Lecture6 7
Lecture6 7
+ + +
( (
= = +
( (
.
So we have
2 2 2 2 1 2
( ) cos( ) sin( ) ( ) ( )
kt kt
kA kA
k k
z t c e t i c e t x t i y t
e e
e | e |
+ +
( (
= + + + = +
( (
as the general
solution. This individually gives solutions
2 2 1
( ) cos( )
kt
kA
k
x t c e t
e
e |
+
= + to the first ODE and
2 2 2
( ) sin( )
kt
kA
k
y t c e t
e
e |
+
= + to the second ODE, and the solution to the first ODE is consistent with what
we derived previously. Here
2 2
k
k
g
e +
= is the gain and tan
k
e
| = determines the lag. This method involving
complex solutions is especially appropriate when considering gain and lag in the solution of higher order linear
ODEs when the inhomogeneity is of the form ( ) cos
at
q t ke t e = or ( ) sin
at
q t ke t e = .
Engineers often plot the gain and lag as functions of the input frequency e . The plots of
| |
log ( ) g e vs. log( ) e
and ( ) | e vs. log( ) e are known as Bode plots. They measure the response to a given signal.
3
Autonomous differential equations
Definition: An first order autonomous differential equation is an ODE of the form ( )
dx
dt
F x = , i.e. an ODE
where the rate
dx
dt
depends only on the value of x. If t represents time, this means that the rate of change is time-
independent.
If we draw the slope field corresponding to an autonomous equation, the slopes will be constant horizontally but
may vary vertically. Two familiar autonomous ODEs are:
(a) Natural (unrestricted) growth:
dx
dt
kx = (exponential growth for 0 k > , exponential decay for 0 k < )
(b) Logistic growth: (1 )
dx x
L dt
kx = (L is the carrying capacity, the relative growth rate
1
(1 )
dx x
x
L dt
k =
decays linearly with increasing population with rate 0 when x L = and negative growth for x L > )
Natural (exponential) growth:
dx
dt
kx = Logistic growth: (1 )
dx x
L dt
kx =
Even if we can solve an autonomous differential equation analytically to get a formula for the solutions, it is
often more important to understand the solutions qualitatively.
Definition: Given an autonomous differential equation ( )
dx
dt
F x = , we call a point
0
x an equilibrium if
0
( ) 0 F x = . The constant solution
0
( ) x t x = will be a solution to the differential equation with initial condition
0
(0) x x = .
As we can see in the illustrations above, some equilibria are such that nearby solutions converge toward the
equilibrium and other equilibria are such that nearby solutions diverge away from the equilibrium.
Definition: If
0
x is an equilibrium of ( )
dx
dt
F x = and if for all initial conditions in some interval around
0
x the
solutions ( ) x t are such that
| |
0
lim ( )
t
x t x
= , then we call
0
x a stable equilibrium. Otherwise we call it an
unstable equilibrium. However, we usually consider an unstable equilibrium to be such that nearby solutions
diverge away from the equilibrium. If we draw only the x-axis and indicate equilibria as points with arrows
indicating the direction of nearby solutions, we refer to this as the phase line.
Theres a simple derivative test for distinguishing stable and unstable equilibria. Suppose
0
x is an equilibrium
for the differential equation ( )
dx
dt
F x = and that ( ) F x is differentiable at
0
x . We learned in Calculus about
linear approximation, and in the vicinity of
0
x well have
0 0 0 0 0
( ) ( ) ( )( ) ( )( ) F x F x F x x x F x x x ' ' ~ + =
4
because
0
( ) 0 F x = . We also know that if we let
0
( ) u x x = , then
0
( )
du d dx
dt dt dt
x x = = , so well have
0 0 0
( )( ) ( )
du
dt
F x x x F x u ' ' ~ = . The differential equation
0
( )
du
dt
F x u ' = yields growth (away from 0 u = or
0
x x = ) if
0
( ) 0 F x ' > , and decay (toward 0 u = or
0
x x = ) if
0
( ) 0 F x ' < . This enables us to distinguish unstable
and stable equilibria. In the case where
0
( ) 0 F x ' = , well have to look at the slope field or use similar analysis.
Note: It may happen that on one side of an equilibrium nearby solutions converge toward the equilibrium but on
the other side they diverge away from the equilibrium. In this case we would call the equilibrium semistable.
Example: Determine the equilibria of the differential equation
2
( 2)
dx
dt
x x = and classify their stability.
Solution: The equilibria will be where
2
( ) ( 2) 0 F x x x = = , i.e. at 0 x = and at 2 x = . The derivative gives
( ) ( 2)(3 2) F x x x ' = . We have (0) 4 0 F' = > so this equilibrium will be unstable. On the other hand,
(2) 0 F' = so we must use other means to determine the stability of this equilibrium. Note that ( ) 0 F x ' > for
2 x > (repelling) and ( ) 0 F x ' < for 2 x < (attracting), so this equilibrium will be semistable.
Analytic solution of the logistic equation
The logistic equation is (1 )
dx x
L dt
kx = where 0 k > is constant. It has an unstable equilibrium at 0 x = and a
stable equilibrium at x L = (the carrying capacity). Suppose
0
(0) x x = is the initial condition. We can write
(1 )
L
x
dx
x
kdt
= and
(1 )
L
x
dx
x
kdt kt C
= = +
} }
. The integral on the left is done using partial fractions. Specifically,
1
(1 ) ( )
( )
L
x
L A B
x
L x x x L x
L A L x Bx
= = + = + . Choosing 0 x = gives AL L = or 1 A = . Choosing x L =
gives BL L = or 1 B = . So
( )
1 1 1 1 1
(1 ) (1 )
ln ln ln
L L
x x
dx x
x x
L x L x L x x x
dx x L x
= + = + = =
} }
.
So ln (1 ) ( )
1
kt
kt kt kt kt
kt
x x
L x L x
LAe
kt C Ae x LAe Axe x Ae x t
Ae
= + = = + = =
+
.
The initial condition gives
0
0 0 0 0 0
0
(0) ( )
1
x LA
x x LA x Ax A L x x A
A L x
= = = + = =
+
.
5
So
0
0 0 0
0 0 0 0 0
0
( )
( ) ( )
1
kt
kt
kt kt
kt
x
L e
L x Lx e Lx
x t
L x x e x L x e x
e
L x
| |
|
\ .
= = =
+ + | |
+
|
\ .
. So the solution is
0
0 0
( )
( )
kt
Lx
x t
x L x e
=
+
.
Note, in particular, that
| |
0
0 0
lim ( ) lim
( )
kt
t t
Lx
x t L
x L x e
(
= =
(
+
, as expected.
Example: Suppose population growth is governed by the logistic differential equation (1 )
dx x
L dt
kx = with 1 k =
and carrying capacity 1000 L = . Further suppose that the initial population is (0) 100 x = . The analytic solution
will then be
100000 1000
( )
100 900 1 9
t t
x t
e e
= =
+ +
. If we would like to know when the population will reach 500, we
have
1000
( ) 500 1000 500 4500 9 ln9 2.197
1 9
t t
t
x t e e t
e
= = = + = = ~
+
.
If we ask when the population will reach 990, we have
1000
( ) 990 1000 990 8910 891 ln891 6.792
1 9
t t
t
x t e e t
e
= = = + = = ~
+
.
We occupied the remainder of Lecture #7 with questions from the Practice Exam.
Notes by Robert Winters