Linear Algebra and Differential Equations
Lecture 23
Nonlinear Systems of Differential Equations
July 28, 2016
Linear Algebra and Differential EquationsLecture 23 Nonlinear Systems
July 28, of
2016
Differential
1 / Equa
20
Nonlinear systems
Consider the nonlinear system
x0 = y
y 0 = 3x 2 1
To understand the dynamics of this system first we will identify the
equilibrium points
x0 = y0 = 0
1
( , 0)
3
1
( , 0)
3
To understand the flow we will also identify the nullclines, where the flow
is purely horizontal or purely vertical. The x nullcline is when x 0 = 0 and
the motion is purely vertical. In this example the x nullcline is the x axis
and the motion
is upwards for 3x 2 1 > 0. The y nullclines are the lines
x = 1/ 3. The motion is positive in the upper half plane and negative
in the lower half plane.
Linear Algebra and Differential EquationsLecture 23 Nonlinear Systems
July 28, of
2016
Differential
2 / Equa
20
This system can be integrated to obtain the trajectories
dy
3x 2 1
=
dx
y
1 2
y = x3 x + C
2
The phase portrait looks like
The equilibrium point at (0, 1/ 3) is a center and (0, 1/ 3) is a saddle.
Linear Algebra and Differential EquationsLecture 23 Nonlinear Systems
July 28, of
2016
Differential
3 / Equa
20
Linearized Stability Analysis
Given the system
x 0 = f (x, y )
y 0 = g (x, y )
Let (x ? , y ? ) be an equilibrium point. As we did in the one dimensional
case, we will expand the equations in a Taylor series about the equilibrium
point. Near the equilibrium point
f (x, y ) f (x ? , y ? ) + fx (x ? , y ? )(x x ? ) + fy (x ? , y ? )(y y ? )
g (x, y ) g (x ? , y ? ) + gx (x ? , y ? )(x x ? ) + gy (x ? , y ? )(y y ? )
The system of equations near (x ? , y ? ) has the form
0
x x?
fx (x ? , y ? ) fy (x ? , y ? )
x x?
=
y y?
y y?
gx (x ? , y ? ) gy (x ? , y ? )
The coefficient matrix is called the Jacobian and it indicates the type of
equilibrium point (x ? , y ? ) is.
Linear Algebra and Differential EquationsLecture 23 Nonlinear Systems
July 28, of
2016
Differential
4 / Equa
20
For the example the Jacobian is
J=
0
6x
1
0
So at (1/ 3, 0)
J=
0
1
2 3 0
p
with eigenvalues 2 3 and it is a saddle, as we deduced from the phase
portrait.
At (1/ 3, 0) the Jacobian is
0 1
J=
2 3 0
with eigenvalues i
p
2 3 and it is a center.
Linear Algebra and Differential EquationsLecture 23 Nonlinear Systems
July 28, of
2016
Differential
5 / Equa
20
The simple pendulum
The oscillations of a mass hanging from a string
are governed by the DE
g
sin = 0
L
We can re express this as a first order system
0
v
=
gL sin
v
00 +
Linear Algebra and Differential EquationsLecture 23 Nonlinear Systems
July 28, of
2016
Differential
6 / Equa
20
The equilibrium points are given by
v =0
sin = 0
= N
The Jacobian matrix is
J=
0
1
gL cos 0
So for even N we have a center, for odd N a saddle.
Linear Algebra and Differential EquationsLecture 23 Nonlinear Systems
July 28, of
2016
Differential
7 / Equa
20
Damped Pendulum
If we consider adding friction, the DE system is modified to be
0
v
=
gL sin cv
v
with c > 0. Calculating the Jacobian
0
1
J=
gL cos c
For n odd the determinant is negative andthe equilibrium is a saddle. For
n even the characteristic equation is
!
r
g
1
4g
2 + c + = 0
=
c c 2
L
2
L
So the equilibria are stable: nodes if c 2 >
4g
L
and spirals if c 2 <
4g
L
Linear Algebra and Differential EquationsLecture 23 Nonlinear Systems
July 28, of
2016
Differential
8 / Equa
20
Phase plane plot
A phase plane plot for c 2 <
4g
L .
Linear Algebra and Differential EquationsLecture 23 Nonlinear Systems
July 28, of
2016
Differential
9 / Equa
20
The SIR model for epidemics
The SIR model consists of the three DEs
x 0 = axy
y 0 = axy by
z 0 = by
x is the number of healthy people, y is the number of sick people, z is the
number of immune people, and a and b are positive constants.
Firstly note we only have to consider the first two equations since
(x + y + z)0 = 0
x +y +z =N
The equilibria are given by
xy = 0
axy by = 0
y =0
Linear Algebra and Differential EquationsLecture 23 Nonlinear July
Systems
28, 2016
of Differential
10 / Equa
20
The Jacobian matrix is
J=
ay
ay
ax
ax b
0
ax
0 ax b
On the y axis the Jacobian is
J(x, 0) =
with eigenvalues 0, ax b so the equilibria are stable if x <
unstable if x > ba
ba
a
and
We can solve for the trajectories using the DEs
dy
axy by
b
=
= 1 +
dx
axy
ax
y = x =
b
ln x + C
a
Linear Algebra and Differential EquationsLecture 23 Nonlinear July
Systems
28, 2016
of Differential
11 / Equa
20
A phase plot for the system is shown below
So the epidemic starts if the number of healthy people is greater than b/a,
the number of sick of people increases until the number of healthy people
reaches b/a then the number of sick people decreases.
Linear Algebra and Differential EquationsLecture 23 Nonlinear July
Systems
28, 2016
of Differential
12 / Equa
20
Another example
Analyze the system
x 0 = 2 cos x cos y
y 0 = cos x 2 cos y
The equilibria are given by
cos x ? = cos y ? = 0
The Jacobian is
J=
x? =
2 sin x ? sin y ?
sin x ? 2 sin y ?
For ( 2 , 2 )
J( , ) =
2 2
2 1
1 2
y? =
with eigenvalues = 1, 3 so it is an unstable node.
Linear Algebra and Differential EquationsLecture 23 Nonlinear July
Systems
28, 2016
of Differential
13 / Equa
20
For ( 2 , 2 )
J( , ) =
2 2
2 1
1 2
the determinant is negative it is a saddle.
For ( 2 , 2 )
J( , ) =
2 2
2 1
1 2
the determinant is negative it is also a saddle.
For ( 2 , 2 )
J( , ) =
2 2
2 1
1 2
with eigenvalues = 1, 3 so it is a stable node.
Linear Algebra and Differential EquationsLecture 23 Nonlinear July
Systems
28, 2016
of Differential
14 / Equa
20
Phase plot
A phase plane plot looks like
Linear Algebra and Differential EquationsLecture 23 Nonlinear July
Systems
28, 2016
of Differential
15 / Equa
20
Another example
The system of equations
x0 = 9 x2 y2
y0 = 3 x y2
has critical points given by
9 x2 y2 = 0
3 x y2 = 0
with solutions (3, 0), (2, 5), (2 5).
The Jacobian is
J=
2x
1
2y
2y
Linear Algebra and Differential EquationsLecture 23 Nonlinear July
Systems
28, 2016
of Differential
16 / Equa
20
At (3,0)
J=
6 0
1 0
with eigenvalues = 0, 6 so a stable node.
At (2, 5)
4 25
J=
1 2 5
which is a saddle since the determinant is negative.
At (2. 5)
4 2 5
J=
1 2 5
with characteristic equation
2
(4 + 2 5) + 10 5 = 0
=2+
q
5
98 5
thus it is an unstable spiral.
Linear Algebra and Differential EquationsLecture 23 Nonlinear July
Systems
28, 2016
of Differential
17 / Equa
20
Phase plane
Linear Algebra and Differential EquationsLecture 23 Nonlinear July
Systems
28, 2016
of Differential
18 / Equa
20
Yet another example: A Limit Cycle
Consider the system
x 0 = x y x(x 2 + y 2 )
y 0 = x + y y (x 2 + y 2 )
The origin is an equilibrium point. The Jacobian is
1 1
J=
1 1
which gives rise to an unstable spiral.
Now multiply the first equation by x and the second by y and add them.
xx 0 + yy 0 = x 2 + y 2 (x 2 + y 2 )2
Linear Algebra and Differential EquationsLecture 23 Nonlinear July
Systems
28, 2016
of Differential
19 / Equa
20
Now change to polar coordinates
rr 0 = r 2 r 4
r 0 = r (1 r 2 )
If the point is inside the circle r = 1, r 0 > 0 and the point moves outwards.
If the point is outside the circle r 0 < 0 and the point moves towards the
origin. So eventually all the trajectories tend to the circle r = 1.
The phase portrait is
Linear Algebra and Differential EquationsLecture 23 Nonlinear July
Systems
28, 2016
of Differential
20 / Equa
20
