Lecture 9

Chaney I Linear systems

Chart 5
2x2systems
autonomous
general
Waste
Rio I
FIFTH
If words
f
f or in other
Itt where
where
4th
É fact yea H Xo glory

of gcacti yet
E is some open
is C E where
As before If Il R2 R2 solution exists
is unique
our
subset of R containing to then
t o
around the initial
in a time interval
now be using phase plane
Visualization will
my

starting at various initial positions

trajectories
XA ily t
never crossby uniqueness
Inerties Trajectories is tangent to
in the plane trajectory
At each point the main Ey
f as stated by
the vector field e
helpful to think y
NthyetDIFlity trajectory passing
x x thru x
field
 to a closed orbit in phase plane
Kablary A loop refers
fixed point of the
It
yP equilibria
is an
Denikin
system if flat
non linear 2 2 systems
to analyze
A powerful method some time
on

we start by spending
is linearization

Systems where
I is linear
I Linasystems
2 2 matrix
ATTA where Ais
RSR
If Flo É
A is also a linear map

PreliminairiestromlinearAlgebran

a
A a d be
d
det
at
A trc A
trace
Sit ARE
WCA PER A if
I 8 is an eigen pair for
Eigen pairs 7,7 A is the eigenvector
eigenvalue
A XI AI X is the
A 717 0 a quadratic ey
To find them
Solve det
sit PE NCA X1 prove
then find I
Practical formulas
I
Ai z
AdetaTffese
trCAI.tt in
Linear
t CA A th
7,72
dit A
y
 A
Examples Find eigenpairs of 4 1
above formula
Wejust apply the
In 1 A
tr A yielding
2 72 3
6
to get det A 5 8
For theeigenvectors F
we solve A 7,57 8 i.e
I 1

É.it 2 2 both
miiiiiih
example I
Agitent statins it B itieline the
Definition
R as
matrix exponential
et R

et It A
I Aft also E.tk
O 1 Note AZ AXA
where AO I I
A A XA XA
AK AxA x XA K times

2 2 matrix of the form

Eek let A be a
diagonal

what is eh
A
show by induction
that Ak q
Skin first we can
Ayu t
AY A's t t t
EA I A

L 9 o ai 1 2 1 t
Mig t

Is
Timpani rÉÉnÉttÉ 2 2 matrix If A possesses 2 real and
theorem Let A be a

ER then the corresponding eigenvectors

distinct eigenvalues 7,472
 there exists invertible matrix
I VI form a basis for R an

I 2 2 Sit P AP Ah where P Y É
i 12
1 diagaid
and
1
find transformation that diagonalizes
3 a a

a
f I
b What's eh
2
Shi a 7 1 72
P
5 1 15 1 yielding
f I
PIC D n
f
1
Check indeed that
A Prp
PhP
New ere e It Put Pip Pep t

PrP D
Pip
III I T K times

PAPI
PIP
PIE't PIE't PIE
P
P Ith I 13 t
Ah t

P eh p P
f
p
e

matrix
carry the
111 g milliniatg
one 2 2 matrix
 Lecture 10 I More Preliminaries
Chad The fundamentaltheorem for
Linear systems

I
a b
2 2 matrix A
Rhin 1 If A is a
of the form b a
cos b sinb
then et e
sin b costs

Joof Exercise foryou Hint First show that I a

V32

J
Red Im X Kati b
where
Im x relay
en ItA B fit
Now what is A what is
g ab
We show to induction
by b
I I
a
base step 12 2
b a
b Lab Rea
star Im X
Imig
Red
Lab
a b Rea Ima
Induction step Assume a Reck
b Imlay
show that
a b t
a
b a
b
by
a
f I
 recant Impey
Imca'D Refry
Now et It At
Agt Agt
Itoi o Rea Ema
t
O Hoi Ima Rea

REP E
D REY
t
1 Eg
E
II
Refittt it
Imfitatisity
Reflathit
i

Im It
Bit
ed Im e cab
Re easing
Im ed Rege Esinb ecosb
O
Rosita 2 If A
f then

eat eel P
Hint write A It B where B
Roof
Exercise a
18
then show that BK 00 K k 32
if 2 matrices M N commute
Here you can use the fact that
em en
then ent
k 2
BE
Very easy to show that J
o o
13 0
by showing o o
B at B
Now e ett eat EB since commute
w

E
7
It B
EIB But t

tgBft
EB It B

E e Itb 9

EL i
Noty
If't
eat It Or It
I e
I I a
92
 I

The fundamental theorem for

linear systems

2 2 matrix for I R the initial value

Let A be e

jajaifohasatun.ge
a

solution for an too

gimby take X
that is indeed a solution for l by using
Proof We first show
eat A ett Claim
Ig
eat
feat fig edge hide
et tf
I th
A At A
eat
es LA ht AE t

d
eat A A eat Remark
this is really
then if Clam is true we
simply fig hi
At
tht I
the
differentiate 4 with respect to t wecaninterchange
Since we can
show
2 bits
Aettxt A Xlt in other words Eg 4 series defining eth
If
or
that

g g
uum
g I I I

Take solution Ict to Eg H

guess any
Let Ict e
At
Ict TW
yrndto show that yet is actually

to thereby showing ICHis ofthe fringe

a constant
indepdt
oft
 At
Ae t E
II
t
dig use pinkfact

Ae
At
A e
At
Attn
8 since Et A commute

est independent of t What constant is that

y
A
Yet o e Io I To to El

Earnestine
Solve d AM A

Sketch the phase portrait

 Proposition 1 Proof

e
Rea Im X Rex
Ima
Im H
Reca t

Imlay Relay
Re at
it Imfitat it

Re 1 7 72 t
Imf 1 7 72 t

eafosb.ge
refe'D Impey esinby
regex sinbecosb sinbas
Inlet
Recall else'tib eh eib Recd e'Receib ecosb
Im e e'sinb BE
 Lectured Linear 2 2systems
Exercises Soling 2 2 systems
sketching phaseportraits
AF Sketch the phase portrait
Equal Solve
If lol

diagonal
values for is itself
I Matrix A has 2 real distinct eigen
A then our
General If A is already diagonal i.e
f
E eats
or
solution is simply
g i

xH E No
t 0

yet edity
as a parameter
parametric curves with time
This is a tasty of
the phase portrait
The corresponding sketch in the plane is
the conditions of
If A is not diagonal but satisfies I above
theoremICL A in Lecture 9 then
that
A PAP which means
by lecturesagain edit o
solution I A elPAP txt petty xp p
our o etat

phpatat
Itai T.soiefsiat.nl
at Exo
Rt yet X Yo
I D
E point
CH
L yet ety
10,0 is an equilibrium Xo o y
x.lt ogen y1Fyt 4 D
Y
fatmilies
on this axis of hyperbolas
gilt 0 so we stay

it Xoco he ox 4 of
 If we start on the axis 0 then
Eo So we stay on this axis
to if yosoggy 4 o
yet
Further if yo o then
fly
i
p
Putting everythingtogether
s
nY 0
I 1
then

Same
A 0
for arrows
in Q2 03
Iggy t o
E ox
Phaseportrait

Itoi gin is called a saddle

is stable subspace
which is a subspaceof R
a
The axis origin
IÉ crated to the
X
if we start anywhere
there are
since since if we
of RR
The axis is an unstable subspace
y repelled by origin
the
start anywhere there
47 we are

12 2 V3
of Xi 1 I
span I x axis span ti y axis
uncoupled
A
Easier o Xo 0 I Y 40
xitketx
A I
g
Solution so
Let
th est yo y.gg
Xo o ly axis XH 0 if yo
015,44
to
fp 1g
hi ya
 Xt to if XoColfaxHk a

Yo
0 x axis yet
0
if Xo o
Ey
The origin is a fixed point
n

theportrait
The originisrepelling

uncoupled
can
the phase
Exercise
for you A you guess
1 portrait

A
Exe coupled
1
vectors are
Eigenvalues

7 1 at
1 2
Ifi
Shin P
i PIL
 Ect P ent p
I
xp

i l L
t et
t Xo e tt yo e

yet get
Coupled system
Back to general
solution for an

man
multiply
T by iii
P on the left

E P A.IE
It P
Transformation
Let
JEDI a
PEPI
then
It siagmtsystem
ftp
 I
is
Solution of Diagonal system
t
É
e
o
simply JA o eat
where I P É
back to us from this
And to go

we simply apply F PJ a

linear transformation
 Source L Perko Chapters

shows the phase

Figured
portrait of y Solution to
the diagonal system where
as
it's easier to express y
rid of t
a
function of x
get
sketch the
portrait
shows the phase portrait
Fig 2

of 7 obtained from g by

the linear transformation

P
PJ
y
 HI Inediyfems Complex eigenvalues

real valued matrix

a 2 2
e A is

values
complexygen
has
treat
we can
D
Due to the formula
Eatin eigenvalues eigenvalues
were to occur they
that if complex i.e we would have
say as conjugates
would occur ib
a ti b I a
D D
2 I has eigenvalues 2 I
Empted A I
1 2

eigenvectors sit
D I tie
I real real
Hint Find AT ati b I complex comple
AT 75
5 ti
Ey
We find 11
It
it that matrix is already in special
E it Of form
2 real
valued
2
Let A be a

Theorem 2 L.A eigenvalues

complex
that has 2 distinct
matrix ib with associated
A at Iita
denoted by
s
 Tattie
O
denoted
eigenvectors by then
I T G R
E A it where
is invertible and
matrix P
he
I at
2 2

a b Not diagonal
P AP but special form
b a how
since we know
to take its exponential

Ég 2 Find eigenpairs use

this matrix wi
them
the
to put
som
E 1
7 2 i 5 2
We find
and associated eigenvectors
D R

roti III
5
11 1
yielding
131
 O
now we have P 21067
Yr Yog
2 I
P AP
s a

CL.AT
T
above
akhypot
matrix
on the
A the solution
to is

given by sinbt
cost É
A P
I g Sinbt cosbt
main theorem
10
1 in Lecture
PI Cf Proposition
AT
for solving É
Exercised Solve def AT
where A
sketch the
f
I is specified
if No
ft L phase portrait
already in special form
matrix is
of that our
b t.PT I P
where a 2 jeg
application
f
If I
E O
TA o e
a proofs
 e cost
Xlt Parameter
curve
yet
e Tsin t
This is a spiral
towards the origin

Sketchingthesolutuncurve
Yf
The origin
Nd
is called
astute
focus
SaresL.Perkolchapters
ht
Note that X'a yet e

tant
4
O
yet
EI
XA
Es
 Lectured Linear 2 2systems
I Complex eigenvalues
A
o Sketch the phase plane
AI with
Exampled Solve
4 I 2i
to find a Ii
We first compute the eigenvalues 9 0 D 2
D I it E
with corresponding eigenvectors
5 we have P
18 9 9
where I 19 1
so that P API
g sin I É
Solution Ect p
us at
cost J
P
g
sin at
sin at
Xlt Xo cos at Ry

A Xo Sui at yo cos at
y
Our trajectories are ellipses
Rct 4yet xEt4y
Stan dart form of an ellipse f F 1

a fixed point
x

III I
Xo 1 Yo
a
around it

II Repeated Eigenvalues 2 repeated twice

X of multiplicity
A has an eigenvalue
real valued 2 2 matrix Let
A be a
Theorem3 L.A Let Then
A with multiplicity 2
I be an eigenvalue of g
O
there exists a
matrix N nilpotent of ord i.e 102 0

Sit A St N N where S N commute

2
A lectures proposition
Iles 1

1 has multiplicity 2
We can easily that 7
check that N o
so that A 9 easily
o

I I with multiplicity
get A
I
2 A we can
I 1
ween a s

Sy f's
also N
1
LA The
theorem 3
theorem Under the hypotheses of
AI Flo Io is given by
solution of JI Combine general
theorem
A e't It Nt I PE
2 in
Lecture 9
Proposition
above
example
Ease Solve AF with gin
dig
solute Et ee 9 LEI I l
It t etc Et
tf It
e

Hit e the tats

END of cases
 0
Stability
that has k negative eigenvalues
let A be 2 2 matrix
a
t
oté's real A ok
2 k positive
We define the following subspaces of R 12 0,1 or 2

E span up
v5
he d
stable cities'pindding to the
subspace pls are the
above eigenvalues
span Fans I
Eunstable
The
subspace
Lecture in
Examples in

Generaclasfcatmtvesnad.su
epa
are possible for
d AX
portraits that
portraits of
describing the phase
We shall do that by first the form
B is a 2 2 matrix of
BE where
II
B
or B or
Bf
1 we shall
for each of these
eases
At the same time
the origin
comment on the stability of
 13 with scorn
LI 1 g
The phase portrait looks like
n
y

fit.tn saddest

7 F
E x axis

E axis
y
B
I
with her co a
B
case

with X O
like
would look
In this case the phase portraits
y

Its

unstable if the above signs

Istablenobeathengind were reversed

with a co
case B g ab
 r Y

I
Im.am eigenvalues
are purely imaginary
Y n
n y

bl o

so
EnTER theoigin

Remand In general the linear system

to have a saddle node
said
If AF is
the origin if the
matrix
focus Center
the matrices B of
A is similar to one
of
respectively i.e
forms
equivalent to
its phase portrait is linearly
drawn in I or
the phase portraits
O