Chapter 1

Introduction to Nonlinear

Dynamical Systems

1.1 Introduction

The nature of the forces acting on various physical systems, along with their

initial conditions, determines how these systems evolve over time. Based on

the form of the forces, we can distinguish between linear and nonlinear forces.

This distinction leads to the study of linear and nonlinear dynamical systems,

particularly in systems with a limited number of degrees of freedom.

The description of the evolution of physical systems is built upon Newton’s

principles. Based on these laws, appropriate mathematical formulations can be

expressed in the form of differential equations (ordinary or partial), difference

equations, or even integral equations, or combinations of these [1, 2].

It is evident from the structure of dynamical equations, or equations of

motion, that the behavior of physical systems is primarily governed by the

structure of the forces (or potentials) acting on them. Recent research has

clearly shown that the presence or absence of linear or nonlinear forces

significantly affects system behavior. When the forces are nonlinear, studies

1
Introduction to Nonlinear Dynamical Systems 2

have led to the development of new and revolutionary concepts in a field

commonly known as nonlinear dynamics.

When modeling dynamical systems, nonlinear ordinary or partial differential

equations (ODEs/PDEs) or difference equations are often used. The choice

of equation depends on whether the system is continuous or discrete.

Generally, the study of nonlinear dynamical systems proceeds along two main

directions, based on the properties of the systems being studied: integrable and

nonintegrable systems.

Compared to the total number of nonlinear systems, the number of

integrable systems is almost negligible. Hence, they are quite rare. Although

many such systems have been identified and their properties studied, they are

considered exceptions rather than the rule. Moreover, many integrable systems

become nonintegrable when small disturbances are introduced. In contrast, as

the degree of nonlinearity increases, most nonintegrable systems exhibit highly

complex behavior.

Nonintegrable systems can display regular, irregular, aperiodic, and complex

behavior in phase space (the abstract space of state variables), depending on

the strength of the nonlinearity. Because these systems cannot be solved

exactly, one must rely on numerical integration. However, even a small error

in the initial conditions or round-off errors during numerical computation

can grow exponentially over time. As a result, the computed solution may

deviate significantly from the true trajectory after some time. This exponential

divergence of nearby trajectories occurs even when using highly accurate and

fast computers.

This raises the question of whether the effect described above is merely a

mathematical or computational artifact, or if it genuinely reflects the physical

behavior of the system. The exponential divergence of nearby trajectories has

1.1. Introduction 3

a clear physical interpretation: it corresponds to the extreme sensitivity of the

system’s behavior to initial conditions - a well - known feature of nonlinear

systems. Such complex behavior, which arises in deterministic nonlinear

systems and shows sensitivity to initial conditions, is referred to as chaotic

motion or simply chaos [1, 2].

Dynamical systems can be broadly categorized into dissipative and

conservative (Hamiltonian) systems. There are characteristic differences

between the nature of chaos exhibited by these two types of systems [3–5].

Most chaotic motions are observed in nonlinear Hamiltonian or conservative

systems, where the phase-space volume remains conserved during time

evolution. For example, in a two-dimensional phase space denoted by (q,p),

where q is the position and p is the conjugate momentum, Hamilton’s equations

evolve the system from an initial time t = t0 to t = t1 . Although the shapes of the

regions may change, their total area (or volume in higher dimensions) remains

the same [6].

Consequently, chaotic orbits in conservative systems are not attractive;

instead, they tend to explore all accessible regions of phase space uniformly.

The dynamics of a nonintegrable conservative system are therefore neither

completely regular nor completely irregular. Its phase space typically contains

a mixture of regular and irregular regions. In the regular regions, motion is

quasiperiodic and the trajectories lie on tori. In the irregular regions, the motion

appears chaotic, but the orbits are non-attractive.

Typical examples of conservative chaotic systems include coupled nonlinear

oscillators [7], the Hénon–Heiles system [8, 9], and the anisotropic Kepler

problem [10]. Quantum manifestations of Hamiltonian chaos have also been

studied extensively [11, 12]. However, this thesis focuses on dissipative

systems, not Hamiltonian chaos.

Introduction to Nonlinear Dynamical Systems 4

In dissipative systems, phase-space areas or volumes typically decrease over

time due to energy loss. As a result, trajectories tend to evolve toward a smaller

subset of phase space known as an attractor [13, 14]. For example, in a

damped harmonic oscillator, the system eventually settles at a point attractor,

corresponding to its rest state. In a driven oscillator, the attractor is a limit

cycle, representing periodic motion. In some systems, the attractor can be

much more complex and irregular, referred to as a strange or chaotic attractor.

Any differential equation, regardless of order, can be rewritten as a system of

first-order differential equations without explicit time dependence. This allows

the qualitative behavior of dynamical systems to be studied using geometric and

analytical methods. These studies introduce the notions of attractors, repellors,

and basins of attraction, each associated with specific types of motion [15, 16].

The simplest type of attractor is a stable equilibrium point. More complex

attractors include limit cycles, quasiperiodic attractors, and strange attractors.

To analyze their behavior, various statistical measures - such as phase space

trajectories, Poincaré sections, and Lyapunov exponents - are used. These

measures help to distinguish the degree of complexity of the attractors and the

underlying dynamics [1, 2].

1.2 Attractors

A dynamical system can be described as a deterministic mathematical

framework that governs the evolution of its state in time [17]. Consider a system

represented by a set of coupled first-order differential equations:

 
d
ẋi = Fi (x1 , x2 , . . . , xn ; µ), i = 1, 2, . . . , n, ˙= (1.1)
dt
∂Fi
where Fi and its partial derivatives are assumed to be bounded and
∂xj
1.2. Attractors 5

continuous functions of their arguments. The parameter µ determines the

behavior of the system. The vector X(t) = (x1 (t), x2 (t), . . . , xn (t)) represents a

trajectory or flow in an n-dimensional phase space.

Any nth -order time-dependent non-autonomous system with period T can

be transformed into an (n + 1)th -order autonomous system by introducing an

additional variable θ = 2πt

T
, so that θ̇ = ω = 2π
T
. Therefore, Eq. (1.1) can be taken

as the general representation of a nonlinear dynamical system. Solving this

system with initial conditions X(0) = (x1 (0), x2 (0), . . . , xn (0)) provides the flow

X(t) for any given t, leading to various kinds of attractors as discussed below.

1.2.1 Point attractors

In dissipative dynamical systems, the time evolution typically contracts the

phase-space volume, causing trajectories to asymptotically approach a specific

set known as an attractor [17, 18]. The simplest form of attractor is a fixed

point or equilibrium point, where all motion ceases.

A classic example is the damped pendulum, which, regardless of its initial

position or velocity, eventually returns to its lowest (vertical) position due to

air resistance and energy dissipation. Small perturbations around the stable

equilibrium do not change the long-term state of the system; the system always

returns to equilibrium once transient effects disappear. Such a point is called

a stable attractor.

Conversely, if nearby trajectories diverge from the equilibrium as t → ∞, the

fixed point is unstable and acts as a repeller. Examples include unstable nodes,

unstable foci, and saddle points. If trajectories neither approach nor diverge

from the equilibrium, it is termed neutrally stable. In dissipative nonlinear

systems, only stable nodes and stable foci act as genuine point attractors.

Fig. 1.1(a) illustrates a typical example of a stable-node point attractor in the

Introduction to Nonlinear Dynamical Systems 6

Duffing system.

Figure 1.1: Different types of attractors: (a) Point attractor (stable node) of the
Duffing system ẋ = y, ẏ = αx − hy − βx3 with α = 0.5, β = h = 1.0; (b) Limit
cycle of the van der Pol oscillator ẋ = y, ẏ = −αx + h(1 − x2 )y with α = 1.0,
h = 1.0; (c) Quasiperiodic orbit of the forced van der Pol oscillator ẋ = y,
ẏ = −αx + h(1 − x2 )y + f cos ωt with α = 1.0, h = 1.0, and f = 1.0; (d) Chaotic
attractor of the forced Duffing oscillator ẋ = y, ẏ = αx − hy − βx3 + f cos ωt with
α = 0.5, β = h = 1.0, ω = 1.0, and f = 0.42.

1.2.2 Periodic attractors

A second important type of attractor is the limit cycle. In two-dimensional

systems, it is possible to have closed, isolated orbits that are stable under

certain conditions. If every trajectory that starts sufficiently close to this closed

curve approaches it as t → ∞, the orbit is a stable limit cycle; if trajectories

approach it as t → −∞, it is unstable.

A stable limit cycle corresponds to a period-T attractor or periodic attractor,

1.2. Attractors 7

where the system’s motion repeats itself with period T [19]. Fig. 1.1(b) shows

an example of such behavior in the van der Pol oscillator [1].

1.2.3 Quasiperiodic attractors

A quasiperiodic attractor arises when the system exhibits motion governed

by multiple incommensurate frequencies. A quasiperiodic trajectory can be

expressed as

X(t) = ϕ(ω1 t, ω2 t, . . . , ωn t),

where ϕ is periodic with period 2π in each argument, and (ω1 , ω2 , . . . , ωn )

represent distinct frequencies satisfying the following conditions:

1. The frequencies are linearly independent, meaning there are no integers

{l1 , l2 , . . . , ln } (not all zero) such that l1 ω1 + l2 ω2 + · · · + ln ωn = 0.

2. There is no rational relationship among the frequencies, i.e., ωi ̸= |l1 ω1 +

l2 ω2 + · · · + ln ωn | for any integers {l1 , l2 , . . . , ln }.

Thus, a quasiperiodic motion can be considered as the superposition of

several periodic motions with incommensurate frequencies. Such motion

occurs on an n-dimensional torus (T n ). When nearby trajectories are attracted

toward this torus, it is referred to as a quasiperiodic attractor [see Fig. 1.1(c)].

1.2.4 Strange and chaotic attractors

Dissipative nonlinear systems can exhibit not only point and periodic attractors

(which are integer-dimensional), but also more intricate attractors with self-

similar geometries and non-integer (fractal) dimensions. These are called

strange attractors [20, 21].

Introduction to Nonlinear Dynamical Systems 8

Initially, strange attractors were believed to occur only in systems with

chaotic dynamics. However, later research demonstrated that nonchaotic

behavior can also correspond to strange attractors. To distinguish between

the two, we briefly introduce both strange chaotic and strange nonchaotic

attractors.

A strange nonchaotic attractor does not exhibit sensitivity to initial

conditions. When the system’s state is slightly perturbed, the trajectory

eventually returns to the attractor. In contrast, a strange chaotic attractor,

as defined by Ruelle and Takens [22], satisfies the following:

1. It is an attractor-meaning it is a bounded region of phase space that

asymptotically attracts trajectories from its basin of attraction-and it is

indecomposable (the trajectory visits every point on it over time).

2. It exhibits sensitive dependence on initial conditions.

A strange chaotic attractor is neither a finite set of points nor a closed

curve; instead, it forms a complex, self-similar, fractal structure. In such

systems, trajectories that start very close together diverge exponentially due

to the combined effects of stretching and folding of trajectories. This repeated

process leads to chaotic motion that is highly sensitive to initial conditions [see

Fig. 1.1(d)].

These attractors are typically found in higher-dimensional systems. For

discrete-time irreversible systems, chaos can occur even in one dimension,

while for reversible maps, at least two dimensions are required. In continuous-

time nonlinear systems, chaos emerges only when the phase space has three

or more dimensions.

When analyzing dissipative systems, it is common to distinguish between

transient behavior, which disappears after a finite time, and asymptotic

1.3. Bifurcation 9

behavior, which persists indefinitely. Depending on the control parameters,

dissipative systems can exhibit fixed-point, periodic, quasiperiodic, or chaotic

behavior.

Chaos in nonlinear systems can emerge through various routes as system

parameters are varied. The most commonly observed routes include the period-

doubling route [23], the intermittency route [24–26], the quasiperiodic route [27],

and the crisis route [28].

1.3 Bifurcation

In nonlinear dynamics, a bifurcation refers to a qualitative change in the

behavior of a system as a parameter is varied. In simpler terms, it means that a

small, smooth change in a system parameter can cause a sudden and major

shift in the system’s long-term behavior (steady states or periodic motions)

[29, 30].

Bifurcations often indicate a transition between different dynamical regimes,

such as moving from a stable equilibrium to oscillations, or from periodic

motion to chaos.

1.3.1 Types of Bifurcations

Bifurcations occur when the stability of an equilibrium point or a periodic orbit

changes as a parameter varies. These can be detected by analyzing the linear

stability near the fixed point.

Common local bifurcations include:

(a). Saddle-Node (Fold) Bifurcation

Two fixed points (one stable and one unstable) collide and annihilate each

other as the parameter passes a critical value [31].

Introduction to Nonlinear Dynamical Systems 10

Example: (ẋ = r + x2 ).

When (r < 0), two fixed points exist; at (r = 0), they merge; and for (r > 0),

no fixed points remain.

(b). Transcritical Bifurcation

Two fixed points exchange their stability as the parameter crosses a critical

value [32].

Example: (ẋ = rx − x2 ).

(c). Pitchfork Bifurcation

A single fixed point splits into three as a parameter changes [33]. There are

two types:

• Supercritical: A stable equilibrium becomes unstable and gives rise to

two new stable equilibria.

• Subcritical: A stable equilibrium disappears and gives rise to two unstable

equilibria.

Example: (ẋ = rx − x3 ).

(d). Hopf Bifurcation

A stable fixed point becomes unstable, and a stable limit cycle (oscillation)

appears (or vice versa) [33, 34]. This type of bifurcation is commonly observed

in systems such as the van der Pol oscillator when a control parameter changes

[35].

1.4 Bifurcation Scenario in the Duffing Oscillator

Next, we examine the bifurcation phenomena and the transition to chaos in

continuous-time (flow) dynamical systems described by ordinary differential

equations. Building upon our discussion of bifurcations and chaos, we begin by

1.4. Bifurcation Scenario in the Duffing Oscillator 11

analyzing the period-doubling route to chaos in a typical nonlinear dissipative

system - the Duffing oscillator [1].

1.4.1 The Duffing Oscillator

The Duffing oscillator represents one of the most fundamental nonlinear

differential equations encountered in physics, engineering, and even biological

systems. This model was first proposed by the Dutch physicist Georg Duffing

[36] to describe the hardening spring behavior observed in many mechanical

systems. The general form of the Duffing equation is given by

ẍ + αẋ + ω02 x + βx3 = f sin(ωt), α > 0, (1.2)

where α denotes the damping coefficient, ω0 is the natural frequency of the

linear oscillator, and β is the nonlinear stiffness parameter. The term f sin(ωt)

represents the external periodic driving force of amplitude f and frequency ω.

Equation (1.2) can also be interpreted as the equation of motion for a particle

of unit mass moving in a potential field V (x) under the influence of viscous

damping and external forcing. The potential is given by

1 1
V (x) = ω02 x2 + βx4 . (1.3)
2 4

Depending on the signs of ω02 and β, the potential can take different forms of

physical interest:

1. ω02 > 0, β > 0: A single-well potential with a minimum at x = 0 (Fig. 1.2a).

p
2. ω02 < 0, β > 0: A double-well potential with minima at x = ± |ω02 |/β and a

local maximum at x = 0 (Fig. 1.2b).

3. ω02 > 0, β < 0: A double-hump potential with a local minimum at x = 0 and

Introduction to Nonlinear Dynamical Systems 12

Figure 1.2: The shape of the potential function given by Eq. (1.3) can be
classified as follows: (a) For (ω02 > 0) and (β > 0), it represents a single-well
potential. (b) For (ω02 < 0) and (β > 0), it forms a double-well potential. (c)
For (ω02 > 0) and (β < 0), it corresponds to a double-hump potential. (d) For
(ω02 < 0) and (β < 0), it gives rise to an inverted single-well potential, which is
not physically significant.

maxima at x = ±(ω02 /|β|) (Fig. 1.2c).

Each of these configurations serves as a prototype for describing a variety

of nonlinear phenomena. For weak nonlinearities (|β| ≪ 1), the system exhibits

nonlinear resonances and jump phenomena, which can be studied analytically

using perturbation methods. However, for larger values of β, the system

displays a rich variety of periodic and chaotic behaviors, depending on the

control parameters f and ω. Since Eq. (1.2) cannot be solved analytically, we

must rely on numerical simulations to study its dynamics.

In this section, we focus on the most interesting case: the double-well

Duffing oscillator (ω02 < 0, β > 0). The single-well and double-hump cases can

be analyzed in a similar way (see, e.g., Ref. [1, 36]).

1.4.2 Period-Doubling Route to Chaos

To explore the bifurcation behavior, we fix ω02 = −1, β = 1, α = 0.5, and ω = 1,

and vary the driving amplitude f . Equation (1.2) is integrated using a standard
1.4. Bifurcation Scenario in the Duffing Oscillator 13

fourth-order Runge–Kutta algorithm (RK4). The system’s behavior is analyzed

through phase portraits in the (x, ẋ) plane and bifurcation diagrams.

For f = 0, the system undergoes damped oscillations that asymptotically

approach one of the stable equilibrium points (see Fig. 1.3(a)). Increasing f

leads to a stable periodic motion (limit cycle) of period T = 2π/ω about the same

equilibrium point, which persists up to f = 0.33 (Fig. 1.3(b)).

At f = 0.33, the system undergoes a bifurcation that gives rise to a new

limit cycle of period 2T (Fig. 1.3(c)). As f is further increased, successive

period-doubling bifurcations occur: at f = 0.357, a period-4T orbit emerges

(Fig. 1.3(d)), followed by period-8T , period-16T , and so on. These bifurcations

accumulate at a critical value fc ≈ 0.3589, beyond which the motion becomes

chaotic. Figure 1.3(e) illustrates a typical one-band chaotic attractor confined

to the left potential well for f = 0.366. Beyond f = 0.366, the one-band chaotic

attractor evolves into a cross-well (double-band) chaotic attractor (f = 0.42),

where the trajectory alternates between the two wells (Fig. 1.3(f)).

As f is increased further (f > 0.42), the system exhibits a complex

sequence of transitions including chaotic regions, periodic windows, reverse

period-doubling, and intermittent chaos. A summary of the main bifurcation

phenomena is presented in Table 1.1.

Table 1.1: Summary of bifurcation phenomena in the double-well Duffing

oscillator (1.2) with α = 0.5, ω02 = −1, β = 1, and ω = 1.
Value of f Nature of motion
f =0 Damped oscillation approaching stable focus
0 < f < 0.3437 Period-T limit cycle
0.3437 ≤ f < 0.355 Period-2T limit cycle
0.355 ≤ f < 0.3577 Period-4T limit cycle
0.3577 ≤ f < 0.3589 Further period doublings
0.3589 ≤ f < 0.3833 One-band chaotic attractor
0.3833 ≤ f ≤ 0.42 Double-band chaotic attractor
f > 0.42 Chaos with periodic windows and reverse bifurcations

The transition from periodic to chaotic behavior in the Duffing oscillator

Introduction to Nonlinear Dynamical Systems 14

Figure 1.3: Phase portraits for various values of forcing amplitude f. (a) f=0, (b)
f=0.33, (c)=0.35, (d)=0.357, (e)=0.366, and (f) f=0.42.
1.4. Bifurcation Scenario in the Duffing Oscillator 15

Figure 1.4: Bifurcation diagram for various values of forcing amplitude f vs

x(t).

illustrates one of the most common routes to chaos - the period-doubling

cascade.

1.4.3 Bi-directionally coupled Duffing oscillators

After that, we will examine two double-well Duffing oscillators that are

connected in a bi-directional manner and are driven continuously. The

equations that control these oscillators are as follows:

ẍ + αẋ − ω02 x + βx3 = F sin(ωt) + k1 (y − x)

ÿ + αẏ − ω02 y + βy 3 = F sin(ωt) + k2 (x − y). (1.4)

In this context, the states of the first and second oscillators are denoted

by the variables x(t) and y(t), respectively. In addition, the amplitude and

frequency of the external periodic force are denoted by the symbols F and ω,

respectively. The damping constant is denoted by α, the natural frequency

Introduction to Nonlinear Dynamical Systems 16

(a) 1

0.5
x(t) / y(t)
0

-0.5

-1

-1.5
0.1 0.2 0.3 0.4 0.5
F
(b) 1

0.5
x(t) / y(t)

-0.5

-1

-1.5
0.1 0.2 0.3 0.4 0.5
F

Figure 1.5: Bifurcation diagrams between F and (x(t), y(t)) for (a) k2 = 0.3 and
(b) k2 = -0.3, respectively, with fixed k1 = 0.3. The maximum value of x(t) and
y(t) are indicated, respectively, as red and blue color points.

is denoted by ω0 , and the nonlinear coefficient is denoted by β. Within the

scope of our investigation, the parameters mentioned above are set at α = 0.5,

ω02 = 1.0, β = 1.0, and ω = 1.0. On the other hand, the forcing parameter F ,

coupling strengths k1 and k2 will be subject to variation. Both the first and

second oscillators are represented by the coupling strengths denoted by the

letters k1 and k2 , respectively. Depending on whether the coefficients k1 and k2

have the same sign or a different sign, the coupling can be characterized as

either attractive or repulsive.

Further, we investigate the nature of oscillations exhibited by the

1.4. Bifurcation Scenario in the Duffing Oscillator 17

bidirectionally coupled system. The representative time evolutions of the two

oscillators are illustrated in Fig. 4.2 for the driving amplitude (F = 0.3). In

panel (a), the coupling strength is chosen as (k2 = +0.3), corresponding to

attractive coupling, while in panel (b) it is (k2 = −0.3), representing repulsive

coupling. The solid violet and dashed red lines denote the temporal behaviors

of (x(t)) and (y(t)), respectively. It is evident from the plots that in both cases

the two oscillators evolve with the same frequency and maintain a fixed phase

relationship, indicating the occurrence of synchronization. For the attractive

coupling (k2 > 0), the two oscillators oscillate almost in phase, whereas for

the repulsive coupling (k2 < 0), they exhibit an approximately anti-phase

relationship. This confirms that the system supports synchronized motion

for both types of coupling, with the coupling sign determining whether the

oscillators are in-phase or anti-phase synchronized. Such behavior reflects

the intrinsic symmetry of the bidirectional interaction, where both attractive

and repulsive couplings stabilize coherent oscillatory states with distinct phase

configurations (see Fig.1.5).

The emergence of synchronization in coupled nonlinear oscillators has

been extensively explored as a fundamental mechanism underlying collective

dynamics and information processing. In particular, Manaoj Aravind et al.

demonstrated that coupling can play a crucial role in enhancing the system’s

response to weak external signals through a phenomenon known as coupling-

induced logical stochastic resonance [?]. Their work showed that appropriate

coupling between noisy nonlinear elements can enable reliable logic operations,

even in the presence of significant fluctuations. Extending this idea, Aravind,

Parmananda, and Sinha later reported the phenomenon of emergent noise-

aided logic through synchronization, where synchronization between coupled

oscillators facilitates the realization of logic behavior without the need for
Introduction to Nonlinear Dynamical Systems 18

external noise tuning. These studies collectively highlight the deep connection

between synchronization, noise, and logical behavior in nonlinear systems. In

the present work, we build upon these insights by examining how bidirectional

coupling influences the oscillatory states of the system and how different

coupling regimes—attractive and repulsive—can lead to distinct synchronized

configurations that may serve as the foundation for implementing logical

operations in coupled oscillator networks.

In recent years, there has been a growing interest in utilizing nonlinear

dynamical systems as potential computing elements due to their rich

repertoire of behaviors such as multistability, chaos, synchronization, and

pattern formation. Unlike conventional digital logic systems that rely solely

on binary states and rigid architectures, nonlinear systems can process

information through their intrinsic dynamics, offering adaptability, parallelism,

and energy efficiency. The inherent sensitivity of nonlinear oscillators

to external perturbations allows them to perform complex tasks such as

signal detection, pattern recognition, and decision-making with minimal

hardware requirements. Moreover, phenomena such as stochastic and

coherence resonance can be harnessed to enhance computational reliability

in noisy environments, mimicking the way biological systems exploit noise

for functional advantage. Thus, nonlinear dynamical computing provides a

promising paradigm for developing unconventional computing architectures

that bridge the gap between traditional electronic logic and biologically inspired

information processing.

1.5 Nonlinear dynamics based computing

Logic gates are the building blocks of any digital circuit and computer

architecture. In general, following Boolean algebra, the logic operations are

1.5. Nonlinear dynamics based computing 19

performed by converting any given two inputs into a single logical output.

For any logic operation, both the inputs and outputs have two states, namely

’ON’ or ’TRUE’ and ’OFF’ or ’FALSE’ states [37, 38]. The reliability of logic

operations depends on the reliable operations of systems chosen. Since the

current demand for miniaturization of logic devices, speed of computation

and low-power consumption devices, it is inevitable to design an appropriate

system which is able to produce noise-immune gates. As a result, for the past

two decades or more, several schemes are being proposed such as DNA/RNA

computing [39, 40], quantum computing [41, 42], nano computing [43] and

nonlinear dynamics based computing in order to replace or complement the

existing computer architecture based on silicon chips [44–66]. Among these

methods, nonlinear dynamics based computing can make reliable and re-

configurable computer architecture because the underlying nonlinear systems

posses a large number of basic functions.

Utilizing the flexibility of nonlinear dynamical systems for storing,

communicating and processing of information in computer architecture has

been an active area of research in nonlinear dynamics. In this connection,

Hopfield had constructed a memory device using artificial neural network

to store and retrieve information [44]. Sinha and Ditto proposed a chaos-

computing scheme to emulate different logic elements [45, 46]. In an optimal

window of noise, the possibility of occurrence of logic behavior was observed

and this phenomenon is termed as logical stochastic resonance (LSR) [48–

51]. Strange nonchaotic attractors can also help to build dynamical logic gates

[64, 66, 67]. Besides, self-sustained oscillators can function as latches and

registers if Boolean logic states are associated with the phases of the oscillator

signals [56]. Heteroclinic computing is another nonlinear phenomenon based

computation using a collective system of nonlinear oscillators [68, 69].

Introduction to Nonlinear Dynamical Systems 20

Not only nonlinear dynamics based computing complements existing silicon

based technology, there also exist several efforts to extend computation

techniques to other domains, including optical [54], chemical [50, 70, 71],

physical [64], mechanical [72], biological [73, 74], molecular [40, 75] and other

areas of science [53, 76–81]. Instead of needing multiple hardwares for different

types of computations, nonlinear dynamical systems can act as processors of

a flexibly configured and reconfigured device to produce different logic gates

[82]. In practice, the generation of nonidealistic and ambient noise restricts

the ability to obtain different logic gates in these systems [49]. Thus it is

essential to choose appropriate nonlinear-dynamics based computing systems

to overcome these odds. As attractors determine the long-term behavior of

dissipative dynamical systems, the concept of attractors is central to the

analysis of many natural systems as well as to the design of artificial systems.

For example, the computational capabilities of neural networks are controlled

by the attractors of their collective dynamics. Thus the nature of the attractors

in the nonlinear systems constitute a focus of current research in nonlinear-

dynamics based computing. In general the state space of nonlinear dynamical

system is partitioned into various basins of attractors which are utilized for

dynamical based computing. How these attractors are tolerant to noise? These

and related questions are addressed in this thesis with reference to specific

nonlinear systems. In particular, we propose the nonlinear dynamics based

computing by utilizing the strange nonchaotic attractors [64, 67] and strange

attractors [66, 67] that appear typically in quasiperiodically driven nonlinear

systems and periodically driven nonlinear systems. In particular, strange

chaotic and strange nonchaotic attractors are able to produce all types of logic

gates and memory latches. The details of various computing schemes are given

in chapter 2.
1.6. Outline of the thesis 21

1.6 Outline of the thesis

Motivated by the above discussion, this thesis explores the implementation of

fundamental logic elements using nonlinear dynamical systems. In particular,

we demonstrate how basic logic operations such as AND/NAND and OR/NOR

can be realized in a bi-directionally coupled Duffing oscillator, as well as

in multiple-input, single-output logic configurations. These logic operations

form the essential building blocks for most digital circuits and computing

architectures. We anticipate that the present study could contribute toward the

development of a simple, fast, and cost-effective set of dynamic logic elements,

potentially paving the way for alternative computational architectures. A brief

outline of the work presented in this thesis is provided below.

1.6.1 Nonlinear dynamics based computing

To start with, in Chapter 2, we discuss the various ways of implementation of

the logic gates. In addition, we discuss various types of nonlinear dynamics

based computation.

1.6.2 Fundamental and Nonlinear dynamical based logical

computing

In Chapter 3, we discussed the Fundamental & Universal Computing in

Nonlinear System. Finally, we illustrate the dynamical mechanism for the

implementation of logic operations.

Introduction to Nonlinear Dynamical Systems 22

1.6.3 Realization of logic gates in bi-directionally coupled

nonlinear oscillators

In Chapter 4, we will investigate the hopping nature of trajectories in different

wells of appropriate nonlinear systems, one can observe that these systems

are capable of producing all types of logic operations. Specifically in the

present work, we demonstrate the design of logic gates in bi-directionally

coupled double-well Duffing oscillators by applying two logic inputs to the drive

system alone along with a fixed bias. The nonlinear system, comprising of

bi-directional components, exhibits varied logic behaviors within an optimal

range of coupling strength. Both attractive and repulsive couplings yield similar

and complementary logic behaviors in the first and second oscillators. These

couplings play a major role in exhibiting fundamental and universal logic gates

in simple nonlinear systems. Furthermore, we validate the robustness of the

bi-directional oscillators against moderate noise.

1.6.4 Realization of multiple-input and single-output logic

gates in nonlinear systems

In Chapter 5, we will investigate the designing of multiple-input and single-

output (MISO) logic gates using a periodically driven double-well Duffing

oscillator system. Our research focuses on the existence of OR and AND

logic gate properties in the output of a single double-well Duffing oscillator

with multiple inputs under the influence of external periodic forcing and bias.

Through our investigation, we demonstrate that the logic gate can be switched

between OR and AND by adjusting the bias value. Our approach successfully

scales the number of logical inputs for both the gates up to ten and even more.

Additionally, we verify that the logical output of this nonlinear system maintains
1.6. Outline of the thesis 23

robustness against noise.

1.6.5 Summary, conclusions and future outlook

Finally, the results presented in the thesis are summarized and concluded in

Chapter 6. Also, the possible extensions of our work and outlook on future

direction are indicated in the last chapter.