HW 9 Solutions

Homework 9 Solutions
Joshua Hernandez
December 6, 2009
6.3 - The Adjoint of a Linear Operator
2b. Let V = C
2
(over C) and linear transformation g : V F dened by the mapping
g(z
1
, z
2
) = z
1
2z
2
,
nd a vector y such that g(x) = x, y for all x V:
Solution: Clearly, y = (1, -2) works. Taking the standard orthonormal basis {e
i
}
n
i=1
,
y =
n
i=1
g(e
i
)e
i
= e
1
2e
2
= (1, -2).
3b. Let V = C
2
(over C) and linear operators T on V, let T : V V be dened by the mapping
T(z
1
, z
2
) = (2z
1
+ iz
2
, (1 i)z
1
).
Evaluate T
at x = (3 i, 1 + 2i).
Solution: Let be the standard orthonormal basis of C
2
.
[T
x]
= [T]
[x]
=
_
2 i
1 i 0
_
_
3 i
1 + 2i
_
=
_
2 1 + i
-i 0
__
3 i
1 + 2i
_
=
_
5 + i
1 3i
_
.
6. Let T be a linear operator on an inner product space V. let U
1
= T + T
and U
2
= TT
. Prove that
U
1
= U
1
and U
2
= U
2
.
Solution: In two lines,
U
1
= (T +T
= T
+ (T
= T
+T = U
1
,
U
2
= (TT
= (T
(T)
= TT
= U
2
.
12. Let V be an inner product space, and let T be a linear operator on V. Prove the following results.
a. R(T
= N(T)
1
Solution:
z R(T
(x), z = x, T(z) = 0
for all x V
z N(T).
The right equivalence is a consequence of inner product property (d).
b. If V is nite-dimensional, then R(T
) = N(T)
.
Solution: Above, we proved that R(T)
= N(T). Taking the perpendicular of both sides,

(R(T
= N(T)
.
By problem (6.2.13c) in the last homework, (R(T
= R(T
). Note that, to apply that result,

we only needed rank(T
) rank(T) < .
18. Let A be an n n matrix. Prove that det(A
) = det(A).
Solution: This is trivial if A =
_
a
_
M
11
:
det(A
) = det(
_
a
_
) = a = det(
_
a
_
) = det(A).
Assume the result is true for (n 1) (n 1) matrices. Then,
det(
ij
) = det((
A
ji
)
) = (det(
A
ji
)).
Expanding along the rst column,
det(A
) = (A
)
11
det(
11
) + + (-1)
n+1
(A
)
n1
det(
n1
)
= A
11
det(
A
11
) + + (-1)
n+1
A
1n
det(
A
1n
),
which is just the conjugate of det(A), expanded along the rst row.
20c. Use the least squares approximation to nd the best ts with both (i) a linear function and (ii) a
quadratic function. Compute the error E in both cases.
{(-2, 4), (-1, 3), (0, 1), (1, -1), (2, -3)}
Solution: In the notation of the least-squares formulation on p.363,
A
1
=
_
_
_
_
_
_
-2 1
-1 1
0 1
1 1
2 1
_
_
_
_
_
_
, A
2
=
_
_
_
_
_
_
4 -2 1
1 -1 1
0 0 1
1 1 1
4 2 1
_
_
_
_
_
_
and y =
_
_
_
_
_
_
4
3
1
-1
-3
_
_
_
_
_
_
,
2
so that
A
1
A
1
=
_
-2 -1 0 1 2
1 1 1 1 1
_
_
_
_
_
_
_
-2 1
-1 1
0 1
1 1
2 1
_
_
_
_
_
_
=
_
10 0
0 5
_
, (A
1
A
1
)
-1
=
_
1
10
0
0
1
5
_
, and
A
2
A
2
=
_
_
4 1 0 1 4
-2 -1 0 1 2
1 1 1 1 1
_
_
_
_
_
_
_
_
4 -2 1
1 -1 1
0 0 1
1 1 1
4 2 1
_
_
_
_
_
_
=
_
_
34 0 10
0 10 0
10 0 5
_
_
, (A
2
A
2
)
-1
=
_
_
1
14
0 -
1
7
0
1
10
0
-
1
7
0
17
35
_
_
.
Therefore
_
c
d
_
=
_
1
10
0
0
1
10
__
-2 -1 0 1 2
1 1 1 1 1
_
_
_
_
_
_
_
4
3
1
-1
-3
_
_
_
_
_
_
=
_
1
10
0
0
1
5
__
-18
4
_
=
_
-1.8
0.8
_
,
and
_
_
c
d
e
_
_
=
_
_
1
14
0 -
1
7
0
1
10
0
-
1
7
0
17
35
_
_
_
_
4 1 0 1 4
-2 -1 0 1 2
1 1 1 1 1
_
_
_
_
_
_
_
_
4
3
1
-1
-3
_
_
_
_
_
_
=
_
_
1
14
0 -
1
7
0
1
10
0
-
1
7
0
17
35
_
_
_
_
6
-18
4
_
_
=
_
_
-1/7
-9/5
38/35
_
_
.
The rst-order least-squares solution is therefore y
1
= -1.8x + 0.8. The error is
E
1
=
_
(4.4 4)
2
+ (2.6 3)
2
+ (0.8 1)
2
+ (-1 (-1))
2
+ (-2.8 (-3))
2
= 0.63246.
The second-order solution is y
2
= -
1
7
x
2
9
5
x +
38
35
. The error in this case is
E
2
=
_
(
144
35
4)
2
+ (
96
35
3)
2
+ (
38
35
1)
2
+ (-
6
7
(-1))
2
+ (-
108
35
(-3))
2
= 0.33806.
2 1 1 2
2
1
1
2
3
4
(a) First-Order Approximation
2 1 1 2
3
2
1
1
2
3
4
(b) Second-Order Approximation
3
6.4 - Normal and Self-Adjoint Operators
2 For each linear operator T on an inner product space V, determine whether T is normal, self-adjoint, or
neither. If possible, produce an orthonormal basis of eigenvectors of T for V and list the corresponding
eigenvalues.
b. V = R
3
and T is dened by
T(a, b, c) = (-a + b, 5b, 4a 2b + 5c)
Solution: Let be the standard orthonormal basis of R
3
. Then
A := [T]
=
_
_
-1 1 0
0 5 0
4 -2 5
_
_
.
This is clearly not self-adjoint. Next,
AA
=
_
_
-1 1 0
0 5 0
4 -2 5
_
_
_
_
-1 0 4
1 5 -2
0 0 5
_
_
=
_
_
2 5 -6
5 25 -10
-6 -10 45
_
_
,
A
A =
_
_
-1 0 4
1 5 -2
0 0 5
_
_
_
_
-1 1 0
0 5 0
4 -2 5
_
_
=
_
_
17 -9 20
-9 30 -10
20 -10 25
_
_
.
These are not equal, so T is not normal. Therefore we can not nd an orthonormal basis of
eigenvectors.
d. V = P
2
(R) and T is dened by T(f) = f
, where
f, g =
_
1
0
f(t)g(t) dt.
Solution: Well start with the standard (not orthonormal) basis = {1, x, x
2
}. Clearly,
A := [T]
=
_
_
0 1 0
0 0 2
0 0 0
_
_
.
This is an upper-triangular matrix, so its eigenvalues, which lie along the diagonal, are all zero.
However, it is clear that dim(E
0
) = nullity(A 0I) = 3 rank(A) = 1, so this matrix does not
have any basis of eigenvectors, let alone an orthonormal one.
f. V = M
22
(R) and T is dened by T
_
a b
c d
_
=
_
c d
a b
_
Solution: Let = {(
1 0
0 0
) , (
0 1
0 0
) , (
0 0
0 1
) , (
0 0
0 1
)}, the standard orthonormal (under the standard,
pointwise, inner product) basis of M
22
(R). Then
A := [T]
=
_
_
_
_
0 0 1 0
0 0 0 1
1 0 0 0
0 1 0 0
_
_
_
_
This is clearly a self-adjoint matrix.
4
Computing eigenvalues:
A
() = det
_
_
_
_
- 0 1 0
0 - 0 1
1 0 - 0
0 1 0 -
_
_
_
_
= -det
_
_
- 0 1
0 - 0
1 0 -
_
_
+ 1 det
_
_
0 1 0
- 0 1
1 0 -
_
_
= -(-
3
+ ) + 1(1
2
) =
4
2
2
+ 1 = (
2
1)
2
= ( + 1)
2
( 1)
2
.
Computing eigenspaces:
E
1
= N
_
_
_
_
-1 0 1 0
0 -1 0 1
1 0 -1 0
0 1 0 -1
_
_
_
_
= span
_
_
_
_
_
_
1
0
1
0
_
_
_
_
,
_
_
_
_
0
1
0
1
_
_
_
_
_
_
,
E
1
= N
_
_
_
_
1 0 1 0
0 1 0 1
1 0 1 0
0 1 0 1
_
_
_
_
= span
_
_
_
_
_
_
1
0
-1
0
_
_
_
_
,
_
_
_
_
0
1
0
-1
_
_
_
_
_
_
.
The orthonormal basis is therefore = {(
1 0
1 0
) , (
0 1
0 1
) , (
1 0
-1 0
) , (
0 1
0 -1
)}, with corresponding eigen-
values 1, 1, -1, -1.
6 Let V be a complex inner product space, and let T be a linear operator on V. Dene
T
1
=
1
2
(T +T
) and T
2
=
1
2i
(T T
)
a. Prove that T
1
and T
2
are self-adjoint and that T = T
1
+ iT
2
.
Solution: First,
(T
1
)
=
1
2
(T +T
=
1
2
(T
+T) = T
1
and (T
2
)
=
1
2i
(T T
= -
1
2i
(T
T) = T
2
.
Next, observe that T
1
+ iT
2
=
1
2
((T +T
) + (T T
)) = T.
b. Suppose also that T = U
1
+iU
2
, where U
1
and U
2
are self-adjoint. Prove that U
1
= T
1
and U
2
= T
2
.
Solution: Subtracting the two equations,
0 = T T = (U
1
T
1
) + i(U
2
T
2
) =: S
1
+ iS
2
.
Taking the adjoint of both sides, we get
0 = S
1
+ (iS
2
)
= S
1
iS
2
.
The only way to reconcile this equation with the previous is to set S
1
= S
2
= 0.
c. Prove that T is normal if and only if T
1
T
2
= T
2
T
1
.
Solution:
If T is normal, then TT
= T
T and
T
1
T
2
=
1
2
(T +T
)
1
2i
(T T
) =
1
4i
(T
2
TT
+T
T + (T
)
2
) =
1
4i
(T
2
+ (T
)
2
)
=
1
4i
(T
2
+TT
T + (T
)
2
) =
1
2i
(T T
)
1
2
(T +T
) = T
2
T
1
.
5
We can write T
= T
1
iT
2
. If T
1
T
2
= T
2
T
1
, then
TT
= (T
1
+ iT
2
)(T
1
iT
2
) = T
2
1
+ i(T
1
T
2
T
2
T
1
) +T
2
2
= T
2
1
+T
2
2
= T
2
1
+ i(-T
1
T
2
+T
2
T
1
) +T
2
2
= (T
1
iT
2
)(T
1
+ iT
2
) = T
T,
so T is normal.
11a. Assume that T is a linear operator on a complex (not necessarily nite-dimensional) inner product
space V with an adjoint T
. Prove that if T is self-adjoint, then T(x), x is real for all x V.

Solution: If T is self-adjoint, then for all x V,
T(x), x = x, T
(x) = x, T(x) = T(x), x,

where we rst applied the denition of the adjoint, then the denition of a self-adjoint matrix,
and then inner product property (c). The LHS is equal to its conjugate, so it must be real.
12. Let T be a normal operator on a nite-dimensional real inner product space V whose characteristic
polynomial splits. Prove that V has an orthonormal basis of eigenvectors of T. Hence prove that T is
self-adjoint.
Solution: (Analogous to the proof of Theorem 6.16)
Since T is a linear operator on a nite-dimensional inner-product space V whose charac-
teristic polynomial splits, We may apply Schurs theorem to obtain an orthonormal basis
= {v
1
, . . . , v
n
} such that A := [T]
is upper-triangular. We know that v

1
is an eigenvec-
tor of T, because T(v
1
) = A
11
v
1
. Supposing all v
1
, . . . , v
k
are eigenvectors, we can show that v
k
is an eigenvector, and so prove the claim. Since A is upper triangular,
T(v
k
) = A
1k
v
1
+ A
2k
v
2
+ + A
lk
.
Since v
i
are orthonormal, we can write A
jk
in terms of Fourier coecients:
A
jk
= T(v
k
), v
j
= v
k
, T
(v
j
) =
v
k
,
j
v
j
)
_
= v
k
, v
j
= 0.
Thus T(v
k
) = A
kk
v
k
, and hence v
k
is an eigenvector of T. By induction, all the vectors in are
eigenvectors of T.
17a. Let T be a self-adjoint linear operators on an n-dimensional inner product space V. Prove that T is
positive denite [semidenite] if and only if all its eigenvalues are positive [nonnegative].
Solution:
Suppose T is positive denite [semidenite]. If (, x) is some eigenpair of T, then
x, x = T(x), x > 0 [ 0].
Since x, x > 0, we have > 0 [ 0].
6
Let = {v
1
, . . . , v
n
} be an orthonormal basis of eigenvalues. If these are positive [nonneg-
ative], then for any x =
c
i
v
i
V,
T(x), x =
_
T
_
c
i
v
i
_
,
c
j
v
j
_
=
_
c
i
T(v
i
),
c
j
v
j
_
=
_
c
i
i
v
i
,
c
j
v
j
_
=
i,j
c
i
c
j
i
v
i
, v
j
and since the v

i
are orthonormal,
=
i
|c
i
|
2
i
> 0 [ 0].
Thus T is positive denite [semidenite].
7

HW 9 Solutions

Uploaded by

Copyright:

Available Formats

HW 9 Solutions

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

HW 9 Solutions

Uploaded by

Copyright:

Available Formats

Homework 9 Solutions

= N(T). Taking the perpendicular of both sides,

). Note that, to apply that result,

. Prove that if T is self-adjoint, then T(x), x is real for all x V.

(x) = x, T(x) = T(x), x,

is upper-triangular. We know that v

and since the v

You might also like