Assignment 2

Problem 1 : -

v is a solution of Ax =b ⇐ Air =b

C- : E is invertible ) ⇐ EAV = Eb

⇐ v is a solution of EAx=Eb .

Problem 2

(a) Let the size of A

,
is rxn and size of Az
be (M -
r )xn .

( ol
Ñ
Choose a matrix E S -
t .

E. A = RREFCA ) ,
=
'

, ,

4- tjxn

(I )
block
Let
by
E °
=
' rxlmr) Then
I
"

cnn.nxcm.rs
(M
-
r
)×r

multiplication

( ) ( ) ( )
A1 EIA -10 Az
O
EA E ,
=
, =

o I Az OE ,
+ I Az

( A.) E.
Az
=

( ¥a ) .
 RREF of EA has atleast as
many pivots as

in Ñ ,
=
t -

Rank ( EA ) It = Rank CA , )

since E is invertible , we have Rank (A) =

Rank ( EA )2RanklA , )

similarly ,
one can show Rank CAI 2 Rank ( Az ) .

(b) Let the size of A

,
is rxn and size of Az
be (M -
r )xn .

Choose invertible matrices and of and

E
,
Ez size rxr

(m -
r ) ✗ Cm -
r ) such that

Ñ
E RREFCA , )
[ }
A =
=
, ,

Or -
txn

(
Ñz
EzAz
}
and = RREF / Az) =

0cm -

n -
sixn

( }
Let E El Orxm block
Then
by
=
.
-
r

0 Ez
m -

rxr

multiplication we have
 ( e.) [ I =/ / { ^¥ )
EA =
E O " ""
,
= 0
o Az
EzAz .

Since the of
piviots decreases
#
by applying row

have
operations ,
we

t + s = # pivots of Ai + # pivots of Az

2 # pivots in RREF ( EA ) =
Rank ( EA )

since E is invertible we have

,

Rank (A) = Rank ( EA ) £ tts =

Rank 1A ) ,
+ Rank ( Az )

an invertible

(c)
=/ Aj J
Choose matrix E such that EA =

RREFCA ) .

Im -

rlxn

( )
Let EB B1 where has
B size rxn
=
.

,
B
,

Thus

J ]
-

( Ag ) (
)
(
E LA + B) = B, =
CA , -1131
'

+
Ba B
,

Rank ( At B) = Rank CECA -113 ) )

I Rank CA , + Bit + Rank ( Ba ) (

by
(b) )
 E # Rows of ( Ai -113 ) ,
1- Rank ( Ba )

=
Rank ( Ai ) + Rank (
Ba )

c- Rank (A) + Rank ( EB ) ( by ( a) )

= Rank (A) + Rank (B)

Rank ( ATB ) I Rank (A) + Rank (B) .

an invertible

(d)
=/ )
Choose matrix E such that EA Ai =

RREFCA ) .

n o
Im -

rlxn

Then

[ Aol ) [ AFB )
ECAB ) = ⇐ A) B
=

B =

Rank ( AB ) =
Rank CEAB ) =
Rank ( AIB )

I # rows of A B # rows of AI
, =

(
°

: A ,
= RREFIA , ) ) = # pivots of A ,

=
Rank (A) '

Rank ( AB ) E Rank (A)

If B is invertible ,
then B is a
square matrix .

"
Rank (A) = Rank ( ABB ) =/ Rank LAB ) £ Rank (A)
 Rank (A) =
Rank ( AB ) .

(e) Rank / A) =
Rank ( At ) E Rank / A- it ) + Rank / Aat )
a

part (a) =
Rank ( Ai ) + Rank ( Az ) .

Alternatively ,
note that A = [ A. 07 -110 AD

hank (A) I rank [ A. 0 ] thank [ 0A

,
]

÷
= rank ( Ai ) + rank (Az) .

Problem 3

( it
""

-1 : : :|
I 0 2
"
A- = • →
3 -
l I

5 - I 5

fr . ,↳, e. → r ,

% ""
, yy | ?)
,

,
' °
, ,

2 to <
o
0 -
I - 5
,
z ,

Hence matrices similar

given are
.

Alternatively ,
one could have show
given
two matrices

have same RREF .

Hen RREFCA ) RREFCB )

A
XRB show =/
 Problem 4 : -

Clearly ,

A ( t☒A+( 1-
7) µ)
= A- (✗ ) * + All -

✗ 1¥

=
✗ Ax + ( I -

d) Ay

= Xb + ( I -

d) b = b

The this that

geometric meaning
of is
every point on

line solution to
.

&
joining ✗ y is a A- ✗ =b .

Problem 5 :
2×1
-3×2 k 3×1

[§)
✓
Size
(a) A
[ 9) = of A is 3×2

I ;] Al ;)
And =
= Al -

i 2)

We know that system Ax=b has unique solution

and
if only if rank A a rank [ A b) =
# columns of A -

Clearly rank [ A b) =
rank A as AK -

i 2) = b .

rank / A) Al
=L ;]
So matrix A s t -

= 2 and in
any
-
-

solution
will have
(9) as for A# b

a-
=L : :/ 1%1
(b) A
[§ ) =
(9) Size of A = 2×3 .

=) rank (A) 52 <

# columns of A .

Thus
,
the
system Ax =

(9) cannot have unique solution .

( As the
possible cases are either Rank (A) < Rank b)

or Rankl A) =
Rank [ A
b) < # columns of
A) .

Problem 6 :

( | ( | }
21

y
2

¥%
1 l 0 0 1
2 2 1 00

2 I 2 O -3 -2 -210
0 I ①

2 2 I
Rz / 2)
- '

o o i O l o l I
l
-

~
/ Rz (3) >

[ | ) ¥-115 )
( | )
1 2 I 0 0
2 I 2 2 10 O

O o ,
¥ ¥ -3-5 0 ① -5 -2 -23
O l -
l 0 -
l l
O l -
l o -

I 1

/ Riz / -2
)

( / / / | -3¥
¥ ¥
2 -2 ' 00
12,2/-41
I 0 4 I

I
0
§ ¥ ¥ 00
0 1 -

→
¥ ¥
1232 ( )
t
O l -

l o , o
o 1 1
¥ ¥
-

F-
-

⇐
1 : : :| 0 I 0
's
F
:-&
-

F F ]
F F F
-
 [ )
-3 2 2

is the inverse
÷
.

2
2 -3

2 2 -3

Problem 7 :

1 : : :| :/ I ? '÷ :/ ÷:|
"
' " "
¥
% /,
-

2)

1%1
f. I ,%| ! }
"
,
0 -51, I 5/2

2k 1-
y -12-2=3 5/2 and
,
-512g + 2- =

12-52=0
l and 2=2
y
-2=0 =
.

,
-

Problem 8 :

"
"
125
' "
Note that E. (2) EDDA Thus B-
B A
EIN E.
= =
.

"
.
=
A E / E) E. (E)
,

calculate
"
A GJE
We
using
now .
 [ to :?) Y=
( ]
' ' ' ' ' ' ' o o

l 2 3 O l 0 0 I 2 -

l l 0

I -

I 2 123,1-11 0 -2 I -

I 0 I

0¥45)[ )
=
2 I 0

(
I
-

2 I I 0 -

2) I
-

I
Rzzl 0 -

→ 0 2 I
0 I 2 -
1 1 01
I -1

Riz / 1) 0 O l 1-
-

0 0 5 -3 2 I - I E-
5 5
5

µ }
3/5 45
-

75

(
-

I 0 0

R , 0 I 0
1/5 45 -45
Rzzl -2)
O O l
-315 2/5 45

(
"

}
-35
'

A =
75 15 and

1/5 45 -215
-

3/5 45 45

( }
" "

710 3/10 45
/ %) / Yz )
-

A
B =
Ez E ,
=

Yao Ho -215
-31,0 2/10 45

Problem gi

Let r= rank (A) = rank / At ) ( °o° column rank = row rank )

[ /
let E be invertible matrix sit .
EAT =
°
= RREFC #-)
On -

r×m
 C has r -

pivots .

EAT
[ §)
Thus E- A EAT RREFCA )
'
= = = is '*

( RREF is ) ( At ) rank / A)
as
unique .

rank __ r =
.

Problem 10 :

(a) We will use GJE .

| /?/ / / )
' ' ' '
1 '

Rat ? ?
? -12
i 2 c
c. ,
4
I 2 2C R2 ,
C- 1) O k g
-

R "" ) Rank / A) =3

[
"

%)
' ' '
-

if 4 '

o o e,

0 -2
Rank (A) =
2

O k -
8
c- 1

if
.

has solution
If -1,1<=16
system then of
equations
c- no .

-
( Rank (A) =
2 < Rank ( Alb ) =3 )
If then
0=1 , Rank (A) =2 and Rank ( Alb ) = 2 ⇐ k=6

Thus soln
system of equations has
infinitely many .

if c=l and 1<=6

then of has
If c -1-1 the
system eqns unique soln
,
.

,
 will have solution -1-1
The
system if
c
unique
.

1: :| =/ ? ]
(b) .
I 1 I

A ☒ 7

:
=

[; / ?) ( %)
l l l l
l
"' l
[ A lb ] =
R "
> o o za

;
2

; Rs ,
1- 1) o , a.

If ) and has
ctyz , then Rank (A) =3 = Rank ( Alb system
unique solution .

If then is inconsistent ② 1) 7=4

system
c-

c=Yz ,
as

have
Thus
given system
of
equations will never

solns
infinitely many
.
 Problem 11 :

we will prove def (A) -1-0 ⇐ Ax=o has no non-trivial solns .

1=>1
0=1--1 (
"
A- ✗ =o AX ) =
(AA ) ✗ = Ix = ✗

So Ax=o ✗ =0 .
Thus Ax=o has no non-trivial solns .

( ⇐ ) Proof by contradiction .

Suppose cletltl =o .

Then rank ( At ) = rank (A) < n -

Thus RREF ( At )

contains a row which is zero

.

At ( ;-) EAT ( 2 ;-) Cn At ( n ;-) 0

C I + + . .
- - . . - .

-1 =
, .

lxn

for icn t all =/

-

some ci.cz , .
-
-

.
s - -
C
; 0 .

( As row
operations always leads to linear combination of rows
)

Atli ;-) = Al -
ji )T

[§ }
h

Al ji )T 0T (1)
Ic ; - =

Nxt
with -1-0
-

Ñ= 1
,

transpose both sides in (1)

get
Taking on we

/? /
2- g. Al ji ) =
0 i e A
0
-

'
-

nx , =
i =
1

Cn
contradicts Ax=o has no nontrivial soln .
 Problem 12 :

1=>1
"

detlAI.de/-lA-')EZAA-'--I=detlA1detCA-'
If A ,
A has
integer entries then

1=1

"
since def (A) det( A- 1) Z det (A) ,=detCA )
,
c- , we
get = -1-1 .

(⇐
'

det (A) = -11 A- =

1
Adj (A) =
±
Adj / A)
detect )

since Adj (A) has so does A.

integer entries ,

Problem 13 :
-

2
(a) 1=det( AAT ) =
det / A) detlat ) =
del (A)
-

det (A) = -1-1

=L ; ;] =L ; ?;)
(b) a- A. and detlt )=i

11.11 :) =L :/
a- a-
: :
.

AA
=
__

be
Question might .

'
det (A) = i -4+-1 but AA = Ata = I .

wrong !
confirm and
( I will

clarity ) .