Volume 6, Issue 8, August – 2021 International Journal of Innovative Science and Research Technology

ISSN No:-2456-2165

The Quantum World and the

Realm of Fractional Calculus
Fractional order Quantum Particle Search Optimization

Ansh Sarkar (Author) Anubhab Swain (Author)

B. Tech Computer Science B. Tech Computer Science
Kalinga Institute of Industrial Technology Kalinga Institute of Industrial Technology
Bhubaneswar , India Bhubaneswar , India

Abstract:- Quantum mechanics emerged as a result of the I. INTRODUCTION

scientific enquiry into the wave nature of light during the
mid-17th century under the contribution of eminent The main concern of engineering problems has always
scientists including Christian Huygens, Robert Hooke been about maximizing yields or minimizing losses. As
and James Maxwell. This was followed by the Young’s science and technology develop, the complexity of
Double Slit experiment [1] and coining of the term optimization problems increases. Engineering problems in
Quanta (small energy packets) by Max Planck in the the areas of energy conversion, conservation and
Planck’s hypothesis for emission and absorption of distribution, mechanical design, logistics, all the way to the
radiation or energy. Several other contributions ranging reload of nuclear reactors often find in themselves an
from Wien’s law and Maxwell’s equations of inherent need for optimization techniques .
Electromagnetic radiation to Schrödinger’s [3] and Klein-
Gordon’s equations [4] have resulted in the continuous The world of quantum mechanics is known for its
development of this seemingly reality defying field of unforgiving use of seemingly complex terminologies and
modern science. mathematical equations, mostly defining concepts which
seem to defy our perception of reality. The following section
𝝏|𝝍𝒏(𝒕)⟩ introduces the reader to a few of the algorithms [23] and
̂ |𝝍𝒏 (𝒕)⟩ = 𝒊ℏ
𝑯
𝝏𝒕 underlying mathematical theorems which can prove useful
𝑺𝒄𝒉𝒓ö𝒅𝒊𝒏𝒈𝒆𝒓 𝑬𝒒𝒖𝒂𝒕𝒊𝒐𝒏 [𝟑] for understanding the application of Fractional order
differential equations ( or fractional order calculus in general
𝟏 𝝏𝟐 ∅𝒏 𝒎𝒄 𝟐 ) and the concept behind Optimization Techniques ,
𝟐 𝟐
− 𝜵𝟐 ∅𝒏 + ( ) ∅𝒏 = 𝟎 particularly the Fractional Order Quantum Particle Swarm
𝒄 𝝏𝒕 ℏ
𝑲𝒍𝒆𝒊𝒏 − 𝑮𝒐𝒓𝒅𝒐𝒏 𝑬𝒒𝒖𝒂𝒕𝒊𝒐𝒏 [𝟒] Optimization technique (FQPSO) [27].

As with most of the other fields in science, quantum II. PARTICLE SWARM OPTIMIZATION [27]
mechanics has infiltrated other fields of scientific interest
including but not limited to nano-materials, possible The Particle Swarm Optimization (PSO) algorithm [6]
teleportation and not to mention, quantum computing is a metaheuristic algorithm, originally proposed by
[12]. The main focus of this chapter revolves around the Kennedy and Eberhart, which is capable of solving various
optimization techniques, their quantum implementations complex engineering problems using the power of swarm
and how they can be made more efficient by introducing intelligence
a slight amount of fractional order calculus which Pseudo Code:
introduces the long term memory required for faster
convergence to optimal values. Most of the concepts 1. Initialising each particle
required to understand this text have been explained in 1.1. Initialise each 𝑋𝑖 randomly
the subsequent sections. Readers should feel free to refer 1.2. Initialize each 𝑉𝑖 randomly
to the reference/bibliography section for further reading. 1.3. Evaluate the fitness of each 𝑋𝑖 or : 𝑓(𝑋𝑖 )
1.4. Initialising 𝑝𝑏𝑒𝑠𝑡𝑖 with a copy of 𝑋𝑖
Keywords:- Fractional Calculus [5] , Particle Swarm [27] 2. Initialize global best or 𝑔𝑏𝑒𝑠𝑡 with a copy of 𝑋𝑖
Optimization [6] , Quantum Computing [12], Optimization 3. Repeat the following until stopping criterion achieved
tools and Techniques , Review :
3.1. For every particle 𝑖:
3.1.1. Update 𝑉𝑖𝑡 and 𝑋𝑖𝑡 according to the formulas :
𝑉𝑖𝑗𝑡+1 = 𝑤𝑉𝑖𝑗𝑡 + 𝑐1 (𝑝𝑏𝑒𝑠𝑡𝑖𝑗 − 𝑋𝑖𝑗𝑡 ) + 𝑐2 𝑟2𝑡 (𝑔𝑏𝑒𝑠𝑡𝑖𝑗 − 𝑋𝑖𝑗𝑡 )
And, 𝑋𝑖𝑗𝑡+1 = 𝑋𝑖𝑗𝑡 + 𝑉𝑖𝑗𝑡+1

IJISRT21AUG770 www.ijisrt.com 1055

Volume 6, Issue 8, August – 2021 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
3.1.2. Evaluate the fitness𝑓(𝑋𝑖𝑡 ) One such optimization algorithm is Particle Swarm
3.1.3. 𝑝𝑏𝑒𝑠𝑡𝑖  𝑋𝑖𝑡 if 𝑓(𝑝𝑏𝑒𝑠𝑡𝑖 ) < 𝑓(𝑋𝑖𝑡 ) Optimization (PSO) [6] which was developed in 1995 by the
3.1.4. 𝑔𝑏𝑒𝑠𝑡  𝑋𝑖𝑡 if 𝑓(𝑔𝑏𝑒𝑠𝑡) < 𝑓(𝑋𝑖𝑡 ) authors Kennedy and Eberhart, inspired by the behaviour of
* ( Please refer to the following section for the social organisms in groups, such as bird and fish schooling
meanings of the symbols used in the above pseudo-code. The or ant colonies. This algorithm mimics the interaction
following section provides an in depth analysis of the PSO between members of a particular group to share information.
algorithm [22] and its working ) The fitness function (also It has been applied to various areas of optimization and in
referred to as the objective function) combination with other existing algorithms. This algorithm
𝑓(𝑋) ∶ 𝑅𝐷 → 𝑅 performs the look for the optimal solution through agents,
also mentioned as particles, whose trajectories are adjusted
On a D-dimensional domain 𝑅𝐷 represents the function by a deterministic and a stochastic component. Each and
which we are interested in minimising. every particle is influenced by its ‘best’ (local attractor [7])
achieved position and the group’s ‘best’ position (global
Particle Swarm Optimization [6] was originally attractor), but tends to move randomly.
intended to be used for simulating social behaviour such as
the movement of organisms in swarms, for example: a flock This algorithm also has a competitive performance
of birds or a school of fishes. New and more complex with the Evolutionary Programming (EP) [29], Genetic
forms/variants of the PSO algorithm [23] are being Algorithm (GA), Evolution Strategies (ES), Genetic
continuously developed for improving [18] the Programming (GP) and many other classic algorithms [23].
optimizational efficiency of these algorithms. The new
variants include: improved genetic algorithms and The goal of an optimization problem is to determine a
Accelerated Particle Swarm Optimization Techniques. variable represented by a vector 𝑋 = [𝑥1 𝑥2 𝑥3 … 𝑥𝑛 ] , that
minimizes or maximizes depending on the proposed
Since the Quantum Particle Swarm Optimization optimization formulation of the function 𝑓(𝑥). The variable
Algorithm [8] as well as the Fractional Quantum Particle vector 𝑋 is known as position vector; this vector represents a
Swarm Optimization Algorithm [9] are both based on solid variable model and it is 𝑛 dimensional vector, where 𝑛
foundations of the PSO concept, we shall be dedicating an represents the number of variables that may be determined
entire section only to contribute to our understanding of the in a problem, that is, the latitude and therefore the longitude
fundamentals of the PSO technique. within the problem of determining some extent to land by a
flock. On the other hand, the function 𝑓(𝑥) is referred to as
A. Attractors the fitness or objective function, which assesses how good
The local attractor (better known as the private guide) or bad a position 𝑋 is, i.e. how good a particular landing
[7] of a particle can be defined as the best position with point a bird thinks it is after the animal finds it, evaluation of
respect to 𝑓 (the objective function) that particle has which is subjected to several survival criteria.
encountered so far. The swarm [26] also shares a standard
memory also called the worldwide attractor (better referred to Consider a swarm constituting P particles, in such a
as the local guide), which is the best position any of the case we know that there exists a position vector 𝑋𝑖𝑡 =
particles within the entire population has found so far. The (𝑥𝑖1 𝑥𝑖2 𝑥𝑖3 … 𝑥𝑖𝑛 )𝑇 and a velocity vector 𝑉𝑖𝑡 =
𝑇
movement of the entire swarm is hence governed by the local (𝑣𝑖1 𝑣𝑖2𝑣𝑖3 … 𝑣𝑖𝑛 ) at 𝑡 iteration for each one of the 𝑖
attractor [7] , the global attractor and the movement particle that composes it. These vectors are updated through
equations (which shall be discussed in the upcoming section). the dimension 𝑗 according to the following equations:
𝑉𝑖𝑗𝑡+1 = 𝑤𝑉𝑖𝑗𝑡+1 + 𝑐1 𝑟1𝑡 (𝑝𝑏𝑒𝑠𝑡𝑖𝑗 − 𝑋𝑖𝑗𝑡 ) + 𝑐2 𝑟2𝑡 (𝑔𝑏𝑒𝑠𝑡𝑗 −
B. Understanding the Particle Swarm Optimization 𝑋𝑖𝑗𝑡 ) Eq.(1)
Algorithm
and
There are several approaches that one could perform to
𝑋𝑖𝑗𝑡+1 = 𝑋𝑖𝑗𝑡 + 𝑉𝑖𝑗𝑡+1 Eq.(2)
maximize or minimize a function in order to find the
optimum value. Though there exists a wide range of where 𝑖 = 1,2,3 … 𝑃 and 𝑗 = 1,2,3 … 𝑛.
optimization algorithms that could be used, there is not a
particular one that is considered to be the best for any case. Eq. (1) denotes the presence of three different
An optimization method that works fine for one problem contributors to the particle’s movement in an iteration,
might not be so for another one; it depends on several giving rise to three terms which shall be further discussed.
features, for example, whether the function is concave or On the other hand, Eq. (2) updates the particle’s positions.
convex, and whether it is differentiable or not. To solve such The inertia weight constant represented as w is a positive
a problem, an individual must understand various constant value in case of classical Particle Swarm [26]
optimization techniques so as to select the algorithm that best Optimization and is important for balancing the global
fits on the optimization problems parameters. search, also referred to as exploration (in case higher values
are set), and local search, known as exploitation (in case
lower values are set).This is one of the main differences
between the classical and the derived versions of PSO.

IJISRT21AUG770 www.ijisrt.com 1056

Volume 6, Issue 8, August – 2021 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
The first term of the velocity update equation is given parameter involved in social learning , which weighs the
by the product between parameter w and the particle’s importance of the global learning of the swarm while 𝑟2
previous velocity, which is how it takes into consideration plays the exact same role as 𝑟1 .
the particle’s previous motion while calculating the current
one. Hence, for example, if 𝑤 = 1, the particle’s motion is
completely influenced by its previous motion and therefore
the particle may keep going on , in the same direction
whereas if 0 ≤ 𝑤 < 1, a partial influence is imparted , which
implies that a particle wanders and goes to other regions in
the search domain. Therefore, a reduction in the inertia
weight parameter makes the swarm explore more areas in the
searching domain, thus enhancing the chances of finding a
global optimum. However, using lower w values, comes with
the price of an increased simulation time as the swarm takes a
longer time to converge.

Fig 2

III. UNDERSTANDING QUANTUM COMPUTERS

AND THE QUANTUM WORLD

A. Quantum Physics
The emergence of Quantum mechanics in the early
1900s was mainly driven by the urge to develop a
framework to explain nature on the nanoscopic scale of
atoms and drive advancements in areas such as transistor
technology, efficient or better laser systems, and more
accurate high resolution magnetic resonance imaging. The
Fig 1 merging of quantum mechanics and information theory was
a debate in and around the 1970s but garnered little
The individual cognition term, (indicated by the 2nd attention. That was until 1982, when physicist Richard
term in Eq.(1) ) , is calculated using the difference between Feynman gave a talk revolving around the fact that
the particle’s own best position, for example, 𝑝𝑏𝑒𝑠𝑡𝑖𝑗 , and its computing based on classical logic could not process
current position 𝑋𝑖𝑗𝑡 . One may notice that the idea behind this calculations describing quantum phenomena. On the other
term is that as the particle gets more distant from the 𝑝𝑏𝑒𝑠𝑡𝑖𝑗 hand, quantum phenomena based computing, configured to
position, the difference (𝑝𝑏𝑒𝑠𝑡𝑖𝑗 − 𝑋𝑖𝑗𝑡 ) must increase; simulate and test other quantum phenomena, could
therefore, this term increases, attracting the particle to its own theoretically act as a great solution as it would not be subject
best position. The parameter 𝑐1 exists as a product in this to the same bottlenecks as the ones faced by modern day
term while being a positive constant and an individual- classical computational systems. This application of
cognition parameter, helping particles to weigh the quantum computing [12] eventually gave rise to the field of
importance of their own previous experiences. The second quantum simulation but failed to spark much research
parameter that is made up of the product of the second term activity or interest at the time.
is 𝑟1 , and this is a random value parameter lying within the
range [0,1] . This random parameter helps in avoiding This lack of interest in quantum computing [12] and
premature convergences, thereby increasing the chances at research related to the field however took a turn in the year
finding the most likely global optima. 1994, when mathematician Peter Shor developed a quantum
algorithm, capable of finding the prime factors of large
Lastly , the third term in the equation is responsible for numbers efficiently. The word, “efficiently” here refers to a
the social learning behaviour of the swarm. It makes it time range of practical relevance, which is currently beyond
possible for all the particles in the swarm to be able to share the capability of even state-of-the-art classical algorithms.
the information of the best point achieved with each other ,
regardless of which particle had found it, for example, B. Quantum Computing
𝑔𝑏𝑒𝑠𝑡𝑗 . It has the same format as that of the term The fundamental motivation behind both Quantum and
classical computers remains the same , i.e. to help and try to
corresponding to the individual learning property of the
solve problems in an efficient manner within a particular
swarm. Hence : (𝑔𝑏𝑒𝑠𝑡𝑗 − 𝑋𝑖𝑗𝑡 ) is the difference that acts as time frame, but at their core , the way they manipulate the
an attraction for the particles to the best point until it is found data to derive answers is almost completely different. In the
at some 𝑡 iteration. In the same way , the variable 𝑐2 is a following parts of this section we aim to provide an

IJISRT21AUG770 www.ijisrt.com 1057

Volume 6, Issue 8, August – 2021 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
explanation targeting the uniqueness of quantum computing D. The building blocks of Quantum Computers: Qubits
[12] systems by introducing the two fundamental underlying A well known fact is that modern day computers are
principles of quantum mechanics crucial for their operation binary digital systems which use bits - or in other words a
namely : superposition and entanglement [21]. stream of electrical or optical pulses representing the two
states of either 1s or 0s , highs or lows and on or off.
Everything in the digital world ranging from our tweets and
emails, to our credit card information , songs and YouTube
videos are nothing more than long strings of these 1s and 0s.

In quantum computing [12] systems the building

blocks consist of a qubit or quantum bit which is also the
basic unit of quantum information - it represents the
quantum version of the classical binary bit usually
physically designed and working as a two-state device. The
design or structure of a qubit essentially represents a
somewhat similar two-state (or two-level) quantum-
mechanical system, often regarded as perhaps one of the
simplest quantum systems having the ability to display the
peculiarity of quantum mechanics. The similarity in
Fig 3 representation of states in a qubit in comparison with a
classical bit can be seen by taking the spin of an electron as
C. Need for Quantum Computing an example in which the two levels can be considered as the
The possibility of someday developing a quantum two spins namely : spin up and the spin down of an electron
computer sophisticated and capable enough to pull off feats ; or even the polarization of a single photon in which the
such as executing Shor's algorithm for large numbers which states can be considered according to whether the light has
might effectively result in a revamp of currently implemented undergone vertical polarization or horizontal polarization.
cryptographic tools and systems , has been one of the The generation and maintenance of stable qubits [13] poses
motivators for advancing the field of quantum computation. huge scientific and engineering challenges. Popular Tech
It is however important to understand that quantum Conglomerates, such as IBM, Google [19], and Rigetti
computers are likely to deliver tremendous speed-ups only Computing, make use of superconducting circuits which are
for specific types of problems which can be simulated in the cooled down to mind cryogenic temperatures , even colder
quantum world. In other words , quantum computing [12] than that in deep space. Other systems like IonQ ( Ion
does not mean faster computers , it just means that we shall Capture Technology ) [14] work by trapping and isolating
be able to solve specific problems , within a reasonable time- individual atoms with the help of strong electromagnetic
frame compared to what would have required traditional or fields which are produced by strong superconducting
classical computers years to compute. Researchers are electromagnets on silicon chips kept inside ultra-high-
working towards enhancing their understanding of which vacuum chambers. Whatever be the methods used , the end
problems are best suited for quantum speed-ups and focusing goal is to isolate the qubits [13] in a controlled quantum
on the development of algorithms to demonstrate them. state.
Scientists working in the field of quantum computing [12]
currently involved in active research believe that quantum Unlike binary bits, qubits have some quirky quantum
computers should be able to help immensely with problems properties due to which a connected group of them can
related to optimization of parameters or resources , which can provide way more processing power than the same number
play key roles in advancements in fields ranging from of binary bits. The properties that enable the qubits [13] to
everything from defence to financial trading. exhibit such extraordinary performance are superposition
and entanglement [21].
There are additional applications focused at exploiting
qubit based systems for fields that are not related to There exists 2 orthogonal z-basis states of a qubit
computing or simulation and are active areas of research. which can be defined as follows :
Research is currently underway on two of the most prominent  |0⟩
application areas or use cases of qubit based computational  |1⟩
systems namely :
The two orthogonal x-basis states are:
1. Quantum Sensing and Metrology , leveraging the |0⟩+|1⟩
1. |+⟩ =
extreme sensitivity of qubits to environmental factors in √2
|0⟩−|1⟩
order to realize and improve sensing capabilities beyond 2. |−⟩ =
√2
the classical shot noise limit, and
2. Quantum Networks and Communications , which are
thought to lead to revolutionary breakthroughs in the
realm of information / data sharing .

IJISRT21AUG770 www.ijisrt.com 1058

Volume 6, Issue 8, August – 2021 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
Similarly , the two orthogonal y-basis states of a qubit The entanglement of two qubits [13] brings into
can be defined as : existence a special connection between them. The effects of
|0⟩+𝑖|1⟩ the entanglement become clear from the observations
1. |𝑅⟩ =
√2 obtained after measurements. Upon measurement , the
|0⟩−𝑖|1⟩
2. |𝐿⟩ = outcome of observing the individual qubits could be either 0
√2
or 1. However , since the qubits are entangled with each
The basis states can be visualized as points which are other , the outcome of the measurement of a single qubit will
located at the opposite points on the Bloch sphere always be correlated or influenced by the measurement on
representation of the current state of a single qubit. the other qubits [13]. This happens , even if the particles are
far apart and separated from each other by large distances.
E. Understanding Superposition The Bell state configuration of qubits [13] can be given as
The property of superposition is what sets apart a an example for the observation of the behaviour of an
quantum bit from a classical bit is the fact that a qubit can be entangled qubit system.
in superposition. Superposition is one of the most weird ,
intriguing and yet among the fundamental principles of As an example, let us consider two particles created in
quantum mechanics. We have read about the Fourier Series such a way that the total spin of the system is zero. A certain
and its applications in signal processing. Using the same reference axis is considered and the spin of the particles is
principle in classical physics, a wave describing a musical measured along that axis. Let us suppose that the spin of one
tone can therefore be seen as a combination or superposition of the generated particles measured with respect to a certain
of several waves with different frequencies that are added axis is found to be counter clockwise, then it can be
together. In the same manner , in case of quantum mechanics guaranteed that a measurement of the spin of the other
a superposed quantum state can be broken down or particle (with reference to the same axis) will exhibit a
decomposed into a linear combination of a number of other clockwise spin. It appears as if one of the entangled particles
distinct quantum states. is able to feel or detect immediately that a measurement has
been performed on the other entangled particle and
The superposition of the Qubits [13] can be broken accordingly locks what the outcome should be, but this is
down and represented as a linear combination of the basis not the case. All this happens instantaneously, without the
states ∣0⟩ and ∣1⟩. When a qubit is measured (only exchange of any form of information between the entangled
observables are measurable), the qubit falls or collapses to particles even if they were located billions of miles away
one of its Eigen states which is then reflected in the measured from each other , this entanglement would still exist.
value. For example, if a qubit exists in an equally weighted
superposed state, measuring it or trying to observe it , would IV. FRACTIONAL CALCULUS
make it collapse to one of its basis states∣0⟩or∣1⟩ with an
equal probability of 50%. The ∣0⟩basis state on measurement Fractional calculus [5] as the name suggests, deals
or observation collapses and always gives the result as 0 and with the analysis of differentials of fractional order. The
the basis state∣1⟩always converts or collapses to 1. topic is in no way a recent development and has seen a lot of
contribution from people from the field of pure mathematics.
It should however be kept in mind that Quantum Only recently has it undergone major, heated development
Superposition is fundamentally different from summation by the contribution of various eminent mathematicians from
and superposition of classical waves. A n qubit containing the field of applied mathematics. Nevertheless there is a
quantum computer can in total exist in a superposition of 2𝑛 wide variety of theorems to choose from in order to navigate
states: from∣000...0⟩ to ∣111...1⟩. But on the other hand , if we the realm of Fractional Calculus [5] and they are usually not
take the example of a classical situation such as compatible (equivalent) to each other. Only in recent years
playing n musical sounds with distinct frequencies, it can be have scientists around the world discovered the use of
observed that we can only obtain a superposition fractional calculus and its theories in understanding various
of n frequencies. Therefore , adding classical waves scales natural frameworks ranging from viscoelastics, signal
linearly, whereas the superposition of quantum states scales processing and of course now in Quantum Particle Swarm
exponentially. Optimization technique [8].

F. Quantum Entanglement: Spooky Action at a Distance [21] The use of Convolutional kernels for evaluating
Entanglement is one among the many counter-intuitive fractional integrals, The Riemann-Liouville Integral /
phenomena in the realm of Quantum Physics. Entanglement Derivative, The Reisz fractional derivative , The Caputo
refers to a phenomenon when the quantum state of a pair or fractional derivative and the Grünwald–Letnikov [2]
group of particles cannot be described independently of the derivative are some of the frameworks in place for
quantum state of the other particle(s). Therefore the quantum understanding / evaluating fractional order differentials. The
system as a whole can be described as being in a definite most popular among these: The Grünwald–Letnikov [2]
state where the measurement of each particle affects every derivative, has been discussed in further details in the
other particle, even though the individual parts of the system upcoming sections.
are not in a definite state.

IJISRT21AUG770 www.ijisrt.com 1059

Volume 6, Issue 8, August – 2021 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
A few terms and functions to know about while Here we assume that the infinitesimally small
discussing fractional calculus have been listed as follows . (differential) element converges synchronously. This can be
further verified by using the Mean Value Theorem or
A. Analytic Functions: M.V.T. We now generalize this by using our knowledge of
Functions which can be represented in the form of a binomial expressions in order to obtain :
power series are known as Analytic functions. Their
importance lies in the fact that they encompass and contain 1 𝑛
most of the commonly encountered mathematical functions. 𝑓 (𝑛) (𝑥) = lim 𝑛
∑ (−1)𝑚 ( ) 𝑓(𝑥 + (𝑛 − 𝑚)𝑐)
𝑐→0 𝑐 𝑚
0≤𝑚≤𝑛

These functions are usually divided into two separate

We now remove the restriction that n should be a
types : Real analytic functions and Complex analytic
positive integer and hence derive the form:
functions (also called holomorphic functions). This class of
functions is closed relative to the fundamental operations of 1 𝑞
arithmetic which are used in algebra, analysis and the other 𝐷𝑞 𝑓(𝑥) = lim ∑ (−1)𝑚 ( ) 𝑓(𝑥 + (𝑞 − 𝑚)𝑐)
fields of analytic functions and applied mathematics.
𝑐→0 𝑐𝑞 𝑚
0≤𝑚≤∞

B. The Euler-Gamma function: This, is the general form of the Grünwald–Letnikov

The Gamma function representable as Γ(𝑛) can be (GL) derivative [2].
defined as an extended version of the factorial function (n!)
which is otherwise, only limited to whole numbers. The The Grünwald–Letnikov (GL) derivative [2] :
gamma function hence extends the factorial function to
include complex as well as fractional values of n.

It is related to the factorial of a number as: Γ(𝑛) =

(𝑛 − 1)! The Euler-Gamma function is analytic everywhere
Where, 𝑓(𝑥) is a differentio-integrable function, [a, x]
except at n = 0 , -1 , -2 , … and residue at 𝑧 = −𝑖 is given by is the domain of the function, Г represents the Euler-Gamma
:
function, α represents the fractional order of the derivative
(−1)𝑖 and 𝐺𝐿𝑎𝐷 𝛼𝑥𝑓(𝑥) represents the GL derivative operator.
𝑅𝑒𝑠𝑧=−𝑖 Γ(𝑧) =
𝑖!
When n becomes large enough, the limit symbol is
C. Grünwald–Letnikov derivative neglected (as we then consider infinitesimal sections) . We
Just like modules enhance/extend the functionality of a can hence rewrite the above equation for large values of n
programming language, various theorems in mathematics act as:
as patches which help to increase the domain of functioning
of a particular field of mathematics. 𝑛−1
𝑑𝛼 𝑥 −𝛼 𝑛𝛼 Γ(𝑘 − 𝛼)
𝛼
𝑓(𝑥) ≅ ∑ 𝑓 (𝑥
The Grünwald–Letnikov (GL) derivative [2] does the 𝑑𝑥 Γ(−𝛼) Γ(𝑘 + 1)
𝑘=0
same for the field of differential calculus and hence extends 𝑘𝑥
the domain of differential equations in order to include − ) ; 𝑓𝑜𝑟 𝑙𝑎𝑟𝑔𝑒 𝒏
𝑛
fractional order differentials. It allows us to take the
derivative of a function, non-integer number of times The above equation can be further generalised for a 1D
signal as follows (Note: the equation given below is not
D. Understanding / Deriving the Grünwald–Letnikov convenient for thorough analysis but is often used for
derivative [2] numerical approximations):
We all are well aware that the differential of a function
representable as 𝑓 ′ (𝑥) is: 𝑑𝛼
𝑓(𝑥 + 𝑐) − 𝑓(𝑥) 𝑓(𝑥) ≅ 𝑓(𝑥) + (−𝛼)𝑓(𝑥 − 1)
𝑓 ′ (𝑥) = lim 𝑑𝑥 𝛼
𝑐→0 𝑐 (−𝛼)(−𝛼 + 1)
+ 𝑓(𝑥 − 2) + ⋯
2
The derivative can be applied recursively on a function (−𝛼)(−𝛼 + 1) … (−𝛼 + 𝑛)
to get derivatives of higher orders. So we can write the 2 nd + 𝑓(𝑥 − 𝑛)
𝑛!
derivative as:
𝑓 ′ (𝑥 + 𝑐) − 𝑓 ′ (𝑥)
𝑓 ′′ (𝑥) = lim
𝑐→0 𝑐
𝑓(𝑥+𝑐1 +𝑐2 )−𝑓(𝑥+𝑐1 ) 𝑓(𝑥+𝑐2 )−𝑓(𝑥)
lim − lim
𝑐2 →0 𝑐2 𝑐2 →0 𝑐2
= lim 𝑐1
𝑐1 →0

IJISRT21AUG770 www.ijisrt.com 1060

Volume 6, Issue 8, August – 2021 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
V. QUANTUM PARTICLE SWARM Where 𝐿𝑖𝑑 (𝑡) denotes the standard deviation, which
OPTIMIZATION describes the search space of each particle. The Monte Carlo
method can now be used in order to obtain the position of
The major/significant difference between classic PSO the particle as follows :
and Quantum PSO is the use of the wave function in QPSO
in order to facilitate optimization or determine the position of 1 1 2|𝑦| 1
a given particle in 3D space. 𝑠 = 𝑢 = 𝑒 − 𝐿 , 𝑢 ~ (0,1)𝑎𝑛𝑑 𝑠 ~ (0, )
𝐿 𝐿 𝐿

Trajectory analysis according to [citation] demonstrated Then by comparing:

that each particle in practice converges to its corresponding 2|𝑦|
𝑢 = 𝑒− 𝐿
local attractor [7] or 𝐶𝑖𝑑 which can be demonstrated by the
following equation:
Let 𝑦 = 𝑥 − 𝑐, we have:
𝐶𝑖𝑑 (𝑡) = 𝑎. 𝑝𝑏𝑒𝑠𝑡𝑖𝑑 (𝑡) + (1 − 𝑎). 𝑔𝑏𝑒𝑠𝑡𝑖𝑑 (𝑡) , 𝑎~𝑈(0,1)
𝑐1 𝑟1 𝐿 1
Where, 𝑎 = (𝑐 𝑥 = 𝑐 ± ln ( )
1 𝑟1 +𝑐2 𝑟2 ) 2 𝑢

The local attractor [7] is thus a randomly distribution or The convergence condition for PSO is given by:
pattern that may be analyzed statistically but may not be 𝑥 → 𝑐 , 𝑤ℎ𝑒𝑛 𝑡 → ∞
predicted precisely or in other words a stochastic attractor of
the 𝑖𝑡ℎ particle that lies inside a hyper-rectangle with 𝑝𝑏𝑒𝑠𝑡𝑖𝑑 Let L be a function of time, then for the above
and 𝑔𝑏𝑒𝑠𝑡𝑖𝑑 as the two ends of its diagonal. condition to hold true, we have:
𝐿 = 𝐿(𝑡) , 𝐿 → 0 , 𝑤ℎ𝑒𝑛 𝑡 → 0
Sun et al. studied the convergence behaviour of PSO
and proposed a novel PSO model from quantum mechanics We can further write the iterative version for the 𝑖𝑡ℎ
abbreviated as QPSO or the Quantum Particle Swarm particle as:
Optimization [8]. Based on the Delta potential, the quantum
behaviour of particles are considered. In the framework of 𝐿𝑖𝑑 1
quantum time-space, the quantum state of a particle is 𝑋𝑖𝑑 (𝑡 + 1) = 𝐶𝑖𝑑 ± ln ( )
2 𝑢
represented by a wave function 𝜓(𝑥, 𝑡) .
In 3D space, 𝜓(𝑥, 𝑡) is given as A global point called mean best position is introduced
to evaluate the value of 𝐿𝑖𝑑 (𝑡). This global point, which is
|𝜓 |2 𝑑𝑥𝑑𝑦𝑑𝑧 = 𝑄𝑑𝑥𝑑𝑦𝑑𝑧 denoted by mbest and can be computed as the mean of
the pbest positions of all the particles present in the swarm,
where Q is the probability of finding the particle at a which can be given s:
particular point in space (with respect to the xyz coordinate
system).Here , |𝜓 |2 represents the probability density of the 𝑚𝑏𝑒𝑠𝑡(𝑡) = (𝑚𝑏𝑒𝑠𝑡1 (𝑡), 𝑚𝑏𝑒𝑠𝑡2 (𝑡), … , 𝑚𝑏𝑒𝑠𝑡𝑑 (𝑡)
𝑛 𝑛 𝑛
particle. 1 1 1
= ∑ 𝑝𝑖1 (𝑡), ∑ 𝑝𝑖2 (𝑡), … , ∑ 𝑝𝑖𝑑 (𝑡),
𝑛 𝑛 𝑛
As a probability density function, we have: 𝑖=1 𝑖=1 𝑖=1

+∞ +∞ The values of 𝐿𝑖𝑑 (𝑡) is calculated by:

∫ |𝜓 |2 𝑑𝑥𝑑𝑦𝑑𝑧 = ∫ 𝑄𝑑𝑥𝑑𝑦𝑑𝑧 = 1
−∞ −∞
𝐿𝑖𝑑 (𝑡) = 2𝛽|𝑚𝑑 (𝑡) − 𝑋𝑖𝑑 (𝑡)|
The normalized version of 𝜓 can be represented as:
And finally, the position of the particle 𝑋𝑖𝑑 can be
1 −|𝑦| given by:
‖𝜓‖ = 𝜓(𝑦) = 𝑒 𝐿
√𝐿 1
𝑋𝑖𝑑 (𝑡 + 1) = 𝐶𝑖𝑑 (𝑡 + 1) ± 𝛽|𝑚𝑏𝑒𝑠𝑡𝑑 − 𝑋𝑖𝑑 (𝑡)| ln ( )
𝑢
As a result, Q and the corresponding distribution
function 𝐹 can be obtained as: Where parameter 𝛽 is step size, which is utilized to
control the convergence speed. U(0,1) is the shorthand for
1 −2|𝑦|
𝑄(𝑦) = |𝜓(𝑦)|2 = 𝑒 𝐿 Standard Uniform Random Variable Distribution in between
𝐿 0 and 1.
And :
2|𝑝𝑖𝑑 (𝑡)−𝑋𝑖𝑑 (𝑡+1)|
−
𝐹(𝑋𝑖𝑑 (𝑡 + 1)) = 1 − 𝑒 𝐿𝑖𝑑 (𝑡)

A. QPSO with the fractional-order position : FQPSO

IJISRT21AUG770 www.ijisrt.com 1061

Volume 6, Issue 8, August – 2021 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
Fractional Calculus [5] is often regarded to have 𝑤ℎ𝑒𝑛 𝑟𝑎𝑛𝑑 < 0.5 𝑎𝑛𝑑 𝑚𝑏𝑒𝑠𝑡𝑑 < 𝑋𝑖𝑑 (𝑡)
excellent long term memory characteristics. From the
definition of Grünwald-Letnikov differential [2] (under And ,
fractional calculus sub-section), it can be seen that the 𝐷𝛼 (𝑋𝑖𝑑 (𝑡 + 1)) =
fractional derivative is computed with the previous/historical 1 1
states which makes it suitable for the iterative procedure of 𝐶𝑖𝑑 (𝑡) − 𝛽. ln ( ) . 𝑚𝑏𝑒𝑠𝑡𝑑 + (𝛽. ln ( ) ± 1) 𝑋𝑖𝑑 (𝑡)
𝑢 𝑢
intelligent optimization algorithms like the Fractional … 𝑒𝑞. 2
Quantum Particle Swarm Optimization algorithm [9] , which
introduces fractional index based swarm optimization. 𝑤ℎ𝑒𝑛 𝑟𝑎𝑛𝑑 > 0.5 , 𝑚𝑏𝑒𝑠𝑡𝑑 < 𝑋𝑖𝑑 (𝑡) 𝑜𝑟 ,
𝑤ℎ𝑒𝑛 𝑟𝑎𝑛𝑑 < 0.5 𝑎𝑛𝑑 𝑚𝑏𝑒𝑠𝑡𝑑 > 𝑋𝑖𝑑 (𝑡)
To further improve the speed and accuracy of
convergence of QPSO algorithm/technique, in this section, Researchers have demonstrated the fact that for
the proposed QPSO with the fractional-order position is ~𝑈(0,1) , smoother variations and a prolonged memory
detailed. Initially, the original position is rearranged to effect can be introduced which might possibly lead to a
modify the order of the position derivative, which can be better accuracy than then the integral-order based model
derived as follows:
To study the behaviour and accuracy of the proposed
Equations corresponding to all possible relations fractional-order strategy, a set of functions are tested with
between rand and mbest id . (rand~U(0,1)) 𝛼~𝑈(0,1) and 𝛥𝛼 = 0.1 which denotes the step size. To
simplify the computational complexity, we usually truncate
𝑪𝒂𝒔𝒆 𝟏 ∶ 𝑟𝑎𝑛𝑑 > 0.5 , 𝑚𝑏𝑒𝑠𝑡𝑑 > 𝑋𝑖𝑑 (𝑡) the simplified form of the Grünwald-Letnikov differential
1 [2] and only use the first four terms, so we have:
𝑋𝑖𝑑 (𝑡 + 1) = 𝐶𝑖𝑑 (𝑡) + 𝛽. ln ( ) . (𝑚𝑏𝑒𝑠𝑡𝑑 − 𝑋𝑖𝑑 (𝑡))
𝑢
𝑎𝑛𝑑 , 𝑟𝑎𝑛𝑑 > 0.5 , 𝑚𝑏𝑒𝑠𝑡𝑑 > 𝑋𝑖𝑑 (𝑡) 𝐷𝛼 (𝑋𝑖𝑑 (𝑡 + 1)) =
1
𝑪𝒂𝒔𝒆 𝟐 ∶ 𝑟𝑎𝑛𝑑 > 0.5 , 𝑚𝑏𝑒𝑠𝑡𝑑 < 𝑋𝑖𝑑 (𝑡) 𝑋𝑖𝑑 (𝑡 + 1) − 𝛼𝑋𝑖𝑑 (𝑡) − 𝛼(1 − 𝛼)𝑋𝑖𝑑 (𝑡 − 1)
1 2
𝑋𝑖𝑑 (𝑡 + 1) = 𝐶𝑖𝑑 (𝑡) + 𝛽. ln ( ) . (𝑋𝑖𝑑 (𝑡) − 𝑚𝑏𝑒𝑠𝑡𝑑 ) 1
𝑢 − 𝛼(1 − 𝛼)(2 − 𝛼)𝑋𝑖𝑑 (𝑡 − 2)
6
𝑎𝑛𝑑 𝑟𝑎𝑛𝑑 > 0.5 , 𝑚𝑏𝑒𝑠𝑡𝑑 < 𝑋𝑖𝑑 (𝑡) 1
− 𝛼(1 − 𝛼)(2 − 𝛼)(3 − 𝛼)𝑋𝑖𝑑 (𝑡
24
𝑪𝒂𝒔𝒆 𝟑 ∶ 𝑟𝑎𝑛𝑑 < 0.5 , 𝑚𝑏𝑒𝑠𝑡𝑑 > 𝑋𝑖𝑑 (𝑡) − 3)
1 … 𝑡𝑟𝑢𝑛𝑐𝑎𝑡𝑒𝑑 𝑡𝑜 4 𝑡𝑒𝑟𝑚𝑠
𝑋𝑖𝑑 (𝑡 + 1) = 𝐶𝑖𝑑 (𝑡) − 𝛽. ln ( ) . (𝑚𝑏𝑒𝑠𝑡𝑑 − 𝑋𝑖𝑑 (𝑡))
𝑢
𝑎𝑛𝑑 𝑟𝑎𝑛𝑑 < 0.5 , 𝑚𝑏𝑒𝑠𝑡𝑑 > 𝑋𝑖𝑑 (𝑡) Using the above result we can redefine / modify eq.1
as :
𝑪𝒂𝒔𝒆 𝟒 ∶ 𝑟𝑎𝑛𝑑 < 0.5 , 𝑚𝑏𝑒𝑠𝑡𝑑 < 𝑋𝑖𝑑 (𝑡)
1 (𝑋𝑖𝑑 (𝑡 + 1)) =
𝑋𝑖𝑑 (𝑡 + 1) = 𝐶𝑖𝑑 (𝑡) − 𝛽. ln ( ) . (𝑋𝑖𝑑 (𝑡) − 𝑚𝑏𝑒𝑠𝑡𝑑 )
𝑢 1 1
𝑎𝑛𝑑 𝑟𝑎𝑛𝑑 < 0.5 , 𝑚𝑏𝑒𝑠𝑡𝑑 < 𝑋𝑖𝑑 (𝑡) 𝐶𝑖𝑑 (𝑡) + 𝛽. ln ( ) . 𝑚𝑏𝑒𝑠𝑡𝑑 − (𝛽. ln ( ) ± 1 − 𝛼) 𝑋𝑖𝑑 (𝑡)
𝑢 𝑢
+ 𝑋𝑋𝑖𝑑 (𝑡)
All the above equations can now be uniformly written
as : We also modify eq.2 as :
𝑋𝑖𝑑 (𝑡 + 1) − 𝑋𝑖𝑑 (𝑡) =
1 1 (𝑋𝑖𝑑 (𝑡 + 1)) =
𝐶𝑖𝑑 (𝑡) + 𝛽. ln ( ) . (𝑚𝑏𝑒𝑠𝑡𝑑 ) − (𝛽. ln ( ) ± 1) 𝑋𝑖𝑑 (𝑡) 1 1
𝑢 𝑢 𝐶𝑖𝑑 (𝑡) − 𝛽. ln ( ) . 𝑚𝑏𝑒𝑠𝑡𝑑 + (𝛽. ln ( ) ± 1 + 𝛼) 𝑋𝑖𝑑 (𝑡)
𝑢 𝑢
The left side of the above equation is the discrete + 𝑋𝑋𝑖𝑑 (𝑡)
version of the derivative with 𝛼 = 1. We can hence extend /
generalize the above equation to get a general fractional order Where 𝑋𝑋𝑖𝑑 (𝑡) is defined as :
differential as follows:
𝑋𝑋𝑖𝑑 (𝑡) =
1 1
𝐷𝛼 (𝑋𝑖𝑑 (𝑡 + 1)) = 𝛼(1 − 𝛼)𝑋𝑖𝑑 (𝑡 − 1) + 𝛼(1 − 𝛼)(2 − 𝛼)𝑋𝑖𝑑 (𝑡 − 2)
1 1 2 6
𝐶𝑖𝑑 (𝑡) + 𝛽. ln ( ) . 𝑚𝑏𝑒𝑠𝑡𝑑 − (𝛽. ln ( ) ± 1) 𝑋𝑖𝑑 (𝑡) 1
𝑢 𝑢 + 𝛼(1 − 𝛼)(2 − 𝛼)(3 − 𝛼)𝑋𝑖𝑑 (𝑡 − 3)
24
… 𝑒𝑞. 1

𝑤ℎ𝑒𝑛 𝑟𝑎𝑛𝑑 > 0.5 , 𝑚𝑏𝑒𝑠𝑡𝑑 > 𝑋𝑖𝑑 (𝑡) 𝑜𝑟 ,

IJISRT21AUG770 www.ijisrt.com 1062

Volume 6, Issue 8, August – 2021 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
With respect to the position updating equations VI. APPLICATIONS OF THE ABOVE
(equations of motion) of the particles it can be seen that the CONCEPTS AND (F)QPSO IN THE MODERN
position of the particles depend not only on the position of WORLD
the previous particle but also on the historical position of the
particle itself at different points in time. The position A. Applications of Quantum Computing [17]
acquired by the particle in subsequent steps or epochs is  As the new technologies have penetrated almost every
hence the result of it’s long term memory. This long term aspect of our lives, artificial intelligence and machine
memory has the ability to protect the population density learning have become some of the prominent areas right
(distribution) and the diversity to a certain extent. now. Some of the widespread applications that we see
every day are in recognition of image, voice and
B. Pseudo Code (Algorithm) handwriting samples. However, it becomes a
1. Initialising the FQPSO parameters ( 𝛼 , t , 𝛽 , 𝑢 ) challenging task for traditional computers, to match up
2. Initialise the swarm of particles the accuracy and speed as the number of applications
3. For each particle 𝑖 in population: have increased. And, that’s where quantum computers
3.1. Computer 𝑓(𝑋𝑖 ) , where 𝑓( ) represents the fitness are better than traditional computers as they can help in
function processing through complex tasks in very less time,
3.1.1. If 𝑓(𝑋𝑖 ) < 𝑓(𝑝𝑏𝑒𝑠𝑡𝑖 ) : which would have taken traditional computers
𝑝𝑏𝑒𝑠𝑡𝑖 = 𝑋𝑖 thousands of years.
End
 Weather forecasting is another such field which
3.1.2. If 𝑓(𝑝𝑏𝑒𝑠𝑡𝑖 ) < 𝑓(𝑔𝑏𝑒𝑠𝑡) : includes several variables to consider, such as
𝑔𝑏𝑒𝑠𝑡 = 𝑝𝑏𝑒𝑠𝑡𝑖 temperature, air pressure and air density, which makes it
End tough for it to be predicted accurately. Applications of
quantum machine learning can help us improve our
3.2. Calculate Q using the equations ability to recognize patterns, which will make it easier
3.2.1. If for scientists and researchers to predict harsh weather
(𝑟𝑎𝑛𝑑 > 0.5 𝑎𝑛𝑑 𝑚𝑏𝑒𝑠𝑡𝑑 < 𝑋𝑖𝑑 (𝑡)) 𝑜𝑟 (𝑟𝑎𝑛𝑑 conditions and potentially save thousands of lives.
< 0.5 𝑎𝑛𝑑 𝑚𝑏𝑒𝑠𝑡𝑑 > 𝑋𝑖𝑑 (𝑡)) Meteorologists will be able to analyse more detailed
: climate models with the help of quantum computers that
(𝑋𝑖𝑑 (𝑡 + 1)) = can provide greater insights into climate change, and
1 helping them if figuring out ways to tackle it.
𝐶𝑖𝑑 (𝑡) + 𝛽. ln ( ) . 𝑚𝑏𝑒𝑠𝑡𝑑
𝑢
1
− (𝛽. ln ( ) ± 1 − 𝛼)  Quantum Chemistry: Even in the smallest of
𝑢 molecule, it is believed that the amount of quantum
𝑋𝑖𝑑 (𝑡) + 𝑋𝑋𝑖𝑑 (𝑡)
states is extremely vast, and therefore tough for
Else
traditional computing memory to process. Meanwhile
quantum computers can focus on the existence of both 0
If
and 1 simultaneously which could provide a great
(𝑟𝑎𝑛𝑑 > 0.5 𝑎𝑛𝑑 𝑚𝑏𝑒𝑠𝑡𝑑 > 𝑋𝑖𝑑 (𝑡))
amount of power to these machines to successfully map
or (𝑟𝑎𝑛𝑑 < 0.5 𝑎𝑛𝑑 𝑚𝑏𝑒𝑠𝑡𝑑 < 𝑋𝑖𝑑 (𝑡)) the molecules which potentially opens up opportunities
in research related to pharmaceuticals. Some of the
(𝑋𝑖𝑑 (𝑡 + 1)) = critical problems that could be solved via them are —
1 improving the nitrogen-fixation process for creating
𝐶𝑖𝑑 (𝑡) − 𝛽. ln ( ) . 𝑚𝑏𝑒𝑠𝑡𝑑 ammonia-based fertilizer; creating a superconductor at
𝑢
1 room-temperature; removing carbon dioxide from the
+ (𝛽. ln ( ) ± 1 + 𝛼) atmosphere for a better climate; and creation of solid-
𝑢
𝑋𝑖𝑑 (𝑡) + 𝑋𝑋𝑖𝑑 (𝑡) state batteries.
End
3.3. 𝑡 =𝑡+1  Financial Modelling: Finding the right mix for fruitful
4. Continue Until termination criterion is satisfied. investments for a finance industry is based on the risk
associated, expected returns and other factors which are
Note : Other algorithms like the Fractional order important for its survival in the market. To achieve that,
Darwinian Particle Swarm Optimization [28] also use the technique of ‘Monte Carlo’ simulations is
Fractional Calculus [5] but since they belong to the category continually being run on traditional computers, which
of evolutionary algorithm [29], we will not be covering it consume an enormous amount of compute time.
here. The reader may feel free to read the articles or However, by applying quantum computing [12] in
resources given in the reference section. You can also refer to performing these huge and complex calculations,
references section for experimental evidence of the efficiency industries not only reduce the time to develop them but
of FQPSO over QPSO. also improve upon the quality of the solutions.

IJISRT21AUG770 www.ijisrt.com 1063

Volume 6, Issue 8, August – 2021 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
 Cybersecurity & Cryptography: The online space C. Applications of Quantum Particle Swarm Optimization
recently has been quite vulnerable due to the increasing (QPSO) [11] and (FQPSO)
number of cyber-attacks occurring across the world.  Use of Quantum mechanics principles instead of
Although organizations are establishing essential using Newtonian laws for governing particle motion
security frameworks, the process has become unsettling unlike normal implementations of the Particle Swarm
and impractical for traditional computers. And, Optimization Algorithm (also known as the classical
therefore, cybersecurity has continued to be an important PSO implementations) [6].
concern round the world. We are becoming even more
vulnerable to these threats with our increasing  Quantum particle swarm optimization [8] can be used in
dependency on digitization. Quantum computers with order to tune antennas more accurately. It
the help of machine learning can help to develop several outperforms traditional PSO implementations and also
techniques to tackle these cybersecurity threats. has a much faster convergence rate than classical PSO.
Additionally, quantum computing [12] can help in
creating encryption methods, also referred to as,  The QPSO algorithm can be used for predicting /
quantum cryptography. designing a more efficient, equivalent circuit for a
given Dielectric Resonator Antenna (DRA) circuit [16].
B. Applications of Particle Swarm Optimization Algorithms This can further make it easier to derive important
(PSO) [10] circuit properties such as the Quality factor of the
 Symbolic regression, which may be a sort of circuit (Q-factor).
multivariate analysis that searches the space of
mathematical expressions to seek out the model that  Both QPSO and the FQPSO algorithms have very few
most closely fits a given dataset, both in terms of parameters which need to be tuned for optimization.
accuracy and ease, may be a crucially important This can be easily done by either checking accuracy at
theoretical and practical problem. different states or values (i.e. trial and error) or by
simple linear variation in the parameters. This also
 Floor planning is to design the layout of equipment in a makes it easier to avoid errors and hence makes the
factory or components on a computer chip to reduce algorithm even more accurate. Classical PSO generally
manufacturing time. involves 4 parameters: 𝑐1 , 𝑐2 , 𝑤 , 𝑉𝑚𝑎𝑥 . On the other
hand, general implementations of the QPSO algorithms
 Weapon target assignment problem is to seek out an involve only one parameter which can be tuned easily
optimal assignment of a group of weapons of varied for satisfactory convergence to optimal solution. This
types to a group of targets so as to maximize the has the obvious advantage of less computational cost as
expected damage done to the opponent. tuning 4 parameters takes much more cycles than tuning
a single parameter.
 Supply chain management is the systemic and strategic
coordination of the business functions and the tactics  The algorithm (QPSO) can also be used to find a set of
across them within a specific organization and across infinitesimal dipoles that produce the same near as that
businesses within the availability chain, for the needs of produced by a circular dielectric resonator antenna or
improving the performance of the individual companies DRA [16].
in the long run and therefore the supply chain as an
entire.  The Dielectric Resonator Antenna or DRA[16] is
often used in microwaves. They generally consist of a
 Nurse scheduling problem is to seek out an optimal block of ceramic material which acts as the dielectric.
path to assign nurses to shifts, typically with a group of This block is mounted on a metal plate. Radiation is
hard constraints during which all valid solutions must introduced inside the ceramic material by using a
follow and a set of sentimental constraints which define Transmitter. This radiation bounces around inside the
the relative quality of valid solutions. ceramic block giving rise to standing radio waves. The
walls of the ceramic block are preferentially transparent
 The mathematical study of waiting lines is named to radio-waves and hence let them escape into space
queuing theory. In queuing theory, a model is giving rise to a radio signal.
constructed so that queue lengths and waiting times can
be predicted to make business decisions about the  These antennas can have lower losses and be more
resources needed. efficient than metal antennas for high microwave and
millimetre wave frequencies. This is where QPSO can
be used in order to optimise frequency values [15]
for finding the most optimum frequency or a range of
optimum frequencies.

IJISRT21AUG770 www.ijisrt.com 1064

Volume 6, Issue 8, August – 2021 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
 The application of Fractional Calculus [5] to the field of [4]. Klein-Gordon Equation , Retrieved from :
Quantum Particle Swarm Optimization [8] results in the https://www.degruyter.com/view/journals/zna/71/9/artic
FQPSO technique. This new algorithm combines the le-p817.xml?language=en
speed of Quantum Particle Swarm Optimization [8] as [5]. Fractional Calculus , Retrieved from :
well as the long-term memory effects and non-locality of https://www.researchgate.net/publication/289991482_S
Fractional Calculus [5] leading to faster convergence of pecial_issue_on_Fractional_calculus_and_applications
particles while avoiding local optimums. It therefore [6]. PSO (Particle Swarm Optimization) , Retrieved from :
significantly enhances the global optimum finding ability https://www.spiedigitallibrary.org/conference-
of the algorithm. Hence it has been experimentally proceedings-of-spie/10322/1/Research-on-particle-
verified that Fractional Order Quantum Particle Swarm swarm-optimization-algorithm-based-on-optimal-
Optimization (FQPSO) outperforms the traditional movement/10.1117/12.2265257.full
integer-based Quantum Particle Swarm Optimization [8] [7]. Local attractor based PSO , Retrieved from :
in most of the cases. https://www.researchgate.net/publication/275066977_P
article_Swarm_Optimization_Based_on_Local_Attracto
VII. SUMMARY AND CONCLUSION rs_of_Ordinary_Differential_Equation_System
[8]. QPSO (Quantum Particle Swarm Optimization) ,
The world of Quantum mechanics might be a new Retrieved from :
addition to the realm of science, but it can be said beyond https://www.hindawi.com/journals/mpe/2019/1805198/
doubt that this new field has given rise to some powerful [9]. FQPSO (Fractional Quantum Particle Swarm
concepts, extendable to various areas of modern life, one of Optimization) , Retrieved form :
them being complex engineering optimization problems. The https://www.researchgate.net/publication/333912195_Fr
Quantum Particle Swarm Optimization Algorithm [8] is cross actional-
between the classical principles of Particle Swarm order_quantum_particle_swarm_optimization
Optimization (PSO) [6] and the probability equations which https://journals.plos.org/plosone/article?id=10.1371/jour
govern the quantum world. nal.pone.0218285
[10]. Applications of Particle Swarm Optimization ,
We further investigated yet another algorithm: The Retrieved from :
Fractional Order Quantum Particle Swarm Optimization, https://www.hindawi.com/journals/mpe/2019/8164609/
which happens to be a cross between its predecessor (the [11]. Applications of QPSO , Retrieved form :
QPSO) and the realm of Fractional Calculus. Together they http://ieeexplore.ieee.org/document/6852313
exhibit a display the potential advantages which fractional [12]. Quantum Computing , Retrieved from :
order systems are able to extend to different areas of modern https://ieeexplore.ieee.org/abstract/document/8991254
life. By replacing the integer-based approach to QPSO with a [13]. Applying Qubits In Real Life , Retrieved from :
fractional index we have thus been able to show the https://www.academia.edu/44373054/Atoms_Communi
superiority of FQPSO algorithm over the QPSO algorithm cation_Quantum_Computing
over various natural, optimization problem. Another variant [14]. Trapped Ion Quantum Computing , Retrieved from :
of this algorithm also known as the Fractional order https://www.academia.edu/44372853/Trapped_Ions_to_
Darwinian Particle Swarm Optimization [27], although not Quantum_Computing
based on quantum mechanical principles, can help us see [15]. Application in Antenna Frequency Optimization ,
how, extending Fractional Calculus [5] to evolutionary Retrieved from :
algorithms [29] can lead to a significant increase in accuracy https://www.researchgate.net/publication/3018797_Qua
in evolutionary algorithms (use of fractional order system in ntum_Particle_Swarm_Optimization_for_Electromagne
natural simulation). The reader can feel free to refer to the tics
references section for further reading and to keep exploring https://core.ac.uk/download/pdf/44172152.pdf
this vast intersection of Physics, Maths and Biology. [16]. DRA (Dielectric Resonator Antenna) , Retrieved from :
https://www.tandfonline.com/doi/abs/10.1080/0252266
REFERENCES 7.2020.1822043
[17]. Quantum Computing Application , Retrieved from
[1]. Young’s Double-Slit Exp , Retrieved from : :https://analyticsindiamag.com/top-applications-of-
https://openphysicsjournal.com/VOLUME/3/PAGE/122/ quantum-computing-everyone-should-know-about/
FULLTEXT/ [18]. Improving Quantum Computers , Retrieved from :
[2]. Grunewald-Letnikov derivative , Retrieved from : https://phys.org/news/2019-04-quantum.html
https://www.researchgate.net/publication/220481880_A_ [19]. Google’s Quantum Supremacy , Retrieved from :
Study_of_the_Grunwald- https://www.bbc.com/news/science-environment-
Letnikov_Definition_for_Minimizing_the_Effects_of_R 50154993
andom_Noise_on_Fractional_Order_Differential_Equati [20]. Google’s Sycamore Chip , Retrieved from :
ons https://www.scientificamerican.com/article/hands-on-
[3]. Schrodinger Equation , Retrieved from : with-googles-quantum-
https://www.researchgate.net/publication/262007560_A_ computer/#:~:text=Google's%20quantum%20computin
Review_of_Schrodinger_Equation_Classical_Wave_Equ g%20chip%2C%20dubbed,some%209.7001992547409
ation 92%20quadrillion%20bit%20strings

IJISRT21AUG770 www.ijisrt.com 1065

 Volume 6, Issue 8, August – 2021 International Journal of Innovative Science and Research Technology
ISSN No:-2456-2165
[21]. Superposition and Entanglement , Retrieved from
https://www.afr.com/technology/quantum-computing-
101-what-s-superposition-entanglement-and-a-qubit-
20191218-p53l2j
https://www.researchgate.net/publication/2196987_Supe
rposition_and_Entanglement_Pillars_of_Quantum_Infor
mation_Processing
[22]. Quantum Computing , Retrieved from :
https://www.technologyreview.com/2019/01/29/66141/w
hat-is-quantum-computing/
https://uwaterloo.ca/institute-for-quantum-
computing/quantum-computing-101
https://www.sciencedaily.com/terms/quantum_computer.
htm
https://www.wired.com/story/quantum-computing-here-
but-not-really/
https://arxiv.org/abs/0708.0261
https://iopscience.iop.org/article/10.1088/0034-
4885/61/2/002/meta
[23]. Quantum Algorithms , Retrieved from :
https://www.quantum-inspire.com/kbase/what-is-a-
quantum-algorithm/
[24]. Why Quantum Computing ? , Retrieved from :
https://www.ncbi.nlm.nih.gov/books/NBK538701/
[25]. Quantum Development Kit , Retrieved from :
https://docs.microsoft.com/en-
us/quantum/overview/overview
[26]. Swarm Intelligence , Retrieved from :
http://www.swarmintelligence.org/tutorials.php#:~:text=
Particle%20swarm%20optimization%20(PSO)%20is,as
%20Genetic%20Algorithms%20(GA).
[27]. Particle Swarm Algorithms , Retrieved from :
http://www.iitg.ac.in/rkbc/CE602/CE602/Particle%20Sw
arm%20Algorithms.pdf
[28]. Darwinian Particle Swarm Optimization , Retrieved
from :
https://www.researchgate.net/publication/220888367_Da
rwinian_Particle_Swarm_Optimization
[29]. Evolutionary Algorithms , Retrieved from :
https://www.researchgate.net/publication/261842296_Ev
olutionary_Algorithms

IJISRT21AUG770 www.ijisrt.com 1066