Math 54 - Homework 8 Solutions

5.4.2
ker(A
T
) = ker
__
1 1 1
1 2 3
__
_
1 1 1
1 2 3
_

_
1 1 1
0 1 2
_

_
1 0 1
0 1 2
_
_
1 0 1
0 1 2
_
_
_
1
2
1
_
_
= 0
Basis for ker(A
T
) =
_
_
_
_
_
1
2
1
_
_
_
_
_
For the sketch illustrating (Im(A))

= ker(A
T
), Im(A) is the plane spanned by
_
_
1
1
1
_
_
,
_
_
1
2
3
_
_
. Notice that
_
_
1
2
1
_
_
is orthogonal to
_
_
1
1
1
_
_
and
_
_
1
2
3
_
_
5.4.7
A is a n n symmetric matrix. This implies that A = A
T
(Im(A))

= ker(A
T
) = ker(A)
Or alternatively, Im(A) = (ker(A))

1
5.4.8
A has dimension mn
y has dimension m1
ker(A) = 0 A
T
A is invertible
Least squares solution of L(x) = y is (A
T
A)
1
A
T
y
(a)
L
+
(y) = (A
T
A)
1
A
T
y
A
+
= (A
T
A)
1
A
T
It is easy to verify that L
+
is linear
L
+
(y +z) = L
+
(y) + L
+
(z)
L
+
(ky) = kL
+
(y)
(b)
If A is invertible, then (A
T
A)
1
= A
1
(A
T
)
1
then, A
+
= (A
T
A)
1
A
T
= A
1
(A
T
)
1
A
T
= A
1
If L is invertible, L
+
= L
1
(c)
L
+
(L(x)
L
+
(Ax)
(A
T
A)
1
A
T
Ax
= x
(d)
L(L
+
(y)
= L((A
T
A)
1
A
T
y)
= A(A
T
A)
1
A
T
y
2
5.4.10 (a)
Let x be a solution to the system x = x
h
+ x
0
where x
h
ker(A), x
0
(ker(A))

Ax = b
A(x
h
+ x
0
) = b
Ax
h
+Ax
0
= b
Ax
0
= b
(b)
Ax
0
= b
Ax
1
= b
where x
0
, x
1
(ker(A))

A(x
0
x
1
)
= Ax
0
Ax
1
= b b = 0
which is a contradiction. It would imply that x
0
x
1
ker(A)
(c)
If x
1
is a solution, it can be written as x
1
= x
ker(A)
+x
ker(A)

However, in (b) we established that x

ker(A)
has only one possible value, x
0
Thus, x
1
= x
ker(A)
+ x
0
Look at x
1

2
= (x
ker(A)
+ x
0
) (x
ker(A)
+ x
0
)
= x
0
x
0
+ something positive = x
0

2
+ something positive
It follows that x
1

2
> x
0

2
3
5.4.15
A is an mn matrix
ker(A) = {0}
A
T
A is invertible by fact 5.4.2b
B A = I
A
T
B
T
= I
A
T
A(A
T
A)
1
= I
Let B = (A(A
T
A)
1
)
T
= (A
T
A)
1
A
T
5.4.22
x

= (A
T
A)
1
A
T
b
=
_
_
_
3 5 4
2 3 5
_
_
_
3 2
5 3
4 5
_
_
_
_
1
_
3 5 4
2 3 5
_
_
_
5
9
2
_
_
=
_
3
2
_
b Ax

=
_
_
0
0
0
_
_
4
5.5.2
< f, g + h >=< g + h, f >=< g, f > + < h, f >=< f, g > + < f, h >
5.5.4 (a)
< A, B >= tr(A
T
B)
= tr
_
_
_[a
1
. . . a
n
]
_

_
b
1
.
.
.
b
n
_

_
_
_
_
= tr([a
1
b
1
+ . . . + a
n
b
n
])
You can think of this as a one by one matrix
The trace of a square matrix is the sum of its diagonal matrix.
Thus, < A, B > is the dot product of A and B
(b)
< A, B >= tr
_
_
_
_

_
a
1
.
.
.
a
m
_

_[b
1
. . . b
m
]
_
_
_
= a
1
b
1
+ . . . a
m
b
m
5.5.10
< f, g >=
1
2
_
1
1
f(t)g(t) dt
< f, g(t) >=
1
2
_
1
1
t(at
2
+ bt + c) dt
=
1
2
_
9t
4
4
+
bt
3
3
+
ct
2
2
_
1
1
set < t, g(t) >= 0, we get
2b
3
= 0 or b = 0
g(t) = at
2
+ c
A basis for the space of all functions in P
2
that are orthogonal to f(t) = t is {1, t
2
}.
Apply Gram-Schmidt to the basis {v
1
, v
2
}
5
u
1
=
v
1
v
1

=
1
_
1
2
_
1
1
dt
= 1
u
2
=
v

2
v

2

where v

2
= v
2
(u
1
v
2
)u
1
= t
2

1
3
v

2

2
=< t
2

1
3
, t
2

1
3
>
=< t
2
, t
2
> +2 < t
2
,
1
3
> + <
1
3
,
1
3
>
=
1
5

2
9
+
1
9
=
1
5

1
9
=
9 5
45
=
4
45
u
2
=
45
4
(t
2

1
3
)
5.5.12
f(t) = |t|
b
k
=< f(t), sin(kt) >
=
1

|t| sin(kt) dt
c
k
=< f(t), cos(kt) >=
1

|t| cos(kt) dt
a
0
=< f(t),
1

2
>=
1

2
_

|t| dt =
1

2
5.5.16
P
1
with inner product < f, g >=
_
1
0
f(t)g(t) dt
(a)
Basis for P
1
= {v
1
, v
2
} = {1, t}
u
1
=
v
1
v
1

=
1
_
_
1
0
dt
= 1
u
2
=
v

2
v

2

=

12(t
1
2
)
where v

2
= v
2
(u
1
v
2
)u
1
= (t
1
2
)
6
v

2
=

_
1
0
t
2
t +
1
4
dt
=

_
t
3
3

t
2
2
+
1
4
t
_
1
0
=
_
1
3

1
2
+
1
4
=
_
1
12
(b)
We need to nd proj
p
1
(f(t))
proj
p
1
(f) =< u
1
, f > u
1
+ < u
2
, f > u
2
=< 1, t
2
> 1 +
_

12
_
t
1
2
_
, t
2
_

12(t
1
2
)
5.2
This is FALSE. (AB)
T
= B
T
A
T
= A
T
B
T
unless B
T
and A
T
commute.
5.4
This is TRUE.
A and S are orthongonal
S
1
ASx = SS
1
ASx
= ASx
= Sx
= x
5.10
This is TRUE.
AA
1
= I
(AA
1
)
T
= I
T
(A
1
)
T
A
T
= I
5.20
ker(A) = 0 only guarantees that the column vectors are linearly independent. How-
ever, by fact 5.3.10, we also necessitate that the column vectors have length 1 and are
7
orthogonal to each other.
5.22 This is TRUE because
n
is an inner product space, and also by the Gram-
Schmidt process.
6.1.2
det
_
2 3
4 5
_
= 2 5 3 4 = 2
6.1.6
det
_
_
6 0 0
5 4 0
3 2 1
_
_
= 6 4 1 = 24
6.1.12
det
_
1 k
k 4
_
= 4 k
2
This is invertible only if k = 2
6.1.22
det
_
_
cos(k) 1 sin(k)
0 2 0
sin(k) 0 cos(k)
_
_
= cos(k)(2 cos(k)) 1 0 sin(k)(2 sin(k))
= 2(cos
2
(k) + sin
2
(k)) = 2
6.1.40
There are lots of zeroes everywhere. This is good.
det = 3 2 4 5
= 120
6.1.44
For n by n matrix A: det(kA) = k
n
det(A).
Proof by induction works here. When n=1 or 2, this is easy to see. Assume true for n=m-1,
show statement is true for n=m.
6.1.55 (a)
Notice that d
4
= d
3
det
_
_
1 1 0
0 1 1
0 1 1
_
_
= d
3
d
2
8
In general, d
n
= d
n1
d
n2
(b)
d
1
= 1; d
2
= 0
The rest follows from the recursive formula.
1, 0, 1, 1, 0, 1
. .
, 1, 0, 1, 1, 0, 1
. .
, . . .
Note that the the sequence repeats
It lows from part (a) that 100 mod 6 = 4 so, d
100
= d
4
= 1
9