Appendix A.

Phase Plane Analysis

We discuss here, only very briey, general autonomous second-order ordinary differen-
tial equations of the form
dx
dt
= f (x, y),
dy
dt
= g(x, y). (A.1)
We present the basic results which are required in the main text. There are many books
which discuss phase plane analysis in varying depth, such as Jordan and Smith (1999)
and Guckenheimer and Holmes (1983). A good, short and practical exposition of the
qualitative theory of ordinary differential equation systems, including phase plane tech-
niques, is given by Odell (1980). Phase curves or phase trajectories of (A.1) are solu-
tions of
dx
dy
=
f (x, y)
g(x, y)
. (A.2)
Through any point (x
0
, y
0
) there is a unique curve except at singular points (x
s
, y
s
)
where
f (x
s
, y
s
) = g(x
s
, y
s
) = 0.
Let x x x
s
, y y y
s
; then, (0, 0) is a singular point of the transformed equation.
Thus, without loss of generality we now consider (A.2) to have a singular point at the
origin; that is,
f (x, y) = g(x, y) = 0 x = 0, y = 0. (A.3)
If f and g are analytic near (0, 0) we can expand f and g in a Taylor series and,
retaining only the linear terms, we get
dx
dy
=
ax +by
cx +dy
, A =

a b
c d

f
x
f
y
g
x
g
y

(0,0)
(A.4)
which denes the matrix A and the constants a, b, c and d. The linear form is equivalent
to the system
502 Appendix A. Phase Plane Analysis
dx
dt
= ax +by,
dy
dt
= cx +dy. (A.5)
Solutions of (A.5) give the parametric forms of the phase curves; t is the parametric
parameter.
Let
1
and
2
be the eigenvalues of A dened in (A.4); that is,

a b
c d

= 0
1
,
2
=
1
2
(a +d [(a +d)
2
4 det A]
1/2
). (A.6)
Solutions of (A.5) are then

x
y

= c
1
v
1
exp[
1
t ] +c
2
v
2
exp[
2
t ], (A.7)
where c
1
and c
2
are arbitrary constants and v
1
, v
2
are the eigenvectors of A correspond-
ing to
1
and
2
respectively; they are given by
v
i
= (1 + p
2
i
)
1/2

1
p
i

, p
i
=

i
a
b
, b = 0, i = 1, 2. (A.8)
Elimination of t in (A.7) gives the phase curves in the (x, y) plane.
The form (A.7) is for distinct eigenvalues. If the eigenvalues are equal the solutions
are proportional to (c
1
+c
2
t ) exp[t ].
Catalogue of (Linear) Singularities in the Phase Plane
(i)
1
,
2
real and distinct:
(a)
1
and
2
have the same sign. Typical eigenvectors v
1
and v
2
are illustrated
in Figure A.1(a). Suppose
2
<
1
< 0. Then, from (A.7), for example, for
c
2
= 0, c
1
= 0,

x
y

= c
1
v
1
exp[
1
t ],
so the solution in the phase plane simply moves along v
1
towards the origin
as t in the direction shown in Figure A.1(a) along PO if c
1
> 0
and along QO if c
1
< 0.
From (A.7) every solution tends to (0, 0) as t since, with
2
<

1
< 0, exp[
2
t ] = o(exp[
1
t ]) as t and so

x
y

c
1
v
1
exp[
1
t ] as t .
Thus, close enough to the origin all solutions tend to zero along v
1
as shown
in Figure A.1(a). This is called a node (Type I) singularity. With
1

2
<
0 it is a stable node since all trajectories tend to (0, 0) as t . If
1
>

2
> 0 it is an unstable node; here (x, y) (0, 0) as t .
Appendix A. Phase Plane Analysis 503
Figure A.1. Typical examples of the basic linear singularities of the phase plane solutions of (A.4). (a) Node
(Type I): these can be stable (as shown) or unstable. (b) Saddle point: these are always unstable. (c) Spiral:
these can be stable or unstable. (d) Centre: this is neutrally stable. (e) Node (Type II): these can be stable or
unstable. (f) Star: these can be stable or unstable.
(b)
1
and
2
have different signs. Suppose, for example,
1
< 0 <
2
then
v
1
exp[
1
t ]v
1
0 along v
1
as t while v
2
exp[
2
t ] 0 along v
2
as
t .
There are thus different directions on v
1
and v
2
: the solutions near (0, 0)
are as shown in Figure A.1(b). This is a saddle point singularity. It is always
unstable: except strictly along v
1
any small perturbation from (0, 0) grows
exponentially.
504 Appendix A. Phase Plane Analysis
(ii)
1
,
2
complex:
1
,
2
= i, = 0. Solutions (A.7) here involve exp[t ]
exp[it ] which implies an oscillatory approach to (0, 0).
(a) = 0. Here we have a spiral, which is stable if < 0 and unstable if
> 0; Figure A.1(c) illustrates a spiral singularity.
(b) = 0. In this case the phase curves are ellipses. This singularity is called
a centre and is illustrated in Figure A.1(d). Centres are not stable in the
usual sense; a small perturbation from one phase curve does not die out
in the sense of returning to the original unperturbed curve. The perturbation
simply gives another solution. In the case of centre singularities, determined
by the linear approximation to f (x, y) and g(x, y), we must look at the
higher-order (than linear) terms to determine whether or not it is really a
spiral and hence whether it is stable or unstable.
Figure A.2. Summary diagram showing how tr A and det A, where A is the linearisation matrix given by
(A.4), determine the type of phase plane singularity for (A.1). Here det A = f
x
g
y
f
y
g
x
, tr A = f
x
+ g
y
,
where the partial derivatives are evaluated at the singularities, the solutions of f (x, y) = g(x, y) = 0.
Appendix A. Phase Plane Analysis 505
(iii)
1
=
2
= . Here the eigenvalues are not distinct.
(a) In general, solutions now involve terms like t exp[t ] and there is only one
eigenvector v along which the solutions tend to (0, 0). The t in t exp[t ]
modies the solution away from (0, 0). It is called a node (Type II) singu-
larity, an illustration of which is given in Figure A.1(e).
(b) If the solutions do not contain the t exp[t ] term we have a star singularity,
which may be stable or unstable, depending on the sign of . Trajectories in
the vicinity of a star singularity are shown in Figure A.1(f).
The singularity depends on a, b, c and d in the matrix A in (A.4). Figure A.2
summarises the results in terms of the trace and determinant of A.
If the system (A.1) possesses a conned set (that is, a domain on the boundary B
of which the vector (dx/dt, dy/dt ) points into the domain) enclosing a single singular
point which is an unstable spiral or node then any phase trajectory cannot tend to the
singularity with time, nor can it leave the conned set. The Poincar eBendixson theorem
says that as t the trajectory will tend to a limit cycle solution. This is the simplest
application of the theorem. If the sole singularity is a saddle point a limit cycle cannot
exist; see, for example, Jordan and Smith (1999) for a proof of the theorem, its general
application and some practical illustrations.
Appendix B. RouthHurwitz Conditions,
Jury Conditions, Descartes Rule of Signs
and Exact Solutions of a Cubic
Appendix B.1 Characteristic Polynomials, RouthHurwitz
Conditions and Jury Conditions
Linear stability of the systems of ordinary differential equations such as arise in inter-
acting population models and reaction kinetics systems (cf. Chapters 3 and 6) is de-
termined by the roots of a polynomial. The stability analysis we are concerned with
involves linear systems of the vector form
dx
dt
= Ax, (B.1)
where A is the matrix of the linearised nonlinear interaction/reaction terms: it is the
Jacobian matrix about the steady statethe community matrix in ecological terms. So-
lutions are obtained by setting
x = x
0
e
t
, (B.2)
in (B.1) where x
0
is a constant vector and the eigenvalues are the roots of the charac-
teristic polynomial
| A I | = 0, (B.3)
where I is the identity matrix. The solution x = 0 is stable if all the roots of the
characteristic polynomial lie in the left-hand complex plane; that is, Re < 0 for all
roots . If this holds then x 0 exponentially as t and hence x = 0 is stable to
small (linear) perturbations.
If the system is of nth order, the characteristic polynomial can be taken in the
general form
P() =
n
+a
1

n1
+ +a
n
= 0, (B.4)
where the coefcients a
i
, i = 0, 1, . . . , n are all real. We tacitly assume a
n
= 0 since
otherwise = 0 is a solution, and the polynomial is then of order n 1 with the
508 Appendix B. Conditions, Rule of Signs and Exact Solutions
equivalent a
n
= 0. We require conditions on the a
i
, i = 0, 1, . . . , n such that the zeros
of P() have Re < 0. The necessary and sufcient conditions for this to hold are the
RouthHurwitz conditions. There are various equivalent forms of these, one of which
is, together with a
n
> 0,
D
1
= a
1
> 0, D
2
=

a
1
a
3
1 a
2

> 0, D
3
=

a
1
a
3
a
5
1 a
2
a
4
0 a
1
a
3

> 0,
D
k
=

a
1
a
3

1 a
2
a
4

0 a
1
a
3

0 1 a
2


0 0 a
k

> 0, k = 1, 2, . . . , n.
(B.5)
These conditions are derived, using complex variable methods, in standard texts on the
theory of dynamical systems (see, for example, Willems 1970). As an example, for the
cubic equation

3
+a
1

2
+a
2
+a
3
= 0
the conditions for Re < 0 are
a
1
> 0, a
3
> 0; a
1
a
2
a
3
> 0.
Frankly it is hard to imagine anyone actually using the conditions for polynomials of
order ve or more.
Although (B.5) are the necessary and sufcient conditions we need, the usual al-
gebraic relations between the roots and the polynomial coefcients can often be very
useful. If
1
, . . . ,
n
are the distinct nonzero roots of (B.4) these are
n

i =1

i
= a
1
,
n

i, j
i =j

j
= a
2
, . . .
1

2
. . .
n
= (1)
n
a
n
. (B.6)
Lewis (1977) gives several useful ways of deriving qualitative results using the
RouthHurwitz conditions together with network concepts directly on the matrices.
In the case of systems of discrete models for interacting populations (cf. Chapter 3),
and with single population models with delay (cf. Chapter 2), stability is again deter-
mined by the roots of a characteristic polynomial (cf. Section 3.10 in Chapter 3). The
linearised systems again give rise to matrix forms like (B.3) and hence polynomials like
(B.4). With delay equations, with a delay of n time-steps say, we have to solve linear
difference equations typically of the form
u
t +1
= b
1
u
t
+ +b
n
u
t n
Appendix B.2 Descartes Rule of Signs 509
((2.37) in Chapter 2 is an example). We solve this by setting u
t

t
which again results
in a polynomial in . Linear stability here however is determined by the magnitude of
: stability requires | | < 1 since in this case u
t
0 as t . So, for the linear
stability analysis of discrete systems we require conditions on the coefcients of the
characteristic polynomial so that the solutions have magnitude less than 1. The Jury
conditions are the conditions for this to be the case.
The Jury conditions are given, for example, in the book by Lewis (1977), who
describes and illustrates several useful, analytical and numerical techniques concerning
the size and signs of the roots of polynomials. For the polynomial P(l) in (B.4), let
b
n
= 1 a
2
n
, b
n1
= a
1
a
n
a
n1
, . . . , b
nj
= a
j
a
n
a
nj
, . . . ,
b
1
= a
n1
a
n
a
1
;
c
n
= b
2
n
b
2
1
, c
n1
= b
n
b
n1
b
1
b
2
, . . . ,
c
nj
= b
n
b
nj
b
1
b
j +1
, . . . , c
2
= b
n
b
2
b
1
b
n1
;
d
n
= c
2
n
c
2
2
, . . . , d
nj
= c
n
c
nj
c
2
c
j +2
, . . . , d
3
= c
n
c
3
c
2
c
n1
;
and so on until we are left with only three elements of the type
s
n
= r
2
n
r
2
n3
, s
n1
= r
n
r
n1
r
n3
r
n2
, s
n2
= r
n
r
n2
r
n3
r
n1
.
The Jury conditions (necessary and sufcient) which ensure that the roots of the poly-
nomial P() in (B.4) all have magnitudes less than 1 are:
P(1) > 0, (1)
n
P(1) > 0,
| a
n
| > 1, | b
n
| > | b
1
|,
| c
n
| > | c
2
|, | d
n
| > | d
3
|, . . . , | s
n
|| s
n2
|.
(B.7)
Appendix B.2 Descartes Rule of Signs
Consider the polynomial (B.4), and, as before, we take without loss of generality a
n
> 0.
Let N be the number of sign changes in the sequence of coefcients {a
n
, a
n1
, . . . , a
0
},
ignoring any which are zero. Descartes Rule of Signs says that there are at most N roots
of (B.4), which are real and positive, and further, that there are N, N 2 or N 4, . . .
real positive roots. By setting = and again applying the rule, information is
obtained about the possible real negative roots. Together these often give invaluable in-
formation on the sign of all the roots, which from a stability point of view is usually all
we require.
As an example consider

3
+a
2

2
a
1
+a
0
= 0, a
i
> 0 for all i = 0, 1, 2. (B.8)
There are two sign changes in the sequence of coefcients, and so there are either two
or zero real positive roots. If we now set = , the equation becomes

3
a
2

2
a
1
a
0
= 0,
510 Appendix B. Conditions, Rule of Signs and Exact Solutions
which has one change of sign in the sequence, and so there is at most one real positive
root . This means there is exactly one negative root of (B.8).
Appendix B.3 Roots of a General Cubic Polynomial
Sometimes it is helpful to have the actual roots of the characteristic polynomial, however
complicated they may be. Although it is possible to nd these for polynomials higher
than order three, the complexity is usually not worth the effort. The roots of a cubic are
probably the most complicated that we would ever wish to have; simple derivations of
these have been given by Miura (1980) and Namias (1985). Table B.1 gives the explicit
forms of the roots of a cubic.
T
a
b
l
e
B
.
1
.
E
x
p
l
i
c
i
t
r
o
o
t
s
o
f
t
h
e
c
u
b
i
c
p
o
l
y
n
o
m
i
a
l
p
(

)
=

3
+
A

2
+
B

+
C
,
w
i
t
h
A
,
B
a
n
d
C
r
e
a
l
.

3
+
A

2
+
B

+
C
=
0
,
A

3
a
,
B

3
b
,

a
2

b
,

2
a
3

3
a
b
+
C

>
0

=
0

1
=

2
=
(
3

)
1
/
2

3
=

(
3

)
1
/
2

a
|

3
/
2

1
=
2

1
/
2
s
i
n

=
(
1
/
3
)
s
i
n

1
{

/
[
2

3
/
2
]
}

2
=

1
/
2
s
i
n
(

/
3
+

/
6

/
6

3
=
2

1
/
2
s
i
n
(

/
3

>
2

3
/
2

1
=

1
/
2
c
o
s
h

=
(
1
/
3
)
c
o
s
h

1
{
|

|
/
[
2

3
/
2
]
}

2
=

1
/
2
c
o
s
h

a
+
i
(
3

)
1
/
2
s
i
n
h

<

3
/
2

1
=

1
/
2
c
o
s
h

2
=

1
/
2
c
o
s
h

a
+
i
(
3

)
1
/
2
s
i
n
h

3
=

1
/
2
c
o
s
h

i
(
3

)
1
/
2
s
i
n
h

=
0

<

<

1
=

1
/
3

2
=

1
/
3
/
2

a
+
3
i

2
/
3
/
4

3
=

1
/
3
/
2

3
i

2
/
3
/
4

<
0

<

<

1
=

2
(

)
1
/
2
s
i
n
h

=
(
1
/
3
)
s
i
n
h

1
{

/
[
2
(

)
3
/
2
]
}

2
=
(

)
1
/
2
s
i
n
h

a
+
i
(

)
1
/
2
c
o
s
h

3
=
(

)
1
/
2
s
i
n
h

i
(

)
1
/
2
c
o
s
h