150 Int. J. Mathematical Modelling and Numerical Optimisation, Vol. 4, No.

2, 2013

A literature survey of benchmark functions for global

optimisation problems

Momin Jamil*
Blekinge Institute of Technology
SE-37179, Karlskrona, Sweden
and
Harman International,
Cooperate Division,
Becker-Goering Str. 16,
D-76307 Karlsbad, Germany
E-mail: [email protected]
*Corresponding author

Xin-She Yang
Middlesex University,
School of Science and Technology,
Hendon Campus, London NW4 4BT, UK
E-mail: [email protected]

Abstract: Test functions are important to validate and compare the

performance of optimisation algorithms. There have been many test or
benchmark functions reported in the literature; however, there is no standard
list or set of benchmark functions. Ideally, test functions should have diverse
properties to be truly useful to test new algorithms in an unbiased way. For this
purpose, we have reviewed and compiled a rich set of 175 benchmark functions
for unconstrained optimisation problems with diverse properties in terms of
modality, separability, and valley landscape. This is by far the most complete
set of functions so far in the literature, and it can be expected that this complete
set of functions can be used for validation of new optimisation in the future.

Keywords: global optimisation; test functions; unimodal; multimodal;

separable; non-separable.

Reference to this paper should be made as follows: Jamil, M. and Yang, X-S.
(2013) ‘A literature survey of benchmark functions for global optimisation
problems’, Int. J. Mathematical Modelling and Numerical Optimisation,
Vol. 4, No. 2, pp.150–194.

Biographical notes: Momin Jamil received his BSc from the University of the
Punjab, Lahore, Pakistan in 1991, BSc in Electrical and Electronic Engineering
from Technical University of Budapest, Hungary in 1996, and Master of
Engineering from the University of Pretoria, Pretoria, South Africa in 1999.
From 2001–2005, he worked as a Development Engineer at Siemens Mobile
Phone Development Center in Ulm, Germany. From 2006–2011, he worked as

Copyright © 2013 Inderscience Enterprises Ltd.

 A literature survey of benchmark functions 151

a Development Engineer at Harman/Becker Automotive System GmbH in

Germany. Presently, he is working as Patent Engineer/Patent Risk Management
at Harman International. He is also enrolled as an industrial PhD student at
Blekinge Institute of Technology, Sweden. His research interests include radio
communication, spread spectrum and optimisation theory.

Xin-She Yang is a Reader in Modelling and Simulation at Middlesex

University. He is an Adjunct Professor at Reykjavik University, Iceland, and a
Distinguished Professor at Xi’an Polytechnic University, China. He was a
Senior Research Scientist at UK’s National Physical Laboratory. He has
authored/edited 15 books and published more than 170 papers.

1 Introduction

The test of reliability, efficiency and validation of optimisation algorithms is frequently

carried out by using a chosen set of common standard benchmarks or test functions from
the literature. The number of test functions in most papers varied from a few to about two
dozens. Ideally, the test functions used should be diverse and unbiased, however, there is
no agreed set of test functions in the literature. Therefore, the major aim of this paper is
to review and compile the most complete set of test functions that we can find from all
the available literature so that they can be used for future validation and comparison of
optimisation algorithms.
For any new optimisation, it is essential to validate its performance and compare
with other existing algorithms over a good set of test functions. A common practice
followed by many researches is to compare different algorithms on a large test set,
especially when the test involves function optimisation (Gordon and Whitley, 1993;
Whitley et al., 1996). However, it must be noted that effectiveness of one algorithm
against others simply cannot be measured by the problems that it solves if the the set of
problems are too specialised and without diverse properties. Therefore, in order to
evaluate an algorithm, one must identify the kind of problems where it performs better
compared to others. This helps in characterising the type of problems for which an
algorithm is suitable. This is only possible if the test suite is large enough to include a
wide variety of problems, such as unimodal, multimodal, regular, irregular, separable,
non-separable and multi-dimensional problems.
Many test functions may be scattered in different textbooks, in individual
research articles or at different websites. Therefore, searching for a single source of
test function with a wide variety of characteristics is a cumbersome and tedious task.
The most notable attempts to assemble global optimisation (GO) test problems can
be found in Ali et al. (2005), Averick et al. (1991, 1992), Branin (1972), Chung
and Reynolds (1998), Dixon and Szegó (1978), Dixon and Price (1989), Fletcher
and Powell (1963), Flouda et al. (1999), Moré et al. (1981), Powell (1962, 1964),
Price et al. (2005), Salomon (1996), Schwefel (1981, 1995), Suganthan et al. (2005),
Tang et al. (2008, 2010) and Whitley et al. (1996). Online collections of test problems
also exist, such as the GLOBAL library at the cross-entropy toolbox (The Cross-Entropy
Toolbox, http://www.maths.uq.edu.au/CEToolBox/), GAMS World (2000) CUTE
(Gould et al., 2001), GO test problems collection by Hedar (n.d.), collection
152 M. Jamil and X-S. Yang

of test functions (Andrei, 2008; GEATbx, http://www.geatbx.com/; Test Problems for

Global Optimization, http://www2.imm.dtu.dk/ kajm/Test ex forms/test ex.html; Mishra,
2006a, 2006b, 2006c, 2006d, 2006e, 2006f, 2006g), a collection of continuous GO test
problems COCONUT (Neumaier, 2003) and a subset of commonly used test functions
(Yang, 2010a). This motivates us to carry out a thorough analysis and compile a
comprehensive collection of unconstrained optimisation test problems.
In general, unconstrained problems can be classified into two categories: test
functions and real-world problems. Test functions are artificial problems, and can be
used to evaluate the behaviour of an algorithm in sometimes diverse and difficult
situations. Artificial problems may include single global minimum, single or multiple
global minima in the presence of many local minima, long narrow valleys, null-space
effects, and flat surfaces. These problems can be easily manipulated and modified to test
the algorithms in diverse scenarios. On the other hand, real-world problems originate
from different fields such as physics, chemistry, engineering, mathematics, etc. These
problems are hard to manipulate and may contain complicated algebraic or differential
expressions and may require a significant amount of data to compile. A collection of
real-world unconstrained optimisation problems can be found in Averick et al. (1991,
1992).
In this present work, we will focus on the test function benchmarks and their
diverse properties such as modality and separability. A function with more than one
local optimum is called multimodal. These functions are used to test the ability of an
algorithm to escape from any local minimum. If the exploration process of an algorithm
is poorly designed, then it cannot search the function landscape effectively. This, in
turn, leads to an algorithm getting stuck at a local minimum. Multi-modal functions with
many local minima are among the most difficult class of problems for many algorithms.
Functions with flat surfaces pose a difficulty for the algorithms, since the flatness of
the function does not give the algorithm any information to direct the search process
towards the minima (Stepint, Matyas, PowerSum).
Another group of test problems is formulated by separable and non-separable
functions. According to Boyer et al. (2005), the dimensionality of the search space is
an important issue with the problem. In some functions, the area that contains that
global minima are very small, when compared to the whole search space, such as
Easom, Michalewicz (m=10) and Powell. For problems such as Perm, Kowalik and
Schaffer, the global minimum is located very close to the local minima. If the algorithm
cannot keep up the direction changes in the functions with a narrow curved valley, in
case of functions like Beale, Colville, or cannot explore the search space effectively,
in case of function like Pen Holder, Testtube-Holder having multiple global minima,
the algoritm will fail in these kinds of problems. Another problem that the algorithms
may suffer is the scaling problem with many orders of magnitude differences between
the domain and the function hyper-surface (Junior et al., 2004), such as Goldstein-Price
and Trid.

2 Characteristics of test functions

The goal of any GO is to find the best possible solutions x∗ from a set X
according to a set of criteria F = {f1 , f2 , · · · fn }. These criteria are called objective
functions expressed in the form of mathematical functions. An objective function is a
 A literature survey of benchmark functions 153

mathematical function f : D ⊂ ℜn → ℜ subject to additional constraints. The set D is

referred to as the set of feasible points in a search space. In the case of optimising a
single criterion f , an optimum is either its maximum or minimum. The GO problems
are often defined as minimisation problems, however, these problems can be easily
converted to maximisation problems by negating f . A general global optimum problem
can be defined as follows:

minimisef (x) (1)

x

The true optimal solution of an optimisation problem may be a set of x∗ ∈ D of

all optimal points in D, rather than a single minimum or maximum value in some
cases. There could be multiple, even an infinite number of optimal solutions, depending
on the domain of the search space. The tasks of any good GO algorithm is to
find globally optimal or at least sub-optimal solutions. The objective functions could
be characterised as continuous, discontinuous, linear, non-linear, convex, non-conxex,
unimodal, multimodal, separable1 and non-separable.
According to Chung and Reynolds (1998), it is important to ask the following two
questions before solving an optimisation problem;

1 What aspects of the function landscape make the optimisation process difficult?
2 What type of a priori knowledge is most effective for searching particular types
of function landscape?

In order to answer these questions, benchmark functions can be classified in terms of

features like modality, basins, valleys, separability and dimensionality (Winston, 1992).

2.1 Modality

The number of ambiguous peaks in the function landscape corresponds to the modality
of a function. If algorithms encounters these peaks during a search process, there is
a tendency that the algorithm may be trapped in one of the peaks. This will have a
negative impact on the search process, as this can direct the search away from the true
optimal solutions.

2.2 Basins

A relatively steep decline surrounding a large area is called a basin. Optimisation

algorithms can be easily attracted to such regions. Once in these regions, the search
process of an algorithm is severely hampered. This is due to lack of information to direct
the search process towards the minimum. According to Chung and Reynolds (1998), a
basin corresponds to the plateau for a maximisation problem, and a problem can have
multiple plateaus.

2.3 Valleys

A valley occurs when a narrow area of little change is surrounded by regions of

steep descent (Chung and Reynolds, 1998). As with the basins, minimisers are initially
154 M. Jamil and X-S. Yang

attracted to this region. The progress of a search process of an algorithm may be slowed
down considerably on the floor of the valley.

2.4 Separability

The separability is a measure of difficulty of different benchmark functions. In general,

separable functions are relatively easy to solve, when compared with their inseperable
counterpart, because each variable of a function is independent of the other variables.
If all the parameters or variables are independent, then a sequence of n independent
optimisation processes can be performed. As a result, each design variable or parameter
can be optimised independently. According to Salomon (1996), the general condition of
separability to see if the function is easy to optimise or not is given as

∂f (x)
= g(xi )h(x) (2)
∂xi

where g(xi ) means any function of xi only and h(x) any function of any x. If this
condition is satisfied, the function is called partially separable and easy to optimise,
because solutions for each xi can be obtained independently of all the other parameters.
This separability condition can be illustrated by the following two examples.
For example, function (f105 ) is not separable, because it does not satisfy the
condition (2)

∂f105 (x1 , x2 )
= 400(x21 − x2 )x1 − 2x1 − 2
∂x1
∂f105 (x1 , x2 )
= −200(x21 − x2 )
∂x2

On the other hand, the sphere function (f137 ) with two variables can indeed satisfy the
above condition (2) as shown below.

∂f137 (x1 , x2 ) ∂f137 (x1 , x2 )

= 2x1 = 2x2
∂x1 ∂x2

where h(x) is regarded as 1.

In Boyer et al. (2005), the formal definition of separability is given as

(
arg minimisef (x1 , ..., xp ) = arg minimisef (x1 , ...), ...,
x1 ,...,xp x1
)
arg minimisef (..., xp ) (3)
xp

In other words, a function of p variables is called separable, if it can written as a sum

of p functions of just one variable (Boyer et al., 2005). On the other hand, a function
is called non-separable, if its variables show inter-relation among themselves or are
not independent. If the objective function variables are independent of each other, then
the objective functions can be decomposed into sub-objective functions. Then, each of
 A literature survey of benchmark functions 155

these sub-objectives involves only one decision variable, while treating all the others as
constant and can be expressed as

∑
p
f (x1 , x2 , · · · , xp ) = fi (xi ) (4)
i=1

2.5 Dimensionality

The difficulty of a problem generally increases with its dimensionality. According to

Winston (1992) and Yao and Liu (1996), as the number of parameters or dimension
increases, the search space also increases exponentially. For highly non-linear problems,
this dimensionality may be a significant barrier for almost all optimisation algorithms.

3 Benchmark test functions for GO

Now, we present a collection of 175 unconstrained optimisation test problems which

can be used to validate the performance of optimisation algorithms. The dimensions,
problem domain size and optimal solution are denoted by D, Lb ≤ xi ≤ U b and
f (x∗ ) = f (x1 , ...xn ), respectively. The symbols Lb and Ub represent lower, upper
bound of the variables, respectively. It is worth noting that in several cases, the
optimal solution vectors and their corresponding solutions are known only as numerical
approximations.

1 Ackley Function 1 (Báck and Schwefel, 1993) (continuous, differentiable,

non-separable, scalable, multimodal)
√ −1 ∑D 2 −1 ∑D
f1 (x) = −20e−0.02 D i=1 xi − eD i=1 cos(2πxi ) + 20 + e

subject to −35 ≤ xi ≤ 35. The global minima is located at origin x∗ = (0, · · · , 0),
f (x∗ ) = 0.
2 Ackley Function 2 (Ackley, 1987) (continuous, differentiable, non-separable,
non-scalable, unimodal)
√ 2 2
f2 (x) = −200e−0.02 x1 +x2

subject to −32 ≤ xi ≤ 32. The global minimum is located at origin x∗ = (0, 0),
f (x∗ ) = −200.
3 Ackley Function 3 (Ackley, 1987) (continuous, differentiable, non-separable,
non-scalable, unimodal)
√ 2 2
f3 (x) = 200e−0.02 x1 +x2 + 5ecos(3x1 )+sin(3x2 )

subject to −32 ≤ xi ≤ 32. The global minimum is located at x∗ = (0, ≈ −0.4),

f (x∗ ) ≈ −219.1418.
156 M. Jamil and X-S. Yang

4 Ackley Function 4 or Modified Ackley Function (Rónkkónen, 2009) (continuous,

differentiable, non-separable, scalable, multimodal)

D (
∑ √ )
f4 (x) = e−0.2 x2i + x2i+1 + 3 (cos(2xi ) + sin(2xi+1 ))
i=1

subject to −35 ≤ xi ≤ 35. It is highly multimodal function with two global

minimum close to origin
x = f ({−1.479252, −0.739807}, {1.479252, −0.739807}), f (x∗ ) = −3.917275.
5 Adjiman Function (Adjiman et al., 1998) (continuous, differentiable,
non-separable, non-scalable, multimodal)

x1
f5 (x) = cos(x1 )sin(x2 ) −
(x22 + 1)

subject to −1 ≤ x1 ≤ 2, −1 ≤ x2 ≤ 1. The global minimum is located at

x∗ = (2, 0.10578), f (x∗ ) = −2.02181.
6 Alpine Function 1 (Rahnamyan et al., 2007a) (continuous, non-differentiable,
separable, non-scalable, multimodal)

D 
∑ 
 
f6 (x) = xi sin(xi ) + 0.1xi 
i=1

subject to −10 ≤ xi ≤ 10. The global minimum is located at origin

x∗ = (0, · · · , 0), f (x∗ ) = 0.
7 Alpine Function 2 (Clerc, 1999) (continuous, differentiable, separable, scalable,
multimodal)

∏
D
√
f7 (x) = xi sin(xi )
i=1

subject to 0 ≤ xi ≤ 10. The global minimum is located at x∗ = (7.917 · · · 7.917),

f (x∗ ) = 2.808D .
8 Brad Function (Brad, 1970) (continuous, differentiable, non-separable,
non-scalable, multimodal)

15 [
∑ ]2
yi − x1 − ui
f8 (x) =
i=1
vi x2 + wi x3

where ui = i, vi = 16 − i, wi = min(ui , vi ) and y = yi = [0.14, 0.18, 0.22,

0.25, 0.29, 0.32, 0.35, 0.39, 0.37, 0.58, 0.73, 0.96, 1.34, 2.10, 4.39]T . It is subject
to −0.25 ≤ x1 ≤ 0.25, 0.01 ≤ x2 , x3 ≤ 2.5. The global minimum is located at
x∗ = (0.0824, 1.133, 2.3437), f (x∗ ) = 0.00821487.
 A literature survey of benchmark functions 157

9 Bartels Conn Function (continuous, non-differentiable, non-separable,

non-scalable, multimodal)
     
f9 (x) = x21 + x22 + x1 x2  + sin(x1 ) + cos(x2 )

subject to −500 ≤ xi ≤ 500. The global minimum is located at x∗ = (0, 0),

f (x∗ ) = 1.
10 Beale Function (Hedar, n.d.) (continuous, differentiable, non-separable,
non-scalable, unimodal)

f10 (x) = (1.5 − x1 + x1 x2 )2 + (2.25 − x1 + x1 x22 )2

+(2.625 − x1 + x1 x32 )2

subject to −4.5 ≤ xi ≤ 4.5. The global minimum is located at x∗ = (3, 0.5),

f (x∗ ) = 0.
11 Biggs EXP Function 2 (Biggs, 1971) (continuous, differentiable, non-separable,
non-scalable, multimodal)
∑
10
( )2
f11 (x) = e−ti x1 − 5e−ti x2 − yi
i=1

where ti = 0.1i, yi = e−ti − 5e10ti . It is subject to 0 ≤ xi ≤ 20. The global

minimum is located at x∗ = (1, 10), f (x∗ ) = 0.
12 Biggs EXP Function 3 (Biggs, 1971) (continuous, differentiable, non-separable,
non-scalable, multimodal)
∑
10
( )2
f12 (x) = e−ti x1 − x3 e−ti x2 − yi
i=1

where ti = 0.1i, yi = e−ti − 5e10ti . It is subject to 0 ≤ xi ≤ 20. The global

minimum is located at x∗ = (1, 10, 5), f (x∗ ) = 0.
13 Biggs EXP Function 4 (Biggs, 1971) (continuous, differentiable, non-separable,
non-scalable, multimodal)
∑
10
( )2
f13 (x) = x3 e−ti x1 − x4 e−ti x2 − yi
i=1

where ti = 0.1i, yi = e−ti − 5e10ti . It is subject to 0 ≤ xi ≤ 20. The global

minimum is located at x∗ = (1, 10, 1, 5), f (x∗ ) = 0.
14 Biggs EXP Function 5 (Biggs, 1971) (continuous, differentiable, non-separable,
non-scalable, multimodal)
∑
11
( )2
f14 (x) = x3 e−ti x1 − x4 e−ti x2 + 3e−ti x5 − yi
i=1

where ti = 0.1i, yi = e−ti − 5e10ti + 3e−4ti . It is subject to 0 ≤ xi ≤ 20. The

global minimum is located at x∗ = (1, 10, 1, 5, 4), f (x∗ ) = 0.
158 M. Jamil and X-S. Yang

15 Biggs EXP Function 6 (Biggs, 1971) (continuous, differentiable, non-separable,

non-scalable, multimodal)

∑
13
( )2
f15 (x) = x3 e−ti x1 − x4 e−ti x2 + x6 e−ti x5 − yi
i=1

where ti = 0.1i, yi = e−ti − 5e10ti + 3e−4ti . It is subject to −20 ≤ xi ≤ 20. The

global minimum is located at x∗ = (1, 10, 1, 5, 4, 3), f (x∗ ) = 0.
16 Bird Function (Mishra, 2006f) (continuous, differentiable, non-separable,
non-scalable, multimodal)
2 2
f16 (x) = sin(x1 )e(1−cos(x2 )) + cos(x2 )e(1−sin(x1 )) + (x1 − x2 )2

subject to −2π ≤ xi ≤ 2π. The global minimum is located at x∗ = (4.70104,

3.15294),(−1.58214, −3.13024), f (x∗ ) = −106.764537.
17 Bohachevsky Function 1 (Bohachevsky et al., 1986) (continuous, differentiable,
separable, non-scalable, multimodal)

f17 (x) = x21 + 2x22 − 0.3cos(3πx1 )

−0.4cos(4πx2 ) + 0.7

subject to −100 ≤ xi ≤ 100. The global minimum is located at x∗ = f (0, 0),

f (x∗ ) = 0.
18 Bohachevsky Function 2 (Bohachevsky et al., 1986) (continuous, differentiable,
non-separable, non-scalable, multimodal)

f18 (x) = x21 + 2x22 − 0.3cos(3πx1 ) · 0.4cos(4πx2 )

+0.3

subject to −100 ≤ xi ≤ 100. The global minimum is located at x∗ = f (0, 0),

f (x∗ ) = 0.
19 Bohachevsky Function 3 (Bohachevsky et al., 1986) (continuous, differentiable,
non-separable, non-scalable, multimodal)

f19 (x) = x21 + 2x22 − 0.3cos(3πx1 + 4πx2 ) + 0.3

subject to −100 ≤ xi ≤ 100. The global minimum is located at x∗ = f (0, 0),

f (x∗ ) = 0.
20 Booth Function (continuous, differentiable, non-separable, non-scalable, unimodal)

f20 (x) = (x1 + 2x2 − 7)2 + (2x1 + x2 − 5)2

subject to −10 ≤ xi ≤ 10. The global minimum is located at x∗ = f (1, 3),

f (x∗ ) = 0.
 A literature survey of benchmark functions 159

21 Box-Betts Quadratic Sum Function (Ali et al., 2005) (continuous, differentiable,

non-separable, non-scalable, multimodal)

∑
D−1
f21 (x) = g(xi )2
i=0
where
−(i+1)
g(x) = e−0.1(i+1)x1 − e−0.1(i+1)x2 − e[(−0.1(i+1))−e ]x3

subject to 0.9 ≤ x1 ≤ 1.2, 9 ≤ x2 ≤ 11.2, 0.9 ≤ x2 ≤ 1.2. The global minimum

is located at x∗ = f (1, 10, 1) f (x∗ ) = 0.
22 Branin RCOS Function (Branin, 1972) (continuous, differentiable, non-separable,
non-scalable, multimodal)
( )2
5.1x21 5x1
f22 (x) = x2 − + − 6
4π 2 π
( )
1
+10 1 − cos(x1 ) + 10
8π

with domain −5 ≤ x1 ≤ 10, 0 ≤ x1 ≤ 15. It has three global minima at

x∗ = f ({−π, 12.275}, {π, 2.275}, {3π, 2.425}), f (x∗ ) = 0.3978873.
23 Branin RCOS Function 2 (Muntenau and Lazarescu, 1998) (continuous,
differentiable, non-separable, non-scalable, multimodal)
( )2
5.1x21 5x1
f23 (x) = x2 − + − 6
4π 2 π
( )
1
+10 1 − cos(x1 ) cos(x2 ) ln(x21 + x22 + 1) + 10
8π

with domain −5 ≤ xi ≤ 15. The global minimum is located at

x∗ = f (−3.2, 12.53), f (x∗ ) = 5.559037.
24 Brent Function (Branin, 1972) (continuous, differentiable, non-separable,
non-scalable, unimodal)

f24 (x) = (x1 + 10) + (x2 + 10) + e−x1 −x2

2 2 2 2

with domain −10 ≤ xi ≤ 10. The global minimum is located at x∗ = f (0, 0),
f (x∗ ) = 0.
25 Brown Function (Begambre and Laier, 2009) (continuous, differentiable,
non-separable, scalable, unimodal)

∑
n−1
2 2
f25 (x) = (x2i )(xi+1 +1) + (x2i+1 )(xi +1)
i=1

subject to −1 ≤ xi ≤ 4. The global minimum is located at x∗ = f (0, · · · , 0),

f (x∗ ) = 0.
160 M. Jamil and X-S. Yang

Bukin functions (Silagadze, 2007) are almost fractal (with fine seesaw edges) in
the surroundings of their minimal points. Due to this property, they are extremely
difficult to optimise by any global or local optimisation methods.
26 Bukin Function 2 (continuous, differentiable, non-separable, non-scalable,
multimodal)

f26 (x) = 100(x2 − 0.01x21 + 1) + 0.01(x1 + 10)2

subject to −15 ≤ x1 ≤ −5 and −3 ≤ x2 ≤ −3. The global minimum is located at

x∗ = f (−10, 0), f (x∗ ) = 0.
27 Bukin Function 4 (continuous, non-differentiable, separable, non-scalable,
multimodal)

f27 (x) = 100x22 + 0.01∥x1 + 10∥

subject to −15 ≤ x1 ≤ −5 and −3 ≤ x2 ≤ −3. The global minimum is located at

x∗ = f (−10, 0), f (x∗ ) = 0.
28 Bukin Function 6 (continuous, non-differentiable, non-separable, non-scalable,
multimodal)
√
f28 (x) = 100 ∥x2 − 0.01x21 ∥ + 0.01∥x1 + 10∥

subject to −15 ≤ x1 ≤ −5 and −3 ≤ x2 ≤ −3. The global minimum is located at

x∗ = f (−10, 1), f (x∗ ) = 0.
29 Camel Function – Three Hump (Branin, 1972) (continuous, differentiable,
non-separable, non-scalable, multimodal)

f29 (x) = 2x21 − 1.05x41 + x61 /6 + x1 x2 + x22

subject to −5 ≤ xi ≤ 5. The global minima is located at x∗ = f (0, 0), f (x∗ ) = 0.

30 Camel Function – Six Hump (Branin, 1972) (continuous, differentiable,
non-separable, non-scalable, multimodal)

x41 2
f30 (x) = (4 − 2.1x21 + )x
3 1
+x1 x2 + (4x22 − 4)x22

subject to −5 ≤ xi ≤ 5. The two global minima are located at x∗ =

f ({−0.0898, 0.7126}, {0.0898, −0.7126, 0}), f (x∗ ) = −1.0316.
31 Chen Bird Function (Chen, 2003) (continuous, differentiable, non-separable,
non-scalable, multimodal)

0.001
f31 (x) = − ⌊ ⌋−
+ (x1 − 0.4x2 − 0.1)2
(0.001)2
0.001
⌊ ⌋
(0.001) + (2x1 + x2 − 1.5)2
2
 A literature survey of benchmark functions 161

subject to −500 ≤ xi ≤ 500 The global minimum is located at

x∗ = f (− 18
7
, − 13 ∗
18 ), f (x ) = −2000.

32 Chen V Function (Chen, 2003) (continuous, differentiable, non-separable,

non-scalable, multimodal)

0.001
f32 (x) = − ⌊ ⌋−
(0.001)2 + (x21 + x22 − 1)2
0.001
⌊ ⌋−
(0.001)2 + (x21 + x22 − 0.5)2
0.001
⌊ ⌋
(0.001)2 + (x21 − x22 )2

subject to −500 ≤ xi ≤ 500 The global minimum is located at

x∗ = f (−0.3888889, 0.7222222), f (x∗ ) = −2000.
33 Chichinadze Function (continuous, differentiable, separable, non-scalable,
multimodal)

f33 (x) = x21 − 12x1 + 11 +

10cos(πx1 /2) + 8sin(5πx1 /2) −
(1/5)0.5 exp(−0.5(x2 − 0.5)2 )

subject to −30 ≤ xi ≤ 30. The global minimum is located at

x∗ = f (5.90133, 0.5), f (x∗ ) = −43.3159.
34 Chung Reynolds Function (Chung and Reynolds, 1998) (continuous, differentiable,
partially-separable, scalable, unimodal)

( )2
∑
D
f34 (x) = x2i
i=1

subject to −100 ≤ xi ≤ 100. The global minimum is located at x∗ = f (0, · · · , 0),

f (x∗ ) = 0.
35 Cola Function (Adorio and Dilman, 2005) (continuous, differentiable,
non-separable, non-scalable, multimodal)

The 17-dimensional function computes indirectly the formula (D, u) by setting

x0 = y0 , x1 = u0 , xi = u2(i−2) , yi = u2(i−2)+1
∑
f35 (n, u) = h(x, y) = (ri,j − di,j )2
j<i

where ri,j is given by

ri,j = [(xi − xj )2 + (yi − yj )2 ]1/2
162 M. Jamil and X-S. Yang

and d is a symmetric matrix given by

 
1.27
 1.69 1.43 
 
 2.04 2.35 2.43 
 
 3.09 3.18 3.26 2.85 
 

d = [dij ] =  3.20 3.22 3.27 2.88 1.55 

 2.86 2.56 2.58 2.59 3.12 3.06 
 
 3.17 3.18 3.18 3.12 1.31 1.64 3.00 
 
 3.21 3.18 3.18 3.17 1.70 1.36 2.95 1.32 
2.38 2.31 2.42 1.94 2.85 2.81 2.56 2.91 2.97
This function has bounds 0 ≤ x0 ≤ 4 and −4 ≤ xi ≤ 4 for i = 1 . . . D − 1. It has
a global minimum of f (x∗ ) = 11.7464.
36 Colville Function (continuous, differentiable, non-separable, non-scalable,
multimodal)

f36 (x) = 100(x1 − x22 )2 + (1 − x1 )2 +

90(x4 − x23 )2 + (1 − x3 )2 +
10.1((x2 − 1)2 + (x4 − 1)2 ) +
19.8(x2 − 1)(x4 − 1)

subject to −10 ≤ xi ≤ 10. The global minima is located at x∗ = f (1, · · · , 1),

f (x∗ ) = 0.
37 Corana Function (Corana et al., 1987) (Discontinuous, non-differentiable,
separable, scalable, multimodal)
{ ( )
2
0.15 zi − 0.05sgn(zi ) di if |vi | < A
f37 (x) =
di x2i otherwise

where

vi = |xi − zi | , A = 0.05
⌊ x  ⌋
 i
zi = 0.2   + 0.49999 sgn (xi )
0.2
di = (1, 1000, 10, 100)

subject to −500 ≤ xi ≤ 500. The global minimum is located at x∗ = f (0, 0, 0, 0),

f (x∗ ) = 0.
38 Cosine Mixture Function (Ali et al., 2005) (Discontinuous, non-differentiable,
separable, scalable, multimodal)

∑
n ∑
n
f38 (x) = −0.1 cos(5πxi ) − x2i
i=1 i=1

subject to −1 ≤ xi ≤ 1. The global minimum is located at x∗ = f (0, 0),

f (x∗ ) = (0.2 or 0.4) for n = 2 and 4 respectively.
 A literature survey of benchmark functions 163

39 Cross-in-Tray Function (Mishra, 2006f) (continuous, non-separable, non-scalable,

multimodal)

f39 (x) = −0.0001[|sin(x1 )sin(x2 )

e|100−[(x1 +x2 )]
2 2 0.5
/π|
| + 1]0.1

subject to −10 ≤ xi ≤ 10.

The four global minima are located at x∗ = f (±1.349406685353340,
±1.349406608602084), f (x∗ ) = −2.06261218.
40 Csendes Function (Csendes and Ratz, 1997) (continuous, differentiable, separable,
scalable, multimodal)

∑
D ( )
1
f40 (x) = x6i 2 + sin
i=1
xi

subject to −1 ≤ xi ≤ 1. The global minimum is located at x∗ = f (0, · · · , 0),

f (x∗ ) = 0.
41 Cube Function (Lavi and Vogel, 1966) (continuous, differentiable, non-separable,
non-scalable, unimodal)
( )2
f41 (x) = 100 x2 − x31 + (1 − x1 )2

subject to −10 ≤ xi ≤ 10. The global minimum is located at x∗ = f (−1, 1),

f (x∗ ) = 0.
42 Damavandi Function (Damavandi and Safavi-Naeini, 2005) (continuous,
differentiable, non-separable, non-scalable, multimodal)
[   ]
 sin[π(x1 − 2)]sin[π(x2 − 2)] 5

f42 (x) = 1 −  
π 2 (x1 − 2)(x2 − 2) 
[ ]
2 + (x1 − 7)2 + 2(x2 − 7)2

subject to 0 ≤ xi ≤ 14. The global minimum is located at x∗ = f (2, 2),

f (x∗ ) = 0.
43 Deb Function 1 (Rónkkónen, 2009) (continuous, differentiable, separable, scalable,
multimodal)

1 ∑ 6
D
f43 (x) = − sin (5πxi )
D i=1

subject to −1 ≤ xi ≤ 1. The number of global minima is 5D that are evenly

spaced in the function landscape, where D represents the dimension of the
problem.
164 M. Jamil and X-S. Yang

44 Deb Function 3 (Rónkkónen, 2009) (continuous, differentiable, separable, scalable,

multimodal)

1 ∑ 6
D
3/4
f44 (x) = − sin (5π(xi − 0.05))
D i=1

subject to −1 ≤ xi ≤ 1. The number of global minima is 5D that are unevenly

spaced in the function landscape, where D represents the dimension of the
problem.
45 Deckkers-Aarts Function (Ali et al., 2005) (continuous, differentiable,
non-separable, non-scalable, multimodal)

f45 (x) = 105 x21 + x22 − (x21 + x22 )2 + 10−5 (x21 + x22 )4

subject to −20 ≤ xi ≤ 20. The two global minima are located at x∗ = f (0, ±15)
f (x∗ ) = −24777.
46 deVilliers Glasser Function 1 (deVillers and Glasser, 1981)(continuous,
differentiable, non-separable, non-scalable, multimodal)

∑
24
[ ]2
f46 (x) = x1 xt2i sin(x3 ti + x4 ) − yi
i=1

where ti = 0.1(i − 1), yi = 60.137 × 1.371ti sin(3.112ti + 1.761). It is subject to

−500 ≤ xi ≤ 500. The global minimum is f (x∗ ) = 0.
47 deVilliers Glasser Function 2 (deVillers and Glasser, 1981) (continuous,
differentiable, non-separable, non-scalable, multimodal)

∑
16
[ ]2
f47 (x) = x1 xt2i tanh [x3 ti + sin(x4 ti )] cos(ti ex5 ) − yi
i=1

where ti = 0.1(i − 1),

yi = 53.81 × 1.27ti tanh(3.012ti + sin(2.13ti )) cos(e0.507 ti ). It is subject to
−500 ≤ xi ≤ 500. The global minimum is f (x∗ ) = 0.
48 Dixon & Price Function (Dixon and Price, 1989) (continuous, differentiable,
non-separable, scalable, unimodal)

∑
D
f48 (x) = (x1 − 1)2 + i(2x2i − xi−1 )2
i=2

subject to −10 ≤ xi ≤ 10. The global minimum is located at x∗ = f (2( 2 2−2

i
i )),

f (x∗ ) = 0.
 A literature survey of benchmark functions 165

49 Dolan Function (continuous, differentiable, non-separable, non-scalable,

multimodal)

f49 (x) = (x1 + 1.7x2 ) sin(x1 ) − 1.5x3 − 0.1x4 cos(x4 + x5 − x1 ) +

0.2x25 − x2 − 1

subject to −100 ≤ xi ≤ 100. The global minimum is f (x∗ ) = 0.

50 Easom Function (Chung and Reynolds, 1998)(continuous, differentiable,
separable, non-scalable, multimodal)

f50 (x) = −cos(x1 )cos(x2 ) exp[−(x1 − π)2

−(x2 − π)2 ]

subject to −100 ≤ xi ≤ 100. The global minimum is located at x∗ = f (π, π),

f (x∗ ) = −1.
51 El-Attar-Vidyasagar-Dutta Function (El-Attar et al., 1979) (continuous,
differentiable, non-separable, non-scalable, unimodal)

f51 (x) = (x21 + x2 − 10)2 + (x1 + x22 − 7)2 +

(x21 + x32 − 1)2

subject to −500 ≤ xi ≤ 500. The global minimum is located at

x∗ = f (2.842503, 1.920175), f (x∗ ) = 0.470427.
52 Egg Crate Function (continuous, separable, non-scalable)

f52 (x) = x21 + x22 + 25(sin2 (x1 ) + sin2 (x2 ))

subject to −5 ≤ xi ≤ 5. The global minimum is located at x∗ = f (0, 0),

f (x∗ ) = 0.
53 Egg Holder Function (continuous, differentiable, non-separable, scalable,
multimodal)

∑
m−1 √
f53 (x) = [−(xi+1 + 47)sin |xi+1 + xi /2 + 47|
i=1
√
−xi sin |xi − (xi+1 + 47)|]

subject to −512 ≤ xi ≤ 512. The global minimum is located at

x∗ = f (512, 404.2319), f (x∗ ) ≈ 959.64.
54 Exponential Function (Rahnamyan et al., 2007b) (continuous, differentiable,
non-separable, scalable, multimodal)
( )
∑ D
f54 (x) = − exp −0.5 2
xi
i=1

subject to −1 ≤ xi ≤ 1. The global minima is located at x = f (0, · · · , 0),

f (x∗ ) = 1.
166 M. Jamil and X-S. Yang

55 EX Function 1 (continuous, differentiable, separable, non-scalable, multimodal)

f55 (x) = 0.1(1 − x21 ) + 0.1 sin(10x1 ) + (11 − x2 )2 + sin(10x2 )

∩
subject to xi ∈ [0, 2] [10, 12]. The global minima is located at
x = f (1.764, 11.150), f (x∗ ) ≈ −1.28186.
56 Freudenstein Roth Function (Rao, 2009) (continuous, differentiable, non-separable,
non-scalable, multimodal)

f56 (x) = (x1 − 13 + ((5 − x2 )x2 − 2)x2 )2 +

(x1 − 29 + ((x2 + 1)x2 − 14)x2 )2

subject to −10 ≤ xi ≤ 10. The global minimum is located at x∗ = f (5, 4),

f (x∗ ) = 0.
57 Giunta Function (Mishra, 2006f) (continuous, differentiable, separable, scalable,
multimodal)

∑2
16
f57 (x) = 0.6 + [sin( xi − 1)
i=1
15
16
+sin2 (xi − 1)
15
1 16
+ sin(4( xi − 1))]
50 15

subject to −1 ≤ xi ≤ 1. The global minimum is located at

x∗ = f (0.45834282, 0.45834282), f (x∗ ) = 0.060447.
58 Goldstein Price Function (Goldstein and Price, 1971) (continuous, differentiable,
non-separable, non-scalable, multimodal)

f58 (x) = [1 + (x1 + x2 + 1)2 (19 − 14x1

+3x21 − 14x2 + 6x1 x2 + 3x22 )]
×[30 + (2x1 − 3x2 )2
(18 − 32x1 + 12x21 + 48x2 − 36x1 x2 + 27x22 )]

subject to −2 ≤ xi ≤ 2. The global minimum is located at x∗ = f (0, −1),

f (x∗ ) = 3.
59 Griewank Function (Griewank, 1981) (continuous, differentiable, non-separable,
scalable, multimodal)

∑n
x2i ∏ ( xi )
f59 (x) = − cos √ + 1
i=1
4000 i

subject to −100 ≤ xi ≤ 100. The global minima is located at x∗ = f (0, · · · , 0),

f (x∗ ) = 0.
 A literature survey of benchmark functions 167

60 Gulf Research Problem (Shanno, 1970) (continuous, differentiable, non-separable,

non-scalable, multimodal)
99 [
∑ ( x3 ) ]2
(ui − x2 )
f60 (x) = exp − − 0.01i
i=1
xi

where ui = 25 + [−50 ln(0.01i)]1/1.5 subject to 0.1 ≤ x1 ≤ 100, 0 ≤ x2 ≤ 25.6

and 0 ≤ x1 ≤ 5. The global minimum is located at x∗ = f (50, 25, 1.5), f (x∗ ) = 0.
61 Hansen Function (Fraley et al., 1989) (continuous, differentiable, separable,
non-scalable, multimodal)

∑
4
f61 (x) = (i + 1)cos(ix1 + i + 1)
i
∑
4
(j + 1)cos((j + 2)x2 + j + 1)
j=0

subject to −10 ≤ xi ≤ 10. The multiple global minima are located at

x∗ = f ({−7.589893, −7.708314}, {−7.589893, −1.425128},

{−7.589893, 4.858057}, {−1.306708, −7.708314},

{−1.306708, 4.858057}, { 4.976478, 4.858057},

{ 4.976478, −1.425128}, { 4.976478, −7.708314}),

62 Hartman Function 3 (Hartman, 1972) (continuous, differentiable, non-separable,

non-scalable, multimodal)
 
∑4 ∑
3
ci exp − aij (xj − pij ) 
2
f62 (x) = −
i=1 j=1

subject to 0 ≤xj ≤ 1, j
∈ {1, 2, 3} with
 constants
 aij , pij and ci are given as
3 10 30 1
0.1 10 35 1.2
A = [Aij ] =    
 3 10 30, c = ci =  3 ,
 0.1 10 35  3.2
0.3689 0.1170 0.2673
 0.4699 0.4837 0.7470
p = pi =   0.1091 0.8732 0.5547


0.03815 0.5743 0.8828

The global minimum is located at x∗ = f (0.1140, 0.556, 0.852),
f (x∗ ) ≈ −3.862782.
168 M. Jamil and X-S. Yang

63 Hartman Function 6 (Hartman, 1972) (continuous, differentiable, non-separable,

non-scalable, multimodal)
 
∑4 ∑
6
ci exp − aij (xj − pij ) 
2
f63 (x) = −
i=1 j=1

subject to 0 ≤xj ≤ 1, j ∈ {1, · · · , 6}  aij, pij and ci are given as

with constants
10 3 17 3.5 1.7 8 1
0.05 10 17 0.1 8 14 1.2
A = [Aij ] =  
 3 3.5 1.7 10 17 8 , c = ci =  3 
 

17 8 0.05 10 0.1 14 3.2

 
0.1312 0.1696 0.5569 0.0124 0.8283 0.5586
0.2329 0.4135 0.8307 0.3736 0.1004 0.9991
p = pi =  0.2348 0.1451 0.3522 0.2883 0.3047 0.6650


0.4047 0.8828 0.8732 0.5743 0.1091 0.0381

The global minima is located at
x = f (0.201690, 0.150011, 0.476874, 0.275332, ...0.311652, 0.657301),
f (x∗ ) ≈ −3.32236.
64 Helical Valley (Fletcher and Powell, 1963) (continuous, differentiable,
non-separable, scalable, multimodal)
[ (√ )]
2
f64 (x) = 100 (x2 − 10θ) + x21 + x22 − 1

+x23

where
 ( )
 1 tan−1 x1 , if x1 ≥ 0
θ=
2π ( x2 )
 1 tan−1 x1 + 0.5 if x1 < 0
2π x2

subject to −10 ≤ xi ≤ 10. The global minima is located at x∗ = f (1, 0, 0),

f (x∗ ) = 0.
65 Himmelblau Function (Himmelblau, 1972) (continuous, differentiable,
non-separable, non-scalable, multimodal)

f65 (x) = (x21 + x2 − 11)2 + (x1 + x22 − 7)2

subject to −5 ≤ xi ≤ 5. The global minimum is located at x∗ = f (3, 2),

f (x∗ ) = 0.
66 Hosaki Function (Bekey and Ung, 1974) (continuous, differentiable,
non-separable, non-scalable, multimodal)

f66 (x) = (1 − 8x1 + 7x21 − 7/3x31 + 1/4x41 )x22 e−x2

subject to 0 ≤ x1 ≤ 5 and 0 ≤ x2 ≤ 6. The global minimum is located at

x∗ = f (4, 2), f (x∗ ) ≈ −2.3458.
 A literature survey of benchmark functions 169

67 Jennrich-Sampson Function (Jennrich and Sampson, 1968) (continuous,

differentiable, non-separable, non-scalable, multimodal)

∑
10
( ( ))2
f67 (x) = 2 + 2i − eix1 + eix2
i=1

subject to −1 ≤ xi ≤ 1. The global minimum is located at

x∗ = f (0.257825, 0.257825), f (x∗ ) = 124.3612.
68 Langerman Function 5 (Bersini et al., 1996) (continuous, differentiable,
non-separable, scalable, multimodal)
 
∑
m ∑D ∑
D
−π j=1 (xj −aij ) cos π (xj − aij )2 
1 2
f68 (x) = − ci e
i=1 j=1

subject to 0 ≤ xj ≤ 10, where j ∈ [0, D − 1] and m = 5. It has a global

minimum value of f (x∗ ) = −1.4. The matrix A and column vector c are given as
The matrix A is given by
 
9.681 0.667 4.783 9.095 3.517 9.325 6.544 0.211 5.122 2.020
 9.400 2.041 3.788 7.931 2.882 2.672 3.568 1.284 7.033 7.374 
 
A = [Aij ] =  
 8.025 9.152 5.114 7.621 4.564 4.711 2.996 6.126 0.734 4.982 
 2.196 0.415 5.649 6.979 9.510 9.166 6.304 6.054 9.377 1.426 
8.074 8.777 3.467 1.863 6.708 6.349 4.534 0.276 7.633 1.567
 
0.806
 0.517 
 
c = ci = 
 1.5 

 0.908 
0.965
69 Keane Function (continuous, differentiable, non-separable, non-scalable,
multimodal)

sin2 (x1 − x2 )sin2 (x1 + x2 )

f69 (x) = √
x21 + x22

subject to 0 ≤ xi ≤ 10.
The multiple global minima are located at x∗ = f ({0, 1.39325},{1.39325, 0}),
f (x∗ ) =−0.673668.
70 Leon Function (Lavi and Vogel, 1966) (continuous, differentiable, non-separable,
non-scalable, unimodal)

f70 (x) = 100(x2 − x21 )2 + (1 − x1 )2

subject to −1.2 ≤ xi ≤ 1.2. A global minimum is located at f (x∗ ) = f (1, 1),

f (x∗ ) = 0.
170 M. Jamil and X-S. Yang

71 Matyas Function (Hedar, n.d.) (continuous, differentiable, non-separable,

non-scalable, unimodal)

f71 (x) = 0.26(x21 + x22 ) − 0.48x1 x2

subject to −10 ≤ xi ≤ 10. The global minimum is located at x∗ = f (0, 0),

f (x∗ ) = 0.
72 McCormick Function (Lootsma, 1972) (continuous, differentiable, non-separable,
non-scalable, multimodal)

f72 (x) = sin(x1 + x2 ) + (x1 − x2 )2 − (3/2)x1 + (5/2)x2 + 1

subject to −1.5 ≤ x1 ≤ 4 and −3 ≤ x2 ≤ 3. The global minimum is located at

x∗ = f (−0.547, −1.547), f (x∗ ) ≈ −1.9133.
73 Miele Cantrell Function (Cragg and Levy, 1969) (continuous, differentiable,
non-separable, non-scalable, multimodal)
( )4
f73 (x) = e−x1 − x2 + 100(x2 − x3 )
6

4
+(tan (x3 − x4 )) + x81

subject to −1 ≤ xi ≤ 1. The global minimum is located at x∗ = f (0, 1, 1, 1),

f (x∗ ) = 0.
74 Mishra Function 1 (Mishra, 2006a) (continuous, differentiable, non-separable,
scalable, multimodal)

( −1
)N −∑N −1
i=1 xi
∑
N
f74 (x) = 1+D− xi
i=1

subject to 0 ≤ xi ≤ 1. The global minimum is f (x∗ ) = 2.

75 Mishra Function 2 (Mishra, 2006a) (continuous, differentiable, non-separable,
scalable, multimodal)

( −1
)N −∑i=1
N −1
0.5(xi +xi+1 )
∑
N
f75 (x) = 1+D− 0.5(xi + xi+1 )
i=1

subject to 0 ≤ xi ≤ 1. The global minimum is f (x∗ ) = 2.

76 Mishra Function 3 (Mishra, 2006f) (continuous, differentiable, non-separable,
non-scalable, multimodal)
√
 √ 
f76 (x) = cos x2 + x2  + 0.01(x1 + x2 )
1 2

The global minimum is located at x∗ = f (−8.466, −10), f (x∗ ) = −0.18467.

 A literature survey of benchmark functions 171

77 Mishra Function 4 (Mishra, 2006f) (continuous, differentiable, non-separable,

non-scalable, multimodal)
√
 √ 
f77 (x) = sin x2 + x2  + 0.01(x1 + x2 )
1 2

The global minimum is located at x∗ = f (−9.94112, −10), f (x∗ ) = −0.199409.

78 Mishra Function 5 (Mishra, 2006f) (continuous, differentiable, non-separable,
non-scalable, multimodal)
[ ]2
f78 (x) = sin2 (cos((x1 ) + cos(x2 )))2 + cos2 (sin(x1 ) + sin(x2 )) + x1
+0.01(x1 + x2 )

The global minimum is located at x∗ = f (−1.98682, −10), f (x∗ ) = −1.01983.

79 Mishra Function 6 (Mishra, 2006f) (continuous, differentiable, non-separable,
non-scalable, multimodal)
[ ]2
f79 (x) = − ln sin2 (cos((x1 ) + cos(x2 )))2 − cos2 (sin(x1 ) + sin(x2 )) + x1
+0.01((x1 − 1)2 + (x2 − 1)2 )

The global minimum is located at x∗ = f (2.88631, 1.82326), f (x∗ ) = −2.28395.

80 Mishra Function 7 (continuous, differentiable, non-separable, non-scalable,
multimodal)
[∏
D ]2
f80 (x) = xi − N !
i=1

The global minimum is f (x∗ ) = 0.

81 Mishra Function 8 (Mishra, 2006f) (continuous, differentiable, non-separable,
non-scalable, multimodal)
[

1 − 20x1 + 180x1 − 960x1 + 3360x1 − 8064x1
f81 (x) = 0.001 x10 9 8 7 6 5



1334x41 − 15360x31 + 11520x21 − 5120x1 + 2624
 ]2
 4 
x2 + 12x32 + 54x22 + 108x2 + 81

The global minimum is located at x∗ = f (2, −3), f (x∗ ) = 0.

82 Mishra Function 9 (Mishra, 2006f) (continuous, differentiable, non-separable,
non-scalable, multimodal)
[ ]2
f82 (x) = ab2 c + abc2 + b2 + (x1 + x2 − x3 )2

where a = 2x31 + 5x1 x2 + 4x3 − 2x21 x3 − 18, b = x1 + x32 + x1 x23 − 22

c = 8x21 + 2x2 x3 + 2x22 + 3x32 − 52. The global minimum is located at
x∗ = f (1, 2, 3), f (x∗ ) = 0.
172 M. Jamil and X-S. Yang

83 Mishra Function 10 (Mishra, 2006f) (continuous, differentiable, non-separable,

non-scalable, multimodal)
[ ]2
f83 (x) = ⌊x1 ⊥ x2 ⌋ − ⌊x1 ⌋ − ⌊x2 ⌋

The global minimum is located at x∗ = f {(0, 0), (2, 2)}, f (x∗ ) = 0.

84 Mishra Function 11 (Mishra, 2006f) (continuous, differentiable, non-separable,
non-scalable, multimodal)

[1 ∑D
  (∏ D
 ) N1 ]2
f84 (x) = xi  − xi 
D i=1 i=1

The global minimum is f (x∗ ) = 0.

85 Parsopoulos Function (continuous, differentiable, separable, scalable, multimodal)
2 2
f85 (x) = cos (x1 ) + sin (x2 )

subject to −5 ≤ xi ≤ 5, where (x1 , x2 ) ∈ R2 . This function has infinite number

of global minima in R2 , at points (κ π2 , λπ), where κ = ±1, ±3, ... and
λ = 0, ±1, ±2, .... In the given domain problem, function has 12 global minima
all equal to zero.
86 Pen Holder Function (Mishra, 2006f) (continuous, differentiable, non-separable,
non-scalable, multimodal)

f86 (x) = − exp[|cos(x1 )cos(x2 )e|1−[(x1 +x2 )] /π| −1

2 2 0.5
| ]

subject to −11 ≤ xi ≤ 11. The four global minima are located at x∗ =

f (±9.646168, ±9.646168), f (x∗ ) = −0.96354.
87 Pathological Function (Rahnamyan et al., 2007a) (continuous, differentiable,
non-separable, non-scalable, multimodal)
 √ 
∑
D−1 sin2 100x2i + x2i+1 − 0.5
f87 (x) = 0.5 + 
i=1
1 + 0.001(x2i − 2xi xi+1 + x2i+1 )2

subject to −100 ≤ xi ≤ 100. The global minima is located x∗ = f (0, · · · , 0),

f (x∗ ) = 0.
88 Paviani Function (Himmelblau, 1972) (continuous, differentiable, non-separable,
scalable, multimodal)

10 [
( 10 )0.2
∑ ] ∏
2 2
f88 (x) = (ln (xi − 2)) + (ln (10 − xi )) − xi
i=1 i=1

subject to 2.0001 ≤ xi ≤ 10, i ∈ 1, 2, ..., 10. The global minimum is located at

x∗ ≈ f (9.351, ...., 9.351), f (x∗ ) ≈ −45.778.
 A literature survey of benchmark functions 173

89 Pintér Function (Pintér, 1996) (continuous, differentiable, non-separable, scalable,

multimodal)

∑
D ∑
D ∑
D
( )
f89 (x) = ix2i + 20isin2 A+ ilog10 1 + iB 2
i=1 i=1 i=1

where

A = (xi−1 sin xi + sin xi+1 )

( )
B = x2i−1 − 2xi + 3xi+1 − cos xi + 1

where x0 = xD and xD+1 = x1 , subject to −10 ≤ xi ≤ 10. The global minima is

located at x∗ = f (0, · · · , 0), f (x∗ ) = 0.
90 Periodic Function (Ali et al., 2005) (Separable)

f90 (x) = 1 + sin2 (x1 ) + sin2 (x2 ) − 0.1e−(x1 +x2 )

2 2

subject to −10 ≤ xi ≤ 10. The global minimum is located at x∗ = f (0, 0),

f (x∗ ) = 0.9.
91 Powell Singular Function (Powell, 1962) (continuous, differentiable, non-separable
Scalable, unimodal)

∑
D/4
2
f91 (x) = (x4i−3 + 10x4i−2 )
i=1
2 4
+5(x4i−1 − x4i ) + (x4i−2 − x4i−1 )
4
+10(x4i−3 − x4i )

subject to −4 ≤ xi ≤ 5. The global minima is located at

x∗ = f (3, −1, 0, 1, · · · , 3, −1, 0, 1), f (x∗ ) = 0.
92 Powell Singular Function 2 (Fu et al., 2006) (continuous, differentiable,
non-separable Scalable, unimodal)

∑
D−2
2
f92 (x) = (xi−1 + 10xi )
i=1
2 4
+5(xi+1 − xi+2 ) + (xi − 2xi+1 )
4
+10(xi−1 − xi+2 )

subject to −4 ≤ xi ≤ 5. The global minimum is f (x∗ ) = 0.

93 Powell Sum Function (Rahnamyan et al., 2007a) (continuous, differentiable,
Separable Scalable, unimodal)
D  i+1
∑  
f93 (x) = xi 
i=1

subject to −1 ≤ xi ≤ 1. The global minimum is f (x∗ ) = 0.

174 M. Jamil and X-S. Yang

94 Price Function 1 (Price, 1977) (continuous, non-differentiable, Separable

Non-Scalable, multimodal)

f94 (x) = (|x1 | − 5)2 + (|x2 | − 5)2

subject to −500 ≤ xi ≤ 500. The global minimum are located at

x∗ =f ({−5, −5},{−5, 5}, {5, −5}, {5, 5}), f (x∗ ) = 0.
95 Price Function 2 (Price, 1977) (continuous, differentiable, non-separable
Non-Scalable, multimodal)

f95 (x) = 1 + sin2 x1 + sin2 x2 − 0.1e−x1 −x2

2 2

subject to −10 ≤ xi ≤ 10. The global minimum is located at x∗ = f (0 · · · 0),

f (x∗ ) = 0.9.
96 Price Function 3 (Price, 1977) (continuous, differentiable, non-separable
Non-Scalable, multimodal)
[ ]2
f96 (x) = 100(x2 − x21 )2 + 6 6.4(x2 − 0.5)2 − x1 − 0.6

subject to −500 ≤ xi ≤ 500. The global minimum are located at

x∗ =f ({−5, −5},{−5, 5}, {5, −5}, {5, 5}), f (x∗ ) = 0.
97 Price Function 4 (Price, 1977) (continuous, differentiable, non-separable,
non-scalable, multimodal)

f97 (x) = (2x31 x2 − x32 )2 + (6x1 − x22 + x2 )2

subject to −500 ≤ xi ≤ 500. The three global minima are located at

x∗ = f ({0, 0},{2, 4}, {1.464, −2.506}), f (x∗ ) = 0.
98 Qing Function (Qing, 2006) (continuous, differentiable, Separable Scalable,
multimodal)

∑
D
f98 (x) = (x2i − i)2
i=1

√
subject to −500 ≤ xi ≤ 500. The global minima are located at x∗ = f (± i),
f (x∗ ) = 0.
99 Quadratic Function (continuous, differentiable, non-separable, non-scalable)

f99 (x) = −3803.84 − 138.08x1 − 232.92x2

+128.08x21 + 203.64x22 + 182.25x1 x2

subject to −10 ≤ xi ≤ 10. The global minimum is located at

x∗ = f (0.19388, 0.48513), f (x∗ ) = −3873.7243.
 A literature survey of benchmark functions 175

100 Quartic Function (Storn and Price, 1996) (continuous, differentiable, separable,
scalable)

∑
D
f100 (x) = ix4i + random[0, 1)
i=1

subject to −1.28 ≤ xi ≤ 1.28. The global minima is located at x∗ = f (0, · · · , 0),

f (x∗ ) = 0.
101 Quintic Function (Mishra, 2006f) (continuous, differentiable, separable,
non-scalable, multimodal)

∑
D
f101 (x) = |x5i − 3x4i + 4x3i + 2x2i − 10xi − 4|
i=1

subject to −10 ≤ xi ≤ 10. The global minimum is located at x∗ = f (–1 or 2),

f (x∗ ) = 0.
102 Rana Function (Price et al., 2005) (continuous, differentiable, non-separable,
scalable, multimodal)

∑
D−2
f102 (x) = (xi+1 + 1)cos(t2 )sin(t1 ) + xi ∗ cos(t1 )sin(t2 )
i=0

√
√ to −500 ≤ xi ≤ 500, where t1 =
subject ∥xi+1 + xi + 1∥ and
t2 = ∥xi+1 − xi + 1∥.
103 Ripple Function 1 (continuous, differentiable, non-separable, non-scalable,
multimodal)

∑
2
xi −0.1 2
f103 (x) = −e-2 ln2( 0.8 ) (sin6 (5πxi ) + 0.1cos2 (500πxi ))
i=1

subject to 0 ≤ xi ≤ 1. It has one global minimum and 252004 local minima. The
global form of the function consists of 25 holes, which forms a 5 × 5 regular grid.
Additionally, the whole function landscape is full of small ripples caused by high
frequency cosine function which creates a large number of local minima.
104 Ripple Function 25 (continuous, differentiable, non-separable, non-scalable,
multimodal)

∑
2
xi −0.1 2
f104 (x) = −e-2 ln2( 0.8 ) (sin6 (5πxi ))
i=1

subject to 0 ≤ xi ≤ 1. It has one global form of the Ripple function 1 without any
ripples due to absence of cosine term.
176 M. Jamil and X-S. Yang

105 Rosenbrock Function 1 (Rosenbrock, 1960) (continuous, differentiable,

non-separable, scalable, unimodal)

∑
D−1
[ ]
f105 (x) = 100(xi+1 − x2i )2 + (xi − 1)2
i=1

subject to −30 ≤ xi ≤ 30. The global minima is located at x∗ = f (1, · · · , 1),

f (x∗ ) = 0.
106 Rosenbrock Modified Function (continuous, differentiable, non-separable,
non-scalable, multimodal)

f106 (x) = 74 + 100(x2 − x21 )2 + (1 − x)2

(x1 +1)2 +(x2 +1)2
−400e− 0.1

subject to −2 ≤ xi ≤ 2. In this function, a Gaussian bump at (−1, 1) is added,

which causes a local minimum at (1, 1) and global minimum is located at
x∗ = f (−1, −1), f (x∗ ) = 0. This modification makes it a difficult to optimise
because local minimum basin is larger than the global minimum basin.
107 Rotated Ellipse Function (continuous, differentiable, non-separable, non-scalable,
unimodal)
√
f107 (x) = 7x21 − 6 3x1 x2 + 13x22

subject to −500 ≤ xi ≤ 500. The global minimum is located at x∗ = f (0, 0),

f (x∗ ) = 0.
108 Rotated Ellipse Function 2 (Price et al., 2005) (continuous, differentiable,
non-separable, non-scalable, unimodal)

f108 (x) = x21 − x1 x2 + x22

subject to −500 ≤ xi ≤ 500. The global minimum is located at x∗ = f (0, 0),

f (x∗ ) = 0.
s
109 Rump Function (Moore, 1988) (continuous, differentiable, non-separable,
non-scalable, unimodal)
x1
f109 (x) = (333.75 − x21 )x62 + x21 (11x21 x22 − 121x42 − 2) + 5.5x82 +
2x2

subject to −500 ≤ xi ≤ 500. The global minimum is located at x∗ = f (0, 0),

f (x∗ ) = 0.
110 Salomon Function (Salomon, 1996) (continuous, differentiable, non-separable,
scalable, multimodal)
( vuD ) v
uD
u∑ u∑
f110 (x) = = 1 − cos 2π t xi + 0.1t
2 x2i
i=1 i=1
 A literature survey of benchmark functions 177

subject to −100 ≤ xi ≤ 100. The global minimum is located at x∗ = f (0, 0),

f (x∗ ) = 0.
111 Sargan Function (Dixon and Szegó, 1978) (continuous, differentiable,
non-separable, scalable, multimodal)
∑ ( ∑ )
f111 (x) = = D x2i + 0.4 xi xj
i=1 j̸=1

subject to −100 ≤ xi ≤ 100. The global minimum is located at x∗ = f (0, · · · , 0),

f (x∗ ) = 0.
112 Scahffer Function 1 (Mishra, 2006g) (continuous, differentiable, non-separable,
non-scalable, unimodal)

sin2 (x21 + x22 )2 − 0.5

f112 (x) = 0.5 +
1 + 0.001(x21 + x22 )2

subject to −100 ≤ xi ≤ 100. The global minimum is located at x∗ = f (0, 0),

f (x∗ ) = 0.
113 Scahffer Function 2 (Mishra, 2006g) (continuous, differentiable, non-separable,
non-scalable, unimodal)

sin2 (x21 − x22 )2 − 0.5

f113 (x) = 0.5 +
1 + 0.001(x21 + x22 )2

subject to −100 ≤ xi ≤ 100. The global minimum is located at x∗ = f (0, 0),

f (x∗ ) = 0.
114 Scahffer Function 3 (Mishra, 2006g) (continuous, differentiable, non-separable,
non-scalable, unimodal)
(  )
 
sin2 cos x21 − x22  − 0.5
f114 (x) = 0.5 +
1 + 0.001(x21 + x22 )2

subject to −100 ≤ xi ≤ 100. The global minimum is located at

x∗ = f (0, 1.253115), f (x∗ ) = 0.00156685.
115 Scahffer Function 4 (Mishra, 2006g) (continuous, differentiable, non-separable,
non-scalable, unimodal)
( )
cos2 sin(x21 − x22 ) − 0.5
f115 (x) = 0.5 +
1 + 0.001(x21 + x22 )2

subject to −100 ≤ xi ≤ 100. The global minimum is located at

x∗ = f (0, 1.253115), f (x∗ ) = 0.292579.
178 M. Jamil and X-S. Yang

116 Schmidt Vetters Function (Lootsma, 1972) (continuous, differentiable,

non-separable, non-scalable, multimodal)
1 ( πx + x )
2 3
f116 (x) = + sin
1 + (x1 − x2 ) 2 2
x1 +x2
−2)2
+e( x2

The global minimum is located at x∗ = f (0.78547, 0.78547, 0.78547), f (x∗ ) = 3.

117 Schumer Steiglitz Function (Schumer and Steiglitz, 1968) (continuous,
differentiable, separable, scalable, unimodal)

∑
D

f117 (x) = x4i

i=1

The global minimum is located at x∗ = f (0, . . . , 0), f (x∗ ) = 0.

118 Schwefel Function (Schwefel, 1981) (continuous, differentiable, partially-separable,
scalable, unimodal)

(∑
D
)α
f118 (x) = x2i
i=1

where α ≥ 0, subject to −100 ≤ xi ≤ 100. The global minima is located at

x∗ = f (0, · · · , 0), f (x∗ ) = 0.
119 Schwefel Function 1.2 (Schwefel, 1981) (continuous, differentiable, non-separable,
scalable, unimodal)
 2
∑ ∑
D
i
f119 (x) =  xj 
i=1 j=1

subject to −100 ≤ xi ≤ 100. The global minima is located at x∗ = f (0, · · · , 0),

f (x∗ ) = 0.
120 Schwefel Function 2.4 (Schwefel, 1981) (continuous, differentiable, separable,
non-scalable, multimodal)

∑
D
f120 (x) = (xi − 1)2 + (x1 − x2i )2
i=1

subject to 0 ≤ xi ≤ 10. The global minima is located at x∗ = f (1, · · · , 1),

f (x∗ ) = 0.
121 Schwefel Function 2.6 (Schwefel, 1981) (continuous, differentiable, non-separable,
non-scalable, unimodal)

f121 (x) = max(|x1 + 2x2 − 7|, |2x1 + x2 − 5|)

subject to −100 ≤ xi ≤ 100. The global minima is located at x∗ = f (1, 3),

f (x∗ ) = 0.
 A literature survey of benchmark functions 179

122 Schwefel Function 2.20 (Schwefel, 1981) (continuous, non-differentiable,

separable, scalable, unimodal)

∑
n
f122 (x) = − |xi |
i=1

subject to −100 ≤ xi ≤ 100. The global minima is located at x∗ = f (0, · · · , 0),

f (x∗ ) = 0.
123 Schwefel Function 2.21 (Schwefel, 1981) (continuous, non-differentiable,
separable, scalable, unimodal)

f123 (x) = max |xi |

1≤i≤D

subject to −100 ≤ xi ≤ 100. The global minima is located at x∗ = f (0, · · · , 0),

f (x∗ ) = 0.
124 Schwefel Function 2.22 (Schwefel, 1981) (continuous, differentiable,
non-separable, scalable, unimodal)

∑
D ∏
n
f124 (x) = |xi | + |xi |
i=1 i=1

subject to −100 ≤ xi ≤ 100. The global minima is located at x∗ = f (0, · · · , 0),

f (x∗ ) = 0.
125 Schwefel Function 2.23 (Schwefel, 1981) (continuous, differentiable,
non-separable, scalable, unimodal)

∑
D
f125 (x) = x10
i
i=1

subject to −10 ≤ xi ≤ 10. The global minima is located at x∗ = f (0, · · · , 0),

f (x∗ ) = 0.
126 Schwefel Function 2.25 (Schwefel, 1981) (continuous, differentiable, separable,
non-scalable, multimodal)

∑
D
f126 (x) = (xi − 1)2 + (x1 − x2i )2
i=2

subject to 0 ≤ xi ≤ 10. The global minima is located at x∗ = f (1, · · · , 1),

f (x∗ ) = 0.
127 Schwefel Function 2.26 (Schwefel, 1981) (continuous, differentiable, separable,
scalable, multimodal)

1 ∑ √
D
f127 (x) = − xi sin |xi |
D i=1
180 M. Jamil and X-S. Yang

subject to −500 ≤ xi ≤ 500. The global minimum is located at

x∗ = ±[π(0.5 + k)]2 , f (x∗ ) = −418.983.
128 Schwefel Function 2.36 (Schwefel, 1981) (continuous, differentiable, separable,
scalable, multimodal)

f128 (x) = −x1 x2 (72 − 2x1 − 2x2 )

subject to 0 ≤ xi ≤ 500. The global minimum is located at x∗ = f (12, · · · , 12),

f (x∗ ) = −3456.
129 Shekel Function 5 (Opačić, 1973) (continuous, differentiable, non-separable,
scalable, multimodal)

∑
5
1
f129 (x) = −
∑
4
2
i=1 (xj − aij ) + ci
j=1

   
4444 0.1
1 1 1 1 0.2
   
where A = [Aij ] =    
8 8 8 8, c = ci = 0.2
6 6 6 6 0.4
3737 0.4
subject to 0 ≤ xj ≤ 10. The global minima is located at x∗ = f (4, 4, 4, 4),
f (x∗ ) ≈ −10.1499.
130 Shekel Function 7 (Opačić, 1973) (continuous, differentiable, non-separable,
scalable, multimodal)

∑
7
1
f130 (x) = −
∑
4
2
i=1 (xj − aij ) + ci
j=1

   
4444 0.1
1 1 1 1 0.2
   
8 8 8 8 0.2
   
where A = [Aij ] = 6 6 6 6, c = ci = 
  
0.4
3 7 3 7 0.4
   
2 9 2 9 0.6
5533 0.3
subject to 0 ≤ xj ≤ 10. The global minima is located at x∗ = f (4, 4, 4, 4),
f (x∗ ) ≈ −10.3999.
131 Shekel Function 10 (Opačić, 1973) (continuous, differentiable, non-separable,
scalable, multimodal)

∑
10
1
f131 (x) = −
∑
4
2
i=1 (xj − aij ) + ci
j=1
 A literature survey of benchmark functions 181
   
4 4 4 4 0.1
1 1 1 1  0.2
   
8 8 8 8  0.2
   
6 6 6 6  0.4
   
3 7 3 7  0.4
where A = [Aij ] =  , c = ci =  
 0.6
2 9 2 9   
5 5 3 3  0.3
   
8 1 8 1  0.7
   
6 2 6 2  0.5
7 3.6 7 3.6 0.5
subject to 0 ≤ xj ≤ 10. The global minima is located at x∗ = f (4, 4, 4, 4),
f (x∗ ) ≈ −10.5319.
132 Shubert Function (Hennart, 1982) (continuous, differentiable, separable,
non-scalable, multimodal)
 
∏
n ∑5
f132 (x) =  cos((j + 1)xi + j)
i=1 j=1

subject to −10 ≤ xi ≤ 10, i ∈ 1, 2, · · · , n. The 18 global minima are located at

x∗ = f ({−7.0835, 4.8580}, {−7.0835, −7.7083},

{−1.4251, −7.0835}, { 5.4828, 4.8580},
{−1.4251, −0.8003}, { 4.8580, 5.4828},
{−7.7083, −7.0835}, {−7.0835, −1.4251},
{−7.7083, −0.8003}, {−7.7083, 5.4828},
{−0.8003, −7.7083}, {−0.8003, −1.4251},
{−0.8003, 4.8580}, {−1.4251, 5.4828},
{ 5.4828, −7.7083}, { 4.8580, −7.0835},
{ 5.4828, −1.4251}, { 4.8580, −0.8003}),

f (x∗ ) ≃ −186.7309.
133 Shubert Function 3 (Adorio and Dilman, 2005) (continuous, differentiable,
separable, non-scalable, multimodal)
 
∑
D ∑ 5
f133 (x) =  jsin((j + 1)xi + j)
i=1 j=1

subject to −10 ≤ xi ≤ 10. The global minimum is f (x∗ ) ≃ −29.6733337 with

multiple solutions.
134 Shubert Function 4 (Adorio and Dilman, 2005) (continuous, differentiable,
separable, non-scalable, multimodal)
 
∑
D ∑ 5
f134 (x) =  jcos((j + 1)xi + j)
i=1 j=1
182 M. Jamil and X-S. Yang

subject to −10 ≤ xi ≤ 10. The global minimum is f (x∗ ) ≃ −25.740858 with

multiple solutions.
135 Schaffer Function F6 (Schaffer et al., 1989) (continuous, differentiable,
non-separable, scalable, multimodal)
√
∑
D sin2 x2i + x2i+1 − 0.5
f135 (x) = 0.5 + [ ]2
i=1 1 + 0.001(x2i + x2i+1 )

subject to −100 ≤ xi ≤ 100. The global minimum is located at x∗ = f (0, · · · , 0),

f (x∗ ) = 0.
136 Sphere Function (Schumer and Steiglitz, 1968) (continuous, differentiable,
separable, scalable, multimodal)

∑
D
f136 (x) = x2i
i=1

subject to 0 ≤ xi ≤ 10. The global minima is located x∗ = f (0, · · · , 0),

f (x∗ ) = 0.
137 Step Function (discontinuous, non-differentiable, separable, scalable, unimodal)

∑
D
f137 (x) = (⌊|xi |⌋)
i=1

subject to −100 ≤ xi ≤ 100. The global minima is located x∗ = f (0, · · · , 0) = 0,

f (x∗ ) = 0.
138 Step Function 2 (Báck and Schwefel, 1993) (discontinuous, non-differentiable,
separable, scalable, unimodal)

∑
D
2
f138 (x) = (⌊xi + 0.5⌋)
i=1

subject to −100 ≤ xi ≤ 100. The global minima is located

x∗ = f (0.5, · · · , 0.5) = 0, f (x∗ ) = 0.
139 Step Function 3 (discontinuous, non-differentiable, separable, scalable, unimodal)

∑
D
( )
f139 (x) = ⌊x2i ⌋
i=1

subject to −100 ≤ xi ≤ 100. The global minima is located x∗ = f (0, · · · , 0) = 0,

f (x∗ ) = 0.
 A literature survey of benchmark functions 183

140 Stepint Function (discontinuous, non-differentiable, separable, scalable, unimodal)

∑
D
f140 (x) = 25 + (⌊xi ⌋)
i=1

subject to −5.12 ≤ xi ≤ 5.12. The global minima is located x∗ = f (0, · · · , 0),

f (x∗ ) = 0.
141 Streched V Sine Wave Function (Schaffer et al., 1989) (continuous, differentiable,
non-separable, scalable, unimodal)

∑
D−1 [ ]
f141 (x) = (x2i+1 + x2i )0.25 sin2 {50(x2i+1 + x2i )0.1 } + 0.1
i=1

subject to −10 ≤ xi ≤ 10. The global minimum is located x∗ = f (0, 0),

f (x∗ ) = 0.
142 Sum Squares Function (Hedar, n.d.) (continuous, differentiable, separable,
scalable, unimodal)

∑
D
f142 (x) = ix2i
i=1

subject to −10 ≤ xi ≤ 10. The global minima is located x∗ = f (0, · · · , 0),

f (x∗ ) = 0.
143 Styblinski-Tang Function (Silagadze, 2007) (continuous, differentiable,
non-separable, non-scalable, multimodal)

1∑ 4
n
f143 (x) = (x − 16x2i + 5xi )
2 i=1 i

subject to −5 ≤ xi ≤ 5. The global minimum is located

x∗ = f (−2.903534, −2.903534), f (x∗ ) = −78.332.
144 Holder Table Function 1 (Mishra, 2006f) (continuous, differentiable, separable,
non-scalable, multimodal)

f144 (x) = −|cos(x1 )cos(x2 )e|1−(x1 +x2 )

0.5
/π|
|

subject to −10 ≤ xi ≤ 10.

The four global minima are located at x∗ = f (±9.646168, ±9.646168),
f (x∗ ) = −26.920336.
145 Holder Table Function 2 (Mishra, 2006f) (continuous, differentiable, separable,
non-scalable, multimodal)

f145 (x) = −|sin(x1 )cos(x2 )e|1−(x1 +x2 )

0.5
/π|
|
184 M. Jamil and X-S. Yang

subject to −10 ≤ xi ≤ 10.

The four global minima are located at x∗ = f (±8.055023472141116,

±9.664590028909654), f (x∗ ) = −19.20850.
146 Carrom Table Function (Mishra, 2006f) (continuous, differentiable, non-separable,
non-scalable, multimodal)

f146 (x) = −[(cos(x1 )cos(x2 )

exp |1 − [(x21 + x22 )0.5 ]/π|)2 ]/30

subject to −10 ≤ xi ≤ 10.

The four global minima are located at x∗ = f (±9.646157266348881,

±9.646134286497169), f (x∗ ) = −24.1568155.
147 Testtube Holder Function (Mishra, 2006f) (continuous, differentiable, separable,
non-scalable, multimodal)
[
f147 (x) = −4 (sin(x1 )cos(x2 )
]
e|cos[(x1 +x2 )/200]| )
2 2

subject to −10 ≤ xi ≤ 10. The two global minima are located at

x∗ = f (±π/2, 0), f (x∗ ) = −10.872300.
148 Trecanni Function (Dixon and Szegó, 1978) (continuous, differentiable, separable,
non-scalable, unimodal)

f148 (x) = x41 − 4x31 + 4x1 + x22

subject to −5 ≤ xi ≤ 5. The two global minima are located at

x∗ = f ({0, 0}, {−2, 0}), f (x∗ ) = 0.
149 Trid Function 6 (Hedar, n.d.) (continuous, differentiable, non-separable,
non-scalable, multimodal)

∑
D
2
∑
D
f149 (x) = (xi − 1) − xi xi−1
i=1 i=1

subject to −62 ≤ xi ≤ 62 . The global minima is located at f (x∗ ) = −50.

150 Trid Function 10 (Hedar, n.d.) (continuous, differentiable, non-separable,
non-scalable, multimodal)

∑
D
2
∑
D
f150 (x) = (xi − 1) − xi xi−1
i=1 i=1

subject to −100 ≤ xi ≤ 100. The global minima is located at f (x∗ ) = −200.

 A literature survey of benchmark functions 185

151 Trefethen Function (Adorio and Dilman, 2005) (continuous, differentiable,

non-separable, non-scalable, multimodal)

f151 (x) = esin(50x1 ) + sin(60ex2 )

+sin(70sin(x1 )) + sin(sin(80x2 ))
1
−sin(10(x1 + x2 )) + (x21 + x22 )
4

subject to −10 ≤ xi ≤ 10. The global minimum is located at

x∗ = f (−0.024403, 0.210612), f (x∗ ) = −3.30686865.
152 Trigonometric Function 1 (Dixon and Szegó, 1978) (continuous, differentiable,
non-separable, scalable, multimodal)

∑
D ∑
D
f152 (x) = [D − cos xj
i=1 j=1

+i(1 − cos(xi ) − sin(xi ))]2

subject to 0 ≤ xi ≤ pi. The global minimum is located at x∗ = f (0, · · · , 0),

f (x∗ ) = 0
153 Trigonometric Function 2 (Fu et al., 2006) (continuous, differentiable,
non-separable, scalable, multimodal)

∑
D
[ ] [ ]
f153 (x) = 1 + 8 sin2 7(xi − 0.9)2 + 6 sin2 14(x1 − 0.9)2 + (xi − 0.9)2
i=1

subject to −500 ≤ xi ≤ 500. The global minimum is located at

x∗ = f (0.9, · · · , 0.9), f (x∗ ) = 1
154 Tripod Function (Rahnamyan et al., 2007a) (discontinuous, non-differentiable,
non-separable, non-scalable, multimodal)

f154 (x) = p(x2 )(1 + p(x1 ))

+|x1 + 50p(x2 )(1 − 2p(x1 ))|
+|x2 + 50(1 − 2p(x2 ))|

subject to −100 ≤ xi ≤ 100, where p(x) = 1 for x ≥ 0. The global minimum is

located at x∗ = f (0, −50), f (x∗ ) = 0.
155 Ursem Function 1 (Rónkkónen, 2009) (continuous, differentiable, separable,
non-scalable, multimodal)

f155 (x) = −sin(2x1 − 0.5π) − 3cos(x2 ) − 0.5x1

subject to −2.5 ≤ x1 ≤ 3 and −2 ≤ x2 ≤ 2, and has single global and local

minima.
186 M. Jamil and X-S. Yang

156 Ursem Function 3 (Rónkkónen, 2009) (continuous, differentiable, non-separable,

non-scalable, multimodal)

2 − |x2 | 3 − |x1 |
f156 (x) = −sin(2.2πx1 + 0.5π). .
2 2
2 − |x2 | 2 − |x1 |
−sin(0.5πx22 + 0.5π). .
2 2
subject to −2 ≤ x1 ≤ 2 and −1.5 ≤ x2 ≤ 1.5, and has single global minimum
and four regularly spaced local minima positioned in a direct line, such that global
minimum is in the middle.
157 Ursem Function 4 (Rónkkónen, 2009) (continuous, differentiable, non-separable,
non-scalable, multimodal)

√
2− x21 + x22
f157 (x) = −3sin(0.5πx1 + 0.5π).
4
subject to −2 ≤ xi ≤ 2, and has single global minimum positioned at the middle
and four local minima at the corners of the search space.
158 Ursem Waves Function (Rónkkónen, 2009) (continuous, differentiable,
non-separable, non-scalable, multimodal)

f158 (x) = −0.9x21 + (x22 − 4.5x22 )x1 x2

+4.7cos(3x1 − x22 (2 + x1 ))sin(2.5πx1 )

subject to −0.9 ≤ x1 ≤ 1.2 and −1.2 ≤ x2 ≤ 1.2, and has single global minimum
and nine irregularly spaced local minima in the search space.
159 Venter Sobiezcczanski-Sobieski Function (Begambre and Laier, 2009) (continuous,
differentiable, separable, non-scalable)

f159 (x) = x21 − 100cos(x1 )2

−100cos(x21 /30) + x22
−100cos(x2 )2 − 100cos(x22 /30)

subject to −50 ≤ xi ≤ 50. The global minimum is located at x∗ = f (0, 0),

f (x∗ ) = −400.
160 Watson Function (Schwefel, 1981) (continuous, differentiable, non-separable,
scalable, unimodal)
[ 4  2 ]2
∑
29 ∑ ∑5
f160 (x) = ((j − 1)ai xj+1 ) − 
j
ai xj+1  − 1 + x21
j

i=0 j=0 j=0

subject to |xi | ≤ 10, where the coefficient ai = i/29.0. The global minimum is
located at x∗ = f (−0.0158, 1.012, −0.2329, 1.260, −1.513, 0.9928),
f (x∗ ) = 0.002288.
 A literature survey of benchmark functions 187

161 Wayburn Seader Function 1 (Wayburn and Seader, 1987) (continuous,

differentiable, non-separable, scalable, unimodal)

f161 (x) = (x61 + x42 − 17)2 + (2x1 + x2 − 4)2

The global minimum is located at x∗ = f {(1, 2), (1.597, 0.806)}, f (x∗ ) = 0.

162 Wayburn Seader Function 2 (Wayburn and Seader, 1987) (continuous,
differentiable, non-separable, scalable, unimodal)

[ ]2
f162 (x) = 1.613 − 4(x1 − 0.3125)2 − 4(x2 − 1.625)2 + (x2 − 1)2

subject to −500 ≤ 500. The global minimum is located at

x∗ = f {(0.2, 1), (0.425, 1)}, f (x∗ ) = 0.
163 Wayburn Seader Function 3 (Wayburn and Seader, 1987) (continuous,
differentiable, non-separable, scalable, unimodal)

x31 [ ]2
f163 (x) = 2 − 8x21 + 33x1 − x1 x2 + 5 + (x1 − 4)2 + (x2 − 5)2 − 4
3

subject to −500 ≤ 500. The global minimum is located at x∗ = f (5.611, 6.187),

f (x∗ ) = 21.35.
164 W / Wavy Function (Courrieu, 1997) (continuous, differentiable, separable,
scalable, multimodal)

1 ∑
D
−x2i
f164 (x) = 1 − cos(kxi )e 2
D i=1

subject to −π ≤ xi ≤ π. The global minimum is located at x∗ = f (0, 0),

f (x∗ ) = 0.
The number of local minima is kn and (k + 1)n for odd and even k respectively.
For D = 2 and k = 10, there are 121 local minima.
165 Weierstrass Function (Suganthan et al., 2005)(continuous, differentiable, separable,
scalable, multimodal)

n [ kmax
∑ ∑
f165 (x) = ak cos(2πbk (xi + 0.5))
i=1 k=0
∑
kmax ]
−n k k
a cos(πb )
k=0

subject to −0.5 ≤ xi ≤ 0.5. The global minima is located at x∗ = f (0, · · · , 0),

f (x∗ ) = 0.
188 M. Jamil and X-S. Yang

166 Whitley Function (Whitley et al., 1996) (continuous, differentiable, non-separable,

scalable, multimodal)

D [
D ∑
∑ (100(x2 − xj )2 + (1 − xj )2 )2
i
f166 (x) =
i=1 j=1
4, 000
]
( )
−cos 100(x2i − xj ) + (1 − xj ) + 1
2 2

combines a very steep overall slope with a highly multimodal area around the
global minimum located at xi = 1, where i = 1, ..., D.
167 Wolfe Function (Schwefel, 1981) (continuous, differentiable, separable, scalable,
multimodal)
4 2
f167 (x) = (x + x22 − x1 x2 )0.75 + x3
3 1
subject to 0 ≤ xi ≤ 2. The global minima is located at x∗ = f (0, · · · , 0),
f (x∗ ) = 0.
168 Xin-She Yang Function 1 (discontinuous, differentiable, separable, scalable,
multimodal)

This is a generic stochastic and non-smooth function proposed in Yang (2010a,

2010b).

∑
D
i
f168 (x) = ϵi |xi |
i=1

subject to −5 ≤ xi ≤ 5. The variable ϵi , (i = 1, 2, · · · , D) is a random variable

uniformly distributed in [0, 1]. The global minima is located at x∗ = f (0, · · · , 0),
f (x∗ ) = 0.
169 Xin-She Yang Function 2 (Yang, 2010a,b) (discontinuous, non-differentiable,
non-separable, scalable, multimodal)
(∑
D ) [ ∑D ]
f169 (x) = |xi | exp − 2
sin(xi )
i=1 i=1

subject to −2π ≤ xi ≤ 2π. The global minima is located at x∗ = f (0, · · · , 0),

f (x∗ ) = 0.
170 Xin-She Yang Function 3 (Yang, 2010a,b) (discontinuous, non-differentiable,
non-separable, scalable, multimodal)
[ ]
2 ∏
∑ ∑D D
− D 2m
−
f170 (x) = e i=1 (x i /β)
− 2e i=1 (x i )
. 2
cos (xi )
i=1

subject to −20 ≤ xi ≤ 20. The global minima for m = 5 and β = 15 is located at

x∗ = f (0, · · · , 0), f (x∗ ) = −1.
 A literature survey of benchmark functions 189

171 Xin-She Yang Function 4 (Yang, 2010a,b) (discontinuous, non-differentiable,

non-separable, scalable, multimodal)
[ ]
2 ∏
∑ ∑D D
− D 2m
− −π)
f171 (x) = e i=1 (x i /β)
− 2e i=1 (x i
. 2
cos (xi )
i=1

subject to −10 ≤ xi ≤ 10. The global minima for m = 5 and β = 15 is located at

x∗ = f (π, · · · , π), f (x∗ ) = −1.
172 Zakharov Function (Rahnamyan et al., 2007a) (continuous, differentiable,
non-separable, scalable, multimodal)
( )2 ( )4
∑
n
1∑
n
1∑
n
f172 (x) = x2i + ixi + ixi
i=1
2 i=1 2 i=1

subject to −5 ≤ xi ≤ 10. The global minima is located at x∗ = f (0, · · · , 0),

f (x∗ ) = 0.
173 Zettl Function (Schwefel, 1995) (continuous, differentiable, non-separable,
non-scalable, unimodal)

f173 (x) = (x21 + x22 − 2x1 )2 + 0.25x1

subject to −5 ≤ xi ≤ 10. The global minima is located at x∗ = f (−0.0299, 0),

f (x∗ ) = −0.003791.
17 Zirilli or Aluffi-Pentini’s Function (Ali et al., 2005) (continuous, differentiable,
separable, non-scalable, unimodal)

f174 (x) = 0.25x41 − 0.5x21 + 0.1x1 + 0.5x22

subject to −10 ≤ xi ≤ 10. The global minimum is located at x∗ = (−1.0465, 0),

f (x∗ ) ≈ −0.3523.
175 Zirilli Function 2 (continuous, differentiable, separable, non-scalable, multimodal)
[ ]
f175 (x) = 0.5x21 + 0.5 1 − cos(2x1 ) + x22

subject to −500 ≤ xi ≤ 500. The global minimum is located at x∗ = (0, 0),

f (x∗ ) = 0.

4 Conclusions

Test functions are important to validate and compare optimisation algorithms, this is
especially true for newly developed algorithms. Here, we attempted to provide the most
comprehensive and concise list of known benchmarks or test functions. Any functions
that is left out is just unintentional. The list is compiled on all the resources all the
literature known to us by the time of writing. It can be expected that majority of these
functions can be used for testing new optimisation algorithms so as to provide a more
complete view on the performance of any algorithms of interest.
190 M. Jamil and X-S. Yang

References
Ackley, D.H. (1987) A Connectionist Machine for Genetic Hill-Climbing, Kluwer, Boston.
Adjiman, C.S., Sallwig, S., Flouda, C.A. and Neumaier, A. (1998) ‘A global optimization
method, aBB for general twice-differentiable NLPs-1, theoretical advances’, Computers
Chemical Engineering, Vol. 22, No. 9, pp.1137–1158.
Adorio, E.P. and Dilman, U.P. (2005) ‘MVF – multivariate test function
library in c for unconstrained global optimization methods’ [online]
http://www.geocities.ws/eadorio/mvf.pdf (accessed 20 January 2013).
Ali, M.M., Khompatraporn, C. and Zabinsky, Z.B. (2005) ‘A numerical evaluation of several
stochastic algorithms on selected continuous global optimization test problems’, Journal of
Global Optimization, Vol. 31, No. 4, pp.635–672.
Andrei, N. (2008) ‘An unconstrained optimization test functions collection’, Advanced Modeling
and Optimization, Vol. 10, No. 1, pp.147–161.
Auger, A., Hansen, N., Mauny, N., Ros, R. and Schoenauer, M. (2007) Bio-Inspired Continuous
Optimization: The Coming of Age, Invited Lecture, IEEE Congress on Evolutionary
Computation, NJ, USA.
Averick, B.M., Carter, R.G. and Moré, J.J. (1991) The MINIPACK-2 Test Problem Collection,
Mathematics and Computer Science Division, Agronne National Laboratory, Technical
Memorandum No. 150.
Averick, B.M., Carter, R.G., Moré, J.J. and Xue, G.L. (1992) The MINIPACK-2 Test Problem
Collection, Mathematics and Computer Science Division, Agronne National Laboratory,
Preprint MCS-P153-0692.
Báck, T. and Schwefel, H.P. (1993) ‘An overview of evolutionary algorithm for parameter
optimization’, Evolutionary Computation, Vol. 1, No. 1, pp.1–23.
Begambre, O. and Laier, J.E. (2009) ‘A hybrid particle swarm optimization – simplex algorithm
(PSOS) for structural damage identification’, Journal of Advances in Engineering Software,
Vol. 40, No. 9, pp.883–891.
Bekey, G.A. and Ung, M.T. (1974) ‘A comparative evaluation of two global search algorithms’,
IEEE Transaction on Systems, Man and Cybernetics, Vol. 4, No. 1, pp.112–116.
Bersini, H., Dorigo, M. and Langerman, S. (1996) ‘Results of the first international contest on
evolutionary optimization’, IEEE International Conf. on Evolutionary Computation, Nagoya,
Japan, pp.611–615.
Biggs, M.C. (1971) ‘A new variable metric technique taking account of non-quadratic behaviour
of the objective function’, IMA Journal of Applied Mathematics, Vol. 8, No. 3, pp.315–327.
Bohachevsky, I.O., Johnson, M.E. and Stein, M.L. (1986) ‘General simulated annealing for
function optimization’, Technometrics, Vol. 28, No. 3, pp.209–217.
Boyer, D.O., Martfnez, C.H. and Pedrajas, N.G. (2005) ’Crossover operator for evolutionary
algorithms based on population features’, Journal of Artificial Intelligence Research, Vol. 24,
No. 1, pp.1–48.
Brad, Y. (1970) ‘Comparison of gradient methods for the solution of nonlinear parametric
estimation problem’, SIAM Journal on Numerical Analysis, Vol. 7, No. 1, pp.157–186.
Branin Jr., F.H. (1972) ‘Widely convergent method of finding multiple solutions of simultaneous
nonlinear equations’, IBM Journal of Research and Development, Vol. 16, No. 5,
pp.504-522.
Chen, Y. (2003) Computer Simulation of Electron Positron Annihilation Processes, Technical
Report SLAC-Report-646, Stanford Linear Accelerator Center, Stanford University [online]
http://www.slac.stanford.edu/pubs/slacreports/slac-r-646.html.
 A literature survey of benchmark functions 191

Chung, C.J. and Reynolds, R.G. (1998) ‘CAEP: an evolution-based tool for real-valued function
optimization using cultural algorithms’, International Journal on Artificial Intelligence Tool,
Vol. 7, No. 3, pp.239–291.
Clerc, M. (1999) The Swarm and the Queen, Towards a Deterministic and Adaptive Particle
Swarm Optimization, IEEE Congress on Evolutionary Computation, Washington DC, USA,
pp.1951–1957.
Corana, A., Marchesi, M., Martini, C. and Ridella, S. (1987) ‘Minimizing multimodal functions
of continuous variables with simulated annealing algorithms’, ACM Transactions on
Mathematical Software, Vol. 13, No. 3, pp.262–280.
Courrieu, P. (1997) ‘The hyperbell algorithm for global optimization: a random walk using
Cauchy densities’, Journal of Global Optimization, Vol. 10, No. 1, pp.111–133.
Cragg, E.E. and Levy, A.V. (1969) ‘Study on supermemory gradient method for the minimization
of functions’, Journal of Optimization Theory and Applications, Vol. 4, No. 3, pp.191–205.
Csendes,T. and Ratz, D. (1997) ‘Subdivision-direction selection in interval methods for global
optimization’, SIAM Journal on Numerical Analysis, Vol. 34, No. 3, pp.922–938.
Damavandi, N. and Safavi-Naeini, S. (2005) ‘A hybrid evolutionary programming method
for circuit optimization’, IEEE Transaction on Circuit and Systems I, Vol. 52, No. 5,
pp.902–910.
deVillers, N. and Glasser, D. (1981) ‘A continuation method for nonlinear regression’, SIAM
Journal on Numerical Analysis, Vol. 18, No. 6, pp.1139–1154.
Dixon, L.C.W. and Price, R.C. (1989) ‘The truncated Newton method for sparse unconstrained
optimisation using automatic differentiation’, Journal of Optimization Theory and
Applications, Vol. 60, No. 2, pp.261–275.
Dixon, L.C.W. and Szegó, G.P. (Eds.) (1978) Towards Global Optimization 2, Elsevier,
Boston/Dordrecht/London.
El-Attar, R.A., Vidyasagar, M. and Dutta, S.R.K. (1979) ‘An algorithm for II-norm minimization
with application to nonlinear II-approximation’, SIAM Journal on Numerical Analysis,
Vol. 16, No. 1, pp.70–86.
Fletcher, R. and Powell, M.J.D. (1963) ‘A rapidly convergent descent method
for minimization’, Computer Journal, Vol. 62, No. 2, pp.163–168 [online]
http://galton.uchicago.edu/∼lekheng/courses/302/classics/fletcher-powell.pdf.
Flouda, C.A., Pardalos, P.M., Adjiman, C.S., Esposito, W.R., Gúmús, Z.H., Harding, S.T.,
Klepeis, J.L., Meyer, C.A. and Schweiger, C.A. (1999) Handbook of Test Problems in Local
and Global Optimization, Kluwer, Boston.
Fraley, C. (1989) Software Performances on Nonlinear Least-Squares Problems, Technical Report
No. STAN-CS-89-1244, Department of Computer Science, Stanford University [online]
http://www.dtic.mil/dtic/tr/fulltext/u2/a204526.pdf.
Fu, M.C., Hu, J. and Marcus, S.I. (2006) ‘Model-based randomized methods for global
optimization’, Proc. 17th International Symp. Mathematical Theory Networks Systems,
Kyoto, Japan, pp.355–365.
GAMS World (2000) GLOBAL Library [online] http://www.gamsworld.org/global/globallib.html.
GEATbx – The Genetic and Evolutionary Algorithm Toolbox for Matlab [online]
http://www.geatbx.com/.
Goldstein, A.A. and Price, J.F. (1971) ‘On descent from local minima’, Mathematics and
Computation, Vol. 25, No. 115, pp.569–574.
Gordon, V.S. and Whitley, D. (1993) ‘Serial and parallel genetic algorithms as function
optimizers’, in S. Forrest (Ed.): 5th Intl. Conf. on Genetic Algorithms, pp.177–183, Morgan
Kaufmann.
192 M. Jamil and X-S. Yang

Gould, N.I.M., Orban, D. and Toint, P.L. (2001) CUTEr, A Constrained and Un-constrained
Testing Environment, Revisited [online] http://cuter.rl.ac.uk/cuter-www/problems.html
(accessed 14 July 2012).
Griewank, A.O. (1981) ‘Generalized descent for global optimization’, Journal of Optimization
Theory and Applications, Vol. 34, No. 1, pp.11–39.
Hartman, J.K. (1972) Some Experiments in Global Optimization [online]
http://ia701505.us.archive.org/9/items/someexperimentsi00hart/someexperimentsi00hart.pdf
(accessed 15 August 2012).
Hedar, A-R. (n.d.) Global Optimization Test Problems [online]
http://www-optima.amp.i.kyoto-u.ac.jp/member/student/hedar/Hedar files/TestGO.htm
(accessed 17 August 2012).
Hennart, J.P. (Ed.) (1982) ‘Numerical analysis’, Proc. 3rd AS Workshop, Lecture Notes in
Mathematics, Vol. 90, Springer.
Himmelblau, D.M. (1972) Applied Nonlinear Programming, McGraw-Hill, New York.
Jennrich, R.I. and Sampson, P.F. (1968) ‘Application of stepwise regression to non-linear
estimation’, Techometrics, Vol. 10, No. 1, pp.63–72
http://www.jstor.org/discover/10.2307/
1266224?uid=3737864&uid=2129&uid=2&uid=70&uid=4&sid=21101664491701.
Junior, A.D., Silva, R.S., Mundim, K.C. and Dardenne, L.E. (2004) ‘Performance
and parameterization of the algorithm simplified generalized simulated
annealing’, Genet. Mol. Biol., Vol. 27, No. 4, pp.616–622 [online]
http://www.scielo.br/scielo.php?script=sci arttext&pid=S1415-47572004000400024
&lng=en&nrm=iso; ISSN 1415-4757 [online]
http://dx.doi.org/10.1590/S1415-47572004000400024.
Lavi, A. and Vogel, T.P. (Eds.) (1966) Recent Advances in Optimization Techniques, John Wliley
& Sons, New York.
Lootsma, F.A. (Ed.) (1972) Numerical Methods for Non-Linear Optimization, Academic Press,
London, New York.
Mishra, S.K. (2006a) Performance of Differential Evolution and Particle Swarm
Methods on Some Relatively Harder Multi-modal Benchmark Functions [online]
http://mpra.ub.uni-muenchen.de/449/ (accessed 14 August 2012).
Mishra, S.K. (2006b) Performance of the Barter, the Differential Evolution and the Simulated
Annealing Methods of Global Optimization on Some New and Some Old Test Functions
[online] http://www.ssrn.com/abstract=941630 (accessed 14 August 2012).
Mishra, S.K. (2006c) Repulsive Particle Swarm Method on Some Difficult Test Problems of
Global Optimization [online] http://mpra.ub.uni-muenchen.de/1742/ (accessed 14 August
2012).
Mishra, S.K. (2006d) Performance of Repulsive Particle Swarm Method in Global
Optimization of Some Important Test Functions: A Fortran Program [online]
http://www.ssrn.com/abstract=924339 (accessed 14 August 2012).
Mishra, S.K. (2006e) Global Optimization by Particle Swarm Method: A Fortran Program,
Munich Research Papers in Economics [online] http://mpra.ub.uni-muenchen.de/874/
(accessed 14 August 2012).
Mishra, S.K. (2006f) Global Optimization By Differential Evolution and Particle Swarm
Methods: Evaluation on Some Benchmark Functions, Munich Research Papers in Economics
[online] http://mpra.ub.uni-muenchen.de/1005/ (accessed 14 August 2012).
Mishra, S.K. (2006g) Some New Test Functions For Global Optimization and Performance of
Repulsive Particle Swarm Method [online] http://mpra.ub.uni-muenchen.de/2718/ (accessed
14 August 2012).
 A literature survey of benchmark functions 193

Moore, R.E. (1988) Reliability in Computing, Academic Press, San Diego, CA, USA.
Moré, J.J., Garbow, B.S. and Hillstrom, K.E. (1981) ‘Testing unconstrained optimization
software’, ACM Trans. on Mathematical Software, Vol. 7, No. 1, pp.17–41.
Muntenau, C. and Lazarescu, V. (1998) ‘Global search using a new evolutionary framework:
the adaptive reservoir genetic algorithm’, Complexity International, Vol. 5 [online]
http://www.complexity.org.au/ci/vol05/munteanu/munteanu.html (accessed 14 August 2012).
Neumaier, A. (2003) COCONUT Benchmark [online] http://www.mat.univie.ac.at/∼neum/
glopt/coconut/benchmark.html (accessed 14 August 2012).
Opačić, J. (1973) ‘A heuristic method for finding most extrema of a nonlinear functional’, IEEE
Transactions on Systems, Man and Cybernetics, Vol. 3, No. 1, pp.102–107.
Pintér, J.D. (1996) Global Optimization in Action: Continuous and Lipschitz Optimization
Algorithms, Implementations and Applications, Kluwer, Hingham, MA, USA.
Powell, M.J.D. (1962) ‘An iterative method for finding stationary values of a function
of several variables’, Computer Journal, Vol. 5, No. 2, pp.147–151 [online]
http://comjnl.oxfordjournals.org/content/5/2/147.full.pdf.
Powell, M.J.D. (1964) ‘An efficient method for finding the minimum of a function for several
variables without calculating derivatives’, Computer Journal, Vol. 7, No. 2, pp.155–162.
Price, K.V., Storn, R.M. and Lampinen, J.A. (2005) Differential Evolution: A Practical Approach
to Global Optimization, Springer-Verlag New York, Inc. Secaucus, NJ, USA.
Price, W.L. (1977) ‘A controlled random search procedure for global
optimisation’, Computer Journal, Vol. 20, No. 4, pp.367–370 [online]
http://comjnl.oxfordjournals.org/content/20/4/367.full.pdf.
Qing, A. (2006) ‘Dynamic differential evolution strategy and applications in electromagnetic
inverse scattering problems’, IEEE Transactions on Geoscience and Remote Sensing,
Vol. 44, No. 1, pp.116–125.
Rónkkónen, J. (2009) Continuous Multimodal Global Optimization With Differential
Evolution-Based Methods, PhD thesis, Lappeenranta University of Technology.
Rahnamyan, S., Tizhoosh, H.R. and Salama, N.M.M. (2007a) ‘A novel population initialization
method for accelerating evolutionary algorithms’, Computers and Mathematics with
Applications, Vol. 53, No. 10, pp.1605–1614.
Rahnamyan, S., Tizhoosh, H.R. and Salama, N.M.M. (2007b) ‘Opposition-based differential
evolution (ODE) with variable jumping rate’, IEEE Symposium Foundations Computation
Intelligence, Honolulu, HI, pp.81–88.
Rao, S.S. (2009) Engineering Optimization: Theory and Practice, John Wiley & Sons, Hoboken,
New Jersey, USA.
Rosenbrock, H.H. (1960) ‘An automatic method for finding the greatest or least
value of a function’, Computer Journal, Vol. 3, No. 3, pp.175–184 [online]
http://comjnl.oxfordjournals.org/content/3/3/175.full.pdf.
Salomon, R. (1996) ‘Re-evaluating genetic algorithm performance under corodinate rotation
of benchmark functions: a survey of some theoretical and practical aspects of genetic
algorithms’, BioSystems, Vol. 39, No. 3, pp.263–278.
Schaffer, J.D., Caruana, R.A., Eshelman, L.J. and Das, R. (1989) ‘A study of control parameters
affecting online performance of genetic algorithms for function optimization’, Proc. 3rd
International Conf. on Genetic Algorithms, George Mason Uni., pp.51–60.
Schumer, M.A. and Steiglitz, K. (1968) ‘Adaptive step size random search’, IEEE Transactions
on Automatic Control, Vol. 13, No. 3, pp.270–276.
Schwefel, H.P. (1981) Numerical Optimization for Computer Models, John Wiley & Sons, New
York, NY, USA.
194 M. Jamil and X-S. Yang

Schwefel, H.P. (1995) Evolution and Optimum Seeking, John Wiley & Sons, New York, NY,
USA.
Shanno, D.F. (1970) ‘Conditioning of Quasi-Newton methods for function minimization’,
Mathematics of Computation, Vol. 24, No. 111, pp.647–656.
Silagadze, Z.K. (2007) ‘Finding two-dimensional peaks’, Physics of Particles and Nuclei Letters,
Vol. 4, No. 1, pp.73–80.
Storn, R. and Price, K. (1996) Differential Evolution – A Simple and Efficient
Adaptive Scheme for Global Optimization over Continuous Spaces, Technical Report
No. TR-95-012, International Computer Science Institute, Berkeley, CA [online]
http://www1.icsi.berkeley.edu/∼storn/TR-95-012.pdf.
Suganthan, P.N., Hansen, N., Liang, J.J., Deb, K., Chen, Y-P., Auger, A. and Tiwari, S.
(2005) Problem Definitions and Evaluation Criteria for CEC 2005, Special Session on
Real-Parameter Optimization, Nanyang Technological University (NTU), Singapore, Tech.
Rep. [online] http://www.lri.fr/∼hansen/Tech-Report-May-30-05.pdf.
Tang, K., Yao, X., Suganthan, P.N., MacNish, C., Chen, Y-P., Chen, C-M. and Yang, Z. (2008)
Benchmark Functions for the CEC 2008 Special Session and Competition on Large Scale
Global Optimization, Tech. Rep. [online] http://nical.ustc.edu.cn/cec08ss.php.
Tang, K., Li, X., Suganthan, P.N., Yang, Z. and Weise, T. (2010) Benchmark Functions for the
CEC 2010 Special Session and Competition on Large-Scale Global Optimization, Tech. Rep.
[online] http://sci2s.ugr.es/eamhco/cec2010 functions.pdf.
Test Problems for Global Optimization [online]
http://www2.imm.dtu.dk/∼kajm/Test ex forms/test ex.html.
The Cross-Entropy Toolbox [online] http://www.maths.uq.edu.au/CEToolBox/.
Wayburn, T.L. and Seader, J.D. (1987) ‘Homotopy continuation methods for computer-aided
process design’, Computers and Chemical Engineering, Vol. 11, No. 1, pp.7–25.
Whitley, D., Mathias, K., Rana, S. and Dzubera, J. (1996) ‘Evaluating evolutionary algorithms’,
Artificial Intelligence, Vol. 85, Nos. 1–2, pp.245–276.
Winston, P.H. (1992) Artificial Intelligence, 3rd ed., Addison-Wesley, Boston, MA, USA..
Yang, X.S. (2010a) ‘Test problems in optimization’, Engineering Optimization: An Introduction
with Metaheuristic Applications, John Wliey & Sons [online] http://arxiv.org/abs/1008.0549.
Yang, X.S. (2010b) ‘Firefly algorithm, stochastic test functions and design optimisation’, Intl. J.
Bio-Inspired Computation, Vol. 2, No. 2, pp.78–84 [online] http://arxiv.org/abs/1008.0549.
Yao, X. and Liu, Y. (1996) ‘Fast evolutionary programming’, Proc. 5th Conf. on Evolutionary
Programming.