Answers to Numerical Problems

CHAPTER 2

2.1a

I 4

1-1

2.2a

2.2d

-1

\~3

L3

-1

l1

y = 24

4J \_z

2.2b

"3

2.2c

'3

L7

-2
r

-2

-7

-1

26

2.1b

1.

1
2.3a

3 - 4

1 -1_

2.2e

2.3b

'if

~2

r
-i_

2.2f

r-3

-5

-2

2.3c

-2

_25_

(5,15,20)

2.3d

"-1

-7]

-3

-ij

2.3e

210

23

-4

2.3f

23/2

-7,8)

2.4b

2.4c

_-2 4 oj
2.4d

0~

-2

- 4 j

2.4e

[-4

12'

L 3

-4_

240

-80

120"

_-80

35

40_

2.4f

2.5a
(i)

(ii)
352

(DE)' = E'D'

D'E'

ea + gb

ec + gd

fa + hb

fc-\- hd_

ae + cf

ag + chl

be + df

bg+dhj

(iii)

E'D' = See (i) above.

12

16

353

CHAPTER 2

2.5b
(i)

(DEy = E'D'=

r 3 0

(iii)

LlO 4_
E ' D ' = See (i) above.

2.6a

F=0;

G^0;

D'E'

(ii)

FG = 0

17

2.6b

-10

30

50

GF =
2.7a

84

2.8a

(-2,6,10)

2.9

2.7b

2.7d

(-3,0,18)

2.8c

28

-30
2.8b

2Jc

a^ + bc-a-2
.
X ^ - X - 2 I = (X + I ) ( X - 2 I ) =
_ac + cd-c

2.10a

2.11a

_8
0

2.12a

148

2.10b

8
13_
2.1l i b
2.12b

["4

2.10c

LO 9_
0

2.11c

-32

2.11d

|A| = 3(10) + 2 - 1 7 = 15

2.14

|M| = ( l ) ( - 8 ) ( - 2 . 7 5 ) ( - 5 . 8 1 8 2 ) = - 1 2 8
SF=33

2.15b
S=

2.15c
C=

R=

2.15e

Ad =

27X2=165

31

38

- 37.5]

31

39.3

33.3

-43.7 1

38.5

33.3

38.3

-38.2

-37.5

-43.7

6.9

5.2

-38.2
50.8J
6.3
- 6.3"

5.2

6.6

5.6

- 7.3

6.3

5.6

6.4

- 6.4
8.5_

--6.4

-13

bc +

bd-b
d^-d-2_
2.10d

28

66

30

48

44

105_

_72

114_

fl2 + Z>^

2X3^-(SZa)^/??! = 50.83

41.5

-6.3
1

2.15d

2.15f

Xi = 4.33

ab +

(0,0,0)

-128

2.12c

2.13

2.15a

2.8d

^<t>

0.76

0.96

-0.82'

0.76

0.85

-0.98

0.96

0.85

-0.86

-0.82

-0.98

-0.86

1
4.2~

[-3.5

-3.3

-3.8

-1.5

-2.3

-0.8

3.2

-2.5

0.7

-1.8

-0.8

1.5

-1.3

0.2

0.2

2.5

2.7

3.2

-2.8

L 3.5

3.7

3.2

- 3 . Jl_

- 3 . 5 - 1 . 5 - 2 . 5 + 1.5 + 2.5 + 3.5 = 0

354

ANSWERS TO NUMERICAL PROBLEMS

CHAPTER 3

3.1a

||a||=V5;

INI = V B ;

3.1b

the plane y= 1 - x for all z

3.1c

a parabola z = x^ for all y

3.Id

a sphere with center (0, 0, 0) and radius 1

3.2a

r5/2l

3.2c

llc|| = V7r' + 7

\\PR\\ = 3 V^l 2 A\RQ\\ = 3y/3/2-A\PQ\\ = 3 ^

3/2
1/2^
3.3a

k = -'\

3.3b

ki = l;

k2 = 2;

k2 = -\

3.3c

linearly independent vectors;

3.4a

3.5

1.414

3.4b

la'cl

Ib'cl

3.6a

a +b

3.6b a ' - b '

3.4c

-^

3.4d

- 6 < a'b < 6

(if a'c and b'c have same sign)

3.6
a

5a'+10b'

3.6d ^ ( a ' - b ' )

3\A0

-l/3>/lO

5/3VT0

xATo

s/y/uo

-ijyjno ejyJVib

\/5450

-20/\/5450

55/V5450

45/\/5450

VnO/2

5/VnO

-7/\/TTo

6/\/nO

8/3\/T0

-43
UUS (7-^y -

3.7a

2^5995 '

"I/VH"

C I C JV

' "

r 33/\/l414l

2/V14 b* = -18/\/l414
L 1/V1414 J
_3/Vl4_
ro"
-I/V5""
"2/V5"]
b* =
C* =
- 2/V5"j
J/\/5"J

c* =

3.7b

3.7c

J'

4/vroi"
7/VlOl
_-6/Vl0l_

The third vector vanishes; in a two-dimensional space no more than two vectors can
constitute a basis.

3.8

1/8"
.5/8.
-2k

3.9

2 V

a'b = 0. The orthogonal vector c is of the form

k
k

355

CHAPTER 4

3.10

3.11

The equation for the circle is independent of parameter ^ , denoting the angle of
rotation. New equations for the ellipse with coordinates (w, v) instead of (jc, y) are
3.10a

^u^ + ^v^-3uv

3.10c

1 -y 13 o 3\/3^
-1/2 + -f-v^ + ^ u v
4
4
2

7 2 ^ 13 2
- w + rlr
4
4

3V3
ruv =4
2

= 4

The vectors ( 4 , - 2 , l , 7 ) a n d ( 2 , - 3 , x , > ' ) are linearly independent for any x , ; ' since
(4, 2) and (2, 3) are linearly independent 4n a two-dimensional space.

3.12
(ei,e2,e3)=

^1^^

n .^^
3.10b

=4

'K -7
a b = 7;

~r(fi,f2,f3)

-1

-1

-13

*'u*
3V2-IIV6
a* b* = ^^^;; ^^
4

3.13b ||a||=V5'; ||b||=VT3; ||a*|| = Vs; ||b*|| = VTI

Vector lengths are preserved under rotation.

4j

Li

_6

3_

10

14.

18

Area of {ABCD) is 4 and area of U ^ C Z ? ) is 8.

y'x2 = m cov(r,X2);

Ily||=\/w^y;

Ilx2ll=\/^^x

CHAPTER 4
4.1a

an identity mapping

4.1b

a projection on z, followed by a reversal of direction along z

4.1c

a reversal of direction along z

4. Id

leaves terminus of vector at same height but moves vector three times as far from z.

4.2a

r 1"
Xi* =

L3/2_
4.2b

X2* =

_4j
"41

r 11
Xi* =

4.2c
X5* =

"5"

"4~[
J

L3/2J
r 1"
;
L5/2_

X2* =

.4J

X6* =

X3* =

X3* =

A
'21

' 5"
;

"21

"21

_3j

_6j

X4* =

.4]

X8* =

Areas are invariant while angles and lengths change.

(i)

3~

10

12

[2
8
(1

i)

3"

~2

10

12

[4

9_

6"

(iii)

15

_2

15_

356

ANSWERS TO NUMERICAL PROBLEMS

4.3b

13

(i)

(ii)

~l/2

. 8

-0.707

0.707

0.35

-0.3^5

0.35.
-2/1 r

L-7/11

5/11.

4.7b

4.7a
1 2

1 2

-0.35

Lo 0 0

A"^ is not defined

d
-1
ad + be L c

b
aj

-7/2

4.8b

3~

4.8c

-1
1

-8/37J

r(A) = 2;

4.9b

r(A) = 3 = r ( B ) = 3 ;

r(B) = 3. Regular inverse does not exist.

A-^ =

-1/3

-2

20/6

1/6

-7/6

1/6

-1/6

r(A) = r(B) = 2. Regular (3 x 3) inverse does not exist.

4.10
-1

4.11

-0.35J

4.9a

4.9c

1.06

0.35

4.7c

1-17/2

0.866

0.35

4.6d

0.5

1.06

4.6b

A ^ is not defined.

5_

1/2

L-0.35

5/11
A-^ =

_11

0.866_ 0
0.866

4.5d

0.35

4.5c

+ (2,-1)

r 0.35

l/2j

12

0 "

4.5b

n "3

0.5

2"

(iii)

LO l/2j L-0.5

.0

4.8a

13

" 5

0.707

3
(Xi*,X2*)=(Xi,X2)

4.6c

L-0.5

4.4c

4.6a

0.866
(Xi*,X2*)=(Xi,X2)

4.5a

0.707

4.4a

4.4b

R"^ =

-3

1/4 I - 1

-1

-13

10.26

-9.74

-9.74

10.26

| A | = 4 ; x i * = - l / 2 ; x 2 * = -l/2;jC3* = 3/2

1.072
|R| = 0.0975;

b* =

-0.128

357

CHAPTER 5

4.12b

-377

4.12a

4.13

X
^1 ^

O-fe

|T| = - 2

X =

1406

4.12c

-5

X '^ =

L^3 * J

" 7

10

_-4

-6.

4.14

X2
o

<
X3 J
X
c

X2

o
o

.^3

^3

0 0 9
T is a stretch transformation.

CHAPTER 5
5.1a

Xi = 5;

X2 = 8;

3/V58
U=

5.1b

Xi = - 4 ;

5.1c

Xi=2;

X2 = - 6 ;

U=

Xi = 5;

X2 = 4;
U=

5.1e

Xi=4;

X2 = 2;

i/Vss 1
"2/V5" 4/V17

Li/Vs" 1/VT7J

V2/2
_>/272

-5/V4i
4/V4I.

~ 1/VIO
_-3/\/lO

l/\/5
-2/V5

X2 = - 7 ;
U=

5.1d

X3 = - 2 ;

i/Vn 3/Vi4 V2/2

5.1f

Xi=3;

X2 = 3;

X3 = 3;

All eigenvectors are of the form k

tf(A)=13;

|A| = 40

5.2c

tr(A) = - 5 ;

|A| = - 1 4

5.3a

0
k

5.3b

5.2b

k
_0_

or

tr(A) = - 1 0 ;

5.2d

tr(A) = 9;

5.3c

|A| = 24
|A| = 20

Any vector remains invariant under

358
5.3d
5.4

ANSWERS TO NUMERICAL

PROBLEMS

There is no vector remaining invariant under this transformation since eigenvalues

The matrix
D= [

13]

has eigenvalues given by the diagonal elements of

4 _ /2 ] ^^^ associated eigenvectors that are

1-V2U2

V2-I.

\/4-2\/2

as columns of U. Then, let A = UDU'.

5.4c

12 + 3V2
0

u
u

"1+ \/2

_ 0

_
5.4g

5.4f

A-^= - 1
^
4
1
Xi = 12;

(4-V2)-'''_
2
1
-2

1 ;

X2 = 6;

0 "
6-V2_ r

rv4"W^

V4-\/2

4-\/2j

0
'1

6 + \/2

r'88 + 50\/2

1
U

_i

5.4d

"(4 + y/iy^^^

5.5

5.6

1-V2J

1
4 + V2

5.4e

5.4b

0
U'
1 2 - 3 >/2j

88-50\/2_

1rr '
1u

U'
"5

ABA'^ = 0

-3

0"

1 0
0

X3 = 6. The last two eigenvalues are not unique. One possible

solution is

U=

1/V6
-2/V6
1/V6

I/V2
0
-I/V2

-1/V3
-l/\/3"
-\l\/3

Then

D = U'AU =
5.7a

5.7b

12

i/\/2

5.8a
A

5.7c

5.7d

i/\/2nri

0"

L1/V2 -1/V2JL0 i_

I/V2 l/\/2'
-I/V2" l/\/2.

5.8b

A=

u'

Q'

-0.650

0.646

-0.616

-0.173

0.430

0.744

"0.290

'2.022

.0

3.86IJ .0.957

-0.957
0.290J

359

CHAPTER 5
0.707 0.707" "1 0"

5.8c

0'

1_

A=
.-0.707 0.707_
5.8d

"0.114

0.634'

0.282

-0.108

A = 0.396

0.526

0.553

-0.551

0.667

0.083_

10.592

0.994

-0.107

.0

2.969

_0.107

0.994

In all four cases, r(A) = 2.

1/V2"

(A'A)^ =

l/\/2"
(5.8b):

1/V^ f l Ol ri/\/2
-1/V2'J Lo i j L1/V2

0.290

2 _

(A'A)^ =

-0.957

0.9571 r i 6.744
0.290J

I/V2
-I/V2.

L0

"0.290

-0.957"

0.957

0.290

227.248

5.9b
(5.8a):

(A'A)"^ =

(5.8b):

(A'A)"^ =
f^V2

5.10a

5.10b

Ai/2

-1/2

5.10c

'I/V2
-I/V2
0.290
-0.957

I/V2
-I/V2
I/V2
L-I/V2
I/V2
=
-I/V2
I/V2
_
-I/V2

,1/2 .

A=U

2.022

O.290J 0

-l/\/2J

~| r0.290

-0.957

3.86IJ L0.957

0.290

0] ri/\/2
-l/\/2
l/\/2 "1
l/\/2
I/V2J Lo V7jLl/\/2
0 1 ri/\/2
-l/\/2"
I/V2I "1
I/V24
1/V2"J -0 I/V7J Ll/\/2
0 ] ri/\/2 - l / \ / 2
I/V2 V6
l/\/2
I/V2 _ 0 2V2JL1/V2
0 i r i / \ / 2 -1/V2"
I/V2 "1/V6
l/\/2
I/V2 _ 0 l/2\/2jLl/\/2

\l\JsA - 0

I/3J L2/\/5"

l/\/5

U', where U is any set of two orthonormal vectors.

V2
0
U'
L 0 \/2

A^/2=U|

0.957]

i/\/2

-l/y/J
I/V5" 2/V5' "2 OI \\I\JS
\/\/5
-llsjs
\l\Js] _0 3J [21^/5
I/V5" 2/V5'1 '1/2 O i r i / V s " -2/\/5"
\_-2l\j5

5.10d

I/V2 "1 01 r i / \ / 2
-I/V2J _0 Ij L1/V2

l/\/2

l/\/2"J

U'

360

ANSWERS TO NUMERICAL PROBLEMS

5.11a

1
5.11b

" 1

5.11c
x;
-2

-2
5.12a

A=

rank 1

s>

-3'

1_
2

5.lid
|x;

rank 2

"1/V2

rar

x;

-3/2~

..-3/2

rank 2

3 _

i/VTl Vi 01 ri/V2
l/\/2"
Lo oj L1/V2 -1/V^J

Rotationprojection on first axis and stretch-rotation

3 + VTy
2
A=U
0

3-yJvf

U'

Rotation-tretch-rotation
5.12c

A=

-3/\/ro i/vr^
i/vro" 3/\/To"

10 0"

-3/Vio i/VTo"
L i/VTo" 3/\/ro

5 + VToT

5.12d

0 0
Rotationprojection on first axis and stretch-rotation

A=U

s-VTo u'

0
Rotation-stretch-rotation

5.13a

5.13b

29.52

19.44

14.44"

C = 19.44

14.19

10.69

14.44

10.69

8.91

Zi

0.755y^j + 0.521X^1 "^ 0.396Xd2; \ = 50.50

Z2
Z3
5.14a

0.2\%Y^ - 0.771Xjjl + 0.598JiQ}2; ^3 = O-^^

A = x'

-3/Vro

\I\/YQ'

10

. i/Vio 3/vro.

"-3/vro i/Vio"

_0

. i/Vio" 3/\/io"-

Positive semidefinite
5.14b

0.542
-0.643
0.542

-0.707

-7.372

0.455

0.542

-0.707

0.765

-3

0.707

0.455

-1.628

-0.643

0.455

Negative definite
5.14c

0.585

0.811"

4.081

0.585

-0.811'

-0.811

0.585

0.919

0.811

0.585_

Positive definite
5.14d

0.707

-0.408

-0.577

0.707

0.408

0.577

0.816

-0.577

0.707

0.707

-0.408

0.408

0.816

-0.577

0.577

-0.577

Positive semidefinite

0.542

0.707

0.765

0.455

361

CHAPTER 6

.15a

0.393

-0.0424

_-0.0424

0.345 _

^ - 1 / 2 ^ ^ - 1 / 2 ^ "21.216

13.203"

_13.203
8.257_
Q2
A
0.849 -0.529] [29.444 0

Y = XdW-^^^Qa = Xd
. 15b

Q2'
0.849 0.529"

0.029.J L-0.529 0.849_

O.849J L 0

_0.529

^
ir

0.311

-0.2441

0.146

0.315j

Both transformations yield same diagonal matrix.

CHAPTER 6
6.1a

6.1b

Y = - 2 . 2 6 3 + 1.550X1-0.239^2

f=-2.263+

C=

1.550X1-0.239X2

14.19

10.69"

.10.69

8.91_

"1
R=

0.951

0.951

b i * = 1.076; b2* =-1.309

Y= 1.125-0.179X1+0.117Xi^
R=

1
_0.978

0.9781
1

/?^ = 0.954

R^ = 0.954 exceeds R'^ = 0.902 for Y regressed on Xi alone.

"1

0.951

_0.951

T=

R=

Zi = 0.707Xsi + 0.707Xs2;

0.707

0.707"

0.707

-0.707_

Xi = 1.951

Z2 = 0.707Xsi-0.707Xs2;

X2 = 0.049
^1/2

F=

0.707

0.707

0.707

-0.707

"1.398

0.988

0.208

0.988

0.147"
-0.147

A 45 rotation is involved. Eigenvectors (and loadings) differ from those obtained

53.25

41.42

41.42

33.08_

Xi = 85.79;
6.2c

T=

X2 = 0.54.

"0.787

0.617"

.0.617

-0.787_

Rotation is 38 .

R^ is the same if both columns of component scores are used, i^^ is lower
(R^ = 0.88) if only the first column of component scores is used.

362

ANSWERS TO NUMERICAL PROBLEMS

6.2d
C=

29.52

19.44

14.44

0.756

-0.618

-0.218

19.44

14.19

10.69

T = 0.521

0.366

0.771

|_14.44

10.69

8.91

0.396

0.696

0.599

Xi = 50.51;
6.3a

X2 = 1.72;

0.937

-0.700

.0.348

0.715

X3 = 0.398
A=

V.=

29.155

0.028

-1

Discriminant weights change under standardization of data matrix.

r0.905

-0.612"

LO.425

0.791

W-^'2Q2 =

As can be seen from Table 6.5, W"^^^Q2 is equal to V, the matrix of eigenvectors
obtained from W A.
6.3c

Xi = 3.768;

ti =

0.868
-0.497

In the two-group case only one discriminant is computed.

C=

14.19

1.46

1.46

0.25

r = - 2 . 4 8 8 + 1 A96Xi
6.4b

R=

0.774

0.774

-1.229^^2

i?^= 0.907

APPENDIX A

A.la
A.2
A.3
A.5

2
2x-x
X = I

1 ^w.
r- e^
, 2 . .x2
A.lc
(;c + 4)
2\/x
(minimum);
x = 1 (maximum)
A.lb

| ^ ( 1 , 3 ) = 4.2; | ^ ( 1 , 3 ) = 4.6
ox
oy
/ ( - 2 / 3 , 4/3) = 16/3 (maximum)

A.4

34
A.6

| ^ ( X ) = 2AJC.

ox

Ifx' = ( 3 , l , 2 ) ,

-|^ =

dx

28
40

A.7

(minimum)

. , , 4(x'-\?(9x'

A.la

min/(jc, >^) = - 1 8

. 3

.5

(2JC^ + 1)'

-I2x')

363

APPENDIX B

APPENDIX B
B.la
B.2a

)^ ( A ) = l

B.lb

"l

B.3a
k-' =

r(B) = 2

: o"

B.lc

B.2b r 1

o~

0_

4/5

-3/5"

L-1/5

2/5J

-3

A-^ = - 1

-3 '
0

-1

-2.11

-1.11

A^ = -1.37

1.40

0.41

-1.22

-0.78

0.55

0.55

-0.67

~-372 + ( 1 / 5 ) 7 5 + 6/5"

B.5

Ji

_
B.6a
A" =

B.ld

" 7

B.3b

2.56

B.4

r(C) = 2

r(D) = 2

2.34~

r^r

72

X2

3/5(7< > + l )

X3

Ts-l

X4

Us.

Ts
r-0.25 3

0.226

().266

L 0.412

- 0.107

- ( ).067

B.6b
A" =

0.2

0.6

0.4

-0.2