A Bifurcation Analysis of The Forced van der Pol Oscillator

Arun Sharma
140281167
MTHM038
2014/15

A thesis presented for the degree of

Master of Mathematics
Supervised by Prof W. Just

Department of Mathematics
Queen Mary University of London

Abstract
In this paper we explore the dynamics of a nonlinear damped differential
equation, the Forced van der Pol Oscillator. We have done this by constructing
the systems bifurcation diagram as it offers a concise and qualitative way of
measuring the systems dependence on certain bifurcation parameters. We go
on to conclude that the system is invariant symmetric about our two subject
bifurcation parameters, as well as convergent to fixed point solutions along one
parameter axis and a stable limit cycle along the other parameter axis. The
system is found to undergo a Saddle-node bifurcation when an eigenvalue becomes zero, as well as a Supercritical Hopf bifurcation due to a pair of complex
conjugate eigenvalues. These both collectively split the bifurcation diagram
into different regions. We also show that one of the two calculated Saddlenode branches is actually a global, Infinite-period bifurcation, which is due to
the oscillations period increasing to infinity, creating a saddle point.

Acknowledgements
I am deeply grateful to my supervising Professor Wolfram Just for his exceptional guidance and support throughout this thesis. I thank him for his
patience and encouragement whenever my progression stalled, and accredit
the completion of this paper to his outstanding teaching ability.

140281167

Introduction

Balthazar van der Pol was a pioneer in telecommunications and an electrical engineer
who when building electronic circuit models of the human heart, began to study the
triode oscillations he encountered in his work. To describe the periodic orbits he
found, Balthazar formulated his now famous equation, the Van der Pol oscillator.
The van der Pol equation is a classical example of a self oscillating system with
nonlinear damping. Energy is generated at low amplitudes and dissipated at high,
and typically gives birth to whats known as a limit cycle. Researching the dynamical
properties these periodic solutions hold have become increasingly popular over recent
years, and due to the nature of the equation, it has become fundamental in describing
oscillating systems. To this date the van der Pol Oscillator has been used in several
applications in diverse fields such as biology, meteorology and sociology [1, 2].
Due to its vast applications and unique characteristics, better understanding the van
der Pol oscillator will shed light on the properties many of these oscillating systems
hold and also the potential for new applications. Therefore this paper sets out to
examine the dynamics of the system by constructing its Bifurcation diagram.

1.1

Method of analysis - Bifurcation diagram

Jules Henri Poincar originally coined the term bifurcation when describing the separating of equilibrium solutions in differential equations. His work showed that given
a system of differential equations that are dependent on a certain parameter, the
topological type of flow can alter as this parameter varies and several branches of
equilibria can come together to form a bifurcation point. Since then many classes

140281167

of flows have been identified, and because of this bifurcation diagrams have become
useful tools in Mathematics for understanding how qualitative changes in flows can
arise due to varying parameters. Fixed points can be created or destroyed and their
stability can be altered, which can all be graphically represent on a single graph
[1, 3]. Hence by constructing this diagram for the van der Pol Oscillator, one can
gain a better understanding for the systems dependence on certain bifurcation parameters and represent the information in a concise, graphical manner. To do this,
this paper will be looking at the Saddle-node, Hopf and Infinite-point bifurcations,
as they sufficiently cover most of the systems dynamics. The mathematical program
Matlab will be used to numerically portray any plots, the bifurcation diagram and
its dynamics, and the built-in function ode45 will generate the phase space diagrams for the systems differential equations. But now we shall move onto the actual
formula.

The Forced van der Pol Oscillator

We shall be using the following form of the system taken from [1]

x + (x)x + x = p(t),

(1)

where (x) is even and < 0 for |x| < 1, and (x) > 0 for |x| > 1. p(t) is periodic for T and , are nonnegative parameters that are  1. Now following [1]s
work, rewriting the equation as an autonomous system and making a
transformation, the system can become

periodic

140281167

u = u v u(v 2 + u2 )

(2)

v = u + v v(v 2 + u2 ) .

(3)

This shall be the Forced van der Pol variant that we use throughout the paper, where
the 2 bifurcation parameters we shall be studying in more detail are and .

Symmetry

We begin by exploring the symmetrical properties of the transformed Forced van der
pol Oscillator (2) and (3) in order to reduce any unnecessary computations. As
and are the bifurcation parameters to be examined, the symmetries for the system
shall be subject to them alone.

By substituting u u and v v into (2) and (3), the first symmetry for
the parameter can be explored, leading to

u = u + v + u(v 2 + u2 )

(4)

v = u v + v(v 2 + u2 ) .

(5)

Equation (4) now represents u while equation (5) represents v with a . Therefore it is evident that the dynamics of u and v are invariant under the substitution
. Hence, the oscillator will resemble a system that as a whole has begun to
rotate as passes through 0 into , yet will retain the same fundamental characteristics.
3

140281167

Now again using a similar analyses for negative time t t and u u, the
symmetry for the parameter can be explored. (2) and (3) then become,

u = u v + u(v 2 + u2 )

(6)

v = u + v v(v 2 + u2 ) .

(7)

By now setting it is evident that (6) becomes u and (7) becomes v.

When = 0 the system will revert its rotation and oscillate in the other direction,
and as the systems flow is continuous, the same will occur for small . Therefore
(2) and (3) also share symmetries and retain their fundamental characteristics as
.
Consequently by combining both symmetries, the overall bifurcation diagram and
analysis can be condensed as each quadrant of the diagram repeats itself in a symmetrical manner. Hence only a single quadrant needs to be constructed to fully
understand the systems bifurcation diagram.
This conclusively means that only values of > 0 and > 0 will be considered
as it will significantly simplify the calculations. We therefore move onto the initial
bifurcation analysis for the system.

Special Cases

We begin by analytically solving the system for certain special parameter values as
this will form the basis for the global bifurcation diagram and analysis. The special
cases give birth to both fixed points as well as a stable limit cycle, and are found

140281167

along the lines = 0 and = 0. However due to the systems spherically dynamics,
we first impose polar coordinates to further discretize the system.
This is begun by making the substitutions u = rcos and v = rsin, along with
using the identity

u2 + v 2 = r2 .

(8)

Differentiating the above formula with respect to time yields

uu + v v = rr,

(9)

where we then substitute in (2) and (3) to obtain

r = r r3

v
r

(10)

or
r = r r3 sin

(11)

This is the first of two formulas and describes the radial velocity of the system, and
we make use of the second identity
in order to calculate the rotational velocity ,
(12) (taken from [3]),

uv v u.

Substituting u = rcos and v = rsin into (12) returns

(12)

140281167

r2 ,

(13)

and we again use identity (12) but now with (2) and (3) to return

r2 u.

(14)

Therefore by equating equations (13) and (14), as they are both derived from identity
(12), we can use them to formulate a function where is the subject,

cos()
=
.
r

(15)

This is now the formula for the rational velocity of the system Forced van der pol
system, and with both polar coordinate equations, we can now explore the special
cases.

4.1

Along the line = 0

Before we begin this section, the reader must fully understand what is meant by a
fixed point and its stability, and therefore we shall summarise [3]s example on fixed
points.

Given that a one-dimensional systems flow is governed by x = f (x), the system

will flow to the right when f (x) > 0 and to the left when f (x) < 0. At any point
at which the flow stops we encounter what is known as a fixed point, an equilibrium
solution defined by f (x ) = 0, and the stability of this equilibrium point is defined
by the flow at sufficiently small distances. If a system converges back to a fixed
6

140281167

point when studying the flow on either side of an equilibrium point, we say this is
a stable fixed point. Conversely, if the system diverges away, we say the fixed point
is unstable. The above summary of fixed points and their stability is portrayed in
Figure 1, where a black circle represents a stable and hollow represents an unstable
fixed point.

Figure 1: The vector field for the one-dimensional system, x = f (x).

Therefore with an understanding for the nature of fixed points we can move onto the
first case = 0, where we find that the system produces a stable limit cycle and can
verify this by setting = 0 in (11) and (15) to yield

h(r) = r = r r3

(16)

= t + constant.

(17)

and

By graphically representing (16) we can show it to have a stable solution at r = 1,

or in other words, a stable fixed point.

140281167

Figure 2: Plot of (16) where the red dot at r = 1 symbolises a stable fixed point.
This however means that as t , r(t) 1, and concludes that there exists a
stable limit cycle.
This emphasises that for any parameter value , the system will always converge
into a stable, symmetrical limit cycle if = 0, as portrayed in Figure 3.

140281167

Figure 3: The Forced van der Pol system converging to a stable, symmetrical limit
cycle of radius length 1, when = 0 and = 0.5

4.2

Along the line = 0

Now by evaluating both r = 0 and = 0, the fixed point solutions along the line
= 0 can be uncovered. Using these with (11) and (15) lead to

sin = r r3

(18)

cos()
= 0,
r

(19)

and

where the solution to (19) for is found to be 2 . Subbing =

yields

into equation (18)

140281167

= r r3 ,

(20)

where Figure 4 graphically represents the different type of fixed point solutions for
.

Figure 4: This figure is a graphical representation of (20) where the full horizontal
lines portray solutions with 1 or 3 fixed points, and the dotted line is the boundary
where the number of fixed point solutions change - there exists a saddle point on this
line. In this graph a red dot represents a stable point and a red circle is unstable.

10

140281167

From the above figure it is clear that if 0 6 <

are stable and 1 is unstable, and if >

,
3 3

,
3 3

there are 3 fixed points where 2

there is 1 stable fixed point. Therefore,

the oscillating dynamics of the system are destroyed along = 0, and depending on
the second bifurcation parameter, will either converge to 1 or 2 stable points (Figure
5).

Figure 5: This graph portrays the boundary parameter conditions for the system.
Along the red line there exists a stable, symmetrical limit cycle, along the green we
have 3 fixed points and along the purple we have 1 fixed point.

11

140281167

Now that the special cases have been considered, the more elusive Saddle-node, Hopf
and Global bifurcations can be explored.

Saddle-node Bifurcations

The Saddle-node bifurcation is a tool for analysing the creation or destruction of fixed
points in a system, the fixed points move closer as a parameter varies and eventually
collide to mutually annihilate each other. At this point the derivative of the Jacobian
is 0, which occurs due to a zero eigenvalue [1, 3]. To illustrate how this works we
shall be highlighting a section from [3]s work on the first order system

x = r + x2 .
[3] demonstrates that for varying r, the system undergoes a Saddle-node bifurcation
where the system goes from having 2 fixed points, x , to a saddle point and then to
none (Figure 6).

Figure 6: On the left we have r < 0, in the middle r = 0 and the right r > 0.
Therefore as r 0 the parabola pushes up, forcing the 2 fixed points to converge
until they form a saddle point, a half stable point at r = 0 (x = 0). As r > 0,
12

140281167

this point vanishes and any fixed points the system had are now destroyed. Hence
we recognise that the system has successful undergone a Saddle-node bifurcation
at r = 0. For the Forced van der pol oscillator we find that system undergoes a
Saddle-node bifurcation along a given trajectory for and . Therefore we begin
by noting that we require the solutions to the system where the fixed points are
mutually annihilated, u and v , and where the system has a 0 eigenvalue leading to
a determinant = 0.

5.1

Formulating u and v

Hence, by equating (2) and (3) to 0, they can be rearranged to yield

0 = u v u r2

(21)

= u + v v r2 ,

(22)

where the fixed points version of (8), u2 + v2 = r2 have been used. Now, rewriting
(21) and (22) in matrix form, they can be solved for u and v to yield the fixed
points solutions.

u =

(1 r2 )2 + 2

(23)

v =

(1 r2 )
.
(1 r2 )2 + 2

(24)

Please see Appendix 10.1 for calculations.

13

140281167

Equations (23) and (24) now describe the systems Saddle-node points, but can be
analytically used to formulate separate trajectory formulas for both and , both
in terms of just r2 .

5.2

Formulating and

Again using relation (8) but now with the newly constructed (23) and (24), we find
that,

r2 =

2
2 + (1 r2 )2

(25)

or

2 = r2 ( 2 + (1 r2 )2 ).

(26)

Now utilizing the zero eigenvalue condition allows us to set the determinant of the
Jacobian equal to 0, which we must do in order to construct an equation in terms of
. Thus

2
2
h i
1 v 3u 2u v
= 0 ,
det
2u
1 u2 3v2

(27)

which can be re-arranged for 2 to produce

2 = 4r2 3r4 1,
14

(28)

140281167

Please see Appendix 10.2 for calculations.

with the positive root

p
4r2 3r4 1.

(29)

Therefore the above formula is the Saddle-node bifurcation trajectory for our first
parameter , and is only in terms of r2 , as earlier stated. Now subbing (28) into (26)
yields the formula for the second parameter, 2 , again only in terms of r2 ,

2 = 2r4 (1 r2 ),

(30)

with positive root

As (29) is strictly > 0 when

1
3

p
2r4 (1 r2 ).

(31)

< r < 1, we only consider these values of r for ,

and as (31) is strictly > 0 for 0 < r < 1, we again only consider these values of r
for (we have neglected negative r values as r is by definition positive). Therefore
plotting (29) against (31) both from 0 < r < 1, while neglecting the imaginary
values for (29), results in Figure 7.

15

140281167

Figure 7: The Forced van der pol Saddle-node bifurcation plot. The trajectories
mark the boundaries where the systems oscillating dynamics fundamentally change.
Figure 7 demonstrates the destructive nature of the bifurcation parameters and splits
the quadrant into different sections where fixed points are either known to exist or
not. Unlike with the special cases, Figure 7 is the first real analysis of the relationship between both parameters when they are both non-zero, and their affect on
the systems fixed points. However Figure 8 expresses how damaging the creation of
fixed points can be to a limit cycles flow, and how this can be achieved with minor
parameters changes.

16

140281167

Figure 8: Here we have the Forced van der pol system when = 0.2, as well as
= 0.15 and = 0.25 from left to right respectively. In the chapter on Global
bifurcations we see that the Saddle-node branch examined in this plot is actually an
Infinite-period bifurcation, which occurs along a saddle-node trajectory.
We now move onto the next section that explores the creation of limit cycles from
stable fixed points, or vice versa, and what is known as a Hopf bifurcation.

Hopf Bifurcation

The Hopf bifurcation is another useful tool for understanding the dynamics of differential equations, and is able to quantify the conditions needed for a stable fixed

17

140281167

point loose stability. Given a two-dimensional system where both eigenvalues have
Re < 0, the system will converge to a stable fixed point. However if a parameter
varies and forces at least one of the eigenvalues to obtain a Re > 0 or both to become complex conjugates (+iw, iw), the fixed point will lost stability and undergo
a Hopf bifurcation.
Obtaining complex conjugate eigenvalues will force the system into a limit cycle;
although the reverse is also possible, a limit cycle can be reverted back into a stable
fixed point by obtaining two eigenvalues with Re < 0 [1, 3].
There are 2 possible types of Hopf bifurcations, a Supercritical or Subcritical. If
a decay parameter moves past a critical threshold to become a growth, the equilibrium state will loose stability and the system will undergone a Supercritical Hopf
bifurcation (Figure 9).

Figure 9: An illustration of the amplitude over time of a system that has undergone
a Supercritical Hopf bifurcation. The full line is the parameter value that falls below
the threshold, decay, and the dotted line falls above the bifurcation threshold, growth.

18

140281167

If the trajectory jumps to an attractor in the form of a fixed point, limit cycle or
infinity after the system bifurcates, the system has undergone a Subcritical Hopf
bifurcation [3].
Nevertheless we can analytically compute the conditions for the Hopf bifurcation for
the van der Pol system by recalling that its eigenvalues must be complex conjugate.
This of course yields a Determinant(Jacobian) > 0 and a T race(Jacobian) = 0.
Therefore by using these conditions, an equation describing the trajectory for the
Hopf bifurcation can be formulated in terms of 2 and 2 , similarly to Saddle-node
section.

6.1

Radial length

We begin by calculating the Hopf bifurcation point using the fixed point solutions for
(2) and (3), again like the saddle-node section. However we then utilise the T race
condition to conclude that

r2 =

1
2

(32)

is the radial length the Hopf bifurcation occurs at. We will use this solution in the
next subsection to solve our formulas for 2 and 2 .

6.2

Formulating and

By now evaluating the Determinant(Jacobian) > 0 condition yields the relation

19

140281167

2 > 4r2 3r4 1,

(33)

Please see Appendix 10.2 for calculations of a similar type.

which is a re-written version of (28). Furthermore by taking (32) and subbing it

into (26), we produce the following hyperbola

1
1
2 = 2 + ,
2
8

(34)

which is portrayed in Figure 10.

Figure 10: This graph is for the entire hyperbola, equation (34), stretching over the
4 bifurcation quadrants.

20

140281167

The above plot, Figure 10, portrays the entire hyperbola plane, however we only
require the North-Eastern quadrant due to the earlier mentioned symmetries.
Finally we use relation (32) with (33) and substitute this value into (34) to return
the bifurcation conditions for the hyperbola. These are,

>

1
2

(35)

>

1
2

(36)

and

Utilising all of the given information we can conclude that the Hopf bifurcation begins at (, ) = ( 21 , 12 ) and continues onwards for growing and . This is represented
in Figure 10 as the blue curve in the north-eastern quadrant.

By graphically representing both the Saddle-node and Hopf bifurcation together

we obtain Figure 11 and 12.

21

140281167

Figure 11: This graph represents both the Saddle-node and Hopf bifurcation. A red
line portrays the Saddle-node trajectory while the blue line is the Hopf trajectory

Figure 12: Zoomed portrait of Figure 11, showing how the Hopf bifurcation begins
at (, ) = ( 12 , 12 ) and crosses through the Saddle-node trajectory.
With the newly found information we can numerically plot the systems behaviour
and actually see the Hopf bifurcation take place. Additionally as [3] suggests, we can
also distinguish what type of Hopf bifurcation occurs.
By studying Figure 13 it is clear that the system has undergone a Supercritical
Hopf bifurcation. This is because the systems amplitude quickly grows into a stable
oscillation as passes beyond the Hopf bifurcation threshold, and does not jump to
any distance attractors.

22

140281167

Figure 13: These 3 plots portray how the Forced van der Pol systems behaves as
we cross the Hopf bifurcation line with = 0.6 and an alternating . On the left
= 0.65, in the middle = 0.68 and on the right = 0.69. Nevertheless despite
somewhat resembling a perfect circle, in reality the system converges to a deformed
circle.
Furthermore as the right hand set of plots in Figure 13 highlights, the system has
successfully formed a limit cycle. From our Special cases analysis we deduced that a
limit cycle must also exist on the boundary axis, therefore no further bifurcations
can occur in-between. This means that the only qualitative change we exhibit is

23

140281167

between a single fixed point and a single limit cycle as the system crosses the Hopf
boundary line right down to the axis boundary, respectively. This once again confirms
a Supercritical Hopf bifurcation.
We now explore the final type of bifurcation, a global bifurcation that can create or
destroy limit cycles throughout large regions of the phase space as opposed to a fixed
point segment.

7
7.1

Global Bifurcations
Infinite-period Bifurcation

The Infinite-period bifurcation is a method that quantifies the conditions needed to

destroy or create a limit cycles with the use of fixed points in a system. They can arise
when the speed of an oscillating system slows down as a parameter reaches a critical
threshold. The oscillation period lengthens and diverges to infinity as the parameter
reaches the critical value; a fixed point appears in result and we say that the system
has undergone an Infinite-period bifurcation. Beyond the critical parameter value, 2
fixed points appear and the system becomes phase locked [3]. This can be a global
phenomenon and is why its important to consider, hence we use [3]s example as an
illustration.
Given the system

r = r(1 r2 )
= sin
24

140281167

where > 0, its concluded that the system approaches the unit circle while rotating
counterclockwise if > 1. However as decreases through 1 the oscillations period
increases, eventually becoming infinite as = 1. Thus a single fixed point is created,
the limit cycle is destroyed, and the system has undergone a Infinite-period bifurcation. If < 1 the single fixed point will split into 2 fixed points, and an example of
the starting and ending states are illustrated in Figure 14.

Figure 14: On the left we have a stable limit cycle with an unstable fixed point,
> 1, and on the right we have a phase locked limit cycle with 3 fixed points, < 1.
Hence we now use the van der Pol system in polar coordinate form and more importantly, equation (15), as this equation describes the orbital velocity of the system.
When examining (15) we find that the system rotates counterclockwise globally when
> 0 or >

for all r (0, ). As decreases through =

cos
,
r

the rotation

changes direction and 2 fixed points are born, an unstable point and a stable point.
Therefore a bottle neck is created around the saddle point = 0, where the oscillations period increases severely as

+
r

(Figure 15). This Saddle point is also

where the Infinite-period bifurcation takes place, and the 3 possible solution states
for (15) are all portrayed in Figure 16.

25

140281167

Figure 15: This represents the diverging length of oscillations for the van der Pol
system as crosses the critical value, = r . In the final plot we can see that the
oscillations period has diverged to infinity.

Figure 16: These 3 plots represent the different solutions for (15). From left to right
we have no fixed point solutions, > r , then a Saddle-point when = r , marked
by a red star, and finally 2 fixed points when < r , the red dot corresponds to a
stable point while a circle corresponds to an unstable point.
From this we can conclude that the Infinite-point bifurcation occurs along one of the
two Saddle-node lines calculated in section 5, which was previously mentioned and
represented in section 5.2 Figure 8.
Now with all of the previously calculated bifurcations, a concluding topological argument can be deduced to examine how the system behaves in the different regions
of the bifurcation diagram.

26

140281167

Topological Analysis

In the section on Special cases we established that there exists a stable limit cycle
along the axis line = 0 for , as well as that the Hopf bifurcation line
above = 0 (please see Figure 11) represents the destruction or creation of a limit
cycle. Therefore we conclude that a limit cycle must exist in-between both boundaries
(Region B Figure 17) as the system cannot exhibit discontinuous jumps when moving
off the = 0 boundary into Region B Figure 17. Because of this we additionally
find that parameter space above the Hopf bifurcation line and upper Saddle-node
branch, represent a region of stable fixed points where the system has ceased to
oscillate (Region A Figure 17). As an additional piece of confirmation, we find that
Figure 13 agrees with the above analysis.
Furthermore, in the section on Global bifurcations it was established that an Infiniteperiod bifurcation occurs on a stable limit cycle at a Saddle-node point. Therefore
this bifurcation must emerge from the bottom of the two Saddle-node branches, as
we know that a limit cycle exists below this line (Black branch Figure 17). Therefore
as the system transitions from Region B up past the lower Saddle-node branch, an
Infinite-period bifurcation takes place and the system becomes phase locked. Now
consider the green axial line in Figure 17 that has 3 fixed points, and the Infiniteperiod bifurcation branch. The area in between (Region C Figure 17) must contain
3 fixed points, 2 of which are created from the Infinite-period bifurcation and 1 from
a negative radial length. For an illustration of where a negative radial length can
arise from, please look at the green solution line in Figure 4.

27

140281167

Figure 17: This is the final bifurcation diagram for the Forced van der Pol oscillator.
The different colours represent the different solutions that the system can hold and
that this paper has covered. While regions A, B and C represent areas of fixed points
and limit cycles respectively.

Conclusion

We conclude that the Forced van der Pol system is invariant under the transformations imposed by negative and values, which lead to simplifying symmetries for
the bifurcation diagram. The Special Cases analysis in section 4 established that
for = 0, there exists a stable limit cycle for all > 0, and that when = 0, the
system either has 3 fixed points ( <
( >

,
3 3

,
3 3

green axis line Figure 17) or 1 fixed point

purple axis line Figure 17). The Saddle-node bifurcation followed and

dissected the graph into 2 regions, expressing branches where fixed points emerge on
limit cycles to phase lock them. Next came the Hopf bifurcation, which begun at
(, ) = ( 12 , 12 ) and increased along the cut-off trajectory of a hyperbola to distinguish

28

140281167

regions where limit cycles loose stability and turn into fixed points, or vice versa.
Finally came the Infinite-period global bifurcation that was found to occur along the
lower of the two Saddle-node branches, a point where the oscillations period diverges
to infinity to create a saddle point. From this it was established that stable limit
cycles exist underneath the Infinite-period and Hopf bifurcation trajectories, and 3
fixed points reside inside the Infinite-period and Saddle-node cusp. Lastly, above
the Saddle-node and Hopf branch the system ceases to oscillate and converges to a
stable fixed point.

29

140281167

References
[1] Guckenheimer, J, Holmes, P. Nonlinear Oscillations, Dynamical Systems, and
Bifurcations of Vector Fields. Applied Mathematical Sciences 42. SpringerVerlag, New York, 1983.
[2] Tsatsos, M. Theoretical and Numerical Study of the Van der Pol equation. Aristotle University of Thessaloniki, Department of Physics, Mechanics. July 2006.
[3] Strogatz, S. Nonlinear Dynamics and Chaos with Applications to Physics, Biology, Chemistry and Eingineering. Studies in Nonlinearity, 1st edition. Westview
Press, United States of America, December 2000.

30

140281167

10
10.1

Appendix
Appendix A

We firstly begin with

2
0
u u r v

=
v v r2 + u

which can be rewritten as,

2
u
0

1 r

=
v

1 r2
This can now be inverted to produce,

1r2
u
2 2
2
(1r
) +
=

v
(1r2 )2 + 2

and thus,

u =

(1 r2 )2 + 2

v =

(1 r2 )
.
(1 r2 )2 + 2

31

0
(1r2 )2 + 2
2
1r

(1r2 )2 + 2

140281167

10.2

Appendix B

Beginning with (27)

h i
1 v2 3u2 2u v

= 0 ,
det
2u
1 u2 3v2
we evaluate the first step of the determinate to yield

(1 v2 3u2 )(1 u2 3v2 ) + ( + 2u v )( 2u v ) = 0,

and then simplify it to

1 4(u2 + v2 ) + 3(u4 + v4 ) + 6u2 v2 + 2 = 0.

Now by making use of the quadratic relation

(u2 + v2 )2 = v4 + u4 + 2u2 v2 ,
we can re-write the above equation as

1 4r2 + 3r4 + 2 = 0,
and then

2 = 4r2 3r4 1.
32