See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/277156623

A Summary of the Korteweg-de Vries Equation

Research · May 2015

DOI: 10.13140/RG.2.1.1579.4083

CITATIONS READS
0 3,878

1 author:

Tian Xiang
Renmin University of China
41 PUBLICATIONS   287 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Patlak-Keller-Segel related systems View project

Analysis on Cross-diffusive systems View project

All content following this page was uploaded by Tian Xiang on 26 May 2015.

The user has requested enhancement of the downloaded file.

 A Summary of the Korteweg-de Vries Equation
Tian Xiang
Institute for Mathematical Sciences, Renmin University of China, Beijing,
100872, China.

Abstract
Traveling waves or solitons as solutions to the Korteweg-deVries equation
(KdV) which is a non-linear Partial Differential Equation (PDE) of third or-
der have been of interest already since 150 years. In this review, we collect
some main features of the KdV equation, including the history and background
of the KdV equation; a derivation of the KdV equation; some elementary prop-
erties of the KdV equation: solitary solutions; the existence and uniqueness
of the initial value problem of the KdV equation, and finally the existence of
an infinite set of independent conservation laws. Some numerical simulations
are provided to illustrate the solitary wave solutions.

1. History and Introduction

The Kortweg-de Vries (KdV) equation is a nonlinear partial differential equation of third
order:

ut + 6uux + uxxx = 0, (1.1)

where u(x, t) denotes the elongation of the wave at place x and time t. The factor 6 is just
conventional but of no great significance. We can transform the KdV equation under the
transform
u → αu, t → βt, x → γx,
where α, β, γ are nonzero real constants, to yield
6αβ β
ut + uux + 3 uxxx = 0.
γ γ

The KdV equation (1.1) is widely recognized as a paradigm for the description of weakly
nonlinear long waves in many branches of physics and engineering. It describes how waves
evolve under the competing but comparable effects of weak nonlinearity and weak dispersion.
Indeed, if it is supposed that the x-derivatives scale as  where  is the small parameter
characterizing long waves (i.e., typically the ratio of a relevant background length scale to a
wavelength scale), then the amplitude scales as −2 and the time evolution takes place on a
scale of −3 .
The KdV equation (1.1) owes its name to the famous paper of Korteweg and de Vries [6],
published in 1895, in which they showed that small-amplitude long waves on the free surface
Email address: [email protected]
Keyword: KdV equation, derivation, exact solution, soliton, conservation law.
AMS Subject Classification: 35Q53, 35B32, 37K10, 37K15.

1
of water could be described by the equation
3c ch2
ζt + cζx + ζζx + δζxxx = 0, (1.2)
2h 6
where ζ(x, t) is the elevation of the free surface relative to the undisturbed depth h, c = (gh)/2
is the linear long wave phase speed, and δ = 1 − 3B, where B = ρ/gh2 is the Bond number
measuring the effects of surface tension (ρσ is the coefficient of surface tension and ρ is the
water density). Transformation to a reference frame moving with the speed c (i.e., (x, t) is
replaced by (x − ct, t), and subsequent rescaling readily establishes the equivalence of (1.1)
and (1.2). Although equation (1.1) now bears the name KdV, it was apparently first obtained
by Boussinesq [1, 1877, p. 360] (see Miles [9, 1970] and Pego and Weinstein [10, 1997] for
more historical discussions on the KdV equation).
The KdV equation was first formulated as part of an analysis of shallow-water waves in
canals, it has subsequently been found to be involved in a wide range of physics phenomena,
especially those exhibiting shock waves, traveling waves, and solitons. Certain theoretical
physics phenomena in the quantum mechanics domain are explained by means of a KdV
model. It is used in fluid dynamics, aerodynamics, and continuum mechanics as a model for
shock wave formation, solitons, turbulence, boundary layer behavior, and mass transport. It
has been studied and applied for many decades. The kdV equation has many remarkable
properties, including the property discovered by Gardner et al [5, 1967]: the KdV equation can
be solved exactly, as an initial-value problem, starting with arbitrary initial data in a suitable
space. This discovery was revolutionary, and it drew the interest of many scolars. We note
especially the work of Zakharov and Faddeev [14, 1971], who showed that the KdV equation
is a nontrivial example of an infinite-dimensional Hamiltonian system that is completely
integrable.
The KdV equation is particularly notable as the prototypical example of an exactly solv-
able model; that is, a non-linear partial differential equation whose solutions can be exactly
and precisely specified. For instance, One way to solve the KdV equation is to use the inverse
scattering transform. The solutions in turn include prototypical examples of solitons. The
mathematical theory behind the KdV equation is rich and interesting, and, in the broad sense,
is a topic of active mathematical research. More story about the history and the underlying
physical properties of the KdV equation can be found for example at Eilbeck [4, 1998] and
Zabusky [15, 2005].
In this note, we shall summarize some main features of the KdV equation. Section two
discusses the derivation of the KdV equation from the Boussinesq equation. Section three
deals with some elementary properties of the KdV equation: solitary solutions, exact solution
with Auto-Backlund transform is mentioned; some numerical simulations are provided to
illustrate the solitray wave solutions or solitons. Section four briefly depicts the existence
and uniqueness of the initial value problem of the KdV equation. The existence of an infinite
set of independent conservation laws is studied in section five.

2. Derivation of the KdV Equation from the Boussinesq E-

quation
The KdV equation was originally derived [6, 1895] as an approximate model of the evolution
of long waves of moderate amplitude, propagating in one direction in shallow water of uniform

2
depth. The derivation is based upon four key hypotheses:

(a) long waves (or shallow water), i.e., the undisturbed depth of water, h  1;

(b) small wave amplitude; a  h;

(c) the waves move primarily in one direction;

(d) all these small effects are comparable in size; this means

a h
= = O(( )2 ),
h λ
where λ is a characteristic horizontal length scale.

This kind of derivation can be found in Drazin and Johnson [3, p.7-12], Segur [12] and Tabor
[13, p. 278-282]. We now follow the Riemann invariants (or characteristics) method applied
by Zabusky and Kruskal [15] to obtain the KdV equation out of the Boussinesq equation.
We begin with the Boussinesq equation, which involves the actual displacements:

utt = (1 + 2ux )uxx + uxxxx (2.3)

First let us consider the PDE:

utt − F 2 (ux )uxx = 0, (2.4)
which is similar to (2.3), up to the dispersive term.
We desire to reduce this equation into two lower order equations - one with right-moving
solution, and one with left-moving solution. Thus we introduce the transforms:

w = ux , v = ut .

Under which (2.4) is equivalent to

wt − vx = 0,

vt − F 2 (w)wx = 0.

Multiplying the first equation by F and then adding and subtracting the second equation
yields
vt + F wt − F vx − F 2 wx = 0,

−vt + F wt − F vx + F 2 wx = 0.
The above two equations can easily be rewritten as
Z w Z w
∂h i ∂ h i
v+ F (ξ)dξ − F v+ F (ξ)dξ = 0, (2.5)
∂t 0 ∂x 0
Z w Z w
∂h i ∂ h i
−v+ F (ξ)dξ + F −v+ F (ξ)dξ = 0. (2.6)
∂t 0 ∂x 0

3
Let us now define the two Riemann invariants to make these equations have a complete
derivative: Rw
r(x, t) = v + 0 F (ξ)dξ,
Rw
s(x, t) = −v + 0 F (ξ)dξ.
Thus equations(2.5) and (2.6) can now be rewritten using the auxiliary parameters p and q
as follows:
dr ∂r ∂t ∂r ∂x ∂r ∂r
= + = −F = 0, (2.7)
dp ∂t ∂p ∂x ∂p ∂t ∂x
ds ∂s ∂t ∂s ∂x ∂s ∂s
= + = +F = 0, (2.8)
dq ∂t ∂q ∂x ∂q ∂t ∂x
where we have defined the parametrization as:
∂t ∂x
∂p = 1, ∂p = −F,

∂t ∂x
∂q = 1, ∂q = F.

Therefore, r is constant along the characteristic dx

dt = −F and s is constant along the char-
dx
acteristic dt = F . We now express F (w) in terms of the Riemann invariants:
Z w
r+s=2 F (ξ)dξ ≡ 2G(w). (2.9)
0

When F (ξ) > 0 the function G is increasing and so the inverse G−1 exists. We therefore
have
r+s
w = G−1 ( ).
2
Consequently, equations (2.7) and (2.8) can now be written fully in terms of r and s :
∂r  r + s  ∂r
− F G−1 ( ) = 0, (2.10)
∂t 2 ∂x

∂s  r + s  ∂s
+ F G−1 ( ) = 0. (2.11)
∂t 2 ∂x
In our case, we have an additional dispersive term in (2.4), which now reads:

utt − F 2 (ux )uxx = uxxxx , (2.12)

1
where F has been specified to be F (s) = (1 + 2s) 2 .
Then Eq.(2.7) and (2.8) correspondingly change to
dr
= −uxxx , (2.13)
dp

ds
= uxxx , (2.14)
dq
which means some mixing between the individual invariant.

4
 If we now take the initial condition s(x; t = 0) = 0 and assume that the dispersive term
is relatively small for all practical purposes, we may assume s(x; t) = s(x; t = 0) = 0. To
1
finally derive the KdV equation, we specialize our F (s) = (1 + 2s) 2 and s = 0 throughout
the evolution. Thus we obtain from (2.9)
Z w
1 2 3 r
G(w) = 2 (1 + 2ξ) 2 )dξ = [(1 + 2w) 2 − 1] = ,
0 3 2

and so
r 1 3r 2 r
w = G−1 ( ) = [(1 + ) 3 − 1] ∼
2 2 2 2
provided that r is small. Therefore,
r 3r 1 r
F (G−1 ( )) = (1 + ) 3 ∼ + 1.
2 2 2
Joining (2.10) and (2.14), we conclude
r 1
rt − ( + 1)rx + rxxx = 0. (2.15)
2 2
Under the transform
r + 2 → −6u, t → 2t, x→x
the above equation becomes the canonical form of the Kurteweg-de Varies eqaution:

ut + 6uux + uxxx = 0.

3. Solitary wave solution of the KdV equation

We now examine how one can readily integrate the KdV equation to obtain a so called soliton
solution. First, consider solutions in which a fixed wave form (given by z(x)) maintains its
shape as it travels to the right at phase speed c. Such a solution is called soliton solution
and is given by u(x, t) = z(x − ct). Substituting it into the KdV equation gives the ordinary
differential equation:
0 0 000
−cz + 6zz + z = 0. (3.1)
Integrating (3.1) once leads to
00
−cz + 3z 2 + z = a, (3.2)
0
where a is a constant. Multiplying (3.2) by 2z and integrating again leads to
0
−cz 2 + 2z 3 + (z )2 = az + b, (3.3)
Recall that we are looking for a solution that is solitary like Russell’s, meaning that away
0
from the heap of water, there is no elevation. Mathematically, that means z(x), z (x), and
00
z (x) should tend to zero as x → ±∞. From these requirements it follows that a = b = 0.
Remark: More general solutions can be found for other choices of a and b. These
solutions can be represented in terms of elliptic integrals, for details see Drazin [2, 1983].

5
 With a = b = 0 equation (3.3) can be written as

dz 2
( ) = z(cz − 2z 2 ). (3.4)
dx
By septation of variable we may write
Z Z
dz
√ = dx. (3.5)
z c − 2z

Setting z = c/2sech2 w, where sech is the hyperbolic secant, i.e., sechx = 2/(ex + e−x ), we
obtain that the solution of (3.5) is

c h √c i
2
z(x) = sech (x − x0 ) , (3.6)
2 2
where x0 is the constant of integration. It decays to zero exponentially as x → ±∞. With
this function z,
c h √c i
2
u(x, t) = z(x − ct) = sech (x − ct − x0 ) (3.7)
2 2
is the soliton, see Figures 1 and 2.

kdv u(x,t) eps = 6

−2
1
10
0.5
0
time 0 −10 x axis

Figure 1: single solition

6
 kdv time =1 eps = 6
5

3
u(x,t)

0
−10 −5 0 5 10
x axis
Figure 2: single soliton at final time.

Remarks: In order to have a real solution the quantity c must be a positive number. As
it is easily seen from (3.7) for c > 0 the solitary wave moves to the right. The second point
is that the amplitude is proportional to the speed which is indicated by the value of c. Thus
larger amplitude solitary waves move with a higher speed than smaller amplitude waves does.
If we put z = −c/2csch2 w in (3.5), where csch is the hyperbolic cosecant, then in the
same way as we derived the solution (3.7) we will get another soliton:

c h √c i
2
u(x, t) = − csch (x − ct − x0 ) (3.8)
2 2
The soliton (3.8) is an irregular solution to the KdV Equation. It has a singularity for
vanishing argument of the cosech-function, i.e., for the line in the x-t-plane with x−ct−x0 = 0.
Remarks: The solutions (3.7) and (3.8) can be obtained by applying Auto-Backlund
transform, see Drazin [3] for details. The KdV equation can support more than one soliton.
This can be shown using the Hirota direct method, or by the auto-Backlund-transform. These
derivations are beyond the scope of this review.
The remarkable stability of the soliton, discovered in the 1960s by computer experimen-
tation by Zabusky and Kruskal, can be roughly described as follows. If we start with two
solitons, the faster one will overtake the slower one and, after a nonlinear interaction, the two
solitons will emerge unscathed as they move to the right, except for a slight delay. In fact, it

7
was observed from the computer output that every solution of the KdV equation, with any
initial function u(x, 0) = φ(x), seems to decompose as t → ∞ into a finite number of solitons
(of various speed) plus a dispersive tail which gradually disappears, see figures 3 and 4.

kdv u(x,t) eps = 6

15

10

−5
1
10
0.5
0
time 0 −10 x axis

Figure 3: double solition at final time

4. Existence and uniqueness of the KdV equation

The existence of solitary solution to the KdV equation has been shown in section 3. In this
section we just briefly describe the initial value problem of the KdV eqution:

ut + 6uux + uxxx = 0,
(4.1)
u(x, 0) = φ(x).

Find a solution to (4.1) is somewhat involved, it needs the so called inverse-scattering-

transform. Schematically, the method is

φ(x) → scattering data at time 0

→ scattering data at time t (4.2)

→ u(x, t).

8
 kdv time =1 eps = 6
10

6
u(x,t)

0
−10 −5 0 5 10
x axis
Figure 4: double solition

The theory (see for example [11, p.369-371]) guarantees that all the three steps in (4.2) have
a unique solution, leading to the existence and unique solution of (4.1) with the prescribed
initial data.

5. Infinite Number of Conservation Laws of the KdV Equa-

tion
5.1. What are conservation laws?
If we think of a PDE, such as the KdV equation, as a dynamical system with infinite number
of degrees of freedom, it is natural to ask, if the equations have any integral of motion. For
PDE the notion of integral of motion is replaced by conservation laws. These are relations
of the form:
∂t N (x, ux , uxx , . . . ) + ∂x J(x, ux , uxx , . . . ) = 0, (5.1)
where N (x, ux , uxx , . . . ) is a ’density’ and J(x, ux , uxx , . . . ) is the associated ’flux’ and u is
the solution of the PDE at hand. For example, for the continuity equation, where N = ρ and
J = ρu, we have
∂t ρ + ∂x (ρu) = 0.

9
Conservation laws can also be associated with constants of the motion. We can see this by
integrating equation (5.1) with respect of x. Then if J decays sufficiently rapidly at the ends,
or if we have periodic boundary conditions, we obtain
Z
d
N dx = 0. (5.2)
dt
This has the consequence that Z
N dx = constant (5.3)

is a constant of the motion.

5.2. Conservation laws of the KdV equation

Historically, the number of conservation laws of the KdV equation was unfolding as time
progressed from the first discovery of this equation for water waves up to the 70’s of the 20’th
century. For the KdV equation, there are three fundamental quantities
R∞
mass = −∞ udx
R∞ 2
momentum = −∞ u dx (5.4)
R∞ 1
energy = −∞ ( 2 ux − u3 )dx.

Let us now check that these quantities are indeed constant of motion. First, we put the
KdV equation in the form
(u)t + (3u2 + uxx )x = 0.
From this conservation law it follows easily that
Z ∞
udx = constant
−∞

which verifies the conservation of mass. Multiplying the KdV equation by 2u and then
integrating from minus infinity to plus infinity yields

d ∞ 2
Z Z ∞ Z ∞
3
u dx + 4 ux dx + 2 uuxxx dx = 0. (5.5)
dt −∞ −∞ −∞

Recall that we have assumed that u and ux decay to zero as |x| → ∞. Then by integration
by parts we obtain from (5.5) that
∞
d ∞
R 2 dx = −4u3 
R∞
dt −∞ u x  − 2 −∞ uduxx
−∞
∞ R∞
= −2uuxx  + 2 −∞ ux uxx dx (5.6)

−∞
∞
= (u2x )x  =0

−∞

10
It follows from (5.6) that Z ∞
u2 dx = constant
−∞

which checks the conservation of momentum said in second line of (5.4).

Multiplying now the KdV equation by 6u2 and then integrating from minus infinity to
plus infinity obtains

d ∞ 3
Z Z ∞ Z ∞
4
2 u dx + 9 ux dx + 6 u2 uxxx dx = 0.
dt −∞ −∞ −∞

Thus, as done before,

d
R∞ 3
R∞ 2
2 dt −∞ u dx = −6 −∞ u duxx
∞ R∞
= −6uuxx 

+ 12 −∞ uux uxx dx (5.7)
−∞
R∞ 2
R∞ 3
=6 −∞ udux = −6 −∞ ux dx.

On the other hand, a differentiation of the KdV equation with respect to x yields

uxt + 6u2x + +6uuxx + uxxxx = 0, (5.8)

Multiplying both sides of (5.8) by 2ux and then integrating from minus infinity to plus infinity,
by using the decaying conditions we find
d ∞
R 2
R∞ 3 R∞ 2
R∞
dt −∞ ux dx = −12 −∞ ux dx − 6 −∞ udux − 2 −∞ ux duxxx
R∞ 3
R∞
= −6 −∞ ux dx +2 −∞ uxx uxxx dx (5.9)
R∞ ∞ R∞
= −6 −∞ u3x dx + u2xx  = −6 −∞ u3x dx.

−∞

Joining (5.7) and (5.9), we arrive at that

d ∞ 3
Z
(u − 2u2 )dx = 0,
dt −∞ x

which implies Z ∞
1
( u3x − u2 )dx = constant.
−∞ 2

This confirms the conservation of energy asserted in third line of (5.4).

For any conserved integral of the Korteweg-de Vries equation, the functional derivative
of the conserved integral evaluated at a solitary wave solution u = z(x − ct) is proportional
to the solitary wave solution, see more information in Lax[7].
Having found three conservation laws, one naturally asks whether there are any more
and indeed, whether there could be an infinite number corresponding to the infinite number
of degrees of freedom? This latter result would imply, in some sense, a type of complete
”integrability”. In 1968, a heroic effort by Miura [8] et al produced a tenth, at that time,

11
 which lead to the conjecture that the KdV equation had an infinite number of conservation
laws. This conjecture was also shown explicitly in 1973 by Whalquist and Estabrook and by
Lamb in 1974 using the Auto-Backlund transformation.

References
[1] Boussinesq, J. (1877), Essai sur la theorie des eaux courantes, Memoires presentes par divers
savants ‘ lcad. des Sci. Inst. Nat. France, XXIII, pp. 1680.

[2] Drazin, P.G. (1983). Solitons. London: Cambridge University Press (London Mathematical So-
ciety, lecture notes series).

[3] Drazin, P. G.; Johnson, R. S. Solitons: an introduction. Cambridge Texts in Applied Mathemat-
ics. Cambridge University Press, Cambridge, 1989, pp7-12.

[4] Eilbeck,c (1998). http:// www.ma.hw.ac.uk chris scott.russell.html.

[5] Gardner,C. S, Greene, J. M and Kruskal, M.D , and R. M. Muira, Method for solving the
Korteweg-de Vries equation, Phys. Rev. Lett. 19 (1967), 1095-1097.

[6] Korteweg, D. J.; de Vries, G. (1895), ”On the Change of Form of Long Waves Advancing in a
Rectangular Canal, and on a New Type of Long Stationary Waves”, Philosophical Magazine 39:
422443.

[7] Lax, Peter D. Integrals of nonlinear equations of evolution and solitary waves. Comm. Pure Appl.
Math. 21 1968 467490

[8] Miura,R.M, Gardner, C. S and Kruskal, M.D, Korteweg-de Vries equation and generalizations.
II. Existance of conservation laws and constants of motion, J. Mat. Phys. 9, 1204, 1968.

[9] Miles J.W. 1980. Solitary waves Annual Review of Fluid Mechanics 12: 11- 43.

[10] Pego, R.L. and Weinstein, M.J. 1997. Convective linear stability of solitary waves for Boussinesq
equations. Studies Appl. Math., 99: 311-375.

[11] Strauss, Walter A. Partial differential equations. An introduction. John Wiley & Sons, Inc., New
York, 1992.

[12] Segur, H, Integrable models of waves in shallow water, MSRI Publ. 55 (2007), 345371.

[13] Tabor, Michael, Chaos and integrability in nonlinear dynamics. An introduction. New York,
1989. xvi+364 pp. ISBN: 0-471-82728-2.

[14] Zakharov, V. E. and Faddeev, L. D, Korteweg-de Vries equation, a completely integrable Hamil-
tonian system, Funct. Anal. Appl. 5 (1971), 280-287.

[15] Zabusky, N. J, Fermi-Pasta-Ulam, solitons and the fabric of nonlinear and computational science:
history, synergetics, and visiometrics, Chaos 15 (2005), no. 1, 015102, 16 pp

[16] Zabusky, N. J and Kruskal, M. D, iteractions of solitons in a collisionless plasma and the recur-
rence of initial states, Phys. Rev. Lett. 15 (1965), 240243.

12

View publication stats