70
Spectrum of A σ(A)=the spectrum of A =the set of all eigenv alues of A

Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

Embed Size (px)

Citation preview

Page 1: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

Spectrum of A

σ(A)=the spectrum of A

=the set of all eigenvalues of A

Page 2: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

PropositionablediagonalizunitarilyisMA n

Aofrseigenvectoofconsisting

basislorthonormaanhasCn

),(,

)()(

,)()(

A

AINAIN

andAINCA

n

Page 3: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

Similarity matrix

If ,then we say that

A is transformed to B under

similarity via similarity matrix P

BAPP 1

Page 4: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

Exercise 1.2.4

If are similar over C,

then A and B are similar over R.

)(, RMBA n

組合矩陣理論 第一章 Exercise.doc

Page 5: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

Proof of Exercise 1.2.4

.

)()(

,sin)(

)()(

)2()1(

)(,sin,)2(

)1(

.

1

1

RoversimilarareBandATherefore

BSSASS

gularnonisandRMSSSince

BSSSSA

BSSBSAAS

RMBAceBSSA

SBAS

SBAS

BASS

tsMS

CoversimilarareBandA

n

n

n

Page 6: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

Schur’s unitary triangularilation Theorem

nMAanyFor

n

UA

0

*2

1

~

unitarily similar

can be in any prescribed

order

Page 7: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

Normal matrix

normalisMA n

AAAA **

e.g Hermitian matrix, real symmetric matrix, unitary matrix, real orthogonal matrix, skew-Hermitian matrix, skew-symmetric matrix.

強調與 complex symmetric matric 作區別

Page 8: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

Remark about normal matrix

Normal matices can not form a

subspace .

Page 9: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

Fact (*) for Normal matrix

normalisMA n

CnormalisIA ,

Proof in next page

Page 10: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

.

)()(

)()(

)()(

))((

))((

*

*

*

**

**

*

*

**

normalisIAHence

IAIA

IAIA

IAIIAA

IIAIAAA

IAAAA

IAIA

IAIA

AAAA

normalisA

Page 11: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

Spectrum Thm for normal matix.normalisMA n

tsUmatrixunitaryan .

Cwhere

UUA

n

n

,,1

*1

0

0

注意

Appling Schur’s unitary triangulariation Theorem to prove.

Page 12: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

Real Version of Spectrum Thm for normal matix

).()( AAAAnormalisRMA TTn

tsQmatrixorthogonalreala .

n

T

A

A

AQQ

0

01

formtheofmatrix

orisAwhere j

22

11

R

,It is normal.The proof is

in next page

Page 13: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

.

(*),

,0

0

0

0

:2

0

0

0

0

:1

2

22

22

22

22

normalis

Factbyandnormalisit

symmetricskewrealisSince

I

Method

Method

Page 14: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

Proposition for eigenvaluenMAanyFor

thentatataatpLet kk ,)( 2

210 k

kAaAaAaIaAp 2210)(

AofseigenvalueareIf n ,,, 21

)(

)(,),(),( 21

Apofseigenvalue

arepppthen n

Page 15: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

Proof of privious Proposition

n

U

A

0

*2

1

~

n

U

pAp

0

*2

1

~)(

)(

)(

)(

~)(

0

*2

1

n

U

p

p

p

Ap

Page 16: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

1.3

Jordan Form and Minimal Polynomial

Page 17: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

Elementary Jordan Block

kk

kJ

0

1

0

10

1

)(

0

0

main diagonal

elementary jordan block

super diagonal

sub diagonal

Page 18: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

kk

kk IJ

0

1

10

)(

0

0

kk

kNLet

0

1

10

0

0

It is Nilpotent matrix.(see ne

xt page)

Page 19: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

kk

kN

0

0

1

00

100

0

0

2

kk

kkN

0

0

0

000

0000

0

1

1

0 kkN

Page 20: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

Jordan Matrix

nnkn

n

n

kJ

J

J

J

)(

)(

)(

0

02

1

2

1

jordan matrix

distinctnecessarynotare

andnnnnwhere

k

k

,,, 21

21

Page 21: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

Jordan Canonical Form Theorem

nMAanyFor

unique up to the ordering of elementary Jordan blocks along the block diagonal.

A is similar to a jordan matrix

If A is real with only real eigenvalues, then the similarity matrix can be taken to be real

By Exercise

1.2.4

Page 22: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

Observation 1 for Jordan matrix

ablediagonalizisA

.matrixdiagonalaisJ A

the jordan matrix of

A

Page 23: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

Observation 2 for Jordan matrix

ofmultiplegeometric

toingcorrespondJinblocksof A#

the proof in next page

Page 24: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

)(

)(

)(

)(

.

11

1

kn

mn

n

n

A

A

A

k

m

m

J

J

J

J

J

formtheintakebecanJThen

toingcorrespond

JinblocksofnumberthebemLet

Page 25: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

kminJrank

andminJrankSince

J

J

J

J

IJ

iin

in

kn

mn

n

n

A

i

i

k

m

m

,,1)(

,,,11)0(

)(

)(

)0(

)0(

11

1

Page 26: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

.

)(

)(

)(

)(dim

)1()(

1

11

toingcorrespond

Jinblocksofnumberthe

mmnn

JIrankn

AIrankn

AIN

ofmultiplegeometric

mnmn

nnJIrank

A

A

k

ii

k

mii

m

iiA

Page 27: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

Observation 3 for Jordan matrix

ofmultipleebraica lg

toingcorrespondJin

blocksofsizestheofsumthe

A

Page 28: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

Observation 4 for Jordan matrix

ofmultiplegeometric

ofmultipleebraica

lg

blocksofareto

ingcorrespondJinblocksall A

11

Page 29: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

Observation 5 for Jordan matrix

Given counter example in next

page

The algebraic and geometric multiple of λ

can not determine completely the

Jordan structure corresponding to λ

Page 30: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

Assume that 1 is an eigenvalue of A and

geometric multiple of 1 is 3

algebraic multiple of 1 is 5

then 3 blocks in AJ corresponding to λ

the sum of sizes of these blocks is 5

Therefore (see next page)

Page 31: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

AJinareJJJ )1(),1(),1( 311

AJinareJJJ )1(),1(),1( 221

or

Page 32: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

Annihilating polynomial for A

In next page we show that A has an annihilating

polynomial.

Let p(t) be a polynomial.

If p(A)=0, then we say p(t) annihilates A and

p(t) is an annihilating polynomial for A

Page 33: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

.)(..

0)(

,)(

0

..,,,

,,,,

,dim

,,,,

2

2

2

2

2

2

2

10

10

10

2

2

2

Aforpolynomialngannihilatianistpei

Ap

thentataatpLet

AaAaIa

tszeroallnotCaaa

dependentlinearlyareAAAI

nMSince

MAAAI

nn

nn

n

n

n

nn

Page 34: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

Minimal polynomial of A

)(tmA

The minimal polynomial of A is

monic polynomial of least degree that

annihilates A and is denoted by

)()(0)( tptmAp A

the proof in next page

Page 35: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

)()(

0)(

0)(

)()()()(

)(deg)(deg0)(

)()()()(

lg

tptm

tr

Ar

ArAmAqAp

tmtrortrwhere

trtmtqtp

orithmadivisionBy

A

A

A

A

Page 36: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

Caley-Hamilton Theorem

0)( AcA

)()( tctm AA

This Theorem implies that

Page 37: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

Minimal Polynomial when A~B

)()(~ tmtmBA BA

the proof in next page

Page 38: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

)()(

)()(,

)()(

0)()(

)(.)(~)(

)(.

)()()(

.

~

11

1

tmtmTherefore

tmtmSimilarly

tmtm

BmAm

tppolyallforApBp

tppolyallfor

QApQAQQpBp

AQQB

tsQ

BA

BA

BA

AB

AA

Page 39: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

Mimimal poly. of Jordan matrix AofseigenvaluedistinctbeLet k ,,, 21

.

arg

iA

i

toingcorrespondJinblock

Jordanestlofsizethem

i

A

mi

k

iJ ttm )()(

1

Given example to

explain in next page

1)()(

..int

h

iAh

iA IJrankIJrank

tshegerenonnegativleastthe

Page 40: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

13

12

12

1

3

2

2

1

1

1

1

1

0

0

10

010

1

1

01

IJ

JLet

A

A

Page 41: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

313

312

212

312

31

213

212

122

12

21

)(

)(

)(3)(

0

0

00

000

)(

)(

)(

)(2)(

0

0

00

100

)(

IJ

IJ

A

A

Page 42: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

Similarly,

0

)(

*)(

)(

)(

*)(

**)(

)(

)(

0

00

)(

)(

*)(

**)(

)(

32

32

31

31

31

31

3

213

221

221

221

221

22

IJ

IJ

A

A

Page 43: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

)()()()(

,0)()()(

6)()(

5)()(

,6)(

3)()(

,4)(

,5)(

32

23

1

32

23

1

223

32

22

2

41

31

21

1

ttttm

andIJIJIJ

IJrankIJrank

IJrankIJrank

IJrank

IJrankIJrank

IJrank

IJrank

AJ

AAA

AA

AA

A

AA

A

A

Page 44: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

Mimimal poly. of Jordan matrixAofseigenvaluedistinctareLet n ,,, 21

.

arg

iA

i

toingcorrespondJinblock

Jordanestlofsizetheismwhere

i

A

mi

k

iJ ttm )()(

1

Proof in next page

1)()(

..int h

iAh

iA

i

IJrankIJrank

tshegerenonnegativleasttheism

Page 45: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

,,1)(0)(

)()()(

,,,,

)(,

,

)()(

.

arg

11

11

1

1

iJpJp

JpJpJp

where

JJandkthen

Jinblocksofnumberthebe

letandxtpLet

toingcorrespondJin

blockestltheofsizethebemLet

inA

ini

ini

A

k

ini

A

A

mi

k

i

iA

i

i

ii

i

i

Page 46: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

)(0

))(()0(

))(())((

))(())((

0)(,

0))((,

,1

,1

1

rj

mnijn

k

rii

mn

rjrj

mnijn

k

rii

mnrjn

mnijn

k

ijn

A

jn

mn

IJJ

rsomeformnandwhere

IJIJ

IJJp

JphaveweprovedisthisOnce

JpjfixedeachforthatshowTo

i

jj

r

j

i

jj

r

jj

i

jjj

j

Page 47: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

jsjs

jj

ii

ii

k

iiA

AA

mnandtssCertainly

mtsjSuppose

kimClaim

ThmHamiltonCayleythegotalreadywe

mwherettm

tpdividestmsoandJforpolynomial

ngannihilatianistpthatprovedhaveWe

i

.

.

,,1:

)(

0)()(

)()(

)(

1

Page 48: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

,0)(

0

))(()0(

))(())((

))((

))((

))((

))(())(()(

,1

,1

1

11

AA

ijn

k

jiim

nijn

k

jiimjjm

mijm

k

i

jmA

snA

inA

l

iin

l

iAAA

Jm

JJ

IJIJ

IJ

Jm

Jm

JmJmJm

i

j

j

j

i

jj

j

jj

i

jj

j

s

ii

Page 49: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

index of eigenvalue p.1

AofseigenvaluedistinctbeLet n ,,, 21

)()(,~ tmtmJASinceAJAA

k

i

miA

ittm1

)()( See next page

Page 50: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

index of eigenvalue p.2

.

arg

iA

i

toingcorrespondJinblock

Jordanestlofsizethem

1)()(

..int

h

ih

i IArankIArank

tshegerenonnegativleastthe

imi

k

iA ttm )()(

1

ii ofindexthecalledism

Page 51: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

index of eigenvalue p.3

nIArankIArank 10 )()(

0 ofindexthe

thenAIf ),(

)(AvbydenotedisofindexThe

Page 52: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

Observation 6 for Jordan matrix p.1

)(ALet

Akk JinsJfnumberthebebLet )(0

andnonegativearebbbThen 321 ,,

bycompletelyederare mindet

32 )(,)(),( IArankIArankIArank

….

Page 53: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

Observation 6 for Jordan matrix p.2

11 )()(2)( kkk

k

IArankIArankIArank

b

the proof in next page

Page 54: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

11

11

1

1

211

211

)()(2)(

.)(~)(,~

)()(2)(

)()(

)()()1()2(

)()()2(

)()()1(

kkkk

AA

kA

kA

kAk

kk

Ak

A

kA

kA

kkk

Ak

A

kkkk

Ak

A

IArankIArankIArankb

ppolyanyforJpApJASince

IJrankIJrankIJrankb

bIJrankIJrank

IJrankIJrank

bbIJrankIJrank

bbbIJrankIJrank

Page 55: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

Observation 7 for Jordan matrix p.2

kk IArankIArank )()( 1

the proof in next page

The number of blocks in

AJ

of size k ≧is

Page 56: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

toingcorrespond

ksizeofJinblocksofnumberthethen

IArankIArankIArank

andmmkb

mk

IArankIArankIArankb

ThenJinsJ

ofnumberthebebandtoingcorrespond

JinblockestltheofsizethebemLet

A

mmm

k

kkkk

Ak

k

A

21

11

)()()(

,,2,10

,,1

)()(2)(

.)(

arg

Page 57: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

kk

mm

kk

m

ki

ii

m

ki

ii

m

ki

ii

m

ki

ii

m

ki

iii

m

kii

IArankIArank

IArankIArank

IArankIArank

IArankIArank

IArankIArank

IArankIArank

IArankIArank

IArankIArankIArank

b

)()(

)()(

)()(

)()(

)()(

)()(

)()(

)()(2)(

1

1

1

1

1

1

1

1

1

11

Page 58: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

thenRMALet n ),(

)()( AA

ofmultipleebraicathe

ofmultipleebraicathe

lg

lg

ofmultiplegeometricthe

ofmultiplegeometricthe

See next page

組合矩陣理論 第一章 Exercise.doc

Page 59: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

Jordan structures for ,

11 )()(2)( kkk IArankIArankIArank

)(# kJ

The Jordan structure of A

corresponding to and that corresponding

to are the same. Because

)(# kJ11 )()(2)( kkk IArankIArankIArank

)(RM n

Page 60: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

0

0)(DLet

bia

11

iiS

),()( 1 baCab

baSSD

denote

The proof is in next page.

Page 61: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

ab

ba

aibi

biai

i

baibaibiabia

biabiabaibai

i

baibia

baibiaii

i

i

i

bia

biaii

iSSD

i

i

iiiiS

iiS

T

22

22

2

1

2

1

112

1

1

1

0

0

112

1)(

1

1

2

111

2

1

11

1

1

Page 62: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

000

100

000

001

)(0

0)(

2

2

J

J

)(0

)( 2~

D

IDP

Permutation similarity

The proof is in next page

Page 63: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

)(0

)(

000

000

100

010

~

000

000

100

001

~

00

100

000

001

)(0

0)(

2

2

2

D

ID

J

J

Page 64: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

..

1000

0010

0100

0001

tsPHence

)(0

)(

)(

)( 2

2

21

D

IDP

J

JP

Page 65: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

),(0

),(

0

0

)(0

)(

0

0 21

12

baC

IbaC

S

S

D

ID

S

S

),(0

),(~

)(0

0)( 2

2

2

baC

IbaC

J

J

similarly

),(

),(

0),(

~)(0

0)(

02

2

3

3

baC

IbaC

IbaC

J

J

Proved in next page

Page 66: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

),(0

),(

)(0

)(

)(0

)(

0

0

0

0

)(0

)(

0

0

2

12

1

1

11

1

12

baC

IbaC

SSD

ISSD

SD

SSD

S

S

S

S

D

ID

S

S

Page 67: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

similarly

kkk

k

J

J

22)(0

0)(

kkbaC

I

baC

IbaC

22

2

2

),(

),(

),(

~

0

0

Page 68: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

Theorem 1.3.4thenRMALet n ),(

)(

)(

),(

),(

~1

11

1

1

rm

m

ppn

n

r

p

J

J

baC

baC

A

Page 69: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

pkforAofeigenvaluerealnon

isibawhere kkk

,,1

Aofseigenvaluerealarer ,,1

)()(~),( knknkkn kkkJJbaC

kk nnkk

kk

kk

baC

I

baC

IbaC

22

2

2

),(

),(

),(

Page 70: Spectrum of A σ(A)=the spectrum of A =the set of all eigenvalues of A

)()( RMJRMA nAn

A

n

JAPP

tsMP

1

..

By Exercise 1.2.4 組合矩陣理論 第一章 Exercise.doc

A

n

JAQQ

tsRMQ

1

..)(