0% found this document useful (0 votes)
15 views14 pages

PS4 Handwritten 3

The document discusses definiteness of matrices and computing matrix minors. It provides steps to determine if a matrix is positive definite, negative definite, or indefinite by examining its minors of different orders. It also discusses using eigendecomposition to represent a matrix as a diagonal matrix of eigenvalues multiplied by its eigenvectors.

Uploaded by

kfcsh5cbrc
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
15 views14 pages

PS4 Handwritten 3

The document discusses definiteness of matrices and computing matrix minors. It provides steps to determine if a matrix is positive definite, negative definite, or indefinite by examining its minors of different orders. It also discusses using eigendecomposition to represent a matrix as a diagonal matrix of eigenvalues multiplied by its eigenvectors.

Uploaded by

kfcsh5cbrc
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 14

PS4 studenthours 1615 17 30

Etraref130
Reminder on definitness

verify Dr De D etc
f
minors
all Dido
if all
PositiveJefinite
iii Dit Negative Definite
if
else we compute arbitrary minors
order 1 diagonal term
if all D'Kyo possiblepositive definite Ps
if all D'a 4 3,0 possible NSD
order 2 cofactorof all diagonalterms
all s'y o possiblePSD
all il D'Kyo Possible NSD

order 3 Matrixdeterminant for 3 3 matrix


D possible PSD
o
IP Bs o possible NSD

all orders should converge to


result not matrix
the same
if
is indefinite
MÜYÜEÏAAÜ
In matrix notation Q X AX

A
fi
Compute the minors Dr 370

Dr 270

B sco
II
D 0 D2 0 D LO S Not Positive nor negative
definite

arbitrary minors

Die3 DÛ 1 DE6 as allD'e 1 Lo cannotbe NSD

D Dz S 20 cannot be PSD

indefinite
hI
1 IA KIL
2 t
2 1
i
K 2
a
ÉÉÇ
I I 3 L
t z 4 il
y
27k 2 t
1 3 1 3 t 1 4 z

4741 3 44 422 21 374 4 2 4 44


242_L 84 412 2 4 64 272
43 242 4 2
L1 4 2 4 1
Et 2 K l

eigenvectors A KI XEO
A Là X O

fait
p f
ex stat o es take za

ÉTÉ ET eigenvectors

wifi Il
EE takgaïctor

ti il
y set ça
2 0 eigenvector

2 B AA

remember A K I X O AX KI XO
A L

Xi Axe Été a
B Xz LAA Xz ALAE AL E A E FAFE

Xz te 1 AXz XB AM Xz
Y
B Xz AAIX A AB AXz Xz
X
f te 2 Ax 24

Bx LAA ALA4 A 2KFZANE44


Bx H
3 Cis st C Ctc

if L is aneigenvalue ther CX

CE CCCX C Xx exacte
K X
C C Ex C CAX 23 X
tex catageaual
III
Exe C C X
K A

if e Int exists det Into


det e l D In o E I cannot betrue

we haveshown that EXEL musthold

IP 1 1 K I cannot be an eigenvalue

Ctrn is invertible
1 Q EAX with x and A
H
2 det A KI n o

Ge 2 C EG a OE 3
SA GEO X
a ESAG O

Ko 2,3 2positive and 1 equaltoo

Thus Qis PSD

fortio
É À
Été
f ÏÏÉ
E set 1
f
e
a

calculate the nom EI E


so Î HIE

fill
E
set c 1
f taff
E fit
e set en
T E
get
IIIÈ
we

A is nonsingular and symétrie


4
if we can use
spectraldecomposit
A êre
c
with E CI à F and A diag ti ti 4
Ô isbydefinitionorthogonalthen Ê Et

Anémié
t
Amené
with N
Gin
ÇI
However
initially we said thatAshouldnotbe singular
Here l o IAKo we cannot apply

5 Using SD A ÊME EEE


Thus A Ê n'ê
as ê is orthonormal Ô é

However we cannot inverse 1

Goo
i
1 det A XI 2 t 1 z
1 2 K z
z 2 S K
2 x 2 K z
t z 2 1 z
2 s g 2 S x 2 X z

2 x 2a s s 4 S K 4 2 2t4 21

2 a 6 7A x2 t 4 44 4

12 42 212 6 4 712 63 51 5
Et 9 2_IS L

C 7 Et 2 L H
K I'll 7
2 AX X ES LA I X O formula ofeigenvectorfor E 1

ai
bg2 giveus
la
II Ii tf
fôdfitktssigüécation
Characteristic polynomial
C IT 4 7
Eisa rootofthis polynomial
The transformation isthusnotlinear
but we have a whole hyperplane

hyperplane

we are asked tochoose 2 vectors that are orthogonal


XTy O

haveshown that at xz 2
for 1 1 we 3 0 E Xz
XIe
Lets take andXz 1 obtain y 1
G
we

XI
normalizingvector ç

For we have again 293 OE


y y yz g Silja
remember that xt.ge
1 1
f ffff
s t se start
t taise E
9F 92 and93
GIF II
Thus g set
f git g
and 5
fifa
3 LA a KO

f 1 1 s z
È O

Ï E fi
set ai 1 we obtain x
E and Effi
4 SD D S A 5

in our case U S and N D


if D S A SES A S'D'S
in our case Sis orthonormale s S

A
Find KS

M is idempotent
M M M

EKH KI II
Il 2K K I K

III r

I 2K KI JIK o

K K O

KE K K is idempotent

KE K

You might also like