Vous êtes sur la page 1sur 87

MA1506

Mathematics II

Chapter 7
Systems of First Order ODEs

1 Chew T S MA1506-14 Chapter 7
7.1 Solving Linear System of ODEs
a,b,c,d constants
How to solve
i.e,
We shall look at an old problem,
which is related to our new problem,
to get ideas to solve our new problem
2 Chew T S MA1506-14 Chapter 7
dx
ax by
dt
= +
dy
cx dy
dt
= +
x a b x
d
y c d y dt
( ( (
=
( ( (

An old problem
Consider ( )
2 ( )
dx t
x t
dt
=
Zero function ( ) 0 x t
is a solution
However we are interested in nonzero solutions
We know that
2
0
( )
t
x t x e =
is the general solution , where
0
x
is any constant
If we want nonzero solutions, then we assume
that
0
x
is nonzero constant
7.1 Solving System of ODEs
3 Chew T S MA1506-14 Chapter 7
So solutions of the old problem are of the
form
0
( )
t
x t x e

=
From this old problem, we may guess that
solutions of
are of the form
0
( )
t
x t x e

=
0
( )
t
y t y e

=
which can be written as
0
0
( )
( )
t
x x t
e
y y t

| | | |
=
| |
\ . \ .
Solutions of systems of ODE
7.1 Solving System of ODEs
4 Chew T S MA1506-14 Chapter 7
x a b x
d
y c d y dt
( ( (
=
( ( (

Now we shall find
0 0
, , x y
How to find? First substitute
0
0
( )
( )
t
x x t
e
y y t

| | | |
=
| |
\ . \ .
into the given ODE,
Note that
7.1 Solving System of ODEs
5 Chew T S MA1506-14 Chapter 7
0
0
( )
( )
t
x x t
d
e
y y t dt

( (
=
( (

x a b x
d
y c d y dt
( ( (
=
( ( (

0 0
0 0
t t
x x
e B e
y y

| | | |
=
| |
\ . \ .
where
a b
B
c d
| |
=
|
\ .
Let
0
0
0
x
u
y
| |
=
|
\ .
We get
0 0
u Bu =
number
2x2 matrix
We can not apply cancellation rule to the above equality
7.1 Solving System of ODEs
6 Chew T S MA1506-14 Chapter 7
0 0
u Bu =
Rewrite
as
2 0 0
I u Bu = where
2
1 0
0 1
I
| |
=
|
\ .
7.1 Solving System of ODEs
7 Chew T S MA1506-14 Chapter 7
Hence we have
2 0
( ) 0 B I u =
We want nonzero solns of the given ODE,
so we want
0
u
to be nonzero vector
2
det( ) 0 B I =
from Chapter 5, Section 5.5.
Therefore
Before we look at
2
det( ) 0 B I =
2 0
( ) 0 B I u =
We shall look at the following equality carefully
This equality can be written as
0 0
Bu u =
eigenvector eigenvalue
So we are looking for eigenvalues
and eigenvectors of B
7.1 Solving System of ODEs
8 Chew T S MA1506-14 Chapter 7
2
det( ) det
a b
B I
c d

| |
=
|

\ .
0 =
Hence
( )( ) 0 a d bc =
Thus
2
1
( ) 4( )
2
a d a d ad bc
(
= + +

| | | | | |
2
1
( ) 4(det )
2
Tr B Tr B B
(
=

Recall
a b
B
c d
| |
=
|
\ .
2
( ) ( ) 0 a d ad bc + + =
Eigenvalue
7.1 Solving System of ODEs
9 Chew T S MA1506-14 Chapter 7
There are three cases
Case 1: Two distinct real roots (eigenvalues)
i.e.,
Suppose two distinct real eigenvalues are
1 2
,
Then find the corresponding eigenvectors
1 2
, u u
by
1 2 1
( ) 0 B I u =
2 2 2
( ) 0 B I u =
1
1
t
u e

2
2
t
u e

and
We have two (linearly independent) solutions
| | ( ) | |
2
4 Tr B Det B >
7.1 Solving System of ODEs
10 Chew T S MA1506-14 Chapter 7
The general solution is
1 2
1 1 2 2
t t
c u e c u e

+
where
1 2
, c c
are any real numbers
..(C1)
7.1 Solving System of ODEs
11 Chew T S MA1506-14 Chapter 7
Case 2: Two complex roots (eigenvalues)
is given by
i.e.,
| | ( ) | |
2
4 Tr B Det B <
If one complex eigenvalue is
then the 2
nd
complex eigenvalue is

which is the conjugate of

i = +
i =
Consider complex eigenvalue

The corresponding complex eigenvector w=u+iv


2
( ) 0 B I w =
7.1 Solving System of ODEs
12 Chew T S MA1506-14 Chapter 7
Then
t
we

is a solution
However it is a complex-valued .
We want real-valued solutions
We can get two (linearly indep) real solutions from
this complex-valued solution. How?
t
we

( )
( ) ( )
i t t i t
u iv e u iv e e
+
= + = +
| |
( ) cos sin
t
u iv e t i t

= + +
| |
cos sin
t
e u t v t

=
| |
sin cos
t
ie u t v t

+ +
7.1 Solving System of ODEs
13 Chew T S MA1506-14 Chapter 7
Real part of the complex solution
| |
cos sin
t
e u t v t


and imaginary part of the complex solution
| |
sin cos
t
e u t v t

+
are two real-valued (linearly indep) solutions
The general solution is
| |
| |
1
2
cos sin
sin cos
t
t
c e u t v t
c e u t v t

+ +
1
c
2
c
are any real numbers
..(C2)
7.1 Solving System of ODEs
14 Chew T S MA1506-14 Chapter 7
Do we need to consider
Since it will induce the same general
solution

ANS: NO
Note:
2
( ) 0 B I w =
( )
2
0 B I w =
2
( ) 0 B I w =
2
( ) 0 B I w =
eigenvalue
eigenvector
7.1 Solving System of ODEs
15 Chew T S MA1506-14 Chapter 7
Case 3: Only one real root (eigenvalue)
i.e.,
| | ( ) | |
2
4 Tr B Det B =
7.1 Solving System of ODEs
16 Chew T S MA1506-14 Chapter 7
we skip this case in this module
Examples
x x
d
B
dt y y
| | | |
=
| |
\ . \ .
Solve
(1)
1 0
1 2
B

| |
=
|

\ .
Eigenvalues are
1
1 =
2
2 =
Corresponding eigenvectors are
1
1
1
u
| |
=
|
\ .
2
0
1
u
| |
=
|
\ .
7.1 Solving System of ODEs
17 Chew T S MA1506-14 Chapter 7
Cont.
The general solution is
2
1 1 2 2
t t
c u e c u e

+
i.e.,
1
( )
t
x t c e

=
2
1 2
( )
t t
y t c e c e

= +
2
1 2
1 0
1 1
t t
c e c e

| | | |
= +
| |
\ . \ .
7.1 Solving System of ODEs
18 Chew T S MA1506-14 Chapter 7
(2)
1 3
3 1
B

| |
=
|
\ .
Eigenvalues are
1 3i = + 1 3i =
The corresponding eigenvector is
Need only to consider 1 3i = +
1 1 0
0 1
w i
i
| | | | | |
= = +
| | |

\ . \ . \ .
7.1 Solving System of ODEs
19 Chew T S MA1506-14 Chapter 7
Cont.
The general solution is
1
1 0
cos3 sin3
0 1
t
c e t t
(
| | | |

(
| |

\ . \ .

2
1 0
sin3 cos3
0 1
t
c e t t
(
| | | |
+ +
(
| |

\ . \ .

Refer to Case 2 solution (C2)
| |
| |
1
2
cos sin
sin cos
t
t
c e u t v t
c e u t v t

+ +
1 3i = +
7.1 Solving System of ODEs
20 Chew T S MA1506-14 Chapter 7
Chew T S MA1506-14 Chapter 7 21
Solving system of ODE by Laplace transform
Let
( )
( )
( )
x t
v t
y t
(
=
(

We shall use the following notations


'( )
'( )
'( )
dx
x t
dv
dt
v t
y t dy dt
dt
(
(
(
= = =
(
(

(
(

dv
Bv
dt
=

Solve the following system of ODE


is 2 2 matrix B
Chew T S MA1506-14 Chapter 7 22
Apply Laplace transform to
dv
Bv
dt
=

get
| |
dv
L BL v
dt
(
=
(

Similar to
( ') ( ) (0) L f sL f f =
we have ( ) (0)
dv
L sL v v
dt
(
=
(


Hence
| |
( ) (0) sL v v BL v =

Chew T S MA1506-14 Chapter 7 23
So
| |
1
( ) (0) L v sI B v

=

Suppose
4 5
2 2
B

(
=
(


Then
| |
4 5
2 2
s
sI B
s

(
=
(
+

| |
1
1
4 5
2 2
s
sI B
s


(
=
(
+

2
2 5
1
2 4 2 2
s
s s s
+
(
=
(
+

0
(0)
2
v
(
=
(

Chew T S MA1506-14 Chapter 7 24


Hence
| |
1
2
( ) (0)
2 5 0
1
2 4 2 2 2
L v sI B v
s
s s s

=
+
( (
=
( (
+


2
10
1
2( 4) 2 2 s s s

(
=
(
+

2
10
1
2( 4) ( 1) 1 s s

(
=
(
+

Chew T S MA1506-14 Chapter 7 25
2
2
2
( ) 10
1
( )
( ) 2( 4) ( 1) 1
10
( 1) 1
2(s 4)
( 1) 1
L x
L v
L y s s
s
s

( (
= =
( (
+

(
(
+
(
=

(
(
+

Chew T S MA1506-14 Chapter 7 26


Hence
2
10
( )
( 1) 1
L x
s

=
+
2 2
2 2
2( 4) 2( 1 3)
(y)
( 1) 1 ( 1) 1
2( 1) 6
( 1) 1 ( 1) 1
s s
L
s s
s
s s

= =
+ +

=
+ +
Thus
( ) 10 sin
t
x t e t =
( ) 2 cos 6 sin
t t
y t e t e t =
Remark:
The above two-dimensional 1
st
order
homogeneous linear system of ODE with
constant coefficients always has TWO
linearly indep (nonparallel) solutions, say
u and v.
x x
d
B
dt y y
| | | |
=
| |
\ . \ .
0
x x
d
B
y y dt
| | | |
=
| |
\ . \ .
i.e.,
Recall 1-dim case
0
dx
px
dt
+ =
cu+dv is again a solution.
Any soln is of this form. Here c and d
are any real numbers
7.1 Solving System of ODEs
27 Chew T S MA1506-14 Chapter 7
Remark: 2
nd
order ODE is a special case of
systems of ODE
Chew T S MA1506-14 Chapter 7 28
Solutions of
is same as solutions of
0 1 x x
d
dt y b a y
( ( (
=
( ( (


7.1 Solving System of ODEs
See next slide for proof
2
2
0
d x dx
a bx
dt dt
+ + =
Chew T S MA1506-14 Chapter 7 29
dx
y
dt
=
dy
bx ay
dt
=
2
2
d x dy
dt dt
=
bx ay =
dx
bx a
dt
=
7.1 Solving System of ODEs
2
2
0
d x dx
a bx
dt dt
+ + =
From the above ODE in matrix form, we get
Hence from above
two equations, get
So
7.2 Stability of equilibrium solution
A soln of the above ODE is said to be an
equilibrium soln(equilibrium point) if it is a
constant function (see Chapter 2)
x x
d
B
dt y y
| | | |
=
| |
\ . \ .
Consider a system of ODE
For example, zero
0
0
x
y
| | | |

| |
\ . \ .
is always a solution, so zero is an equilibrium soln
30 Chew T S MA1506-14 Chapter 7
Are there other constant solutions?
For some cases, we may have other
constant solutions

For example, when
0 0
1 1
B
| |
=
|

\ .
Eigenvalues are
1
0 =
2
1 =
The general soln is
0
1 1 2 2 1 1 2 2
t t t
c u e c u e c u c u e + = +
(See Case 1 in 7.1)
7.2 Stability of equilibrium
31 Chew T S MA1506-14 Chapter 7
Hence
1 1
c u
is a soln, which is constant
1
c is any real number
So we have many constant solns
However , in this section, for convenience,
we assume that
0
0
x
y
| | | |

| |
\ . \ .
is the only constant soln
i.e., we only have ONE equilibrium solution
0
0
x
y
| | | |

| |
\ . \ .
Equivalently, we assume that
inverse of B exists
7.2 Stability of equilibrium
32 Chew T S MA1506-14 Chapter 7
An equilibrium soln (point) is said to be stable
if any soln with an initial point close to the
equilibrium soln stays close to the
equilibrium soln (point)
Now we shall discuss stability of equilibrium
soln of Case1 and Case 2 in 7.1.
We shall not discuss Case 3 and zero eiganvalue
You may refer to text book Chapter 6 for Case 3
(optional)
7.2 Stability of equilibrium
33 Chew T S MA1506-14 Chapter 7
Recall: eigenvalues are given by
Case 1 Two real eigenvalues
The general soln is
1 2
1 1 2 2
t t
c u e c u e

+
We can feel that zero equilibrium soln is stable
iff two eigenvalues are less than or equal to zero
7.2 Stability of equilibrium
34 Chew T S MA1506-14 Chapter 7
a b
Tr a d
c d
(
= +
(

| | | | | |
2
1
( ) 4(det )
2
Tr B Tr B B
(


2
( ) 4det TrB B >
Case 1
From
we know that two eigenvalues are less than or
equal to zero iff
2
(1) ( ) ( ( )) 4det( ) 0 Tr B Tr B B +
2
( ( )) 4det( ) ( ) Tr B B Tr B
2 2
( ( )) 4det( ) ( ( )) Tr B B Tr B
det( ) 0 B
7.2 Stability of equilibrium
35 Chew T S MA1506-14 Chapter 7
| | | | | |
2
1
( ) 4(det )
2
Tr B Tr B B
(
=

| | | | | |
2
(2) ( ) 4(det ) 0 Tr B Tr B B | | | | | |
2
( ) 4(det ) Tr B Tr B B
( ) 0 Tr B < det( ) 0 B < suppose
( ) 0 Tr B
Then
Contrad.
Hence for Case 1:
Zero equilibrium soln is stable iff
and
det( ) 0 B
Note: In fact det( ) 0 B >
since we assume that inverse of B exits, so
det( ) 0 B
( ) 0 Tr B <
7.2 Stability of equilibrium
36 Chew T S MA1506-14 Chapter 7
2
( ) 4det TrB B >
Case 2: Two complex eigenvalues
The general soln is
| |
| |
1
2
cos sin
sin cos
t
t
c e u t v t
c e u t v t

+ +
where
1
( )
2
Tr B =
(See Case 2 in 7.1)
7.2 Stability of equilibrium
37 Chew T S MA1506-14 Chapter 7
2
( ) 4det TrB B <
Hence zero equilibrium soln is stable iff
1
( )
2
Tr B =
Note that in this case
So det( ) 0 B >
i.e., 0
( ) 0 Tr B
7.2 Stability of equilibrium
38 Chew T S MA1506-14 Chapter 7
2
( ) 4det TrB B <
| |
| |
1
2
cos sin
sin cos
t
t
c e u t v t
c e u t v t

+ +
For Case1 and Case 2
stable
unstable
unstable
unstable
7.2 Stability of equilibrium
39 Chew T S MA1506-14 Chapter 7
Dash line
( ) 0 Tr B det( ) 0 B >
detB
TrB
For case1 ( ) 0 Tr B <
We will say more about unstable cases in next section
7.3 Phase plane and classification of zero solution
A solution of a 2-dim system of ODE is given by
( )
( )
x t
y t
| |
|
\ .
In this section, a phase plane is a x(t)-y(t) plane
The trajectories in the phase plane are
the parametric curves described by x(t) and y(t)
Trajectories on a phase plane
create a phase portrait
40 Chew T S MA1506-14 Chapter 7
Phase planes are useful in visualizing
the behavior (including stability of equilibrium soln)
of physical system.
Again we shall only consider
Case1 and Case 2 in 7.1
We shall not consider Case 3
You may refer to text book Chapter 6 for Case 3
(optional)
7.3 Phase plane
41 Chew T S MA1506-14 Chapter 7
Case 1: Two real eigenvalues
The general soln is
1 2
1 1 2 2
t t
c u e c u e

+
eigenvector eigenvector
First note that if
2
0 c =
we have
1 1
1 1 1
( )
( )
t t
x t g
c u e c e
y t h

| | | |
= =
| |
\ . \ .
7.3 Phase plane
42 Chew T S MA1506-14 Chapter 7
So
( ) ( )
h
y t x t
g
=
1 1
1 1
( ) y( )
t t
x t c ge t c he

= =
Hence the st. line through 0 with gradient h/g
is one of trajectories in the phase plane
So the gradient of this st. line equals to
the gradient of the eigenvector
We should have two basic trajectories
induced by two eigenvectors
7.3 Phase plane
43 Chew T S MA1506-14 Chapter 7
( ) ( )
h
y t x t
g
=
1
g
u
h
| |
=
|
\ .
Recall
Examples
(1)Two negative eigenvalues
2 1
1 2
B

| |
=
|

\ .
Eigenvalues are
1 2
1, 3 = =
Corresponding Eigenvectors are
1
1
1
u
| |
=
|
\ .
2
1
1
u
| |
=
|

\ .
7.3 Phase plane
44 Chew T S MA1506-14 Chapter 7
Two basic trajectories induced by two eigenvectors

Trajectories
tend to zero
since solns
tend to zero
as t tends to infinity
Trajectories tend
to zero, but never
reach zero
Corresponding to eigenvalue -3
7.3 Phase plane
45 Chew T S MA1506-14 Chapter 7
1 2
1, 3 = =
1
1
1
u
| |
=
|
\ .
2
1
1
u
| |
=
|

\ .
Corresponding to eigenvalue -1
Trajectories tend to zero, but never reach zero
Why?
By
no crossing principle
(See chapter 3)
Zero function is a soln, a trajectory is a soln,
no crossing principle can be applied here
Next we shall keep using no crossing principle
to sketch other trajectories

7.3 Phase plane
46 Chew T S MA1506-14 Chapter 7

All trajectories
tend to zero
since solns
tend to zero
as t tends to infinity
All trajectories tend
to zero, but never
reach zero
Gradient of every
trajectory near zero
is almost equal to 1
( )
lim 1
( )
t
y t
x t

=
No crossing for
trajectories
Equilibrium soln zero
is stable
For this case ,
zero is called
a nodal sink
7.3 Phase plane
47 Chew T S MA1506-14 Chapter 7
Gradient of
bigger eigenvector
Corresp to bigger eigenvalue -1
Proof next slide
3
1 2
3
1 2
( )
lim lim
( )
t t
t t
t t
y t c e c e
x t c e c e


=
+
2
1 2
2
1 2
lim 1
t
t
t
c c e
c c e

= =
+
7.3 Phase plane
48 Chew T S MA1506-14 Chapter 7
3
1 2
( ) 1 1
( ) 1 1
t t
x t
c e c e
y t

| | | | | |
= +
| | |

\ . \ . \ .
3
1 2
( ) 1
( ) 1
t t
x t g
c e c e
y t h

| | | | | |
= +
| | |

\ . \ . \ .
To have general idea, we write
3
1 2
3
1 2
( )
lim lim
( )
t t
t t
t t
y t c he c e
x t c ge c e


=
+
2
1 2
2
1 2
lim
t
t
t
c h c e h
c g c e g

= =
+
=gradient of eigenvector with bigger eigenvalue
( )
lim 1
( )
t
y t
x t

=
Proof of

Remark: Trajectories (x(t),y(t))
(not parallel to eigenvectors)
are tangent to the eigenvector
with bigger eigenvalue
7.3 Phase plane
49 Chew T S MA1506-14 Chapter 7
g
h
| |
|
\ .
h
y x
g
=
( )
lim
( )
t
y t h
x t g

=
Trajectory according
to eigenvector g
h
| |
|
\ .
Web Application
http://www.aw-bc.com/ide/idefiles/media/J avaTools/lnclmtrx.html
2 1
1 2
B

| |
=
|

\ .
Another Example:
7.3 Phase plane
50 Chew T S MA1506-14 Chapter 7
4 3
2 1
B

(
=
(


2
1 2
1 3
1 2
t t
x
c e c e
y

( ( (
= +
( ( (

Eigenvector
1
1
has eigenvalue -1 which is bigger
Trajectories (not
parallel to
eigenvectors)
are tangent to
eigenvector with
bigger eigenvalue
near pt zero
7.3 Phase plane
51 Chew T S MA1506-14 Chapter 7
y x =
(2/ 3) y x =
has eigenvalue -1 which is bigger
In general, suppose two eigenvalues are negative
Then the phase portrait is similar to the
previous example.
1 2
0 < <
All trajectories (not parallel to eigenvectors)
are tangent, near zero, to the eigenvector
corresponding to bigger eigenvalue
7.3 Phase plane
52 Chew T S MA1506-14 Chapter 7
2

Web Application
http://www.aw-bc.com/ide/idefiles/media/J avaTools/lnclmtrx.html
4 3
2 1
B

(
=
(


(2) Two positive eigenvalues
Eigenvalues are 2, 1
Eigenvectors are
Two eigenvectors induce two basic trajectories,
namely, x-axis and y-axis
General soln is
7.3 Phase plane
53 Chew T S MA1506-14 Chapter 7
2
dx
x
dt
dy
y
dt
=
=
2 0
0 1
B
(
=
(

1
1
0
u
(
=
(

2
0
1
u
(
=
(

2
1 2
( ) 1 0
( ) 0 1
t t
x t
c e c e
y t
| | | | | |
= +
| | |
\ . \ . \ .
Chew T S MA1506-14 Chapter 7 54
1
1
1
1
0
g
u
h
( (
= =
( (

2
2
2
0
1
g
u
h
( (
= =
( (

2
1 2
( ) 1 0
( ) 0 1
t t
x t
c e c e
y t
| | | | | |
= +
| | |
\ . \ . \ .
2
1 1 2 2
( )
t t
x t c g e c g e = +
Let
2
1 1 2 2
( )
t t
y t c h e c h e = +
1
1
( ) 0
lim 0
( ) 1 t
h y t
x t g
= = =
So tangents of all trajectories
(not parallel to eigenvectors),
are almost horizontal, when t tends to infinity .

= the gradient of eigenvector
with bigger eigenvalue
Zero is called
nodal Source

All trajectories never touch zero
Equilibrium
soln zero
is unstable
No crossing for trajectories
All trajectories tend to infinity
as t tends to infinity
7.3 Phase plane
55 Chew T S MA1506-14 Chapter 7
Tangents of all trajectories
(not parallel to eigenvectors),
are almost horizontal,
when t tends to infinity

with bigger eigenvalue
Web Application
http://www.aw-bc.com/ide/idefiles/media/J avaTools/lnclmtrx.html
7.3 Phase plane
56 Chew T S MA1506-14 Chapter 7
2 0
0 1
B
(
=
(

If
lim

()
()
=

1

1
0, ,
all trajectories (not parallel to eigenvectors),
when t tends to infinity ,are tangent to the
eigenvector
then
1
1
g
h
(
(

with bigger eigenvalue
(3) Two real eigenvalues with opposite signs
Suppose we have eigenvectors as follows
Corresponding to negative eigenvalue
Corresponding to positive eigenvalue
7.3 Phase plane
57 Chew T S MA1506-14 Chapter 7
Hence we have two basic trajectories
induced by eigenvectors as follows
7.3 Phase plane
58 Chew T S MA1506-14 Chapter 7
Chew T S MA1506-14 Chapter 7 59
1 2
1 2
1 2
( )
( )
t t
g g x t
c e c e
h h y t

| | | | | |
= +
| | |
\ . \ . \ .
, 0 >
1
1
( )
lim
( )
t
y t h
x t g

=
When t large, trajectories
tangent to the trajectory
induced by
bigger eigenvalue i.e., positive
eigenvalue
Zero is called
a saddle point
By no crossing principle, we can
sketch trajectories as follows

All trajectories never touch zero
No crossing for trajectories
Equilibrium soln zero
is not stable
Most of trajectories
away from equilibrium.
In fact, all except one
7.3 Phase plane
60 Chew T S MA1506-14 Chapter 7
When t large, trajectories
tangent to the trajectory
induced by
bigger eigenvalue
bigger eigenvalue
Web Application
http://www.aw-bc.com/ide/idefiles/media/J avaTools/lnclmtrx.html
7.3 Phase plane
61 Chew T S MA1506-14 Chapter 7
1 2
3 1
B

(
=
(


Summary of Case 1
Opp signs : Saddle
Both > 0 : Nodal source
Both < 0 : Nodal sink
Two Real Eigenvalues
Next we shall use
( ),det( ) Tr B B
to classify the above three cases,
which is easier
7.3 Phase plane
62 Chew T S MA1506-14 Chapter 7
Classification of zero solution
Eigenvalues are given by
Real roots
Nodal Source Both > 0
Nodal Sink Both < 0
Saddle Opp Signs
Recall: we do not discuss zero eigenvector
7.3 Phase plane
63 Chew T S MA1506-14 Chapter 7
| | | | | |
2
1
( ) 4(det )
2
Tr B Tr B B
(
=

2
( ) 4det TrB B >
0 TrB >
0 TrB <
det 0 B >
det 0 B >
det 0 B <
Then from above , we have (proof omitted)
Case 2: Complex Eigenvalues
Rotating
involving
angle
Stretch / Shrink
see following
slides
The general solution is
( )
( )
1
2
cos sin
sin cos
t
c u t v t
e
c u t v t



(
(
+ +
(

2
( ( )) 4det( ) Tr B B <
i =
t
7.3 Phase plane
64 Chew T S MA1506-14 Chapter 7
2
1
( ) 4det( ) ( ( ))
2
r Tr B i B Tr B
(
=

Spiral Sink: (Clockwise or anticlockwise)
Trajectories spiraling towards equilibrium soln zero
(1) ( ) 0 Tr B <
65 Chew T S MA1506-14 Chapter 7
( )
( )
1
2
cos sin
( )
( )
sin cos
t
c u t v t
x t
e
y t
c u t v t



(
| |
=
(
|
+ +
\ . (

1
( )
2
Tr B =
x
y
x
y
2) Spiral Source: (Clockwise or anticlockwise)
Trajectories spiraling away from equilibrium soln zero
(2) ( ) 0 Tr B >
66 Chew T S MA1506-14 Chapter 7
( )
( )
1
2
cos sin
( )
( )
sin cos
t
c u t v t
x t
e
y t
c u t v t



(
| |
=
(
|
+ +
\ . (

1
( )
2
Tr B =
x
x
y
y
Centre
Trajectories orbiting around equilibrium soln zero
(3)
( ) 0 Tr B =
67 Chew T S MA1506-14 Chapter 7
( )
( )
1
2
cos sin
( )
( )
sin cos
t
c u t v t
x t
e
y t
c u t v t



(
| |
=
(
|
+ +
\ . (

1
( )
2
Tr B =
x
y
( )
( )
1
2
cos sin
( )
( )
sin cos
c u t v t
x t
y t
c u t v t


(
| |
=
(
|
+ +
\ . (

( (0), (0)) x y

cos cos( 2 ) cos[ (t 2 / )] t t = + = +


sin sin[ ( 2 / )] t t = +
( (2 / ), (2 / )) x y
Clockwise or anticlockwise
discuss soon
Spiral
Source
Spiral
Sink
Centre
SUMMARY of Case 2
Complex eigenvalues
2
( ( )) 4det( ) Tr B B <
7.3 Phase plane
68 Chew T S MA1506-14 Chapter 7
0 TrB >
0 TrB <
0 TrB =
Classification of zero solution (cont.)
How to determine clockwise or anticlockwise
We shall use one example to illustrate it
Example:
2
( ( )) 4det( ) 4 4( 8 10) 4 Tr B B = + =
Complex Eigenvalues
( ) 2 Tr B =
7.3 Phase plane
69 Chew T S MA1506-14 Chapter 7
4 5
2 2
B

(
=
(


( ) 2 Tr B = Spiral Source
In the above, which one we shall choose
7.3 Phase plane
70 Chew T S MA1506-14 Chapter 7
How to determine clockwise or anticlockwise
It means, near the positive x-axis, the value of y
increases;
near the negative x-axis, the value of y decreases
Hence anticlockwise
7.3 Phase plane
71 Chew T S MA1506-14 Chapter 7
2 2
dy
x y
dt
=
0
2
y
dy
x
dt
=
=
Why choose y=0 and x-axis?
ANS: convenient
We may also look at
0
5
x
dx
y
dt
=
=
It means, near the positive y-axis, the value of x
decreases;
near the negative y-axis, the value of x increases
Hence anticlockwise
The graph just indicates
zero is a spiral source
anticlockwise, for more
precise graph, see next slide
7.3 Phase plane
72 Chew T S MA1506-14 Chapter 7
4 5
dx
x y
dt
=
7.3 Phase plane
73 Chew T S MA1506-14 Chapter 7
Web Application
http://www.aw-bc.com/ide/idefiles/media/J avaTools/lnclmtrx.html
4 5
2 2
B

(
=
(


Summary
Method 1: Find Eigenvalues
7.3 Phase plane
74 Chew T S MA1506-14 Chapter 7
Classification of zero solution
Method 2:
7.3 Phase plane
75 Chew T S MA1506-14 Chapter 7
| | | | | |
2
1
( ) 4(det )
2
r Tr B Tr B B
(
=

2nd method is easier
7.4 Romeo and J uliet
R(t), Romeos feelings J(t), J uliets feelings
a, b >0
The above system of equations says that
J uliets love decreases in proportion to Romeos
love for her
Romeos love grows in proportion to J uliets
love for him
How does the relationship evolve?
76 Chew T S MA1506-14 Chapter 7
, (0)
, (0)
dR
aJ R
dt
dJ
bR J
dt

= =
= =
The story of Romeo and J uliet in this module
is different from Shakespeares
0
0
a
B
b
| |
=
|

\ .
( ) 0 Tr B = det( ) B ab =
2
( ( )) 4det( ) 4 Tr B B ab =
0 <
Complex eigenvalues
7.4 Romeo and J uliet
77 Chew T S MA1506-14 Chapter 7
We shall not solve this ODE,
only look at the phase plane,
R
J
phase plane
CENTRE
Clockwise since
7.4 Romeo and J uliet
78 Chew T S MA1506-14 Chapter 7
0, when consider 0
dR
aJ J
dt
= > >

Therefore at red point , R is increasing
, (0)
, (0)
dR
aJ R
dt
dJ
bR J
dt

= =
= =
R
J
Phase plane
CENTRE
7.4 Romeo and J uliet
79 Chew T S MA1506-14 Chapter 7
R love increases,
J s love decreases R hate decreases,
J s love increases
Rs hate increases
J hate decreases,
J hate increases ,
Rs love decreases
Web Application
http://www.aw-bc.com/ide/idefiles/media/J avaTools/lnclmtrx.html
7.4 Romeo and J uliet
80 Chew T S MA1506-14 Chapter 7
0 1
2 0
B
(
=
(


G(0)=15000
M(0)= 11000
G(t)= #of Gondorians M(t)= #of Mordorians
7.5 Warfare
81 Chew T S MA1506-14 Chapter 7
dM
G
dt
=
0.75
dG
G M
dt
=
7.5. Warfare
The above equation says Mordorians die
(killed by Gondorians) at a rate
proportional to the number of Gondorians
7.5. Warfare
82 Chew T S MA1506-14 Chapter 7
dM
G
dt
=
every Gondorian kills one Mordorian per day.
1
dM
dt
G
=
More precisely, the above equation says
Suppose unit of t is one day.
Rewrite the above eq as
The above equation says one Mordorian
kills 0.75 Gondorians per day
The death rate per capita of Gondorians
due to disease is 1
We shall not solve this ODE,
only look at the phase plane
7.5. Warfare
83 Chew T S MA1506-14 Chapter 7
0.75
dG
G M
dt
=
Saddle
Eigenvectors
Eigenvalues 1/2 -3/2
induce two
basic trajectories
7.5. Warfare
84 Chew T S MA1506-14 Chapter 7
0.75
dG
G M
dt
=
dM
G
dt
=
1 0.75
1 0
B

(
=
(


1, det 0.75 TrB B = =
1
2
(
(


1
2
3
(
(
(

( )
2
( ) 4det( ) 0 Tr B B >
G
M
induces
induces
This trajectory tends to infinity
since eigenvalue is positive
This trajectory
tends to 0
since eigenvalue
is negative
M will win


G will win All will be killed

7.5. Warfare
85
Chew T S MA1506-14 Chapter 7
This trajectory is very crucial
1
2
(
(


1
2
3
(
(
(
2
3
M G =
2 M G =
1/2
-3/2
G
M
M will win


G will win All will be killed

7.5. Warfare
86 Chew T S MA1506-14 Chapter 7
2
3
M G =
15000
9999
(
(

15000
10001
(
(

Suppose G(0)=15000 is fixed
Then min M(0) that M will
win is 10001
G
M
M will win


G will win All will be killed

7.5. Warfare
87 Chew T S MA1506-14 Chapter 7
Suppose M(0)=10000 is fixed
Then min G(0) that G will win
is 15001
2
3
M G =
15001
10000
(
(

14999
10000
(
(

End
Chapter 7

Vous aimerez peut-être aussi