 2 7 4 5 1 3 6
Cu 2 7 3 5 1 4 6
Average Rank 2 7 3.67 5 1 3.33 6
The results for different benchmark functions are listed in Table 5, where the best results
are set in boldface. We note that PSO gives the best result for f8, GA gives the best result for
f2, CFA gives the best results for f1, f3 and f4, FFA gives the best average result for f3, BFO
gives the best and best average results for f6 and f7, and CSA gives the best result for f5 and
the best average results for f1, f2, f4, f5 and f8.
Each BIO-inspired algorithm may get different performance on different type of opti-
mization application, so the general optimization performance of each algorithm can be
described by its average result rank of a set of general benchmark functions. From the
average result rankings given in Table 6, we find that the average performance of CSA is
superior to that of the other algorithms considered, and that the performance of PSO and
FFA is similar and better than that of GA, BFO, KHA, or CFA.
JOURNAL OF ELECTROMAGNETIC WAVES AND APPLICATIONS 1889
Table 10. Optimal parameter values achieved by the tested optimization algorithms in the dielectric
function fitting.
PSO GA CFA FFA CSA KHA BFO
Au 2.10E+00 1.00E+01 3.32E+00 6.85E+00 4.79E+00 6.92E+00 6.85E+00
e Ag 4.49E+00 4.03E+00 6.61E+00 4.36E+00 1.32E+00 4.85E+00 9.42E+00
Cu 4.01E+00 9.99E+00 1.00E+00 8.93E+00 6.49E+00 7.67E+00 1.00E+00
d ci
N
ε(s) = e + + , s = jω , (1)
s s − pi
i=1
where ω is the angular frequency, N is the model order, e and d are real parameters, pi
and ci are poles and residues, respectively, which are either real (for i = 1) or arise in
complex-conjugate pairs (for i ≥ 2). We will use primes and double-primes to denote the
real and imaginary parts of complex numbers, respectively, hence ε = ε + jε . Given
a set of samples {sk = jωk , ε̂(sk )}M k=1 , we wish to determine the set of real parameters
x = {e, d, c1 , p1 , c2 , c2 , p2 , p2 , . . . , cN , cN , pN , pN } that minimizes the cost (fitness) function
M
1 ε̂k − εk (x) 2 ε̂k − εk (x) 2
f (x) =
+
,
(2)
M ε̂k ε̂k
k=1
where we use the notation ε̂k = ε̂(sk ), and similarly for εk . We apply the model (1) to three
noble metals, viz. gold (Au), silver (Ag), and copper (Cu), in the optical frequency range,
where it is convenient to express ω in terms of the photon energy given in electron-volts
(eV). The experimental tabulated data are taken from [65] and the search and initiation
ranges of the parameters are based on the guidelines given in [63] and are listed in Table
7. As in the synthetic problems tests, each algorithm was run 100 times with the maximum
number of function evaluations set to 2000 and the population size set to 20, with the initial
parameter values selected randomly within the specified ranges.
We next summarize the results of fitting the PF model (1) to the experimental data for
the three noble metals using different optimization algorithms with N = 5. The optimal
cost function (2) values achieved by the tested algorithms are listed in Table 8, where the
best overall results are set in boldface. The average performance ranking of the algorithms
is given in Table 9, from which we conclude that CSA is the best, followed by PSO, KHA,
BFO, FFA, CFA, and GA. The optimum dielectric function parameters obtained by different
JOURNAL OF ELECTROMAGNETIC WAVES AND APPLICATIONS 1891
50 0
smp
0 PSO −5
GA
−10
−50 CFA
−15
ε’’
ε’
FFA
−100
CSA −20
−150 KHA
−25
BFO
−200 −30
0 1 2 3 4 5 6 7 0 1 2 3 4 5 6 7
Photon energy [eV] Photon energy [eV]
Figure 1. Plots of the dielectric function fit for Au obtained by different optimization algorithms.
0 smp 0
PSO −1
−50 GA −2
CFA −3
−100
ε’’
ε’
FFA −4
CSA −5
−150
KHA −6
−200 BFO −7
1 2 3 4 5 6 1 2 3 4 5 6
Figure 2. Plots of the dielectric function fit for Ag obtained by different optimization algorithms.
0 smp
0
PSO
−5
−50 GA
−10
CFA
−15
ε’
−100 FFA
−20
ε’’
CSA
−25
−150 KHA
−30
BFO
−35
1 2 3 4 5 6 0 1 2 3 4 5 6 7
Figure 3. Plots of the dielectric function fit for Cu obtained by different optimization algorithms.
optimization algorithms are listed in Table 10. We have used these parameters in (1) to
plot the real and imaginary parts of ε for the three noble metals. These plots are given
in in Figures 1–3, where the dielectric function fits obtained by the tested optimization
algorithms are compared with the experimental data indicated by solid lines.
function fitting problem, the best performing algorithms were CSA for Au, PSO for Cu, and
BFO for Ag, with CSA emerging as the best method on the average.
Disclosure statement
No potential conflict of interest was reported by the authors.
Funding
This work was supported by the “Fundamental Research Funds for the Central Universities” under
[grant number 2682014CX026]; the programme of China Scholarships Council [No. 201307005027];
the “National Natural Science Foundation of China” under [grant number 61271090].
ORCID
K. A. Michalski http://orcid.org/0000-0001-5753-7333
References
[1] Holland JH. Genetic algorithms. Sci. Am. 1972;267:66–72.
[2] Storn R, Price KV. Minimizing the real functions of the ICEC’96 contest by differential evolution.
In: Proceedings of the IEEE International Conference on Evolutionary Computation; May 20–22;
Nagoya, Japan; 1996. p. 842–844.
[3] Kennedy J, Eberhart R. Particle swarm optimization. Vol. 4, In: Proceedings of the 1995 IEEE
International Conference Neural Network; 1995 27 Nov–1 Dec; Perth, Australia; 1995. p. 1942–
1948.
[4] Dorigo M, Maniezzo V, Colorni A. Ant system: optimization by a colony of cooperating agents.
IEEE Trans. Syst. Man Cybern. B, Cybern. 1996;26:29–41.
[5] Song S, Murch RD. An efficient approach for optimizing frequency reconfigurable pixel antennas
using genetic algorithms. IEEE Trans. Antennas Propag. 2014;62:609–620.
[6] Michalski KA. Electromagnetic imaging of dielectric cylinders by differential evolution and
single integral equation. Progress Electromagn. Res. B. 2013;50:19–36.
[7] Qing A. Differential evolution: fundamentals and applications in electrical engineering.
Singapore: Wiley; 2009.
[8] Lin C, Qing A, Zang J. Design of wideband multilayer planar absorber using a new differential
evolution algorithm. Proc. SPIE. 2015;9521:952117-1–952117-7.
[9] Liu D, Jiang Q, Chen JX. Binary inheritance learning particle swarm optimisation and its
application in thinned antenna array synthesis with the minimum sidelobe level. IET Microwaves
Antennas Propag. 2015;9:1386–1391.
[10] Shahpari M, Thiel DV, Lewis A. An investigation into the gustafsson limit for small planar
antennas using optimization. IEEE Trans. Antennas Propag. 2014;62:950–955.
[11] Deb A, Roy JS, Gupta B. Performance comparison of differential evolution, particle swarm
optimization and genetic algorithm in the design of circularly polarized microstrip antennas.
IEEE Trans. Antennas Propag. 2014;62:3920–3928.
[12] Sun Y, Verschuur DJ. A self-adjustable input genetic algorithm for the near-surface problem in
geophysics. IEEE Trans. Evol. Comput. 2014;18:309–325.
[13] Guo SM, Yang CC, Hsu PH, et al. Improving differential evolution with successful-parent-
selecting framework. IEEE Trans. Evol. Comput. 2014;19:717–730.
[14] Hu W, Yen GG. Adaptive multiobjective particle swarm optimization based on parallel cell
coordinate system. IEEE Trans. Evol. Comput. 2015;19:1–18.
[15] Liao T, Socha K, Montes de Oca MA, et al. Ant colony optimization for mixed-variable
optimization problems. IEEE Trans. Evol. Comput. 2014;18:503–518.
JOURNAL OF ELECTROMAGNETIC WAVES AND APPLICATIONS 1893
[16] Passino KM. Biomimicry of bacterial foraging for distributed optimization and control. IEEE
Control Syst. Mag. 2002;22:52–67.
[17] Panigrahi BK, Pandi VR. Bacterial foraging optimisation: Nelder–Mead hybrid algorithm for
economic load dispatch. IET Gener. Transm. Distrib. 2008;2:556–565.
[18] Hota PK, Barisal AK, Chakrabarti R. Economic emission load dispatch through fuzzy based
bacterial foraging algorithm. Int. J. Electr. Power Energy Syst. 2010;32:794–803.
[19] Ali ES, Abd-Elazim SM. BFOA based design of PID controller for two area load frequency control
with nonlinearities. Int. J. Electr. Power Energy Syst. 2013;51:224–231.
[20] Ali ES, Abd-Elazim SM. Bacteria foraging optimization algorithm based load frequency controller
for interconnected power system. Int. J. Electr. Power Energy Syst. 2011;33:633–638.
[21] Niu B, Fan Y, Xiao H, et al. Bacterial foraging based approaches to portfolio optimization with
liquidity risk. Neurocomputing. 2012;98:90–100.
[22] Tangt WJ, Lit MS, Het S, et al. Optimal power flow with dynamic loads using bacterial foraging
algorithm. In: Proceedings of the IEEE International Conference Power Systems Techlnology;
22–26 Oct; Chongqing, China; 2006. p. 1–5.
[23] Das S, Dasgupta S, Biswas A, et al. On stability of the chemotactic dynamics in bacterial-foraging
optimization algorithm. IEEE Trans. Syst. Man Cybern. A, Syst. Humans. 2009;39:670–679.
[24] Dasgupta S, Das S, Abraham A, et al. Adaptive computational chemotaxis in bacterial foraging
optimization: an analysis. IEEE Trans. Evol. Comput. 2009;13:919–941.
[25] Kim DH, Abraham A, Cho JH. A hybrid genetic algorithm and bacterial foraging approach for
global optimization. Inf. Sci. 2007;177:3918–3937.
[26] Biswas A, Dasgupta S, Das S, et al. Synergy of PSO and bacterial foraging optimization – a
comparative study on numerical benchmarks. In: Corchado E, Corchado JM, Abraham A, editors.
Innovations in hybrid intelligent systems. Vol. 44, Advances in soft computing; Berlin: Springer;
2007. p. 255–263.
[27] Sarasiri N, Suthamno K, Sujitjorn S. Bacterial foraging-tabu search metaheuristics for
identification of nonlinear friction model. J. Appl. Math. 2012;2012. doi:10.1155/2012/238563.
[28] Yang XS. Firefly algorithm, stochastic test functions and design optimisation. Int. J. Bio-Insp.
Comput. 2010;2:78–84.
[29] Lukasik S, Żak S. Firefly algorithm for continuous constrained optimization tasks. Semantic
web. Social networks and multiagent systems. In: Nguyen NT, Kowalczyk R, Chen SM, editors.
Computational collective intelligence. Vol. 5796, Lecture notes in computer science. Berlin:
Springer-Verlag; 2009. p. 97–106.
[30] Yang XS. Multiobjective firefly algorithm for continuous optimization. Eng. Comput.
2013;29:175–184.
[31] Sayadi MK, Ramezanian R, Ghaffari-Nasab N. A discrete firefly meta-heuristic with local search
for makespan minimization in permutation flow shop scheduling problems. Int. J. Ind. Eng.
Comput. 2010;1:1–10.
[32] Khadwilard A, Chansombat S, Thepphakorn T, et al. Application of firefly algorithm and its
parameter setting for job shop scheduling. J. Ind. Technol. 2012;8:1–10.
[33] Nouri BV, Fattahi P, Ramezanian R. Hybrid firefly-simulated annealing algorithm for the flow
shop problem with learning effects and flexible maintenance activities. Int. J. Prod. Res.
2013;51:3501–3515.
[34] Marichelvam MK, Prabaharan T, Yang XS. A discrete firefly algorithm for the multi-objective
hybrid flowshop scheduling problems. IEEE Trans. Evol. Comput. 2014;18:301–305.
[35] Jati GK, Suyanto S. Evolutionary discrete firefly algorithm for travelling salesman problem. In:
Bouchachia A, editor. Adaptive and intelligent systems. Vol. 6943, Lecture notes in computer
science. Berlin: Springer-Verlag; 2011. p. 393–403.
[36] Yang XS. Nature-inspired metaheuristic algorithms. 2nd ed. Frome: Luniver Press; 2010.
[37] Yang XS, Deb S. Engineering optimisation by cuckoo search. Int. J. Math. Model. Numer. Opt.
2010;1:330–343.
[38] Yang XS, Deb S. 2009. Cuckoo search via Lévy flights. In: World Congress on Nature Biologically
Inspired Computing; Dec 9–11; Coimbatore; p. 210–214.
1894 D. LIU AND K. A. MICHALSKI
[39] Yang XS, Deb S. Cuckoo search: recent advances and applications. Neural Comput. Appl.
2014;24:169–174.
[40] Gandomi AH, Alavi AH. Krill herd: a new bio-inspired optimization algorithm. Commun.
Nonlinear Sci. Numer. Simul. 2012;17:4831–4845.
[41] Wang GG, Gandomi AH, Alavi AH. A chaotic particle-swarm krill herd algorithm for global
numerical optimization. Kybernetes. 2013;42:962–978.
[42] Wang GG, Guo L, Gandomi AH, et al. Chaotic krill herd algorithm. Inf. Sci. 2014;274:17–34.
[43] Wang GG, Gandomi AH, Alavi AH. An effective krill herd algorithm with migration operator in
biogeography-based optimization. Appl. Math. Model. 2014;38:2454–2462.
[44] Wang GG, Gandomi AH, Alavi AH. Stud krill herd algorithm. Neurocomputing. 2014;128:363–
370.
[45] Wang G, Guo L, Wang H, et al. Incorporating mutation scheme into krill herd algorithm for
global numerical optimization. Neural Comput. Appl. 2014;24:853–871.
[46] Wang GG, Guo L, Gandomi AH. Simulated annealing-based krill herd algorithm for global
optimization. Abstract Appl. Anal. 2013;2013. doi:10.1155/2013/213853.
[47] Guo L, Wang GG, Gandomi AH, et al. A new improved krill herd algorithm for global numerical
optimization. Neurocomputing. 2014;138:392–402.
[48] Eesa AS, Brifcani AMA, Orman Z. Cuttlefish algorithm – a novel bio-inspired optimization
algorithm. Int. J. Sci. Eng. Res. 2013;4:1978–1986.
[49] Eesa AS, Brifcani AMA, Orman Z. A new tool for global optimization problems – cuttlefish
algorithm. Int. J. Math. Comput. Phys. Quant. Eng. 2014;8:1167–1171.
[50] Eesa AS, Orman Z, Brifcani AMA. A novel feature-selection approach based on the cuttlefish
optimization algorithm for intrusion detection systems. Exp. Syst. Appl. 2015;42:2670–2679.
[51] Singh GP, Singh A. Comparative study of krill herd, firefly and cuckoo search algorithms for uni-
modal and multimodal optimization. Int. J. Intell. Syst. Appl. 2014;2. doi:10.18201/ijisae.31981.
[52] Liu D, Feng Q, Wang WB, et al. Synthesis of unequally spaced antenna arrays by using inheritance
learning particle swarm optimization. Progress Electromagn. Res. B. 2011;118:205–221.
[53] Liang JJ, Qin AK, Suganthan PN, et al. Comprehensive learning particle swarm optimizer for
global optimization of multimodal functions. IEEE Trans. Evol. Comput. 2006;10:281–295.
[54] The Mathworks, Inc. MATLAB 8.5, Natick (MA); 2015.
[55] Yang XS. Firefly algorithm, 2010. Available from: http://www.mathworks.com/matlabcentral/
fileexchange/29693-firefly-algorithm.
[56] Yang XS. Cuckoo search (CS) algorithm, 2010. Available from: http://www.mathworks.com/
matlabcentral/fileexchange/29809-cuckoo-search-cs-algorithm.
[57] Dash B. Bacteria foraging optimization (BFO), 2014. Available from: http://www.mathworks.
com/matlabcentral/fileexchange/45774-bacteria-foraging-optimization-bfo-.
[58] Eesa AS. Cuttlefish optimization algorithm, 2015. Available from: https://sites.google.com/site/
cuttlefishalgortihm/home/download-cfa.
[59] Shi Y, Eberhart R. A modified particle swarm optimizer. In: Proceedings of the IEEE International
Conference Evolutionary Computation; May 4–9; Anchorage (AK); 1998. p. 69–73.
[60] Taflove A, Hagness SC. Computational electrodynamics: the finite-difference time-domain
method. Norwood (MA): Artech House; 2005.
[61] Vial A. Implementation of the critical points model in the recursive convolution method for
modelling dispersive media with the finite-difference time domain method. J. Opt. A: Pure
Appl. Opt. 2007;9:745–748.
[62] Han L, Zhou D, Li K, et al. A rational-fraction dispersion model for efficient simulation of
dispersive material in FDTD method. J. Lightwave Technol. 2012;30:2216–2225.
[63] Michalski KA. On the low-order partial-fraction fitting of dielectric functions at optical
wavelengths. IEEE Trans. Antennas Propag. 2013;61:6128–6135.
[64] Grosse C. A program for the fitting of Debye, Cole-Cole, Cole-Davidson, and Havriliak-Negami
dispersions to dielectric data. J. Colloid Interface Sci. 2014;419:102–106.
[65] Johnson PB, Christy RW. Optical constants of noble metals. Phys. Rev. B. 1972;6:4370–4379.