Linear Algebra and Differential Equations
Lecture 5
Qualitative methods for Autonomous Equations
June 20, 2015
Linear Algebra and Differential EquationsLecture 5Qualitative methods
June 20, for
2015
Autonomous
1 / 17Eq
An example of an autonomous equation
Consider the IVP
x 0 = sin x
x(t0 ) = x0
we can separate variables
Z
csc x dx =
x0
which integrates to
dt
t0
csc x + cot x
= e (tt0 )
csc x0 + cot x0
The trouble with this solution is that it is hard to interpret its qualitative
behavior. For example what is the
lim x(t)
Linear Algebra and Differential EquationsLecture 5Qualitative methods
June 20, for
2015
Autonomous
2 / 17Eq
A Geometric way of thinking
An equilibrium or fixed point of an autonomous DE is a constant
solution of the equation
x 0 = f (x)
Suppose x = x ? is a equilibrium solution then
f (x ? ) = 0
If we imagine x describes the motion of a particle on the real line, the DE
dictates the direction the particle moves. If f (x) > 0 the particle moves to
the right and if f (x) < 0 it moves to the left.
An equilibrium point is stable (also called an attractor or sink) if the flow
of the particle is towards it. An equilibrium is unstable (also called a
repellor or source) if the flow is away from the point.
Linear Algebra and Differential EquationsLecture 5Qualitative methods
June 20, for
2015
Autonomous
3 / 17Eq
Our example revisited
The DE
x 0 = sin x
Graphing x 0 vs x
The equilibrium points are at x ? = n. Examining the graph shows that
the stable equilibria are at the points x ? = (2n + 1) (marked by the filled
dots) and the unstable equilibria are at x ? = 2n (marked by the open
dots).
Linear Algebra and Differential EquationsLecture 5Qualitative methods
June 20, for
2015
Autonomous
4 / 17Eq
For example if x0 = 4 the particle initially accelerates to the right until
x = 2 then the particle moves to the right but at a steadily decreasing
rate, tending eventually to an asympotote at x = .
For general initial conditions the solutions to the DE have the qualitative
behavior shown below
Linear Algebra and Differential EquationsLecture 5Qualitative methods
June 20, for
2015
Autonomous
5 / 17Eq
So given the DE
x 0 = f (x)
Graphing x 0 vs x allows the equilibria to be found as the zeros of f (x).
The only possibilities for equilibria are illustrated below
The possibility not exhibited by our example happens when f (x) is tangent
to the x axis at x = x ? . The equilibrium is called semistable because for
values of x < x ? then x x ? as t , but if x > x ? then x as
t .
Linear Algebra and Differential EquationsLecture 5Qualitative methods
June 20, for
2015
Autonomous
6 / 17Eq
Linear Stability Analysis
To determine the stability of an equilibrium point, we will examine x
values near the equilibrium point, x ? . Let = x x ? and expand f (x) in a
Taylor series about x ?
f (x) = f (x ? ) + f 0 (x ? ) + O( 2 ) = f 0 (x ? ) + O( 2 )
Substituting into the DE
0 = f 0 (x ? ) + O( 2 )
If is small enough so that terms of 2 or higher can be neglected, the DE
becomes
0 = f 0 (x ? )
with solution
= ef
0 (x ? )t
Linear Algebra and Differential EquationsLecture 5Qualitative methods
June 20, for
2015
Autonomous
7 / 17Eq
If f 0 (x ? ) > 0 then increases exponentially, in other words x moves away
from x ? i.e. the equilibrium is unstable.
If on the other hand f 0 (x ? ) < 0 then decreases exponentially and x
moves towards x ? . The equilibrium is stable.
Note If f 0 (x ? ) = 0 this analysis doesnt tell us anything about the stability
of the equilibrium.
Example For our original example
f 0 (x) = cos x
then since x ? = n
0
f (x ) = cos(n) =
+1 n is even equilibrium is unstable
1 n is odd equilibrium is stable
which agrees with our earlier analysis.
Linear Algebra and Differential EquationsLecture 5Qualitative methods
June 20, for
2015
Autonomous
8 / 17Eq
Equilibria in physics
If the force depends only on position, Newtons second law has the form
mx 00 = F (x)
x ? is an equilibrium point if
F (x ? ) = 0
Using the same kind of reasoning as for the first order equations, we can
classify the equilibrium as stable if for small disturbances the force tends to
push the particle towards the equilibrium point and unstable if it pushes it
away.
Linear Algebra and Differential EquationsLecture 5Qualitative methods
June 20, for
2015
Autonomous
9 / 17Eq
Potential Energy
The potential energy is defined by
Z
U(x) = F (x) dx
F (x) = U 0 (x)
So in terms of the potential energy the equilibrium points are critical
points for U(x)
and negative slope for the graph of U(x) corresponds to a positive force
and positive slope corresponds to negative force. This means that stable
equilibria correspond to minima for U(x) and unstable correspond to
maxima.
Linear Algebra and Differential EquationsLecture 5Qualitative June
methods
20, 2015
for Autonomous
10 / 17Eq
Linear stability analysis
Expand U(x) in a Taylor series about an equilibrium point x = x ?
1
U(x) = U(x ? ) + U 00 (x ? )(x x ? )2 + higher order terms
2
There is no linear term since it is a critical point. Letting = x x star
and ignoring terms higher than 2 we get
1
U(x) = U(x ? + U 00 (x ? ) 2
2
Plugging into the DE for points near equilibrium
m 00 = U 00 (x ? )
This has the same form as a particle on a spring, So as long as U 00 > 0,
i.e. U has a minimum, the particle oscillates about the equilibrium value.
Linear Algebra and Differential EquationsLecture 5Qualitative June
methods
20, 2015
for Autonomous
11 / 17Eq
Logistic Growth
To model the growth of a population, Verhulst suggested the DE
p
p 0 = rp 1
K
The equilibrium points are given by
p
rp 1
=0
K
p ? = 0, K
To check their stability
f 0 (p) = r
f 0 (0) = r
2rp
K
f 0 (K ) = r
So p = 0 is unstable and p = K is stable. So whatever the initial
condition, the population tends to K which is called the carrying capacity.
Linear Algebra and Differential EquationsLecture 5Qualitative June
methods
20, 2015
for Autonomous
12 / 17Eq
Logistic Growth with Harvesting
The logistic equation can be modified by adding a term representing
harvesting
p
h
p 0 = rp 1
K
To simplify the subsequent algebra introduce the dimensionless quantities
P=
p
K
T = rt
H=
h
rK
The DE becomes
dP
= P(1 P) H
dT
The equilibrium points are given by
P(1 P) H = 0
1
P ? = (1 1 4H)
2
Linear Algebra and Differential EquationsLecture 5Qualitative June
methods
20, 2015
for Autonomous
13 / 17Eq
To check their stability
f 0 (P) = 1 2P
?
f 0 (P
) = f 0 (1 1 4H) = 1 4H
?
f 0 (P+
) = f 0 (1 + 1 4H) = 1 4H
? is stable and P ? is unstable. The plots of x vs t will look like
So P+
As H is increased the equilibrium point approach one another until at
H = 14 they coalesce into one semistable equilibrium.
Linear Algebra and Differential EquationsLecture 5Qualitative June
methods
20, 2015
for Autonomous
14 / 17Eq
Writing the DE in the form
dP
1
1 2
H
= P
dT
2
4
It is clear that when H > 14 , P decreases to zero. This behavior can be
summarized in a graph of P ? vs H called a bifurcation diagram.
Bifurcation refers to the fact that the system evolves completely differently
depending on whether H is greater or less than 14
The stable equilibrium points are represented by the solid curve, the
unstable by the dashed.
Linear Algebra and Differential EquationsLecture 5Qualitative June
methods
20, 2015
for Autonomous
15 / 17Eq
Another bifurcation example
If we consider the DE
x 0 = kx x 3
The equilibrium points are given by
0 = kx x 3
0 = x(k x 2 )
If k < 0 there is only one equilibrium point x ? = 0
If k > 0 there are three equilibrium points x ? = 0, k, k
To check stability
f 0 (x) = k 3x 2
Linear Algebra and Differential EquationsLecture 5Qualitative June
methods
20, 2015
for Autonomous
16 / 17Eq
For k < 0
f 0 (x ? ) = f 0 (0) = k
So x ? = 0 is stable
For k > 0
f 0 (0) = k
f 0 ( k) = 2k
So x ? = 0 is unstable and x ? = k are stable. The bifurcation diagram
is shown below. It is called the pitchfork bifurcation because of its
appearance.
Linear Algebra and Differential EquationsLecture 5Qualitative June
methods
20, 2015
for Autonomous
17 / 17Eq
