You are on page 1of 18

Analytical Insights into Nonlinear Wave

Propagation and Integrable Systems: The


Korteweg-de Vries Equation
By Berk Bozkurt

September 12, 2023

Abstract — Nonlinear wave phenomena have been widespread in various physical systems,
characterized by complex dynamics that defied conventional linear theories. This research report
delves into the captivating realm of nonlinear wave propagation, with a primary focus on the
Korteweg-de Vries (KdV) equation, an emblematic model governing weakly dispersive and
weakly nonlinear water waves. Originating in Boussinesq's implicit investigations in 1872, it
found explicit form in de Vries's dissertation in 1894, eventually culminating in the pioneering
work of Korteweg and de Vries in 1895. This essay embarks on an illuminating exploration,
contrasting the contributions of Boussinesq and KdV while uncovering intriguing connections in
the realm of integrable systems and the equation's resurgence through the Fermi, Pasta & Ulam
problem and inverse-scattering theory in the mid-20th century.

INTRODUCTION

This paper concerns how a unique class of partial differential equations (PDEs) arise and
the methods that were devised to solve them. Waves are a fundamental aspect of physics,
with their behavior extensively studied in linear systems. However, the emergence of
nonlinear wave phenomena has introduced captivating complexities, leading to deeper
investigations. Typical linear waves obey the principle of superposition, meaning that when
multiple waves coexist, their overlap results in a disturbance equal to the vector sum of the
individual wave amplitudes. On the other hand, nonlinear waves, which arise from the
nonlinear terms present in wave equations, don’t obey this principle of linearity. A notable
example of nonlinear waves that will be investigated in this paper is the soliton - a self-
reinforcing solitary wave that emerges from the balance between dispersion and nonlinear
effects. Mathematically speaking, a soliton is a solution to a PDE which is localized (with time)
and is preserved under interactions with other solitons (they can pass through one another
with no scattering). Solitons were originally a phenomenon peculiar to hydrodynamics, but,
as it turns out, they also appear in many other areas of physics.

Within the confines of this essay, our attention is directed towards the origin of the KdV
equation. Beginning with experiments conducted by Scott-Russel in 1834 to later
investigations of Boussinesq and Rayleigh around 1870 and finally ending with the article by
Korteweg and De Vries (in 1895) where they derived the one-soliton solution [1]. The
Korteweg-de Vries equation reads as,
𝜕𝑢 𝜕𝑢 𝜕3 𝑢 (1.1)
− 6𝑢 + =0
𝜕𝑡 𝜕𝑥 𝜕𝑥 3

and it is the result of extensive research on long waves in shallow water, where the variables
x and t represent position and time, respectively, and u = u (x, t) signifies the wave surface.

VISUALISING THE SOLITON

Earlier the definition of the soliton was given to be a self-reinforcing wave that emerges from
the balance between dispersion and nonlinear effects. Before looking at how solitons behave,
it is instructive to see what they’re made of and how these traits lead to their final formalism
Firstly by removing the non-linear term from the KdV equation (𝑢𝑡 + 𝑢𝑥×𝑥 = 0), the solution
to the equation yields dispersive waves – waves that spread and diffuse away energy. A plot
of this wave is shown,

Fig. 1: Plot of a dispersive wave evolving in time (left to right) and losing its form over time [2].

Dispersion refers to the phenomenon where different wavelengths of a wave travel at


different speeds. In the case of the KdV equation, dispersion causes the longer wavelengths
to travel faster than the shorter wavelengths, which leads to the dispersion of wave packets
over time.
The second option is to remove the dispersive term, leaving a non-linear differential equation
(𝑢𝑡 + 6𝑢𝑢𝑥 = 0). Removing the dispersive term from the KdV equation reduces the
equation's ability to account for the varying speeds of different wave components. As a result,
the equation tends to exaggerate the effects of nonlinearity, causing waves to focus their
energy into localized regions. This focusing effect can lead to an increase in wave amplitude
and steepening of wave crests, which are precursors to wave breaking. This effect is visually
shown in figure 2.

Fig. 2: Plot of a wave breaking.


The intricate dynamics of wave breaking hold paramount significance in shaping the ever-
changing coastal landscapes and facilitating erosion control strategies (details on the
applications of wave-breaking is discussed in section 4).

A soliton represents a remarkable equilibrium achieved by harmonizing two opposing


influences: the concentrating and dispersing effects within a wave system. This intricate
balance allows the soliton to retain its distinctive shape and energy content while propagating
through a medium. PDEs that form such solutions are referred to as integrable. To avoid
possible confusion, the number of available travelling wave solutions does not constitute its
integrability. Practically, as is the case for the KdV, complete integrability means one is able to
integrate such PDEs for a broad range of initial or boundary conditions. In short, an integrable
model is one which is exactly solvable.

Analytically solving the KdV equation yields travelling wave solutions which are referred to as
the KdV soliton. Travelling wave solutions are expressions of the form u (x, t) = f (x − vt), which
by substitution into the KdV eventually yields,

𝑣 √𝑣
𝑢(𝑥, 𝑡) = 𝑓(𝑥 − 𝑣𝑡) = 𝑠𝑒𝑐ℎ2 ( (𝑥 − 𝑥0 − 𝑣𝑡)) (2.1)
2 2

This solution has the properties shown in the table below.


speed 𝑣
height 𝑣
2
width 1
√𝑣

The plot of equation 2.1 gives us the image of a soliton evolving with time, shown in figure 3.

Fig. 3: Plot showing the time evolution of the soliton from its initial condition at t=0 (left) to some
subsequent time (right) whilst maintaining its fixed profile.

At first glance, the results don’t seem particularly fascinating as it’s some fixed profile
travelling at constant speed v. But once multiple soliton solutions are considered, one is able
to then visualize the fact that soliton interactions don’t obey the superposition principle.
Two-soliton solution

By assuming there exists an initial profile 𝑢(𝑥, 𝑡 = 0) = 6 𝑠𝑒𝑐ℎ2 (𝑥), one can solve for the
time-evolved form of the solitary wave which can be shown to be two solitons with amplitudes
4 and 16 which happen to overlap at the origin that seems to appear as one wave. This is more
clearly shown in the figure below which shows the waves before, during, and after the
interaction.

Fig. 4a: At t<0, the two solitons Fig. 4b: At t=0, the two solitons Fig. 4c: At t>0, the solitons
travelling as two separate have combined to appear as have split as the taller one
entities with the taller (faster) one soliton in an instant in surpassed the shorter one in
soliton yet to catch up with the time. position.
shorter one.

The solution shows the taller wave (a taller wave corresponds to a faster moving wave)
catches the shorter which then coalesces to form a single wave (the initial profile) and then
reappears as two separate waves with the taller one overtaking the shorter one. As a result
of their nonlinear interaction, a phase shift has occurred. This phase shift results from the
changes in amplitude and speed. When the soliton with a higher amplitude catches up with
another soliton, it transfers energy to the lower amplitude soliton. This energy exchange
causes the trailing soliton to slow down and stretch, effectively leading to a phase shift in the
overall waveform [3].

3. PROPERTIES OF THE KDV EQUATION

Conservation laws

The Korteweg-de Vries equation is the prototype of a completely integrable Hamiltonian


system with an infinite number of conserved functionals (temporal invariants) [1]. A
conservation law associated with an equation such as the KdV in one spatial dimension is
expressed by a partial differential equation of the form,
𝑇𝑡 + 𝑋𝑥 = 0 (3.1)

Where T, the conserved density, and X, the flux of T, are functionals of u. By assuming that u
and its spatial derivatives vanish rapidly enough at the infinities, each polynomial
conservation law yields a constant integral of motion [4],

𝑄𝑛 = ∫ 𝑇𝑛 ⅆ𝑥 (3.2)

Solving the KdV for 𝑇1 is a trivial case since it’s essentially in the conservation form already,
with 𝑇1 = 𝑢. As u is the displacement from the undisturbed level, the constant 𝑄1 is called
the mass (where unit density has been assumed). Multiplying eq. 1.1 by u, the second
conserved KdV quantity can be obtained. This quantity, 𝑇2 = 3𝑢 2 , corresponds to the
constant momentum. The factor ensures that the time derivative of the moment returns the
constant momentum, as shown below [5].

∞ ∞

[ ∫ 𝑥𝑢 ⅆ𝑥 ] = ∫ 3𝑢 2 ⅆ𝑥 = 𝑃 (3.3)
ⅆ𝑡
−∞ −∞

This mathematical relation in fact relates to symmetries of the Lagrangian and Noether’s
theorem which states that any continuous symmetry of a physical system gives rise to a
conservation law, and vice versa. Symmetries of the Lagrangian refer to specific
transformations or changes one can make to the variables in the Lagrangian without altering
the underlying physics. The type of symmetry associated with the constant momentum for
the KdV equation is the space translation symmetry. This means that if the spatial coordinate
x is shifted by a constant amount ε, the KdV equation remains unchanged. Mathematically,
this symmetry can be expressed as [2]:

𝑢 (𝑥, 𝑡) → 𝑢(𝑥, 𝑡 + 𝜀) = 𝑢(𝑥, 𝑡) + 𝜀𝑢𝑥 (3.4)

An inspection of the 𝑢𝑥𝑥 - dependent action (given without proof in eq. 3.5),

ⅆ 𝜕ℒ 𝜕ℒ ⅆ 𝜕ℒ ⅆ 𝜕ℒ
( + 𝑢𝑥𝑥 − ( ) 𝑢𝑥 − ℒ ) + ( 𝑢 )=0 (3.5)
ⅆ𝑥 𝜕𝑢𝑥 𝜕𝑢𝑥𝑥 ⅆ𝑥 𝜕𝑢𝑥𝑥 ⅆ𝑡 𝜕𝑢𝑡 𝑥

gives the conserved current,


𝜕ℒ
𝑃=6 𝑢 = 3𝑢 2 (3.6)
𝜕𝑢𝑡 𝑥

which is a conserved current because the action is space-translation invariant. Conversely, an


explicit space dependence in the Lagrangian would break space translation invariance, and
the current would no longer be conserved. The Lagrangian for the KdV eq. used to give the
result in eq. 3.6 is,
1 1
ℒ = 𝑤𝑥 𝑤𝑡 + (𝑤𝑥 )3 − (𝑤𝑥𝑥 )2 (3.7)
2 2
The numerical performance of the solitons in interacting with a strong permanence was
crucial evidence that there would be an infinite number of these constants of the motion. The
infinite set of conserved densities can be generated by using a Gardner transform. The
presence of these infinite conserved quantities gives numerous restrictions on the behaviour
of the solitary wave, limiting the possibilities for chaotic dynamics.

Inverse Scattering Transform (IST)

A standard scattering problem involves determining the spectrum {𝜓, E} with a given potential
u. In contrast, inverse scattering problems work by determining the potential given the
scattering data. The scattering data consists of the wavenumber, 𝑘𝑛 , the amplitude of the
discrete spectrum, 𝑐𝑛 (0), the time-independent reflection coefficient, 𝑅(𝑘, 0), and the
amplitude of the static transmission coefficient, 𝑇 (𝑘, 0). The scattering data corresponds to,

𝜓𝑥𝑥 + [𝑢 (𝑥, 𝑡) + 𝜆]𝜓 = 0 (3.8)

The IST method is used to solve an initial-value problem for the KdV equation within a class
of initial conditions; assuming that these ICs approach a constant sufficiently rapidly as x →
±∞. There is no loss of generality in choosing that constant to be zero. Since the 1950s it has
been well-established that the potential of the SE can be completely recovered from the
scattering data using the Gelfand-Levitan-Marchenko (GLM) equation. In other words, if you
know how particles scatter due to a certain potential, you can use the GLM equation to figure
out what the potential itself looks like [6]. This is significant because it provides a way to gain
insight into the behaviour of quantum systems and the forces that affect them. The GLM
equations act as a mathematical bridge connecting the scattering data (information about
how particles scatter due to the potential) to the potential itself.

An introduction to this technique, mathematically speaking, is beyond the scope of this


report. Consequently, the report postulates the equations required for it. The function of the
scattering data is defined as follows,

𝑁 ∞
1
𝐹(𝑥, 𝑡) = ∑ 𝑐𝑛2 𝑒 −κ𝑛 𝑥 + ∫ 𝑅(𝑘, 𝑡)𝑒 𝑖𝑘𝑥 ⅆ𝑘 (3.9)
2𝜋
𝑛=1 −∞

Where the first and second terms on the RHS represent the discrete spectrum data and
continuum data, respectively. The discrete spectrum gives the solitons (N of them), whereas
the continuous part gives a dispersive wave. The potential, u(x,t), can then be restored from
the equation,
𝜕
𝑢(𝑥, 𝑡) = 2 𝐾(𝑥, 𝑥, 𝑡) (3.10)
𝜕𝑥

Where the function K(x,y,t) is found via the linear integral-differential GLM equation,

𝐾(𝑥, 𝑦; 𝑡) + 𝐹(𝑥 + 𝑦) + ∫ 𝐾(𝑥, 𝑧; 𝑡)𝐹(𝑦 + 𝑧; 𝑡) ⅆ𝑧 = 0 (3.11)


𝑥

These steps have helped to transform a difficult nonlinear problem into a sequence of linear
problems. In addition to this, the IST method can be argued to be a nonlinear analogue of the
Fourier transform for linear problems [7]; the two diagrams in figure 5 illustrate this.

Fig. 5a: Diagram showing the steps involved Fig. 5b: Analogous diagram showing the steps
when solving for the time evolution of an involved when solving for the time evolution of an
initial profile for a linear equation [7]. initial profile for a nonlinear equation [7].

The Fourier method won’t be discussed here due to its triviality. An example of solving for the
time evolution for two bound states in a reflectionless potential using the same IC seen in
section 2 (𝑢(𝑥, 0) = 6 𝑠𝑒𝑐ℎ2 (𝑥)), will be discussed. Without proof, it can be shown that the
time-dep amplitudes are 𝑐1 (0) = √6𝑒 4𝑡 and 𝑐2 (𝑡) = 2√3𝑒 32𝑡. Since R(k)=0, the scattering
data is simply given by,

𝐹(𝑥, 𝑡) = ∑ 𝑐𝑛2 𝑒 −𝑘𝑛 𝑥 = 6𝑒 8𝑡−𝑥 + 12𝑒 64𝑡−2𝑥 (3.12)


𝑛=1

Using the ansatz, 𝐾(𝑥, 𝑦, 𝑡) = 𝑀1 (𝑥, 𝑡)𝑒 −𝑦 + 𝑀2 (𝑥, 𝑡)𝑒 −2𝑦 , and solving for M1 and M2 and
then lastly using 3.10, the time-dependent two-bound state solution can be shown to be,

𝜕 [3 + 4 𝑐𝑜𝑠ℎ(2𝑥 − 8𝑡) + 𝑐𝑜𝑠ℎ(4𝑥 − 64𝑡)]


𝑢2 = 2 (𝑀1 𝑒 −𝑥 + 𝑀2 𝑒 −2𝑥 ) = 12 (3.13)
𝜕𝑥 [3 𝑐𝑜𝑠ℎ(𝑥 − 28𝑡) + 𝑐𝑜𝑠ℎ(3𝑥 − 36𝑡)]2

The visual of this solution is shown in figure 4 (asymptotically for 𝑡 → ∞, two solitons emerge
having amplitudes 4 and 16 [8]) and it is mathematically seen that when t=0, eq. 3.13 becomes
the initial condition, hence verifying our result.

4. APPLICATIONS OF SOLITONS

A number of applications of solitary waves, with emphasis on a couple of cases where there
is extensive experimental confirmation of the predictions of KdV theory, will now be
examined. Unsurprisingly, many of these applications are in fluid mechanics - not surprisingly,
because fluid mechanics continues to maintain its seminal role in physics as the discipline in
which many key nonlinear structures were first discovered.

Plasma Physics

As mentioned in the abstract, the KdV equation was encountered for the first time since 1895
in 1966 in the work of Washimi and Taniuti [9] on ion-acoustic waves in a cold plasma.
Ordinary fluids can support the propagation of sound (acoustic) waves. Plasmas are
composed of a sea of ionized gas particles. This makes the propagation of sound in a plasma
more complex, due to the motion of electrons and ions. For a compressed region of the
plasma, the electrons (being much lighter than the ions) will react rapidly by moving towards
the denser region, creating an excess of negative charge in that area. On the other hand, the
ions, being heavier, will not respond as quickly and will lag behind the electrons, creating a
deficit of positive charge in the compressed region. A similar effect occurs for an expansion
of the plasma – a deficit of positive charge. This separation of charges (due to the difference
in the electrons and ions relative response times) produces an electric field in the plasma. The
electric field acts as a restoring force, and as the electrons and ions oscillate about their
equilibrium positions, they create pressure fluctuations that move through the plasma. These
pressure fluctuations are the ion acoustic waves.

Focusing on 1-D ion-acoustic waves in a plasma with cold ions1, negligible ion pressure,
isothermal, and in the absence of any current, one has the set of equations,

𝜕𝑛𝑖 𝜕
+ 𝜕𝑥 (𝑛𝑖 𝑣𝑖 ) = 0, (4.1)
𝜕𝑡

𝜕𝑣𝑖 𝜕𝑣𝑖
+ 𝑣𝑖 = 0, (4.2)
𝜕𝑡 𝜕𝑥

𝜕𝑛𝑒
𝑛𝑒 𝐸 + = 0, (4.3)
𝜕𝑥

𝜕𝐸
= 𝑛𝑖 − 𝑛𝑒 , (4.4)
𝜕𝑥

For the electric field E, ion velocity 𝑣𝑖 , and the electron and ion number density, 𝑛𝑒 and 𝑛𝑖
respectively. The four equations listed above are the mass and momentum conservation for
the ions, a force balance equation for the electrons, whose inertia in a low-frequency wave is
negligible, and Poisson's equation for the electric field given the charges respectively [10].
The unperturbed state here assumes an equal number density and charge magnitude for the
two species of particles and static ions (due to 𝑚𝑖 >>𝑚𝑒 ). The dispersion relation from (4.1) -
(4.4) is,
𝜔2
= 1 − 𝑘 2 + 𝑂(𝑘 4 ) (4.5)
𝑘2

1
Cold ions - ions with relatively low KE compared to typical plasma energies.
Which is valid for w<<𝑤𝑝 (ion-plasma frequency). In the situation where the waves are
weakly-nonlinear and weakly-dispersive, the dispersion relation, along with the terms
involving the convective derivatives2 of the wave in equations (4.1) and (4.2), it is evident that
fluctuations in the ion number density must satisfy KdV [10].

Hershkowitz, Romesser, and Montgomery [11] generated solitons from various square-wave
input voltages in a narrow column of cold plasma and were able to predict accurately the
number of solitons produced in any given case from the well-known solutions to the
Schrodinger equation for square-well potentials. The calculation also gave the soliton
amplitudes in reasonable agreement with the experiment, and the experiment further gave
good agreement with the theory on the speed-amplitude-width relation for KdV solitons
(similar to the table seen in section 2).

Ion-acoustic plasma waves provided the first experimental evidence of cylindrical KdV solitons
[12], and since these first experiments, the cylindrical KdV has been proved integrable [10].

Waves On the Shore

As waves approach the shoreline, their interaction with the ocean floor becomes increasingly
pronounced, leading to the transformation of orderly waveforms into turbulent whitecaps.
This occurs when deep water waves move towards shore and hence shallow water. When the
wave comes in contact with the sea floor, friction causes the bottom of the wave to slow down
before the top (as the bottom encounters the sea floor before the top). This results in the
wave crest leading ahead of the bottom, but since there is no water underneath to support it,
the wave becomes unstable and forms a breaker [13].

Fig. 6.1: Illustration showing how as the deep water waves approach the shore, the friction the
bottom of the wave experiences as a result leads to the wave breaking [13].

2
Convective derivatives - a type of derivative that takes into account changes in a quantity as it moves along
with a fluid flow.
This phenomenon, known as wave-breaking, generates immense hydrodynamic forces that
can result in the scouring of sediments and the reshaping of coastlines over time. Coastal
engineers harness a profound understanding of wave-breaking patterns to design and
implement robust protective measures such as breakwaters, seawalls, and groynes. These
engineered structures leverage the principles of wave breaking to dissipate wave energy,
prevent coastal erosion, and ensure the long-term stability of infrastructure and ecosystems
along coastlines. By intricately examining the mechanisms underlying wave breaking and its
implications for sediment transport and nearshore dynamics, coastal engineers contribute to
the sustainable management of coastal regions, effectively safeguarding communities and
vital resources against the forces of the ocean.

Optical fibres
Solitons have their primary practical application in optical fibres. Optical fibres are circular
dielectric waveguides that can transport optical energy and information [14].

Fig 6.2: Schematic diagram of the cross-section of two pulses travelling along the inner core of an
optical fibre. Random ellipticity and stresses along the fibre introduce small refractive index
variations. This causes one polarization state to travel faster than the other, resulting in a distorted
signal at the output of the fibre known as polarization mode dispersion (PMD) [15].

These fibres exhibit nonlinear and dispersion effects. The dispersion effects, where different
wavelengths of light travel at slightly different speeds, cause a pulse of light containing various
wavelengths to spread out over time leading to signal attenuation. In contrast, the nonlinear
properties give a focusing effect. This occurs primarily when there is a high-intensity light
beam, which causes the refractive index of the material (of the fibre) to have a more complex
and nonlinear relationship with the intensity of light. Consequently, there is a change in the
way the beam interacts with itself. This interaction creates a feedback loop: regions of higher
intensity cause an increased refractive index, and a higher refractive index leads to more
confinement of light [16]. This confinement causes the light pulse to become more compact
or concentrated as it travels, effectively self-focusing on its own core.

This self-focusing effect counteracts the dispersion caused by the varying speeds of different
wavelengths of light, as explained earlier. It helps the pulse maintain its shape and prevents
it from spreading out too much over long distances. The combination of this self-focusing
effect and the linear dispersion of the fibre contributes to the formation and stability of
solitons, allowing them to travel long distances while preserving their shape and amplitude
[17].

Nonlinear Acoustics of Bubbly Liquids

No account of applications of KdV should omit reference to theoretical and experimental


work on the propagation of nonlinear acoustic waves 3 in liquids with small volume
concentrations of gas bubbles. Such a suspension has remarkable acoustic properties, even
for very small volume concentrations of gas bubbles.

Consider, as an illustrative instance, the case of air bubbles submerged within water under
standard pressure and temperature conditions. In such scenarios, the sound speed at low
frequencies, denoted as 𝑐0 , propagates at approximately 40 𝑚𝑠 −1 for a concentration (of air
bubbles within the liquid) denoted as 𝛼0 =0.1. This value is not merely appreciably lower than
the sound speed of pure water 𝑐1 =1500 𝑚𝑠 −1 , but also notably inferior to that of the pure
gas phase (340 𝑚𝑠 −1 ). A discernible rationale emerges when inspecting the relationship 𝑐02 =
(bulk modulus)/(density). The presence of bubbles combines the low modulus of the gas
phase with the high density of the liquid phase. As a result, the low values of c₀ indicate that
nonlinear effects in bubbly liquids surpass those in single-phase liquids, even with minimal
bubble concentrations.

In addition, the bubbly liquid demonstrates significant dispersion characteristics. In the


absence of losses, for a suspension composed of bubbles of uniform size, the acoustic phase
speed (𝑐𝑝 (𝜔)) diminishes parabolically below 𝑐0 at lower frequencies and eventually reaches
zero at the resonance frequency of bubble monopoles (𝜔0 ). A distinctive phenomenon
emerges in the form of a restricted frequency band (𝜔0 < 𝜔 <𝜔𝑔 ), within which 𝑐𝑝 (𝜔))
becomes entirely imaginary, representing reactive motion. Beyond this band, 𝑐𝑝 (𝜔))
becomes real again, decreasing from positive infinity to essentially 𝑐1 as frequency ranges
from 𝜔1 to infinity. In the presence of losses, a marginal real component emerges within the
forbidden band, alongside a prominent increase in the imaginary component, peaking at
resonance. Experimental evidence notably supports these findings. An illuminating review by
van Wijngaarden [18] illustrates these trends, where he recorded phase speeds ranging from
as low as 500 𝑚𝑠 −1 to as high as 2800 𝑚𝑠 −1 across distinct frequencies, all measured on the
same bubbly mixture with an exceedingly modest bubble concentration, 𝛼0 = 2 × 10−4 .

These experiments have thus verified that in scenarios like air bubbles in water, the speed of
sound is much lower compared to pure water due to a combination of the bubbles'
characteristics.

5. THE NONLINEAR SCHRODINGER EQUATION

3
Acoustic waves - longitudinal waves generated as a result of the vibration from any source.
The Nonlinear Schrodinger Equation (NSE) emerges in a class of equations that describe the
evolution of slowly varying packets of quasi-monochromatic waves in weakly nonlinear media
that have dispersion. Its principal applications concern the propagation of light in nonlinear
optical fibres (see section 4) and planar waveguides.

If both dispersion and self-phase modulation effects act simultaneously on a wave pulse, the
propagation of the field envelope (the overall shape of a wave pulse) obeys the equation,

𝜕𝑢 1 𝜕2 𝑢 (5.1)
−ⅈ = 2
+ |𝑢|2 𝑢
𝜕𝑧 2 𝜕𝑡

This is the cubic Nonlinear Schrodinger Equation in one space (plus one time) dimension,
where u(z,t) is complex and proportional to the light field and t represents the retarded time
(the transit time delay of a pulse at the central frequency is accounted for/subtracted off).
Though the name is similar to the well-known Schrodinger equation of quantum mechanics
(QM), the NLS in fact has no relation to QM. Rather, it’s an adaptation of Maxwell’s equations
to field propagation in single-mode optical fibres. These single-mode fibres only exhibit one
possible spatial behaviour in the transverse dimensions x and y, meaning that averaging the
field quantities over those dimensions is appropriate. This allows examining the behaviour of
the pulse along the direction of propagation, z, only. Consequently, the term on the LHS
contains a spatial partial derivative with respect to the z-axis. The second-order time
derivative (the first term on the RHS) describes the effects of chromatic dispersion. The
nonlinear term (second term on the RHS) is simply u multiplied by the beam’s intensity
envelope, which ties in with the fact the refractive index is light-intensity dependent. This
term acts to stretch or compress the pulse in the frequency domain.

Eq. 5.1 is appropriate to use to describe behaviour in short and very long fibres (where signal
boosters are implemented) where loss/gain can be neglected. Zakharov and Shabat [19, 20]
showed, using the IST, that the general solution of eq. 5.1 displayed solitons accompanied by
smaller dispersive fields referred to as “radiation”. The general solution to eq. 5.1,

1
𝑢(𝑧, 𝑡) = 𝑠𝑒𝑐ℎ(𝑡) 𝑒 2𝑖𝑧 (5.2)

is named the “fundamental soliton” and is nondispersive (since the phase term is time-
independent).

6. THE FERMI-PASTA-ULAM PROBLEM

In 1955, E. Fermi, J. Pasta, and S. Ulam studied the effects of exciting a 1D array of 64
oscillators which were coupled by nonlinear springs (see fig 6a) in the first mode. The
experiment was designed to mimic how heat is conducted into solids [21]. They expected that
the system’s nonlinearity would lead to the equipartition of energy between all modes (i.e.
ergodic behaviour). However, to their surprise, their numerical experiments instead showed
that although energy did start to transfer from the first mode to higher modes, over time it
actually appeared to return to the first mode. This contradicted the common belief that
nonlinear systems would naturally evolve toward equilibrium and randomness. This
discrepancy was a "problem" because it highlighted a gap in our understanding of nonlinear
dynamics and the complexities of energy transfer and dissipation in such systems.

The equation of motion for a quadratically-coupled4 1D FPU lattice is,

𝑥̈ 𝑖 = 𝑥𝑖+1 + 𝑥𝑖−1 − 2𝑥𝑖 + 𝛼[(𝑥𝑖+1 − 𝑥𝑖 )2 − (𝑥𝑖 − 𝑥𝑖−1 )2 ] (6.1)

Where the index “i” denotes the ith oscillator, and 𝛼 is a parameter referred to as the
nonlinear coupling strength and takes on values from 0 to 1.

Fig. 6: (A) The first 4 energy modes for a system of 64 masses coupled via nonlinear springs in a
similar manner to that used by Fermi, Pasta and Ulam in their numerical experiments.
(B) FPU recurrences in a chain of 64 oscillators (see Eq. 6.1) with a coupling strength of α = 0.25 [22].

Fig, 6b shows how the first 4 energy modes for a system of 64 oscillators vary with time. It is
clear that the energy of each mode is largely different. Furthermore, the energy distribution

4
The potential energy associated with the interaction between two adjacent particles is determined by a term
that involves the square (quadratic) of the difference in their positions.
returned to the initial condition after a period of time. This phenomenon constitutes a
“recurrence” of energy between modes. Such so-called “FPU recurrences” have been the
subject of much research in the past half-century, and the FPU numerical experiments have
led to a wealth of work on recurrences and nonlinear lattice systems [22]. In fact, it was the
nonlinear interaction between molecules in a crystal lattice – due to the strong and weak
force- that sparked Fermi’s interest in using nonlinear springs to model this interaction.

The FPU problem marked the beginning of both the new field of non-linear physics and the
age of computer simulations of scientific problems.

7. SUPERLUMINAL SOLITONS?

This final section will delve into the significance of solitons in the compact mechanisms
contributing to superluminal motion within the framework of general relativity (GR). The
comprehensive mathematical framework of GR will not be extensively examined within the
confines of this paper due to its scope. General Relativity, conceptualized by Albert Einstein
in 1915, is a theory (of gravity) that reconceptualizes gravity as the curvature of spacetime
rather than a conventional force. In this formulation, instead of exerting an attractive force,
he reasoned that each object curves the fabric of space and time around them (the heavier
the object; the greater the influence), forming a sort of well that other objects — and even
beams of light — fall into, giving rise to the gravitational interactions observed in the cosmos
[23].

Before delving into the relevance of solitons in GR, pertinent terminology and definitions shall
be presented. It has been long known that the transportation of time-like observers5 at faster-
than-light velocities violates the weak, strong, and dominant energy conditions of GR. These
energy conditions are mathematically imposed boundary conditions (rather than physical
constraints) and are a generalization of the statement "the energy density of a region of space
cannot be negative" [24]. Violating these conditions could lead to inconsistencies, paradoxes,
and violations of causality. Theoretical approaches have been devised in order to overcome
these barriers, which involve constructing a class of soliton solutions that are capable of
superluminal motion and sourced by purely positive energy densities [25].

Previous soliton solutions in general relativity have long been tied to the requirement of
negative energy densities. This has posed a challenge as there are no known macroscopic
sources in particle physics that could generate such negative energy densities. The negative
energy sources required for these solitons have been thought to be created through energy-
intensive uncertainty principle processes.

The solitons presented by Erik W. Lentz overcome the need for negative energy sources by
using a hyperbolic relation between the components of the space-time metric's shift vector.

5
Time-like - information that can travel between two events at less than the speed of light (maintaining a
cause-and-effect order).
Shift vectors represent the changes in position as an observer moves through space and time.
The solitons have a central region with minimal tidal forces, where observers remain
stationary relative to the soliton. The energy density of the solitons is positive and satisfies
the weak energy condition. The solitons can be sourced from the stress energy of a conducting
plasma and classical electromagnetic fields, making them rooted in conventional physics.

It is worth noting that the energy requirements for constructing a self-sustaining soliton of a
specific size are immense, exceeding the scale of the visible universe. The energy is estimated
to be on the order of magnitude of -6 × 10^62 vs/c kg mass equivalent. Despite the significant
energy demands, the article mentions that advances have been made in this area, which has
lowered the required energy to around -10^30 vs/c kg mass equivalent. Furthermore, there
have been attempts to reduce the energy requirement to the scale of kilograms and grams.
These developments are promising as they bring the energy requirements closer to the
human technological scale. Additionally, it has been predicted that highly magnetized and
energetic atmospheric plasmas, such as those found in magnetars, may provide natural
environments for observing signatures of positive-energy soliton geometries even before
advances in energy reduction are achieved.

8. CONCLUSION

The mathematical findings reported by Korteweg-de Vries in 1895 have led to an extensive
body of computational, theoretical, and experimental work in nonlinear systems. This paper
explores various aspects of this research, from Fermi, Pasta, and Ulam's 1949 experiments
demonstrating energy recurrence in lattices to recent advancements in optical fibre
technology.

This paper began by visualizing how these solitary solutions behaved in solitude and their
interactions with one another. Their trait to remain as a fixed profile during their lifetime was
attributed to their dispersive and non-linear effects effectively balancing.

The pivotal aspect of the soliton ‘miracle’ is the mathematical revelation it presents. It is not
solely the balance of dispersion and nonlinearity, yielding travelling wave solutions - a
phenomenon observed in various systems—that underscores the significance. Rather, the
profound intrigue lies in the solitons' remarkable capacity to endure mutual interactions
essentially unaltered, apart from a phase shift. The solution to the two-soliton solution was
then later derived via the IST method, which was shown to collapse to the initial condition
(shown in section 2). The IST method is used to solve non-linear initial value problems by
transforming them into linear steps, and a worked-through example of this was discussed.

The attention of the paper then turned to the properties of the KdV equation and the
applications of solitary waves. The existence of infinitely many conservation laws being a
significant indicator of the integrability of the system was explained and the first number of
constants of the motion were derived. The infinite number of conserved quantities provides
the underlying mathematical framework that supports the fixed shape of a soliton during
interactions. These quantities, specifically the constant momentum, were shown to relate to
a symmetry in the Lagrangian approach.

The connections in the realm of integrable systems and the KdV equation's resurgence
through the Fermi, Pasta & Ulam problem were explored. The behaviour of the energy used
to excite a 1D array of 64 oscillators which were coupled by nonlinear springs was investigated
in this experiment and defy FPU’s prediction that over a long enough period of time, the
energy would be distributed evenly amongst all modes. Their intuition behind this ansatz was
that nonlinear systems would naturally evolve toward equilibrium and randomness. Rather,
the results revealed that the energy distribution returned to the initial condition, this
phenomenon constituted a “recurrence” of energy between modes (which later became
coined as the FPU recurrence).

Lastly, potential future research of solitons in the field of general relativity was discussed,
highlighting the promising avenues for further exploration in understanding these unique
wave patterns and their potential implications for advanced physics and cosmology.

ACKNOWLEDGEMENTS

I would like to thank my supervisor for this research project, Dr. Frank Berkshire, for his
guidance, insightful feedback, and unwavering support throughout the entirety of this
research endeavor.

REFERENCES

[1] E. De Jager, “On the Origin of the Korteweg-de Vries Equation,” 2011. Available:
https://arxiv.org/pdf/math/0602661.pdf

[2] Durham University, “Solitons,” 2006. Available:


https://www.maths.dur.ac.uk/users/S.A.Abel/Solitons/solitons.pdf

[3] Drazin, P.G. and Johnson, R.S. (1989) Solitons: An Introduction. Cambridge University
Press. https://doi.org/10.1017/CBO9781139172059

[4] R. M. Miura, C. S. Gardner, and M. D. Kruskal, “Korteweg-de Vries Equation and


Generalizations. II. Existence of Conservation Laws and Constants of Motion,” Journal of
Mathematical Physics, vol. 9, no. 8, pp. 1204–1209, Aug. 1968, doi:
https://doi.org/10.1063/1.1664701.

[5] F. Berkshire, “Nonlinear Waves Lecture 1”, Imperial College London, 2000.

[6] A. S. Fokas and V. E. Zakharov, Important Developments in Soliton Theory. 1993. doi:
https://doi.org/10.1007/978-3-642-58045-1.

[7] University of São Paulo, “KdV equation: Solitons.”. Available:


https://www.ifsc.usp.br/~laf/shnir/lecture_2.pdf

[8] F. Berkshire, “Nonlinear Waves Lecture 2”, Imperial College London, 2000.

[9] H. Washimi and T. Taniuti, “Propagation of Ion-Acoustic Solitary Waves of Small


Amplitude,” Physical Review Letters, vol. 17, no. 19, pp. 996–998, Nov. 1966, doi:
https://doi.org/10.1103/physrevlett.17.996.

[10] D. G. Crighton, “Applications of KdV,” Acta Applicandae Mathematicae, vol. 39, no. 1–3,
pp. 39–67, Jun. 1995, doi: https://doi.org/10.1007/bf00994625.

[11] N. Hershkowitz, T. Romesser, and D. Montgomery, “Multiple Soliton Production and the
Korteweg—de Vries Equation,” Physical Review Letters, vol. 29, no. 24, pp. 1586–1589, Dec.
1972, doi: https://doi.org/10.1103/physrevlett.29.1586.

[12] N. Hershkowitz and T. Romesser, “Observations of Ion-Acoustic Cylindrical Solitons,”


Physical Review Letters, vol. 32, no. 11, pp. 581–583, Mar. 1974, doi:
https://doi.org/10.1103/physrevlett.32.581.

[13] P. Webb, “10.3 Waves on the Shore,” rwu.pressbooks.pub.


https://rwu.pressbooks.pub/webboceanography/chapter/10-3-waves-on-the-shore/

[14] “Photonics Technical Note #21 Fiber Optics Fiber Optics: Fiber Basics.” Available:
https://www.newport.com/medias/sys_master/images/images/hd5/ha2/8797095919646/F
iber-Basics.pdf.

[15] P. Chou, “Fiber Optics Part 3: Fiber Dispersion Will Change The Way You See Your Links,”
Cisco Blogs, Sep. 03, 2020. Available: https://blogs.cisco.com/sp/fiber-optics-part-3-fiber-
dispersion-will-change-the-way-you-see-your-links.

[16] G. P. Agrawal, Nonlinear Fiber Optics. Academic Press, 2012.

[17] “The KdV Equation and Solitons,” COMSOL. https://www.comsol.com/model/the-kdv-


equation-and-solitons-85.

[18] L. van Wijngaarden, “One-Dimensional Flow of Liquids Containing Small Gas Bubbles,”
vol. 4, no. 1, pp. 369–396, Jan. 1972, doi:
https://doi.org/10.1146/annurev.fl.04.010172.002101.

[19] A. B. Shabat, “Inverse-scattering problem for a system of differential equations,”


Functional Analysis and Its Applications, vol. 9, no. 3, pp. 244–247, 1979, doi:
https://doi.org/10.1007/bf01075603.

[20] V. E. Zakharov and A. B. Shabat, “A scheme for integrating the nonlinear equations of
mathematical physics by the method of the inverse scattering problem. I,” Functional Analysis
and Its Applications, vol. 8, no. 3, pp. 226–235, 1975, doi:
https://doi.org/10.1007/bf01075696.

[21] N. Staff, “Mathematicians Partially Solve Fermi-Pasta-Ulam Problem | Sci.News,”


Sci.News: Breaking Science News, Mar. 23, 2015. Available:
https://www.sci.news/othersciences/mathematics/science-fermi-pasta-ulam-problem-
02628.html

[22] H. Nelson, M. A. Porter, and B. Choubey, “Variability in Fermi-Pasta-Ulam-Tsingou arrays


can prevent recurrences,” Physical Review E, vol. 98, no. 6, Dec. 2018, doi:
https://doi.org/10.1103/physreve.98.062210.

[23] J. Deaton, “Einstein Showed Newton Was Wrong about gravity. Now Scientists Are
Coming for Einstein.,” NBC News, Aug. 03, 2019.
https://www.nbcnews.com/mach/science/einstein-showed-newton-was-wrong-about-
gravity-now-scientists-are-ncna1038671.

[24] Curiel, E. (2014). "A Primer on Energy Conditions". arXiv:1405.0403.

[25] E. W. Lentz, “Breaking the warp barrier: hyper-fast solitons in Einstein–Maxwell-plasma


theory,” Classical and Quantum Gravity, vol. 38, no. 7, p. 075015, Mar. 2021, doi:
https://doi.org/10.1088/1361-6382/abe692.

You might also like