International Journal of Emerging Trends & Technology in Computer Science (IJETTCS)

Web Site: www.ijettcs.org Email: [email protected]

Volume 5, Issue 4, July - August 2016

ISSN 2278-6856

A comparative study of continuous multimodal

global optimization using PSO, FA and Bat
Algorithms
Radha A. Pimpale1, P.K. Butey2
1

PhD Scholar, Rashtrasant Tukdoji Maharaj Nagpur University, Nagpur, Maharashtra India
2

Associate Professor and Head, Kamla Nehru Mahavidhalaya,

Rashtrasant Tukdoji Maharaj Nagpur University, Nagpur, Maharashtra India

Abstract
This paper introduces the nature-inspired metaheuristic
algorithms for optimization of standard Benchmark function,
including Firefly algorithm, PSO algorithms and Bat
algorithm. We have implemented these algorithms in
MATLAB. We have considered here how this algorithms work
on continuous multimodal global optimization benchmark
function. All these are the evolutionary nature inspired
optimization metaheuristic algorithms and are inspired by the
nature. to find out optimal solutions of
continuous,
multimodel Benchmark function. The stimulation results of
this experiment were analyzed and compared to the best
solutions found and compare with time. So the Bat algorithm
in each continuous, multimodel
Benchmark function
optimization function seems to perform better and efficient.

Keywords:-Firefly algorithm, Metaheuristic algorithm,

PSO, fireflies, bat.

1.Introduction
The goal of global optimization is to find the best possible
possible solution from the given
constraints. The
objective function is minimize or maximize and the
constraints are the relationships [10, 11, 13].
Optimization algorithms can be divided in two basic
classes: deterministic and stochastic (or probabilistic).
Deterministic algorithms are almost all local search
algorithms, and they are quite efficient and aim in finding
the local optima. Stochastic (or probabilistic) algorithms
are optimization methods that generate and use random
variables and have ability to local as well as global
solution. There are three Metaheuristic algorithms like
PSO, Firefly algorithm, Bat algorithms having aim to find
the optimal solution.
Optimization
algorithms
have exploration
and
exploitation problem. In the optimization term,
exploration means visiting new areas of the search space
which have not been investigated before. Exploration is a
procedure by which we try to find novel and better
solution states. Parameters such as convergence, elapse
time, stability of algorithms are considered

Volume 5, Issue 4, July August 2016

Particle swarm optimization (PSO)

Particle swarm optimization (PSO) is a population based
stochastic optimization technique developed by Dr.
Eberhart and Dr. Kennedy in 1995, inspired by social
behavior of bird flocking or fish schooling which aims at
finding the food [1]. The particle swarm optimization
concept consists of, at each time step, changing the
velocity of (accelerating) each particle toward its pbest
and lbest locations. Each swarm moves in the search space
in a cooperative search procedure. Acceleration is
weighted by a random term, with separate random
numbers being generated for acceleration toward pbest
and lbest locations. The group of possible solutions is a set
of particles, called swarms.
PSO is initialized with a group of random particles
(solutions) and then searches for optima by updating
generations. In every iteration, each particle is updated by
following two "best" values. The first one is the best
solution (fitness) it has achieved so far. (The fitness value
is also stored). This value is called pbest. Another "best"
value that is tracked by the particle swarm optimizer is the
best value, obtained so far by any particle in the
population. This best value is a global best and called
gbest. When a particle takes part of the population as its
topological neighbours, the best value is a local best and is
called lbest.
After finding the two best values, the particle updates its
velocity and positions with the PSO. The PSO has
algorithmic parameters:(a) Vmax or maximum velocity
which restricts Vi(t) within the interval [Vmax, Vmax]
(b) An inertial weight factor (c) Two uniformly
distributed random numbers 1 and 2 respectively that
determine the influence of p(t) and g(t) on the velocity
update formula.
(d) Two constant multiplier terms C1
and C2 known as self-confidence and swarm
confidence, respectively following equation 1(a) and 1(b).
v t+1 = v t + c 1 r 1 (g x t ) c 2 r 2 (p x t )

1(a)

Page 137

International Journal of Emerging Trends & Technology in Computer Science (IJETTCS)

Web Site: www.ijettcs.org Email: [email protected]
Volume 5, Issue 4, July - August 2016
x t+1 = x t + v t+1

1(b)

PSO Algorithm
For each particle
Initialize particle
END
Do
For each particle
Calculate fitness value
If the fitness value is better than the best fitness value
(pbest) in history set current value as the new pbest
End
Choose the particle with the best fitness value of all the
particles as the gbest
For each particle
Calculate particle velocity according equation (a)
Update particle position according equation (b)
End
While maximum iterations or minimum error criteria is
not attained Particles' velocities on each dimension are
clamped to a maximum velocity Vmax. If the sum of
accelerations would cause the velocity on that dimension
to exceed Vmax, which is a parameter specified by the
user. Then the velocity on that dimension is limited to
Vmax.
Firfly Algorithm
Firefly algorithm is a population based algorithms, The
Firefly algorithm was developed by Xin-She [Yang XS
(2008) Nature-Inspired Metaheuristic Algorithms and it is
based on idealized behaviour of the flashing
characteristics of fireflies. For simplicity, we can
summarize these flashing characteristics as the following
three rules [2]:
a) All fireflies are unisex, so that one firefly is
attracted to other fireflies regardless of their sex.
b) Attractiveness is proportional to their brightness,
thus for any two flashing fireflies, the less bright one will
move towards the brighter one. The attractiveness is
proportional to the brightness and they both decrease as
their distance increases. If no one is brighter than a
particular firefly, it will move randomly.
c) The brightness of a firefly is affected or determined by
the landscape of the objective function to be optimized
For simplicity we can assume that the attractiveness of a
firefly is determined by its brightness or light intensity
which in turn is associated with the encoded objective
function. In the simplest case for an optimization problem,
the brightness I of a firefly at a particular position X can
be chosen as
. However the attractiveness
is relative, it should vary with the distance rij between
firefly i and firefly j. As light intensity decreases with the
distance from its source and light is also absorbed in the

Volume 5, Issue 4, July August 2016

ISSN 2278-6856

media, so we should allow the attractiveness to vary with

degree of absorption [3].
Step 1: Attractiveness and light intensity
In the firefly algorithm, there are two important issues: the
variation of the light intensity and the formulation of the
attractiveness. We know, the light intensity I(r) varies
with distance r monotonically and exponentially, that is:
(2)
Where I0 the original light intensity and is the light
absorption coefficient. As firefly attractiveness is
proportional to the light intensity seen by adjacent
fireflies, we can now define the attractiveness of a firefly
by Equation (2)
(3)
Where r is the distance between each two fireflies and 0
is their attractiveness at r = 0 i.e., when two fireflies are
found at the same point of search space. The value of
determines the variation of attractiveness with increasing
distance from communicated firefly.
Step 2: Distance
The distance between any two fireflies i and j at xi and xj
respectively, the Cartesian distance is determined by
equation (3) where xi,k is the kth component of the spatial
coordinate of xi the ith firefly and n is the number of
dimensions.

(4)
Step 3: Movement
The firefly i movement is attracted to another more
attractive (brighter) firefly j is determined by:

( 5)
Where the second term is due to the attraction while the
third term is randomization with being the
randomization parameter and r and is a random number
generator uniformly distributed in [0, 1].
Why firefly algorithms is special [4]
1.The first response that can be made regarding this
question is the moment strategy of the fireflies. Eqn (1) is
the moment equation and is consist of 2 main parts. We
can mention them as information based movement and
random moment. These two parts are responsible for
providing proper exploitation and the exploration over the
search space for finding optimal solutions.
Page 138

International Journal of Emerging Trends & Technology in Computer Science (IJETTCS)

Web Site: www.ijettcs.org Email: [email protected]
Volume 5, Issue 4, July - August 2016
2.The given fireflys brightness and its distance with
another brighter firefly are taken into the account when
modifying its solution. Random part will create some
random solution which can lead the solution to a good or
bad situation.
Fireflies Algorithms [5]
initialize n fireflies to random positions
loop max Epochs times
for i := 0 to n-1
for j := 0 to n-1
if intensity(i) < intensity(j)
compute attractiveness
move firefly(i) toward firefly(j)
update firefly(i) intensity
end for
end for
sort fireflies
end loop
return best position found
Bat Algorithms
The bat algorithm is a metaheuristic algorithm for global
optimization. It was inspired by the echolocation behavior
of microbats, with varying pulse rates of emission and
loudness. [6] [7]
The following approximate or idealized rules were used.
1. All bats use echolocation to sense distance and they also
know the difference between food/prey and background
barriers in some magical way;
2. Bats fly randomly with velocity vi at position xi with a
frequency fmin , varying wavelength and loudness A0 to
search for prey. They can automatically adjust the
wavelength (or frequency) of their emitted pulses and
adjust the rate of pulse emission r [0,1], depending on
the proximity of their target;
3. Although the loudness can vary in many ways, we
assume that the loudness varies from a large (positive) A0
to a minimum constant value Amin how their positions xi
and velocities vi in a d-dimensional search space are
updated. The new solutions xtiand velocities vti at time
step t are given by
fi = fmin + (fmax+fmin),
(6)
vit + 1 = vti + (xitx0)fi,
(7)
xit+1=xit+vit
(8)
Initially, each bat is randomly assigned a frequency which
is drawn uniformly from [fmin, fmax]. For the local
search part, once a solution is selected among the current
best solutions, a new solution for each bat is generated
locally using random walk
Where is a random number vector drawn from [1,1],
while At = <At> is the average loudness of all the bats at
this time step. T

Volume 5, Issue 4, July August 2016

ISSN 2278-6856

Loudness and Pulse Emission

The loudness Ai and the rate ri of pulse emission have to
be updated accordingly as the iterations proceed. As the
loudness usually decreases once a bat h as found its prey,
while the rate of pulse emission increases, the loudness
can be chosen as any value of convenience. For example,
we can use A0= 100 and Amin= 1. For simplicity, we can
also use A0 = 1 and Amin = 0, assuming Amin = 0 means
that a bat has just found the prey and temporarily stop
emitting any sound. Now we have
At+1i = Ati, rt i= r0i[1exp(t)]

(10)

where and are constants. In fact, is similar to the

cooling factor of a cooling schedule in the simulated
annealing (Kirkpatrick et al., 1983). For any 0 < < 1
and
>0,
Ati0, rtir0i, as t
Bat Algorithms [12]
Objective function
f(x), x = (x1, ..., xd)T
Initialize the bat population
xi (i= 1,2, ..., n) and vi
Define pulse frequency fi at xi
Initialize pulse rates ri and the loudness Ai
while (t < Max number of iterations)
Generate new solutions by adjusting frequency, and
updating velocities and locations/solutions [equations
(9) and (10)]
if (rand> ri)
Select a solution among the best solutions
Generate a local solution around the selected best
solution
end if
Generate a new solution by flying randomly
if (rand < Ai & f(xi)< f(xi+1))
Accept the new solutions
Increase ri and reduce Ai
end if
Rank the bats and find the current best xi
end while
RESULTS AND DISCUSSION
Consider unconstrained global optimization benchmark
function in order to demonstrate how the firefly, PSO and
Bat algorithm works, we have implemented it in matlab
and the results are shown below. In order to show that
both the global optima and local optima can be found
simultaneously, Simulations were carried out by assigning
equal size of population to be 100,300 and 500 for the all
the algorithms. All the mentioned algorithms are run with
1000 iterations,, then the results are recorded. The
stopping criteria is set to maximum number of iterations.
For fair analysis and comparison all the algorithms are

Page 139

International Journal of Emerging Trends & Technology in Computer Science (IJETTCS)

Web Site: www.ijettcs.org Email: [email protected]
Volume 5, Issue 4, July - August 2016
well tested on standard unconstrained global optimization
benchmark function.
In particle swarm optimization parameters are considered
as Maximum Number of Iterations are 1000,population
size of particle are considered as 100 and 300 and 500,
Inertia Weight w=1, Inertia Weight Damping Ratio
wdamp=0.99 and Personal Learning Coefficient
c1=1.5,and
Global
Learning
Coefficient=2.0.The
dimensions of the problems are set to be 30, for every
population the results are recorded and presented in two
different ways viz. a) best solution i.e global minimum b)
elapsed time
In Firefly Algorithms optimization parameter Maximum
Number of Iterations are 1000, population size of fireflies
are considered as 100 and 300 and 500,
alpha=0.5,
beta=0.1 and gamma =1 , The dimensions of the problems
are set to be 30, for every population the results are
recorded and presented in two different ways viz. a) best
solution i.e global minimum b) elapsed time
In Bat Algorithms parameter considered as Maximum
Number of Iterations are 1000, population size of fireflies
are considered as 100 and 300 and 500, Loudness =0.5
(constant or decreasing), Pulse rate (constant or
decreasing)=0.5 , Lb=-2*ones(1,d),Ub=2*ones(1,d); The
dimensions of the problems are set to be 30, for every
population the results are recorded and presented in two
different ways viz. a) best solution i.e global minimum b)
elapsed time
Table 1 Comparative performance of Algorithms
(Population=100)
Benchmark
function
Sphere Function

PSO

FA

Bat

4.3267
(34.1273)

7.136
(24.94)

Ackley function

8.881
(22.8763)

Rastrigin function

0
(31.4657)

Griewank
function

0.007396
(30.4459)

1.1878
(63.5433
)
8.8818
(18.1611
)
0
(36.9632
)
0

Rosenbrock
Function

0.2898
(47.1184)

Schwefel
Function

798.5128
(32.1073)

(0.03925
)
1.5731
(62.3811
)
801.3398
(31.4254
)

Volume 5, Issue 4, July August 2016

0.00216
(37.69)
57.7098
(42.49)
5.942
(32.11)

29.40
(46.25)

ISSN 2278-6856

Table 2 Comparative performance of Algorithms

(Population=300)
Benchmark
function
sphere Function

PSO

FA

Bat

Ackley function

0
(110.3773)
8.8818e-16

Rastrigin
function
Griewank
function
Rosenbrock
Function
Schwefel
Function

0
(99.24.2)
0
(95.975335)
4.0364
(122.3400)
798.5128
(79.837988)

1.1671
(556.952)
8.8818(451.71
7)
0
(386.0997)
0
(0.3090)
1.2441
(560.4124)
808.3921
(279.2434)

4.7428
(30.10)
0.002579
(38.95)
43.7803
(43.35)
5.942
(32.11)
4.4691
(46.74)
5.7732
(49.64)

Table 3 Comparative performance of Algorithms

(Population=500)
Benchmark
function
Sphere Function

PSO

FA

Bat

0(136.2281)

Ackley function

8.8818(175.3
049)
0(151.844925
)
0.007396(30.
445990)
0.0729(222.7
411)
798.5128(167
.4429)

1.0810
(1560.2457)
8.8818
(451.7176)
0
(1319.7696)
0
(0.8410)
1.4270
(1533.43)
808.3333
(1557.870)

2.5845
(30.67)
0.002277
(41.56)
45.7697
(44.04)
5.0217
(34.77)
0.6223
(47.38)
2.8866
(49.66)

Rastrigin function
Griewank function
Rosenbrock
Function
Schwefel Function

Performance of Multimodel function with given

algorithm
Ackleys function has one global optimum and many
minor local optima. It is better for the Bat algorithm
among the six standard benchmark function as its local
optima are shallow. It has been observed that Ackley
function has best performance than other function.
Griewanks function is that it is good for higher
population for best cost . Rastrigins function is a complex
multimodal problem with a large number of local optima.
When attempting to solve Rastrigins function, algorithms
may easily fall into a local optimum. Sphere function is
differentiated in every algorithum. Firefly algorithms
reaches towards optima in Sphere Benchmark function.
Schwefels function is due to its deep local optima being
far from the global optimum

3.2196
(48.50)

Page 140

International Journal of Emerging Trends & Technology in Computer Science (IJETTCS)

Web Site: www.ijettcs.org Email: [email protected]
Volume 5, Issue 4, July - August 2016

Comparison Graph for Time required to run Different

Algorithms
Comparing the coverage and the execution speed , among
these six multimodal, n dimensional global optimization
benchmark function with PSO algorithms, fireflies and
Bat Algorithms it has been observed that Bat algorithms is
better for convergence and execution speed.

ISSN 2278-6856

[5]. ztok Fister Jr.a, MatjazPercb,c,d, Salahuddin M.

Kamalc, Iztok Fister ,A review of chaos-based firefly
algorithms: Perspectives and research challenges, .
Fister Jr. et al./Applied Mathematics and
Computation 252 (2015) 155165
[6]. J. D. Altringham, Bats: Biology and Behaviour,
Oxford University Press, (1996).
[7]. P. Richardson, Bats. Natural History Museum,
London, (2008), Yang, X. S. (2010). "A New
Metaheuristic Bat-Inspired Algorithm, in: Nature
Inspired Cooperative Strategies for Optimization
(NISCO
2010)". Studies
in
Computational
Intelligence 284: 6574. arXiv:1004.417
[8]. .X. S., Yang, A New Metaheuristic Bat -Inspired
Algorithm, in: Nature Inspired Cooperative Strategies
for Optimization (NISCO 2010) (Eds. Cruz, C.;
Gonzalez, J. R.; Pelta, D. A.; Terrazas, G), Studies in
Computational Intelligence Vol. 284, Springer Berlin,
pp. 65 74, 2010.
[9]. [9] P. Agarwal and S. Mehta, Nature-Inspired
Algorithms: State-of-Art, Problems and Prospects,
International Joournal of Computer Applications,
2014, Vol. 100, No.14, pp. 14-21.
[10]. I. Fister, X. S. Yang, J. Brest, an d D. Fister, A
Brief Review of Nature-Inspired Algorithms for
Optimization, ElektrotehniSki Vestnik, Jul, 2013.
[11]. WengKee Wong, Nature-Inspired Metaheuristic
Algorithms for Generating Optimal Experimental
Designs, 2011

Fig 2 Comparison of Time required to run Different

Algorithms
References
[1]. Kennedy, J. and Eberhart, R. Particle swarm
optimization. In Proceedings of the IEEE
International Conference on Neural Networks , Vol.
4. 1995, 19421948
[2]. Yang XS (2008) Nature-Inspired Metaheuristic
Algorithms. Luniver Press.
[3]. Yang, X. S., (2011). Bat algorithm for multi-objective
optimisation,Int. J. Bio-Inspired Computation, Vol. 3,
No. 5, pp. 267274
[4]. MKA Ariyaratne, WPJ Pemarathne, A Review of
Recent Advancements of Firefly Algorithm; A
Modern Nature Inspired Algorithm , Proceedings of
8 th International Research Conference, KDU,
Published November 2015

Volume 5, Issue 4, July August 2016

[12]. Yazan A. Alsariera, Hammoudeh S. Alamri,

Abdullah M. Nasser, Mazlina A. Majid, Kamal Z.
Zamli, Comparative Performance Analysis of Bat
Algorithm and Bacterial Foraging Optimization
Algorithm using Standard Benchmark Functions,
2014 8th Malaysian Software Engineering
Conference (MySEC).
[13]. S K Subhani Shareef E. Rasul Mohideen Layak Ali
,Directed Firefly algorithm for multimodal
problems, 2015 IEEE International Conference on
Computational Intelligence and Computing Research
AUTHORS
Prof. Radha A. Pimpale is
presently working as Assistant
Professor at
Priyadarshini
Institute of
Engineering and
Technology Nagpur. She has
11 research papers published in
reputed
national
and
international
journals. She is
research scholar at Rashtrasant Tukadoji Maharaj Nagpur
University, Nagpur. She perusing Ph. D in under the
Page 141

International Journal of Emerging Trends & Technology in Computer Science (IJETTCS)

Web Site: www.ijettcs.org Email: [email protected]
Volume 5, Issue 4, July - August 2016

ISSN 2278-6856

guidance of Dr P.K. Butey. She has total experience of 15

years in teaching and research in the field of Computer
Science
Dr. Pradeep Butey is resently
working as associate professor and
Head of the Department of
Computer Science, Kamla Nehru
Mahavidyalaya, Nagpur, He has
done his doctorate in Computer
Science from RTM
Nagpur
University, Nagpur, in Mar 2011.
His areas of interest include data
mining, fuzzy logic, optimization, RDBMS. He has 18
Research papers published in reputed International
Journals. He is Chairman Board of Studies in Computer
Science of the University, He has vast experience of 12
years in the field of teaching and research in Computer
Science. He has worked as a member of selection
committee at many instances. He is a Member of
Academic Council RTM Nagpur University.

Volume 5, Issue 4, July August 2016

Page 142