Academic Journal of Nawroz University (AJNU), Vol.10, No.

3, 2022
This is an open access article distributed under the Creative Commons Attribution License
Copyright ©2017. e-ISSN: 2520-789X
https://doi.org/10.25007/ajnu.v11n3a1414

The Genetic Algorithm (GA) in Relation to Natural Evolution

Pawan Shivan Othman 1, Rasheed Reber Ihsan2, Reving Masoud Abdulhakeem 3
1,3Department
of Computer Science, Nawroz University, Kurdistan Region – Iraq
2Department of Computer Engineering and Communication, Nawroz University, Kurdistan Region –Iraq

ABSTRACT
For optimizing search global solution for complicated issues, the Genetic Algorithm (GA)is a famous evolutionary
computation technique that plays an important role in finding meaningful solutions to hard problems with a huge
search space could be a process based on genetic selection ideas. In addition, it supports machine learning causes,
as well as study and evolution. However, developing genetic processes that were formerly significant to a random
population, which might be started by biology for chromosomal production with factors like selection, crossover,
and mutation. The aim of going through this GA process is to find a solution for consecutive generations. In
individual production there has been an extent success instantly in ratio to fitness which is suited for it, as a result
successive generation will be better in one condition, which is ensuring the quality. Furthermore, John Holland is
considered as being the funding father of the initial genetic algorithm, with a funding date in the 1970s. in this
paper we have explained what a genetic algorithm is, its key operations, and how it works as well as its features
and applications.

KEYWORDS: Optimization Natural Evolution, Genetic Algorithm, Mutation, Crossover, Selection.

1. Introduction
Evolutionary computation is a branch of computer Method.
science that deals with computational issues and The Genetic Algorithm (GA) is an adaptive search
biological progress. Most of those issues need a lot of algorithm that replicates parts of the evolution
research work due to having a huge number of processes: selection, fitness, reproduction, crossover
possibilities, such as hardware circuit layouts for (also known as recombination), and mutation. It is
configuration that are responsible of producing desire based on Darwin's theory of evolution [1][3].
behavior, to find a set of equations that will predict the In an artificial environment, a population of people
ups and downs of financial market, or a set of rules that (organisms, strategies, individuals, objects...) is
controls a robot as it navigates its environment. These characterized by a genetic sequence that induces
kinds of problems always require a process to be physical features.
controlled that is, to continue in a changing Mutations and exchanges of genetic material search
environment and perform well [2]. new individual qualities, whereas the fittest people
Behavior that are complicated and hard to be resolved reproduce and spread their genes. the majority of GAs
such as intelligence could be manifested from aligned has been progressed for optimization purpose; despite
application and the reaction of being originally established for the analysis of
the rules. One of the examples of this philosophy is adaptation in natural systems (J. Holland, 1975).
Neural Networks, and evolutionary commutation is (Whitley, 1994). They've been used in a wide range of
another one. An irresistible source of inspiration for fields, as shown in table 1.
representing demanding computational problems is
Biological Evolution. Also, searching through out a
vast number of possibilities is called the Evolution

243
Academic Journal of Nawroz University (AJNU), Vol.10, No.3, 2022

TABLE 1: GA in various fields hand is maintained by an evolutionary algorithm.

Research Domains: Name Of Authors: After that, a collection of stochastic operators is

Biology Street,1999 applied iteratively to the population to evolve it [15].
Economics
Chatterjee Et Al., 2018 Mutation, recombination, and selection, or something
Finance similar, are commonly included in the set of operators.
Lwin Et Al., 2014
Operational Research We introduce genetic algorithms, their main
Della Croce Et Al.,
principles, and algorithm processes in this study [13].
Game Theory Axelrod Et Al., 1987, Vi ́e,
2020b We begin by discussing their properties as a search
Deep Learning and
Stan- Ley Et Al., 2019, Chung strategy in order to comprehend the rationale for their
Neural Networks
and Shin, 2020
wide-ranging and effective applications in
Forecasting Packard, 1988, C. Ahn and
Ramakrishna, 2003 optimization and modeling.
Optimization Wiransky, 2020, Dhunny Et
Al., 2020 2. Literature Review
Computer Science and GA was reviewed according to the PRISMA criteria. A
Koza, 1992
Algorithms
Data Science Yang And Honavar, 1998 comprehensive search of Google Scholar and PubMed
Healthcare Tao Et Al., 2019 for research papers connected to GA was conducted.
Aside from the hardness of the application problem to
This document also includes the important research
be solved with Genetic Algorithms (GAs), there's also
works discovered during the manual search.
the issue of the solution's quality, or the computational
Keywords like "Genetic Algorithm," "Application of
resources required to find it [5]. it, is dependent on the
GA," "operators of GA," "representation of GA," and
properties of the Genetic Algorithm. Genetic
"variants of GA" were used during the search.
Algorithms (GAs) are a powerful technique for
Let's have a look at some of the publications that have
resolving optimization issues. To use Genetic
been published in the past that are relevant to the
Algorithms to solve a problem, you must first answer
Genetic Algorithm. In the early 1960s, in Holland's
a few questions concerning the starting population, the
papers (e.g. Holland 1962) the idea of Genetic
likely and type of crossover, the likelihood and type of
Algorithm was found in a brief sight, and it is a great
mutation, the terminating criteria, the type of selection
use for finding out the path encoding in graph which
operator, and the fitness function to be used [7].
changes into chromosomes through shorted path via
All of these characteristics and operators have an
application [8].
impact on the performance of a GA and are
Cavicchio's thesis (Cavicchio 1970) observed these
interconnected, forming a system. Plus, it's showed
concepts as a type of adaptive search and examined
that it's a highly recommended that the ''rules'' of
them experimentally on challenging search problems
evolution are remarkably simple Genetic algorithms
including subroutine selection and pattern
(GAs), which will be the main focus of this review, that
recognition. Some of the early research on elitist types
is virtually widely used from of evolutionary Along
of selection and concepts for modifying the rates of
with genetic programming, evolution methods, and
crossover and mutation may be found in his work[2].
evolutionary programming, the genetic algorithm is
The increased interest in GAs prompted a series of
part of the evolutionary algorithm family [9]. The term
conversations and preparations for the inaugural
"evolutionary algorithm" refers to a large category of
International Conference on Genetic Algorithms
stochastic optimization approaches.
(ICGA), which took place in Pittsburgh, Pennsylvania,
A population of candidate solutions for the problem at
in 1985. Around 75 people attended, and they

244
Academic Journal of Nawroz University (AJNU), Vol.10, No.3, 2022

presented and discussed a wide spectrum of recent could be increased by these mutations [14]. To have a
developments in GA theory and implementation full optimization the GA serves as inherent solution
(Grefenstette 1985) [9]. while each chromosome is individually generated.
The International Society for Genetic Algorithms
(ISGA) is a non-profit organization whose mission is to
provide a vehicle for conference funding as well as to
assist in the coordination and facilitation of GA-related
activities. One of its first business moves was to
support a proposal for a FOGA (Foundations of
Genetic Algorithms) theory workshop in Bloomington,
Indiana (Rawlins 1991) [4].
Jang Sung Chun investigated the use of genetic
algorithms as a means of searching for optimization
problems in a variety of ways (1998). On several
Figure 1. Basic flow chart of a genetic algorithm
optimization tasks, the evolutionary algorithm and
4. HOW DO GENETIC ALGORITHMS WORK?
genetic algorithm were compared, and the results
For genetic algorithm their fitness classification is what
revealed that the genetic algorithm outperformed the
gives them the capacity to live and survive despite all
evolutionary algorithm.
the environmental risks. Each generation is made up
From 1990 to the present, the GA community has
by parent population and their offspring, which has all
experienced tremendous growth and diversity, as
the surviving individuals (chromosomes) from the
evidenced via the numerous conference events for
previous generations. The two genetic operators that
example; ICGA, and FOGA [6]. New GA applications
activate and create offspring or progeny are Crossover
are being developed all the time, covering a wide
and Mutation, and apparently this is known as fresh
spectrum of topics from engineering design to
solution [12].
operations research to autonomous programming.
Parents are determined for creating a new generation
3. GENETIC ALGORITHM AND ITS WORKFLOW
so that their chance of being picked up could be
According to Darwin principal of survival of the fittest,
corresponding near their fitness value. The better this
Genetic Algorithm (GA) that is relevant to the types
fitness rating goes, the more opportunity of life and
that are endangered by predators or environmental
surviving to reproduce is given. The table 2 represents
changes and risks [10]. Those who are not suited and
the GA's characteristics [15]. Mostly regular GA starts
fittest cannot always survive this. On the other hand,
with some random individuals to generate a
those who are fittest have a better chance to survive,
population of some possible solutions. After
because they have the ability to change situations, and
measuring the individual fitness, according to their
also mostly their traits and skills are inherited by their
fitness marks, some of them will be chosen as parents.
kids and even sometimes more developed, which
Table 2: GA characteristics.
means the future generations could be even fitter
General Algorithm Genetic Algorithm
which is mentioned in figure 1.
Process Of
In addition, some species randomly get genetic
Generating New Genetic Operators
mutations, and their fit individuals and their
Solution
evolutionary descendants surviving for a long time

245
Academic Journal of Nawroz University (AJNU), Vol.10, No.3, 2022

Selection Surviving Parents number of alternative solutions. A matrix of

Initial Solution Random Chromosome chromosomes of size 𝑀 × 𝑁 shows the population of

Fitness Function Quality Of Individual created alternative solutions:

Best Solution Elite

New Solution Children (Offspring)
Old Solution Parent
Chromosome
Solution
(Individual)
Decision Variable Gene Of Chromosome

Where 𝑋𝐽 is the 𝑗th solution (or chromosome), 𝑥𝑖,𝑗 is the

5. Creating an Initial Population
𝑖th solution's 𝑗th decision variable (or gene), and 𝑀 is
One of the most significant problems to consider in
the population size.
evolutionary computation is population scaling [10],
[18]. Researchers frequently claim that a "small"
5.1 Selection
population size can lead to bad solutions [17], [18],
Selection (Reproduction); The phase of selection is
[12], and that a "big" population size can cause the
dedicated to offer additional reproductive chances to
algorithm to take longer to find a solution [15], [14],
those population members that are the fittest,
[10], [12]. So, we're dealing with a trade-off that may
identifying the proper selection technique is a vital
be approximated by feeding the algorithm "enough"
process [19].
chromosomes [24] to get "excellent" results. For us,
Many selection techniques are introduced in the
"enough" is linked to instances in the search space and
literature, including: Elitism, Rank, Roulette Wheel,
diversity.
Tournament and Stochastic Universal Sampling (SUS),
One of the chosen solutions to the GA's optimization
see Table 3.
problem is chromosome. According mathematical
Table 3: GA selection operators
formulation and also the studies have shown us that
Selection
each Pros Cons
Techniques
Best Individual
A chromosome (also known as a genotype) is a group
Preserve Best Can Be Lost Due to
of factors that define a possible solution to the problem Elitism Individual in Crossover and
Population Mutation
the genetic algorithm is trying to solve. The population Operators
is the collection of all solutions [10]. Slow Convergence
Free From Bias
Sorting Required
Rank Preserve
In an N-dimensional optimization problem, a Computationally
Diversity
Expensive
chromosome is a 1N-dimensional array. The following Risk Of Premature
is the array's definition: Convergence
Easy To
Roulette Depends Upon
Implement
Wheel Variance Present
Simple
Chromosome = 𝑋=(𝑥1 , 𝑥2, … , 𝑥𝑖 , … . . , 𝑥𝑛 ) (1) in The Fitness
Function
Preserve
If X indicates a practical optimization problem Diversity Loss Of Diversity
Parallel When the
solution, 𝑥𝑖 indicates the 𝑖𝑡ℎ decision variable (or gene) Tournament
Implementation Tournament Size
of solution 𝑋, and 𝑁 denotes the number of decision No Sorting Large
Required
variables. 𝑀 stands for the population size, or the

246
Academic Journal of Nawroz University (AJNU), Vol.10, No.3, 2022

Stochastic To come up with the new solutions the chosen

Premature
Universal Fast Method
Convergence parents should take the responsibility, and the new
Sampling (Sus)
solutions are the children (offspring) relying on their
5.2 Population fitness, the selected population of the parents use R
(Whitley, 1989) conducted that we take population into solution.
account by using Selective Pressure and Diversity that Using the crossover probability (PC) which is a user-
are the GA's search methods for population. The two defined parameter of method, parents are randomly
named methods are contrarily connected, and as a chosen for this population. Each solution in the parent
result raising one lowers the other one. The two population needs an un-systemic Rand number that is
methods can highly affect the chromosome in different produced from the range [0,1].
ways [20]. 5.4 Crossover
An optimization search to stall can be caused by the When the solution of two parents start to interact
low selective pressure, on the other hand, premature together, Crossover will start to happen. And then new
convergence could happen because of high selective offspring are produced by transforming genes
pressure [21]. As we explained in the previous between parents.. In other words, swapping parts of
subsection, Convergence mistakes may be initiated the solution with another in chromosomes or solution
throughout proportionate selection. representations [23].
5.3 Reproduction The main role is to provide mixing of the solutions
The Genetic algorithm should create new solutions by and convergence in a subspace. Goldberg (1989) and
proceeding into a perfect solution. There is two ways, Michalewicz (1996) have conducted various ways of
either a portion of the following generation or the crossover, in addition to all of that, they have
parents give birth to offspring who make up the entire mentioned; one-point crossover, two-point crossover,
[22]. As a result, a mixture of parents might be and uniform crossover. Figure 3 shows us the three last
produced for the next generation. kinds of crossings.

Figure 2. The procedure of a new generation being formed

since the preceding one.

The parent-to-offspring ratio is a user-located metric.

The technique for producing the next generation from
Figure 3. Crossover methods include (a) one-point
the previous generation is shown, and explained in
crossover, (b) two-point crossover, and (c) uniform
Figure 2. To create the next generation, there are many crossover.
different approaches and methods, which define GA
variations, have been devised [24]. If we look into the figure 3 we will notice that a
crossover point C in figure is chosen at random, which
247
Academic Journal of Nawroz University (AJNU), Vol.10, No.3, 2022

is the result of using one-point crossover. Figure 4. An example of the mutation operator
In the figure on one side of the point it shows that
5.6 Termination Criteria
some of a child's genes are originate from one parent,
When the algorithm's iterations should come to a stop
and the balance of the child's genetic code.
the termination criteria are specified. Choosing the
Meanwhile, on the other side of the point, it originates
genuine termination has a huge impact over the
from the other parent. The figure 1.5b shows that when
algorithm's proper convergence. Depending on the
using two-point crossover two children are born to
number of iterations, the amount of progress made in
each couple, also it shows that two crossover sites are
the target function between iterations, and the run
created at random and are designated by C and C′.
time, the GA usually reaches the conclusion of its
In the same position the genes between the two
journey[11].
positions in the parent solutions are kept in the genetic
6. Applications
make-up of the offspring. As indicated in figure 3 b,
With high accuracy rates, genetic algorithms have been
the genes are located outside, and the two boundaries
applied to a variety of NP-hard tasks. GAs have been
are exchanged to produce two children. Also, the
used successfully in a few different application areas,
uniform crossover schemes are presented in the figure
which is mentioned in table 4.
3 c, and that is called self-explanatory. In the rage [1,N]
Table 4: GA Application Areas
some integer random numbers are used to create
Operation Wireless
crossover spots. Multimedia
Management Networking
5.5 Mutation Facility Information Load
Mutation delivers new genetic materials substance Layout Security Balance
Image
into population that's why it's a really important Scheduling Localization
Processing
method to know about. The mutation operator starts Bandwidth
Inventory Video
to substitutes a random number of genes in an & Channel
Control Processing
Allocation
offspring. To make it easy to understand, look into the
Forecasting
figure of 4 which shows that, in new solution random Medical
& Network
Imaging
values are replaced with the values of one or more Design
decision variables . Meanwhile, the values of the other Precision
Agriculture
choice variables stay unchangeable. For real-valued
Gaming
variables we have two methods of mutation which are
known as uniform and non-uniform mutations. 7. CHANLLENGES AND FUTURE POSSIBILITES
Uniform mutation is a method of replacing a parent The key challenges observed during the installation of
gene with a randomly produced gene that is within the GAs are highlighted in this section, followed by
solutions' feasible space [5]. relevant research topics.
7.1 Challenges
Despite the numerous benefits, some problems must
be overcome in order for genetic algorithms to
progress and evolve in the future. The following are
some important challenges [25].
▪ selection of efficient fitness function
▪ selection of initial population

248
Academic Journal of Nawroz University (AJNU), Vol.10, No.3, 2022

▪ selection of encoding schemes mutation, and so on.

▪ degree of mutation and crossover The application of GA approach has been
▪ premature convergence demonstrated including; search, optimization,
7.2 Future research directions decision making, machine learning, robotics, and a
GAs have been used in a variety of fields by altering variety of other technologies.
their basic structure. By addressing the current Furthermore, the evolutionary algorithm's main work
difficulty, the optimality of a solution found using GA flow diagram, as well as all of its functions, are covered
can be improved [25]: in detail in this paper, followed by several advantages
• Genetic algorithms imitate the process of natural of utilizing the genetic algorithm.
evolution. Additionally , the concerns and challenges of this
• The mapping from genotype to phenotype in real- study are highlighted at the conclusion of this
world conditions is difficult. publication.
• A method for selecting the optimal degree of
crossover and mutation operators should be REFERENCE
available. 1. Abdelghany A, Abdelghany K, Azadian F (2017) Airline
flight schedule planning under competition. Comput
• Future work can be considered for reducing the Oper Res 87:20–39
problem of early convergence. 2. Abdulal W, Ramachandram S (2011). Reliability-aware
genetic scheduling algorithm in grid environment.
8. CONCLUSION
International Conference on Communication Systems
The paper has presented a study of the overview of and Network Technologies, Katra, Jammu, pp 673– 677
3. Abdullah J (2010) Multiobjectives ga-based QoS routing
natural evolution-based GA approach to solve
protocol for mobile ad hoc network. Int J Grid Distrib
complicate problems. The genetic algorithm (GA) Comput 3(4):57–68
approach to optimization is based upon the concept of 4. Abo-Elnaga Y, Nasr S (2020) Modified evolutionary
algorithm and chaotic search for Bilevel program- ming
survival of the fittest The GA emulates the processes of problems. Symmetry 12:767
evolution and is therefore an evolutionary algorithm. 5. Afrouzy ZA, Nasseri SH, Mahdavi I (2016) A genetic
algorithm for supply chain configuration with new
In such a process the strongest elements become product development. Comput Ind Eng 101:440–454
stronger while the weakest elements are eliminated. 6. Aiello G, Scalia G (2012) La, Enea M. A multi objective
genetic algorithm for the facility layout problem based
Consequently, the ideas of Genetic Algorithms are
upon slicing structure encoding Expert Syst Appl
covered in depth in the first section of the introduction. 39(12):10352–10358
7. Alaoui A, Adamou-Mitiche ABH, Mitiche L (2020)
And in the terms of chromosomal production, it's
Effective hybrid genetic algorithm for removing salt and
similar to biology, including factors like selection, pepper noise. IET Image Process 14(2):289–296
8. Alkhafaji BJ, Salih MA, Nabat ZM, Shnain SA (2020)
crossover, and mutation, all of which combine to
Segmenting video frame images using genetic
generate genetic operations that will be applied to a at algorithms. Periodicals of Engineering and Natural
initially, the population was random. Sciences 8(2):1106–1114
9. Al-Oqaily AT, Shakah G (2018) Solving non-linear
further, the genetic algorithm can be compared to optimization problems using parallel genetic algo-
natural evolution in that it is a population-oriented rithm. International Conference on Computer Science
and Information Technology (CSIT), Amman, pp. 103–
search and optimization tool. The operation of genetic 106
algorithms is slowly inspired by two fundamental 10. Alvesa MJ, Almeidab M (2007) MOTGA: A
multiobjective Tchebycheff based genetic algorithm for
natural evolution ideas; one is genetic dynamics, and
the multidimensional knapsack problem. Comput Oper
the other is natural selection, which is concerned with Res 34:3458–3470
11. Arakaki RK, Usberti FL (2018) Hybrid genetic algorithm
various genetic operations such as crossover,
for the open capacitated arc routing problem. Comput
249
Academic Journal of Nawroz University (AJNU), Vol.10, No.3, 2022

Oper Res 90:221–231

12. Arkhipov DI, Wu D, Wu T, Regan AC (2020) A parallel
genetic algorithm framework for transportation
planning and logistics management. IEEE Access
8:106506–106515
13. Azadeh A, Elahi S, Farahani MH, Nasirian B (2017) A
genetic algorithm-Taguchi based approach to inventory
routing problem of a single perishable product with
transshipment. Comput Ind Eng 104:124– 133
14. Baker JE, Grefenstette J (2014) Proceedings of the first
international conference on genetic algorithms and their
applications. Taylor and Francis, Hoboken, pp 101–105
15. Bolboca SD, JAntschi L, Balan MC, Diudea MV, Sestras
RE (2010) State of art in genetic algorithms for
agricultural systems. Not Bot Hort Agrobot Cluj
38(3):51–63
16. Bonabeau E, Dorigo M, Theraulaz G (1999) Swarm
intelligence: from natural to artificial systems. Oxford
University Press, Inc
17. Burchardt H, Salomon R (2006) Implementation of path
planning using genetic algorithms on Mobile robots.
IEEE International Conference on Evolutionary
Computation, Vancouver, BC, pp 1831–1836
18. Burdsall B, Giraud-Carrier C (1997) Evolving fuzzy
prototypes for efficient data clustering," in second
international ICSC symposium on fuzzy logic and
applications. Zurich, Switzerland, pp. 217-223.
19. Burkowski FJ (1999) Shuffle crossover and mutual
information. Proceedings of the 1999 Congress on
Evolutionary Computation-CEC99 (Cat. No. 99TH8406),
Washington, DC, USA, 1999, pp. 1574–1580
20. Chaiyaratana N, Zalzala AM (2000) "Hybridisation of
neural networks and a genetic algorithm for friction
compensation," in the 2000 congress on evolutionary
computation, vol 1. San Diego, USA, pp 22–29
21. Chen R, Liang C-Y, Hong W-C, Gu D-X (2015)
Forecasting holiday daily tourist flow based on seasonal
support vector regression with adaptive genetic
algorithm. Appl Soft Comput 26:434–443
22. J.R. Cheng and M. Gen (2020) Parallel genetic algorithms
with GPU computing. Impact on Intelligent
23. Katoch, S., Chauhan, S. S., & Kumar, V. (2021). A review
on genetic algorithm: past, present, and
future. Multimedia Tools and Applications, 80(5), 8091-
8126.
24. Logistics and Manufacturing. Cheng H, Yang S (2010)
Multi-population genetic algorithms with immigrants’
scheme for dynamic shortest path routing problems in
mobile ad hoc networks. Applications of evolutionary
computation. Springer, In, pp 562–571
25. Cheng H, Yang S, Cao J (2013) Dynamic genetic
algorithms for the dynamic load balanced clustering
problem in mobile ad hoc net-works. Expert Syst Appl
40(4):1381–1392.

250