eduKateSG · How Science Works
Choose your reading route
Connect dispersion and nonlinearity, verify a travelling KdV pulse, and distinguish ideal solitons from driven or imperfect experimental waves.
Build the dispersion–nonlinearity balance · Chapters 1–3 · Start this route
Verify profiles and interactions · Chapters 4–6 · Start this route
Compare optical systems and evidence · Chapters 7–10 · Start this route
Full contents · Check your understanding · How Science Works master guide
A short wave pulse usually changes as it travels. Different parts of its spectrum can advance at different rates, spreading the pulse or reshaping it. Yet some nonlinear wave systems support a localised pulse whose profile moves almost unchanged. Its persistence comes from a particular balance between dispersion and nonlinearity.
Such a pulse is called a solitary wave. The stricter mathematical idea of a soliton includes how pulses survive interactions, recovering their characteristic forms after a collision, possibly with shifts in position or phase. A single impressive photograph cannot establish all those properties.
This guide develops the balance beneath the Fluid Dynamics collection, then compares it with an optical pulse. The water-wave and optical models share an organising idea, but their variables, equations and physical assumptions must remain distinct.
Contents
Build the dispersion–nonlinearity balance · Chapters 1–3
1. A pulse contains many wavelengths
Verify profiles and interactions · Chapters 4–6
4. Two original pulse comparisons
Compare optical systems and evidence · Chapters 7–10
7. An optical pulse uses a different balance
8. Real solitons need an energy and perturbation account
CHAPTER 1 OF 10 · Build the dispersion–nonlinearity balance
1. A pulse contains many wavelengths
A perfectly repeating sinusoidal wave has one spatial wavelength. A localised pulse, such as a smooth isolated hump, needs a range of spatial Fourier components to describe its shape. Their relative phases determine where those components reinforce one another and where they cancel.
Dispersion means that the frequency–wavenumber relation is not simply proportional over the relevant range. Components then accumulate different phase changes during propagation. The carefully arranged combination that initially made a compact pulse can evolve into a broader or more complicated pattern.
Spreading does not by itself mean that wave energy has been destroyed. In an ideal conservative dispersive model, energy can remain in the wave field while being redistributed over a wider region. Dissipation, which transfers mechanical or optical energy to other forms, is a separate effect.
A simple timing illustration makes the distinction visible. Imagine two narrow-band groups initially centred at the same position, with group speeds of 2.0 and 2.2 metres per second. After ten seconds, their centres are two metres apart. This calculation illustrates differential propagation; it is not a complete prediction of an arbitrary pulse’s width.
Nor must every dispersive pulse broaden monotonically. A pulse with a suitable initial frequency–time arrangement can first compress. The general statement is that dispersion changes the relative arrangement of spectral components. Predicting the resulting shape requires the initial state and the dispersion relation together.
CHAPTER 2 OF 10 · Build the dispersion–nonlinearity balance
2. Nonlinearity makes the wave affect its own evolution
In a linear wave equation, doubling a solution’s amplitude gives another solution, and suitable solutions can be added. In a nonlinear equation, the disturbance itself changes its rate of evolution. The effect of a large-amplitude part cannot always be obtained by scaling the small-amplitude result.
For weakly nonlinear shallow-water waves, one relevant effect is an amplitude-dependent propagation speed. A higher portion of an elevation wave tends to advance differently from a lower portion. Without an opposing effect, that tendency can steepen the profile.
Dispersion changes the profile in another way. Under an appropriate model, the two effects can cooperate to support a travelling shape: neither is absent, but their shape-changing contributions balance in a frame moving with the pulse.
This is not a balance between two ordinary forces pulling a solid object in opposite directions. It is a balance between terms controlling the evolution of a wave field. The distinction matters because a soliton is a moving pattern, not a rigid packet of material glued together.
A useful diagram has three panels. The first shows a pulse evolving under dispersion alone. The second shows nonlinear steepening when dispersion is omitted. The third shows an appropriately matched pulse translating under both effects. Label the panels as comparisons between models, rather than three guaranteed stages experienced by every real wave.
The nonlinear pulse formation and interaction problem was investigated computationally in Zabusky and Kruskal’s original 1965 study. The important insight was not simply that a hump can be long-lived, but that the resulting localised structures could interact and re-emerge with their identities largely preserved.
CHAPTER 3 OF 10 · Build the dispersion–nonlinearity balance
3. A model that makes the balance calculable
One classic equation is the Korteweg–de Vries equation, or KdV. In a standard dimensionless form it is uₜ + 6u uₓ + uₓₓₓ = 0. Here u describes the wave field, x is a scaled spatial coordinate and t is a scaled time. Subscripts indicate partial derivatives.
The first term describes change with time. The product 6u uₓ provides the nonlinear contribution. The third spatial derivative supplies dispersion. The numerical coefficients belong to this chosen normalisation; other books can use different coefficients while describing an equivalent rescaled model.
In a suitable water-wave derivation, KdV approximates long, weakly nonlinear waves travelling predominantly in one direction, with the relevant weak dispersion retained. Constant depth, negligible losses and the treatment of surface tension must be specified. It is not the full equation for breaking waves, turbulent surf or arbitrary ocean disturbances.
The MIT soliton lecture notes present this normalised equation and its travelling solitary solution. We will use it as an exact mathematical model, while keeping any application to real water conditional.
For positive κ, that solution is u(x,t) = 2κ² sech²[κ(x − 4κ²t − x₀)]. The parameter x₀ sets the initial centre. The function sech(y) is 1/cosh(y), where cosh(y) = (exp(y) + exp(−y))/2.
The expression already reveals three linked properties: amplitude 2κ², speed 4κ² and width scale 1/κ. Once κ is selected, we cannot choose all three independently and still claim this exact solution. The wave’s shape is a consequence of the balance, not an arbitrary hump carried unchanged by wishful thinking.
CHAPTER 4 OF 10 · Verify profiles and interactions
4. Two original pulse comparisons
Choose κ = 1 and x₀ = 0. Our first pulse is u = 2 sech²(x − 4t). Its peak value is two, its centre advances at four scaled distance units per scaled time unit, and its width scale is one.
Now choose κ = 2. The second pulse is u = 8 sech²[2(x − 16t)]. It has four times the peak value, four times the speed and half the width scale. These are dimensionless predictions; the number 16 is not a speed in metres per second until a physical scaling is supplied.
A precise width measure is the full width at half maximum. Solving sech²(κξ) = 1/2 gives a full width of 2 arcosh(√2)/κ, approximately 1.76275/κ. The two widths are therefore about 1.763 and 0.881 scaled distance units.
At t = 0.5, the first centre is at x = 2 and the second at x = 8. Each isolated exact solution retains its own profile. A detector fixed at one position sees the pulse rise and fall because it passes the detector; that time variation does not mean the travelling profile is changing shape.
The calculation also suggests a useful test. If supplied data claim to follow this normalised KdV family, a fourfold amplitude increase should accompany a halving of the width. A pulse with four times the amplitude and unchanged width does not match this particular solitary-wave family.
The test is conditional, not a universal law of all solitary waves. Another governing equation, a different background or a dissipative driving mechanism can have a different amplitude–width relation. Always attach the relation to the model that produced it.
CHAPTER 5 OF 10 · Verify profiles and interactions
5. Checking the solution rather than admiring its shape
Let ξ = x − ct − x₀ and write the travelling profile as f(ξ). Substituting into KdV gives −cf′ + 6ff′ + f‴ = 0. If the profile and its derivatives approach zero far away, integrating once gives f″ = cf − 3f².
For f = 2κ² sech²(κξ), differentiation gives f″ = 4κ²f − 3f². Comparing the two expressions requires c = 4κ², precisely the speed used above. This short verification shows that the translating profile satisfies the chosen equation.
An original centre-point check is especially simple for κ = 1. At the peak, f = 2 and c = 4, so cf − 3f² = 8 − 12 = −4. The curvature is negative, as expected for the top of a smooth hump. A wrong sign in the equation would fail this check.
One can also compare the area under each isolated profile. Since the integral of sech²(κξ) over the whole line is 2/κ, the integral of u is 4κ. The κ = 1 and κ = 2 examples therefore have areas four and eight in their scaled units.
That integral is a mathematical property of u. Calling it physical energy without knowing how u was defined would be a mistake. Surface elevation, optical field amplitude and optical intensity have different meanings and different energy relations.
A numerical simulation can be checked against the same exact profile. Halve the spatial step and time step, inspect whether shape error decreases, and distinguish boundary reflections from genuine pulse evolution. A smooth animation alone cannot show that a numerical method solved the intended equation accurately.
CHAPTER 6 OF 10 · Verify profiles and interactions
6. What surviving a collision means
Two widely separated KdV solitons can approach because the taller one travels faster. During their interaction, the combined field changes shape. The exact multi-soliton solution is not simply the sum of two permanently independent humps at every instant.
Far before and far after the encounter, the pulses can again be identified by their characteristic amplitudes, widths and speeds. Their eventual positions differ from the positions obtained by extrapolating uninterrupted isolated trajectories. Such shifts are part of the interaction, not evidence that the waves never influenced one another.
The 1965 numerical study compared emerging pulse widths and speeds with solitary-wave predictions and tracked their interactions. Its original plots are a reminder that the evidence included trajectories and evolving fields, rather than a single snapshot. Mathematical soliton theory sharpens those properties within the ideal equation.
A solitary wave is the broader description of a localised travelling wave. Calling a measured pulse a strict soliton adds claims about its governing dynamics and interaction behaviour. In experimental writing, the term is also used for approximately soliton-like structures or driven dissipative states, so the intended definition should be stated.
This prevents an attractive but weak inference: “The pulse looked unchanged at two locations, therefore it must be a soliton.” A nearly nondispersive linear wave could also retain its shape over that distance. The nonlinear amplitude relation, dispersion scale and interaction tests help distinguish explanations.
A collision experiment also needs appropriate geometry. Two waves moving in opposite directions are not automatically described by a one-direction KdV approximation. The model must cover the encounter being tested; a familiar equation should not be extended merely because the video looks similar.
CHAPTER 7 OF 10 · Compare optical systems and evidence
7. An optical pulse uses a different balance
In an optical fibre, a pulse envelope contains a range of frequencies. Group-velocity dispersion changes their relative timing. The Kerr effect makes the refractive index depend on intensity, producing an intensity-dependent phase change called self-phase modulation.
In a standard lossless fibre model with positive Kerr nonlinearity and anomalous second-order dispersion, these effects support a bright fundamental soliton. Its field envelope has a hyperbolic-secant shape, while its power profile is proportional to sech². The field and power should not be confused.
MIT’s original ultrafast-optics notes derive this nonlinear Schrödinger model and its fundamental pulse. The model assumes a suitable slowly varying envelope and neglects losses and higher-order effects in the simplest form.
For an invented parameter comparison, define a dispersion length Lᴅ = T₀²/|β₂| and a nonlinear length Lₙₗ = 1/(γP₀). Here T₀ is the sech-envelope time parameter, β₂ is second-order dispersion, γ is the fibre nonlinear coefficient and P₀ is peak power.
Take T₀ = 1 picosecond, β₂ = −20 picoseconds squared per kilometre, γ = 2 per watt per kilometre and P₀ = 10 watts. Both lengths are 0.050 kilometres, or 50 metres. Their ratio is one, consistent with the fundamental-soliton balance for the specified unchirped sech input and ideal model.
Equal length scales alone do not prove any arbitrary pulse is a soliton. Input shape and phase, dispersion sign and the governing assumptions still matter. Halving T₀ while retaining the same material coefficients requires four times the peak power, 40 watts, to keep that fundamental balance. These are paper calculations, not instructions for operating a laser.
CHAPTER 8 OF 10 · Compare optical systems and evidence
8. Real solitons need an energy and perturbation account
The conservative models above omit friction, absorption and external driving. Real water waves encounter viscosity, boundaries and changing depth. Real optical pulses can encounter loss, higher-order dispersion, Raman effects and imperfect preparation. These perturbations can change speed, width, spectrum or amplitude.
Some systems deliberately include both driving and loss. Dissipative optical solitons can persist because the input replenishes energy while dispersion and nonlinearity organise the pulse. Their continued existence does not imply a loss-free material or a source of unlimited energy.
A 2017 microresonator communication experiment explicitly used this double balance. The researchers used frequency-comb tones associated with dissipative Kerr solitons as communication carriers. The result should not be simplified into a claim that any soliton automatically carries information without noise or engineering limits.
A more recent 2025 study of driven bright solitons on a mid-infrared laser chip demonstrated pulses in an integrated active resonator system. Its reported operation depended on the device’s driven nonlinear dynamics. It was not an isolated optical pulse travelling forever without an energy supply.
These examples extend the concept while preserving the accounting. Name the stored field energy, the input, the output and the dissipative channels. Then ask which balance fixes the pulse. The same word can cover related mathematical structures without making all experiments governed by the same equation.
A useful failure test is to change a suspected perturbation. If a pulse’s width drifts as loss increases, compare that with a loss-inclusive model. If the waveform recovers only while a pump remains on, include the pump explicitly rather than calling the recovery evidence that dissipation has vanished.
CHAPTER 9 OF 10 · Compare optical systems and evidence
9. How to test shape preservation fairly
Suppose a supplied dataset contains three measured profiles taken at successive positions. Before comparing them, define the measured quantity, calibrate the axes and identify the pulse centre. A spatial elevation profile and a time trace at a fixed sensor are different objects unless speed is used to relate them appropriately.
Shift the profiles into a common moving coordinate so their centres coincide. Compare their absolute amplitudes and widths before normalising anything. Normalising every maximum to one would conceal amplitude decay, while rescaling every width would erase the spreading you are supposed to test.
For a fictional example, the first profile has amplitude 2.00 and half-maximum width 1.76, while the next has amplitude 1.80 and width 1.90 in the same scaled units. The amplitude has fallen by 10%, and the width has increased by about 8%. Calling the profiles identical after separately stretching their axes would be misleading.
Whether those differences matter depends on measurement uncertainty and propagation distance. A two-percent width uncertainty would make an eight-percent change worth investigating; a much larger uncertainty might leave the comparison inconclusive. Report the resolution and uncertainty rather than using visual similarity as a substitute.
To identify nonlinear balance, compare several amplitudes and predict their speeds and widths before inspecting the data. To test collision persistence, record profiles and trajectories on both sides of the interaction. To test a simulation, check convergence and boundary effects independently of how persuasive the animation looks.
The exercise can be completed entirely with supplied measurements, plotted analytic functions or a validated numerical dataset. There is no need to generate waves in hazardous water, approach a tsunami or operate an intense optical source. A single large pulse in nature should not be labelled a soliton without the appropriate dynamical evidence.
CHAPTER 10 OF 10 · Compare optical systems and evidence
10. Check your understanding
Does dispersion necessarily destroy wave energy? No. It changes the relative propagation of spectral components and can reshape a pulse without dissipating energy. Loss must be represented separately when it matters.
Why does the taller pulse in our KdV example travel faster? The exact family links amplitude 2κ² with speed 4κ² in the chosen normalisation. It is a model-specific nonlinear relation, not a universal speed rule for every kind of wave.
Can two solitons interact even if they recover their shapes? Yes. Their combined profile changes during the encounter, and their later positions or phases can be shifted. Recovery does not mean absence of interaction.
Is a long-lived solitary hump automatically a strict soliton? No. Its persistence could have another explanation, and the stricter claim involves the governing model and interaction properties. State the definition and identify which evidence has actually been collected.
Why do driven optical solitons not violate energy conservation? Their input compensates losses. Their stable form concerns a maintained dynamical state, not a supply of energy created from nothing.
The central idea is a tested balance: dispersion and nonlinearity can support special travelling structures, while real boundaries, losses and driving determine how closely an experiment follows the ideal. Continue through Fluid Dynamics or the How Science Works master to compare other examples where a stable pattern emerges from ongoing physical processes.