You are on page 1of 152

Functional Analysis Problems with Solutions

ANH QUANG LE, Ph.D.

September 14, 2013


www.MATHVN.com - Anh Quang Le, Ph.D

Contents

Contents 1

1 Normed and Inner Product Spaces 3

2 Banach Spaces 15

3 Hilbert Spaces 27
3.1 Hilbert spaces . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.2 Weak convergence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

4 Linear Operators - Linear Functionals 45


4.1 Linear bounded operators . . . . . . . . . . . . . . . . . . . . . . . . 45
4.2 Linear Functionals . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63

5 Fundamental Theorems 73

6 Linear Operators on Hilbert Spaces 87

7 Compact Operators 99

8 Bounded Operators on Banach Spaces and Their Spectra 113

9 Compact Operators and Their Spectra 131

10 Bounded Self Adjoint Operators and Their Spectra 143

1
www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
2 CONTENTS

Notations:
• B(X, Y ): the space of all bounded (continuous) linear operators from X to Y .
• Image (T ) ≡ Ran(T ): the image of a mapping T : X → Y .
w
• xn −→ x: xn converges weakly to x.
• X ∗ : the space of all bounded (continuous) linear functionals on X.
• F or K: the scalar field, which is R or C.
• Re, Im: the real and imaginary parts of a complex number.

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D

Chapter 1

Normed and Inner Product Spaces

Problem 1.
Prove that any ball in a normed space X is convex.

Solution.
Let B(x0 ; r) be any ball of radius r > 0 centered at x0 ∈ X, and x, y ∈ B(x0 ; r).
Then
kx − x0 k < r and ky − x0 k < r.

For every a ∈ [0, 1] we have

kax + (1 − a)y − x0 k = k(x − x0 )a + (1 − a)(y − x0 )k


≤ akx − x0 k + (1 − a)ky − x0 k
< ar + (1 − a)r = r.

So ax + (1 − a)y ∈ B(x0 ; r). ¥.

Problem 2.
Consider the linear space C[0, 1] equipped with the norm
Z 1
kf k1 = |f (x)|dx.
0

Prove that there is no inner product on C[0, 1] agreed with this norm.

3
www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
4 CHAPTER 1. NORMED AND INNER PRODUCT SPACES

Solution.
We show that the norm k.k1 does not satisfy the parallelogram law. Let
f (x) = 1 and g(x) = 2x.
Then Z Z
1 1
kf k1 = 1.dx = 1, kgk1 = |2x|dx = 1,
0 0
while Z Z
1 1
1
kf − gk1 = |1 − 2x|dx = , kf + gk1 = |1 + 2x|dx = 2.
0 2 0
Thus,
17
kf − gk21 + kf + gk21 = 6= 2(kf k21 + kgk21 ) = 4. ¥
4

Problem 3.
Consider the linear space C[0, 1] equipped with the norm

kf k = max |f (t)|.
t∈[0,1]

Prove that there is no inner product on C[0, 1] agreed with this norm.

Solution.
We show that the parallelogram law with respect to the given norm does not hold
for two elements in C[0, 1].
Let f (t) = t, g(t) = 1 − t, t ∈ [0, 1]. Then f, g ∈ C[0, 1] and
kf k = max t = 1, kgk = max (1 − t) = 1,
t∈[0,1] t∈[0,1]

and
kf + gk = max 1 = 1, and kf − gk = max | − 1 + 2t| = 1.
t∈[0,1] t∈[0,1]

Thus,
kf − gk21 + kf + gk21 = 2 6= 2(kf k21 + kgk21 ) = 4. ¥

Problem 4.
Prove that:
If the unit sphere of a normed space X contains a line segment [x, y] where x, y ∈
X and x 6= y , then x and y are linearly independent and kx + yk = kxk + kyk .

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
5

Solution.
Suppose that the unit sphere contains a line segment [x, y] where x, y ∈ X and
x 6= y. Then
kax + (1 − a)yk = 1 for any a ∈ [0, 1].
Choose a = 1/2 then we get k 12 (x + y)k = 1, that is kx + yk = 2. Since x and y
belong to the unit sphere, we have kxk = kyk = 1. Hence
kx + yk = kxk + kyk.
Let us show that x, y are linearly independent. Assume y = βx for some β ∈ C. We
have
1 = kax + (1 − a)βxk = |a + (1 − a)β|.
For a = 0 we get |β| = 1 and for a = 1/2 we get |1 + β| = 2. These imply that
β = 1, and so x = y, which is a contradiction. ¥

Problem 5.
Prove that two any norms in a finite dimensional space X are equivalent.

Solution.
Since equivalence of norms is an equivalence relation, it suffices to show that an
arbitrary norm k.k on X is equivalent to the Euclidian norm P k.k2 . Let {e1 , ..., en } be
a basis for X. Every x ∈ X can be written uniquely as x = nk=1 ck ek . Therefore,
n
à n !1/2 à n !1/2
X X X
kxk ≤ |ek |kek k ≤ |ck |2 kek k2 ≤ Akxk2 ,
k=1 k=1 k=1
P 1/2
where A = ( nk=1 |ek |2 ) is a non-zero constant. This shows that the map x 7→ kxk
is continuous w.r.t. the Euclidian norm. Now consider S = {x : kxk2 = 1}. This is
just the unit sphere in (X, k.k2 ), which is compact. The map
S → R defined by x 7→ kxk
is continuous, so it attains a minimum m and a maximum M on S. Note that m > 0
because S 6= ∅. Thus, for all x ∈ S, we have
m ≤ kxk ≤ M.
x
Now, for x ∈ X, x 6= 0, kxk2
∈ S, so

kxk
m≤ ≤ M.
kxk2

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
6 CHAPTER 1. NORMED AND INNER PRODUCT SPACES

That is
mkxk2 ≤ kxk ≤ M kxk2 .
Hence, the two norms are equivalent. ¥

Problem 6.
Let X be a normed space.
(a) Find all subspaces of X which are contained in some ball B(a; r) of X.
(b) Find all subspaces of X which contain some ball B(x0 ; ρ) of X.

Solution.
(a) Let Y be a subspace of X which is contained in some ball B(a; r) of X. Note
first that the ball B(a; r) must contain the vector zero of X (and so of Y ); otherwise,
the question is impossible. For any number A > 0 and any x ∈ Y , we have Ax ∈ Y
since Y is a linear space. By hypothesis Y ⊂ B(a; r), so we have Ax ∈ B(a; r). This
implies that kAxk < r + kak. Finally

r + kak
kxk < .
A
A > 0 being arbitrary, it follows that kxk = 0, so x = 0. Thus, there is only one
subspace of X, namely, Y = {0}, which is contained in some ball B(a; r) of X.
(b) Let Z be a subspace of X which contain some ball B(x0 ; ρ) of X. Take any
x ∈ B(0; ρ). Then x + x0 ∈ B(x0 ; ρ) and so x + x0 ∈ Z since Z ⊃ B(x0 ; ρ). Now,
since x0 ∈ Z, x + x0 ∈ Z and Z is a linear space, we must have x ∈ Z. Hence
B(0; ρ) ⊂ Z.
ρx
Now for any nonzero x ∈ X, we have 2kxk ∈ B(0; ρ) ⊂ Z. Hence x ∈ Z. We can
conclude that Z = X. In other words, the only subspace of X which contains some
ball B(x0 ; ρ) of X is X itself. ¥

Problem 7.
Prove that any finite dimensional normed space :
(a) is complete (a Banach space),
(b) is reflexive.

Solution.
Let X be a finite dimensional normed space. Suppose dim X = d.

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
7

(a) By Problem 5, it suffices to consider the Euclidian norm in X. Let {e1 , ..., ed }
be a basis for X. For x ∈ X there exist numbers c1 , ..., cd such that
d
à d !1/2
X X
x= ck ek and kxk = |ck |2 .
k=1 k=1
P (n)
Let (x(n) ) be a Cauchy sequence in X. If for each n, x(n) = dk=1 ak ek then
à d !1/2
X (n) (m)
kx(n) − x(m) k = |ak − ak |2 → 0 as n, m → ∞.
k=1

Hence, for every k = 1, ..., d,


(n) (m)
|ak − ak | → 0 as n, m → ∞.
(n)
Therefore, each sequence of numbers (ak ) is a Cauchy sequence, so
(n) (0)
ak → ak as n → ∞ for every k = 1, 2, ..., d.
Pd (0)
Let a = k=1 ak ek then x(n) → a ∈ X.

(b) Let f ∈ X ] where X ] is the space of all linear functionals on X. We have


à d ! d d
X X X
f (x) = f ck ek = ck f (ek ) = ck α k ,
k=1 k=1 k=1

where αk = f (ek ). Let us define fk ∈ X ] by the relation fk (x) = ck , k = 1, ..., d.


For any x ∈ X and f ∈ X ] , we get
d
X d
X
f (x) = fk (x)αk , i.e., f = αk fk .
k=1 k=1

Hence,
Pdim X ] ≤ d. Pd
d
Let α f = 0. Then, for any x ∈ X, αk fk (x) = 0, and by taking
Pd k=1 k k k=1
x = k=1 ᾱk ek , we obtain fk (x) = ᾱk , and
d
X d
X
αk fk (x) = |αk |2 = 0.
k=1 k=1

Hence, αk = 0 for all k = 1, ..., d and thus, dim X ] = d. For the space X ∗ we have
X ∗ ⊂ X ] , so dim X ∗ = n ≤ d and dim(X ∗ )] = n. From the relation X ⊂ (X ∗ )∗ ⊂
(X ∗ )] we conclude that d ≤ n. Thus, n = d, and so X = (X ∗ )∗ . ¥

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
8 CHAPTER 1. NORMED AND INNER PRODUCT SPACES

Problem 8. ( Reed-Simon II.4)


(a) Prove that the inner product in a normed space X can be recovered from the
polarization identity:
1h i
hx, yi = (kx + yk2 − kx − yk2 ) − i(kx + iyk2 − kx − iyk2 ) .
4
(b) Prove that a normed space is an inner product space if and only if the norm
satisfies the parallelogram law:

kx + yk2 + kx − yk2 = 2(kxk2 + kyk2 ).

Solution.
(a) For the real field case, the polarization identity is
1
hx, yi = (kx + yk2 − kx − yk2 ). (∗)
4
We use the symmetry of the inner product and compute the right hand side of (∗):
1 1£ ¤
(kx + yk2 − kx − yk2 ) = hx + y, x + yi − hx − y, x − yi
4 4
1£ ¤
= hx, yi + hy, xi
2
= hx, yi.
For the complex field case, we again expand the right hand side, using the relation
we just established:
1h 2 2 2 2
i
(kx + yk − kx − yk ) − i(kx + iyk − kx − iyk )
4
1£ ¤ i£ ¤
= hx, yi + hy, xi − hx, iyi + hiy, xi
2 2
1 1 i2 i2
= hx, yi + hy, xi − hx, yi + hy, xi
2 2 2 2
= hx, yi.
(b) If the norm comes from an inner product, then we have
kx + yk2 + kx − yk2 = hx + y, x + yi + hx − y, x − yi
= 2hx, xi + 2hy, yi + hx, yi + hy, xi − hx, yi − hy, xi
= 2(kxk2 + kyk2 ).

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
9

Now suppose that the norm satisfies the parallelogram law. Assume the field is
C, and define the inner product via the polarization identity from part (a). If
x, y.z ∈ X, we write
y+z y−z y+z y−z
x+y =x+ + , x+z =x+ − ,
2 2 2 2
and we have
1¡ ¢
hx, yi + hx, zi = kx + yk2 + kx − yk2 − kx − yk2 − kx − zk2
4
i¡ ¢
+ kx + iyk2 + kx + izk2 − kx − iyk2 − kx − izk2

° °2 ° ° ° °2 ° °!
1 ° y + z ° ° y + z °2 ° y + z ° ° y − z °2
= °x + ° +° ° ° ° ° °
2 ° 2 ° ° 2 ° − °x − 2 ° − ° 2 °
ð °2 ° ° ° °2 ° °!
i ° y + z ° ° y + z °2 ° y + z ° ° y − z °2
− °x + i ° + °i ° ° ° ° °
2 ° 2 ° ° 2 ° − °x − i 2 ° − °i 2 °
ð °2 ° ° ° ° ° °2 !
1 ° y + z ° ° y + z °2 ° y − z °2 ° y + z °
= °x + ° +° ° −° ° − °x − °
2 ° 2 ° ° 2 ° ° 2 ° ° 2 °
ð °2 ° ° ° ° ° °2 !
i ° y + z ° ° y + z °2 ° y − z °2 ° y + z °
− °x + i ° + °i ° − °i ° − °x − i °
2 ° 2 ° ° 2 ° ° 2 ° ° 2 °
1¡ ¢
= kx + y + zk2 + kxk2 − kx − (y + z)k2 − kxk2
4
i¡ ¢
− kx + i(y + z)k2 + kxk2 − kx − i(y + z)k2 − kxk2
4
= hx, y + zi.

This holds for all x, y, z ∈ X, so, in particular,

hx, nyi = nhx, yi for n ∈ N.

And it also satisfies


hx, ryi = rhx, yi for r ∈ Q.
Moreover, again by the polarization identity, we have
1¡ ¢ i¡ ¢
hx, iyi = kx + iyk2 − kx − iyk2 − kx − yk2 − kx + yk2
4 4
= ihx, yi.

Combining these results we have

hx, αyi = αhx, yi for α ∈ Q + iQ.

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
10 CHAPTER 1. NORMED AND INNER PRODUCT SPACES

Now, if α ∈ C, by the density of Q + iQ in C, there exists a sequence (αn ) in Q + iQ


converging to α. It follows that
hx, αyi = αhx, yi for α ∈ C.
Thus the h., .i is linear.
Since ki(x − iy)k = kx − iyk, we have
hy, xi = hx, yi,
and
1 i¡ ¢
hx, xi = (k2xk2 ) − |1 + i|kxk2 − |1 − i|2 kxk2 = kxk2 .
4 4
So this shows that the norm is induced by h., .i and that it is also positive definite,
and thus it is an inner product. ¥

Problem 9. (Least square approximation. Reed-Simon II.5)


Let X be an inner product space and let {x1 , ..., xN } be an orthonormal set.
Prove that ° °
° XN °
° °
°x − cn xn °
° °
n=1

is minimized by choosing cn = hxn , xi.

Solution.
For every x ∈ X, we write
N
X
x= hxn , xixn + z, for some z ∈ X. (∗)
n=1

We observe that for all n = 1, ..., N ,


N
X
hxn , zi = hxn , xi − hxn , xihxn , xk i
k=1
= hxn , xi − hxn , xi = 0.
Therefore z⊥xn . Then due to (∗) we can write
N
X N
X ¡ ¢
x− c n xn = hxn , xi − cn xn +z.
n=1
|n=1 {z }
zN

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
11

Since z⊥zN , we have


° °2
° N
X °
° °
° x − c x
n n° = kzN k2 + kzk2
° °
n=1
N
X ¯ ¯
= ¯hxn , xi − cn ¯2 + kzk2 ,
n=1

which attains its minimum if

cn = hxn , xi for all n = 1, ..., N. ¥

Review: Quotient normed space.


• Let X be a vector space, and let M be a subspace of X. We define an equivalence relation on
X by
x ∼ y if and only if x − y ∈ M.
For x ∈ X, let [x] = x + M denote the equivalence class of x and X/M the set of all equivalence
classes. On X/M we define operations:

[x] + [y] = [x + y]
α[x] = [αx], α ∈ C.

Then X/M is a vector space.


If the subspace M is closed, then we can define a norm on X/M by

k[x]k = inf kyk = inf kx + mk = inf kx − zk = d(x, M ).


y∈[x] m∈M m∈M

What a ball in X/M looks like?

B([x0 ]; r) := {[x] : k[x] − [x0 ]k < r} = {x + M : kx − x0 + M k < r}.

• Suppose that M is closed in X. The canonical map (the natural projection) is defined by

π : X → X/M, π(x) = [x] = x + M.

It can be shown that kπ(x)k ≤ kxk, ∀x ∈ X, so π is continuous.

Problem 10.
Let X be a normed space and M a closed subspace of X. Let π : X → X/M
be the canonical map. Show that the topology induced by the standard norm on
X/M is the usual quotient topology, i.e. that O ⊂ X/M is open in X/M if and
only if π −1 (O) is open in X.

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
12 CHAPTER 1. NORMED AND INNER PRODUCT SPACES

Solution.
• If O is open in X/M , then π −1 (O) is open in X since π is continuous.
• Now suppose that O ⊂ X/M and that π −1 (O) is open in X. We show that O is
open in X/M . Consider an open ball B(0; r), r > 0 in X. Let x ∈ B(0; r). Then
kxk < r, and so
k[x]k ≤ kxk < r.
On the other hand, if k[x]k < r, then there is an y ∈ M such that kx + yk < r.
Hence x + y ∈ B(0; r), and so
¡ ¢
[x] = π(x + y) ∈ π B(0; r) .

If [x0 ] ∈ O, then x0 ∈ π −1 (O). Since π −1 (O) is open in X, there is an r > 0 such


that
B(x0 ; r) ⊂ π −1 (O).
This implies that
¡ ¢ ¡ ¢
O = ππ −1 (O) ⊃ π B(x0 ; r) = π x0 + B(0; r) = {x + M : kx − x0 + M k < r}.

The last set is the open ball of radius r > 0 centered at [x0 ] ∈ O. Thus O is open
in X/M. ¥

Problem 11.
Let X = C[0, 1], M = {f ∈ C[0, 1] : f (0) = 0}. Show that X/M = C.

Solution.
Given [f ] ∈ X/M , let ϕ([f ]) = f (0). Then the map ϕ : X/M → C is well-defined.
Indeed, if [f ] = [g], then (f − g)(0) = 0 so f (0) = g(0). It is clearly linear.
If f ∈ X, then g = f − f (0) ∈ M , and so f − g = f (0) is constant, which tells us
that
k[f ]k = |f (0)| = |ϕ([f ])|,
so ϕ is an isometry (and thus injective and continuous). Finally, constants are in
X = C[0, 1], so ϕ is surjective and thus an isometric isomorphism. ¥

Problem 12. P
If 0 < p < 1 then `p is a vector space but kxkp = ( n |xn |p )1/p is not a norm for
`p .

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
13

Solution.
Recall that if x = (x1 , x2 , ...), y = (y1 , y2 , ...) ∈ `p and α ∈ C then
x + y = (x1 + y1 , x2 + y2 , ...) and αx = (αx1 , αx2 , ...).
It is clear that αx ∈ `p . We show that x + y ∈ `p . For t ≥ 0 it not hard to see that
(1 + t)p ≤ 1 + tp , 0 < p < 1. This implies that
(a + b)p ≤ ap + bp , 0 < p < 1 and a, b ≥ 0.
Therefore,
kx + ykpp ≤ kxkpp + kykpp .
Since both kxkpp and kykpp are bounded, kx + ykpp is bounded. Hence x + y ∈ `p .
To show k.kp is not a norm for `p , let us take an example: If
x = (1, 0, ...) and y = (0, 1, 0, ...)
then kxkp = kykp = 1 but kx + ykp = 21/p > 2 since 1/p > 1. Therefore,
kx + ykp > kxkp + kykp . ¥

Problem 13.
Suppose that X is a linear space with inner product h., .i. If xn → x and yn → y
as n → ∞, prove that

hxn , yn i → hx, yi as n → ∞.

Solution.
Using the Cauchy-Schwarz and triangle inequalities, we have
|hxn , yn i − hx, yi| ≤ |hxn − x, yn i| + |hx, yn − yi|
≤ kxn − xkkyn k + kxkkyn − yk
≤ kxn − xk(kyn − yk + kyk) + kxkkyn − yk
≤ kxn − xkkyn − yk + kxn − xkkyk + kxkkyn − yk.
Since kxn − xk → 0 and kyn − yk → 0 as n → ∞, we see that
hxn , yn i → hx, yi as n → ∞. ¥

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
14 CHAPTER 1. NORMED AND INNER PRODUCT SPACES

Problem 14.
Prove that if M is a closed subspace and N is a finite dimensional subspace of a
normed space X, then M + N := {m + n : m ∈ M, n ∈ N } is closed.

Solution.
Assume dim N = 1, say N = Span{x}. The case x ∈ M is trivial. Suppose x ∈ / M.
Consider the sequence zk := αk x + mk , where mk ∈ M, αk ∈ C, and suppose
zk ∈ M + N → y as k → ∞. We want to show y ∈ M + N . The sequence (αk )
is bounded; otherwise, there exists a subsequence (αk0 ) such that 0 < |αk0 | → ∞ as
k 0 → ∞. Then
zk 0 mk 0
and → 0 as k 0 → ∞,
αk0 αk0
so x must be 0, which is in M . This is a contradiction. Consequently, (αk ) is bounded
and therefore it has a subsequence (αk0 ) which is converging to some α ∈ C. Thus

mk0 = zk0 − αk0 x → y − αx as k 0 → ∞.

Hence, y − αx is in M since M is closed. Thus y ∈ M + N .


The solution now follows by induction. ¥

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D

Chapter 2

Banach Spaces

Problem 15.
Let
P∞X be a normed space. Prove that X is a Banach space if andPonly if the series

n=1 an converges, where (an ) is any sequence in X satisfying n=1 kan k < ∞.

Solution. P∞
Suppose that
Pn X is complete. Let (an ) be a sequence in X such that n=1 kan k < ∞.
Let Sn = i=1 ai be the partial sum. Then for m > n,
° m °
°X ° Xm
° °
kSm − Sn k = ° ai ° ≤ kai k.
° °
i=n+1 i=n+1

P Pm
By hypothesis, the series ∞ n=1 kan k converges, so i=n+1 kai k → 0 as n → ∞.
Therefore, (Sn ) is P
a Cauchy sequence in the Banach space X. Thus, (Sn ) converges,
that is, the series ∞ P an converges.
n=1 P∞
Conversely, suppose ∞ n=1 an converges in X whenever n=1 kan k < ∞. We show
that X is complete. Let (yn ) be a Cauchy sequence in X. Then

1
∃n1 ∈ N : kyn1 − ym k < whenever m > n1 ,
2
1
∃n2 ∈ N : kyn2 − ym k < 2 whenever m > n2 > n1 .
2
Continuing in this way, we see that there is a sequence (nk ) strictly increasing such
that
1
kynk − ym k < k whenever m > nk .
2
15
www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
16 CHAPTER 2. BANACH SPACES

In particular, we have
1
kynk+1 − ynk k < for all k ∈ N.
2k
Set xk = ynk+1 − ynk . Then
n
X n
X Xn
1
kxk k = kynk+1 − ynk k < k
.
k=1 k=1 k=1
2
P∞
It
Pmfollows that k=1 kxk k < ∞. By hypothesis, there is an x ∈ X such that
k=1 xk → x as m → ∞. But we have

m
X m
X
xk = (ynk+1 − ynk )
k=1 k=1
= ynm+1 − yn1 .

Hence ynm → x + yn1 in X as m → ∞. Thus, the sequence (yn ) has a convergent


subsequence and so must itself converges. ¥

Problem 16.
Let X be a Banach space. Prove that the closed unit ball B(0; 1) ⊂ X is compact
if and only if X is finite dimensional.

Solution.
• Suppose dim X = n. Then X is isomorphic to Rn (with the standard topology).
The result then follows from the Heine-Borel theorem.

• Suppose that X is not finite dimensional. We want to show that B(0; 1) is not
compact. To do this, we construct a sequence in B(0; 1) which have no convergent
subsequence.
We will use the following fact usually known as Riesz’s Lemma: (See the proof below)
Let M be a closed subspace of a Banach space X. Given any r ∈ (0, 1), there exists an x ∈ X such
that
kxk = 1 and d(x, M ) ≥ r.

Pick x1 ∈ X such that kx1 k = 1. Let S1 := Span {x1 }. Then S1 is closed. According
to Riesz’s Lemma, there exists x2 ∈ X such that
1
kx2 k = 1 and d(x2 , S1 ) ≥ .
2

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
17

Now consider the subspace S2 generated by {x1 , x2 }. Since X is infinite dimensional,


S2 is a proper closed subspace of X, and we can apply the Riesz’s Lemma to find
an x3 ∈ X such that
1
kx3 k = 1 and d(x3 , S2 ) ≥ .
2
If we continue to proceed this way, we will have a sequence (xn ) and a sequence of
closed subspaces (Sn ) such that for all n ∈ N
1
kxn k = 1 and d(xn+1 , Sn ) ≥ .
2

It is clear that the sequence (xn ) is in B(0; 1), and for m > n we have
1
kxn − xm k ≥ d(xm , Sn ) ≥ .
2

Therefore, no subsequence of (xn ) can form a Cauchy sequence. Thus, B(0; 1) is not
compact. ¥

Proof of Riesz’s Lemma:


Take x1 ∈/ M . Put d = d(x1 , M ) = inf m∈M km − x1 k. Then d > 0 since M is closed. For any
ε > 0, by definition of the infimum, there exists m1 ∈ M such that

0 < km1 − x1 k < d + ε.


x1 −m1
Set x = kx1 −m1 k . Then kxk = 1 and

1 ° °
kx − mk = °x1 − (m1 + kx1 − m1 km)°
kx1 − m1 k | {z }
∈M

This implies that

inf m∈M kx1 − mk d


d(x, M ) = inf kx − mk = ≥ .
m∈M kx1 − m1 k d+ε
d
By choosing ε > 0 small, d+ε can be arbitrary close to 1. ¥

Problem 17.
Let X be a Banach space and M a closed subspace of X. Prove that the quotient
space X/M is also a Banach space under the quotient norm.

Solution.
We use criterion established above (in problem 15). Suppose that ([xn ]) is any

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
18 CHAPTER 2. BANACH SPACES
P∞
sequence in X/M such that n=1 k[xn ]k < ∞. We show that
k
X
∃[x] ∈ X/M : [xn ] → [x] as k → ∞.
n=1

For each n, k[xn ]k = inf z∈M kxn + zk, and therefore there is zn ∈ M such that
1
kxn + zn k ≤ k[xn ]k +
2n
by definition of the infimum. Hence

X ∞
X 1
kxn + zn k ≤ k[xn ]k + < ∞.
n=1 n=1
2n

But (xn + zn ) is a sequence in the Banach space X, and so



X
(xn + zn ) = x for some x ∈ X.
n=1

Then we have
° k ° ° k °
°X ° °X °
° ° ° °
° [x n ] − [x]° = ° [x n − x]°
° ° ° °
n=1 n=1
° °
°X k °
° °
= inf ° (xn − x + z)°
z∈E ° °
n=1
° °
°X k
¢°
° ¡ °
≤ ° (xn − x) + zn °
° n=1 °
° °
°X k °
° °
= ° (xn + zn ) − x° → 0 as k → ∞.
° n=1 °
Pk
Hence n=1 [xn ] → [x] as k → ∞. ¥

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
19

SPACE `p
(Only properties concerning to norms and completeness will be considered. Other
properties such as duality will be discussed later.)

Problem 18.
Show that `p , 1 ≤ p < ∞ equipped with the norm k.kp is a Banach space.

Solution.
(i) (i)
Let x(i) = (x1 , ..., xk , ...) for i = 1, 2, ... be a Cauchy sequence in `p . Then
kx(i) − x(j) kp → 0 as i, j → ∞.
(i) (j)
Since kx(i) − x(j) kp ≥ |xk − xk | for every k, it follows that
(i) (j)
|xk − xk | → 0 for every k as i, j → ∞.
¡ (i) ¢
This tells us that the sequence xk is a Cauchy sequence in F, which is complete,
¡ (i) ¢
so that xk converges to xk ∈ F as i → ∞ for each k. Set x = (x1 , ..., xk , ...). We
will show that
(∗) kx(i) − xkp → 0 as i, j → ∞ and x ∈ `p .
Given ε > 0, for any M ∈ N, there exists N ∈ N such that
ÃM ! p1 Ã ∞ ! p1
X (i) (j)
X (i) (j)
|xk − xk |p ≤ |xk − xk |p < ε if i, j > N.
k=1 k=1

Letting j → ∞, for i > N we get


ÃM ! p1
X (i)
(∗∗) |xk − xk |p < ε.
k=1

By Minkowski’s inequality,
ÃM ! p1 ÃM ! p1 Ã M ! p1
X X (N ) X (N )
|xk |p ≤ |xk − xk |p + |xk |p
k=1 k=1 k=1
à M
! p1 ̰ ! p1
X (N )
X (N )
≤ |xk − x k |p + |xk |p
k=1 k=1
̰ ! p1
X (N )
< ε+ |xk |p .
k=1

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
20 CHAPTER 2. BANACH SPACES

Letting M → ∞, since the last sum is finite, we see that


̰ ! p1
X
kxkp = |xk |p < ∞.
k=1

This shows that x ∈ `p . Finally letting M → ∞ in (∗∗), for i > N we get


̰ ! p1
X (i)
kx(i) − xkp = |xk − xk |p < ε.
k=1

This shows that x(i) → x in `p as required. ¥

Problem 19.
(a) Show that `∞ equipped with the norm k.k∞ is a Banach space.
(b) Let c0 be the space of sequences converging to 0. Show that c0 is a closed
subspace of `∞ .

Solution.
(a) We need to show that `∞ is complete. Assume that the sequence (x(n) ) is Cauchy
in `∞ . That is, for all ² > 0, there exists N ∈ N such that
(1) n, m ≥ N ⇒ kx(n) − x(m) k∞ < ².
(n) (n)
For a fixed n, we write x(n) = (x1 , x2 , ...). Then for N = N (²) as above,
(n) (m)
(2) |xj − xj | ≤ kx(n) − x(m) k∞ < ² for all j.
(n)
So, for a fixed j, the sequence (xj ) is Cauchy, and therefore convergent in C.
Denote
(n)
xj := lim xj , and x := (x1 , x2 , ...).
n→∞

We need to show x ∈ `∞ and x(n) → x as n → ∞. In (2), for a fixed j, letting


n → ∞ yields
(m)
|xj − xj | < ² for all m ≥ N.
Therefore
(m)
sup |xj − xj | := kx − x(m) k∞ ≤ ² for all m ≥ N.
j

That is x(m) → x as m → ∞ in `∞ . Now for all j ∈ N, n ≥ N ,


(n) (n)
|xj | ≤ |xj − xj | + |xj | ≤ kx − x(n) k∞ + kx(n) k∞ ≤ ² + kx(n) k∞ < ∞.

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
21

This shows that kxk∞ < ∞ and so x ∈ `∞ .


(b) Of course c0 ⊂ `∞ . Assume (x(n) ) ∈ c0 that converges in `∞ to x. We have to
show that x ∈ c0 . Let ² > 0 be arbitrary. Since x(n) → x in `∞ , we can choose
N ∈ N such that
²
kx(N ) − xk∞ < .
2
(N ) (N ) (N )
Since x(N ) = (x1 , x2 , ...) ∈ c0 we have xj → 0 as j → ∞. Therefore, choose
J ∈ N such that
(N ) ²
j ≥ J ⇒ |xj | < .
2
Then, for j ≥ J,
(N ) (N ) (N ) ² ²
|xj | ≤ |xj − xj | + |xj | ≤ kx − x(N ) k∞ + |xj | < + = ².
2 2
Therefore, x = (x1 , x2 , ...) ∈ c0 ¥.

Problem 20.
(a) Let c00 be the space of sequences such that if x = (xn )n∈N ∈ c00 then xn = 0
for all n ≥ n0 , where n0 is some integer number. Show that c00 with the norm
k.k∞ is NOT a Banach space.
(b) What is the closure of c00 in `∞ ?

Solution.
We observe that c00 ⊂ c0 ⊂ `∞ .
(a) Consider the sequence x(n) defined by
³ 1 1 1 ´
x(n) = 1, , , ..., , 0, 0, ... ∈ c00 .
2 3 n
Then, for n, m ≥ N ,
(
1
if n ≥ m,
kx(n) − x(m) k∞ = m+11
n+1
if n ≤ m.
In both cases we have for any N > 0
1
kx(n) − x(m) k∞ ≤ for n, m ≥ N.
N +1
So the sequence x(n) is a Cauchy sequence in c00 . Evidently, it is also a Cauchy
sequence in `∞ . Now consider the sequence
³ 1 1 1 1 1 ´
x = 1, , , ..., , , , ... ∈ `∞ .
2 3 n n+1 n+2

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
22 CHAPTER 2. BANACH SPACES

We see that x(n) ∈


/ c00 , and
1
lim kx(n) − xk∞ = lim = 0.
n→∞ n→∞ n + 1

This tells us that there is a Cauchy sequence in c00 which does not converge to
something in c00 . Therefore, c00 equipped with the k.k∞ norm is not a Banach
space.
(b) We claim that c00 = c0 (closure taken in `∞ ).
According to Problem 19, c0 is closed, so we have c00 ⊂ c0 . We show the inverse
inclusion. Take an arbitrary sequence x = (x1 , x2 , ...) ∈ c0 . We build a sequence a(n)
from x as follows:
a(n) = (x1 , x2 , .., xn , 0, 0....).
It is clear that a(n) ∈ c00 . Now since the sequence x converges to 0, given any ε > 0,
there exists N ∈ N such that

|xi | < ε for i ≥ N.

Then
ka(n) − xk∞ = sup |xi | ≤ ε.
i≥N

hence a(n) converges to x (in `∞ ). Hence x ∈ c00 . Thus, c0 ⊂ c00 . ¥

Problem 21.
Prove that:
p < q < ∞, then `p ⊂ `q and kxkq ≤ kxkp .
(a) If 1 ≤ S
(b) If x ∈ 1≤p<∞ `p then kxkp → kxk∞ as p → ∞.

Solution.
(a) Let x = (x1 , x2 , ...) ∈ `p . Then, for n large enough, we have |xn | < 1 and hence
|xn |q ≤ |xn |p since 1 ≤ p < q < ∞. That implies x ∈ `q . Now we want to show
³X ´1/q ³X ´1/p
|xn |q ≤ |xn |p .

q
Let an = |xn |p and α = p
> 1. The above inequality is equivalent to
X ³X ´α
aαn ≤ an ,

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
23

which follows by
X X ³X ´α−1 X ³X ´α
α α−1
an ≤ (max an ) an ≤ an an = an .

(b) Let x = (x1 , x2 , ...) ∈ `p0 for some p0 . Clearly, kxkp ≥ maxn |xn | = kxk∞ for any
finite p. On the other hand,
à !1/p à !1/p
X X p−p0 p0
p→∞
kxkp = |xn |p ≤ kxkp−p ∞
0
|x n | p0
= kxk ∞
p
kxk p
p
0 −
−−→ kxk∞ .
n n

Problem 22.
Prove that:
(a) If 1 ≤ p < ∞ then `p is separable.
(b) `∞ is not separable.

Solution.
p
(a) First we show that E := {x P∞∈ ` : xn = 0, n ≥ N for some N } is dense
p p
in ` . Indeed, if x ∈ ` , x = k=1 xk ek , where ek is the sequence such that the
k-component is 1 and the others are zero, then
° ° Ã ∞ !1/p
° Xn ° X
° °
°x − xk ek ° = |xk |p → 0 as n → ∞.
° °
k=1 p k=n+1
Pn
But k=1 xk ek ∈ E, so E is dense in `p . Now let A ⊂ E consisting of elements
x = (x1 , x2 , ..., xn , 0, 0, ...) ∈ E such that xk = ak + ibk , ak , bk ∈ Q. Since Q is dense
in R, A is dense in E. Hence A is dense in `p . Since A is countable, `p is separable.
(b) We now show that it is not the case for `∞ .
Let F := {x ∈ `∞ : ∀k ≥ 1, xk = 0 or xk = 1}. Then F is uncountable. Note
that for x ∈ F, kxk∞ = 1. Moreover, x, y ∈ F, x 6= y ⇒ kx − yk∞ = 1. Assume
that `∞ is separable. Then there is a set A = {a1 , a2 , ...} dense in `∞ . So, for all
x ∈ F , there exists k ∈ N such that kx − ak k∞ ≤ 31 . Let F be the family of closed
balls B(x; 13 ), x ∈ F . If B 6= B 0 then B ∩ B 0 = ∅. This allows us to construct an
injection f : F → A which maps each B ∈ F with an element a ∈ B ∩ A. This is
impossible since F is uncountable and A is countable. ¥

∗ ∗ ∗∗

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
24 CHAPTER 2. BANACH SPACES

Problem 23. (The space C[0, 1])


Let C[0, 1] be the space of all continuous functions on [0, 1].
(a) Prove that if C[0, 1] is equipped with the uniform norm

kf k = max |f (x)|, f ∈ C[0, 1]


x∈[0,1]

then C[0, 1] is a Banach space.


(b) Give an example to show that C[0, 1] equipped with the L1 -norm
Z 1
kf k1 = |f (x)|dx, f ∈ C[0, 1]
0

is not a Banach space.

Solution.
(a) Let (fn ) be a Cauchy sequence in C[0, 1] with respect to the uniform norm. Then
for any ε > 0, there exists N ∈ N such that

kfm − fn k < ε for m, n ≥ N.

Therefore

(∗) |fm (x) − fn (x)| < ε for m, n ≥ N and x ∈ [0, 1].


¡ ¢
This shows that for every x ∈ [0, 1], the sequence fn (x) is a Cauchy sequence of
numbers and therefore converges to a number which depends on x, say, f (x). In
(∗), fix n and let m → ∞, we have

(∗∗) |f (x) − fn (x)| < ε for n ≥ N and x ∈ [0, 1].

Thus the sequence (fn ) converges uniformly to f on [0, 1] so that f is continuous on


[0, 1], that is, f ∈ C[0, 1]. From (∗∗) we obtain

max |f (x) − fn (x)| = kf − fn k ≤ ε for n ≥ N.


x∈[0,1]

This shows that


lim kf − fn k = 0.
n→∞

(b) For each n ∈ N, consider the function


(
nx if 0 ≤ x < n1
fn (x) =
1 if n1 ≤ x ≤ 1.

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
25

One can check that the sequence (fn ) is a Cauchy sequence with respect to the
L1 -norm, but it converges to the function
(
0 if x = 0
f (x) =
1 if 0 < x ≤ 1,

which is not continuous, that is, f ∈


/ C[0, 1]. Thus the space C[0, 1] is not complete. ¥

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
26 CHAPTER 2. BANACH SPACES

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D

Chapter 3

Hilbert Spaces

3.1 Hilbert spaces

Problem 24.
Let M be a closed subspace of a Hilbert space H. Prove that:
(a) (M ⊥ )⊥ = M.
(b) If codim M := dim H/M = 1 then dim M ⊥ = 1.

Solution.
(a) In general, if M is a subset of H then M ⊂ (M ⊥ )⊥ . Indeed,

M ⊥ := {x ∈ X : x⊥M }.

So we have
x ∈ M ⇒ x⊥M ⊥ ⇒ x ∈ (M ⊥ )⊥ .
Now suppose M is a closed subspace of H and x ∈ (M ⊥ )⊥ . Since x ∈ H = M ⊕M ⊥ ,
we have
x = u + v, u ∈ M, v ∈ M ⊥ .
Since M ⊂ (M ⊥ )⊥ we have

x = u + v, u ∈ (M ⊥ )⊥ , v ∈ M ⊥ .

Since x − u ∈ (M ⊥ )⊥ and v ∈ M ⊥ and v = x − u we obtain

v ∈ M ⊥ ∩ (M ⊥ )⊥ ,

27
www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
28 CHAPTER 3. HILBERT SPACES

which implies v = 0. Hence, x = u ∈ M .


(b) Assume that there are two linearly independent vectors x, y ∈ M ⊥ . Recall that
M is the zero vector of the linear space X/M . Consider the cosets [x], [y]. Assume
α[x] + β[y] = M , for some scalars α, β. Then αx + βy ∈ M and, since αx + βy ∈ M ⊥
as well, we conclude that αx + βy = 0. Hence, α = β = 0 and therefore [x], [y] are
linearly independent. This contradicts the hypothesis codim M = 1. ¥

Problem 25.
Let T : H1 → H2 be an isometry of two Hilbert spaces H1 and H2 , i.e., kT xk =
kxk for every x ∈ H1 . Prove that

hT x, T yi = hx, yi for every x, y ∈ H1 .

Solution.
By hypothesis, we have

kT (x + y)k2 = kx + yk2 for every x, y ∈ H1 .

By opening up the norm using the inner product, we obtain

RehT x, T yi = Rehx, yi.

Similarly, kT (x + iy)k2 = kx + iyk2 gives that

ImhT x, T yi = Imhx, yi.

Hence,
hT x, T yi = hx, yi. ¥

Problem 26.
Let C be a closed convex set in a Hilbert space H. Show that C contains a unique
element of minimal norm.

Solution.
Let η = inf z∈C kzk.

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
3.1. HILBERT SPACES 29

• If η = 0, then, by definition of infimum, there is a sequence (zj ) in C such that


kzj k → 0 as j → ∞. Therefore, zj → 0 as j → ∞. Since C is closed, 0 ∈ C, and 0
is the unique element of minimal norm.
• Suppose η > 0. First we show that C contains an element of minimal norm.
Take (zj ) in C such that kzj k → η as j → ∞. The convexity of C implies that
1
(z + zk ) ∈ C, so that
2 j

1
kzj + zk k2 = 4. kzj + zk k2 ≥ 4η 2 .
4
Recall now the parallelogram law:
¡ ¢
kx − yk2 + kx + yk2 = 2 kxk2 + kyk2 .

Applying this we have:


¡ ¢
kzj − zk k2 = 2 kzj k2 + kzk k2 − kzj + zk k2
¡ ¢
≤ 2 kzj k2 + kzk k2 − 4η 2 → 4η 2 − 4η 2 = 0 as j, k → ∞.

Thus the sequence (zj ) is Cauchy, so converges to some z ∈ H. Since C is closed,


z ∈ C. The norm function is continuous, so kzk = η. This shows that C contains
an element of minimal norm.
Assume that there are two elements a1 , a2 ∈ C such that ka1 k = ka2 k = η. By the
above we have
ka1 + a2 k2 ≥ 4η 2 .
By the parallelogram law we have
¡ ¢
ka1 − a2 k2 = 2 ka1 k2 + ka2 k2 − ka1 + a2 k2 ≤ 4η 2 − 4η 2 = 0.

Hence, a1 = a2 . ¥

Problem 27.
Let 1 ≤ p < ∞. Prove that `p is a Hilbert space if and only if p = 2.

Solution.
• Suppose p = 2. In the space `2 the inner product is defined by

X
hx, yi = x̄i yi for x = (xi ), y = (yi ) ∈ `2 .
i=1

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
30 CHAPTER 3. HILBERT SPACES

This inner product gives rise to the norm


̰ ! 12
p X
kxk2 = hx, xi = |xi |2 .
i=1

According to Problem 18, the normed space `2 is complete. So `2 is a Hilbert space.


• Consider the general case where 1 ≤ p < ∞. Assume that `p with the corre-
sponding inner product is a Hilbert space. Consider two elements en , em ∈ `p with
m 6= n defined as follows:

en = (0, .., 0, 1, 0, ...),


| {z }
n−1

em = (0, .., 0, 1, 0, ...).


| {z }
m−1

Since `p is a Hilbert space, by the parallelogram law we have


¡ ¢
ken + em k2p + ken − em k2p = 2 ken k2p + kem k2p ,

That is
22/p + 22/p = 22 .
The unique solution of this equation is p = 2. We conclude that `p is a Hilbert space
if and only if p = 2. ¥

Problem 28.
Consider the Hilbert space H = L2 [−1, 1] equipped with the usual scalar product:
Z 1
hx, yi = x(t)y(t)dt, x, y ∈ H.
−1
R1
Let M = {x ∈ H : −1 x(t)dt = 0}.
(a) Show that M is closed in H. Find M ⊥ .
(b) Calculate the distance from y to M for y(t) = t2 .

Solution.
(a) Let 1 ∈ H be the function 1(t) = 1, ∀t ∈ [−1, 1]. Define the map T : H → C
by
x 7→ h1, xi.

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
3.1. HILBERT SPACES 31

Then T is linear. We show that T is bounded, so continuous.


¯Z 1 ¯ Z 1
¯ ¯
¯
|T x| = ¯ ¯
1(t)x(t)dt¯ ≤ |1(t)| |x(t)|dt
−1 −1
µZ 1 ¶ 21 µZ 1 ¶ 12
≤ 1dt |x2 (t)|2 dt
−1 −1

= 2kxk2 .
By definition, M = Ker T = T −1 (0). Since T is continuous, M is closed. Further-
more,
x ∈ M ⇔ h1, xi = 0 ⇔ M = (Span{1})⊥ .
Since M is closed, (see Problem 23 a).
M ⊥ = Span{1}.
(b) The distance from y ∈ H to M is the length of the projection vector of y on
M ⊥ . We have
µZ 1 ¶ µZ 1 ¶−1/2 √
|h1, yi| 2 2
d(y, M ) = = t dt 1dt = . ¥
k1k2 −1 −1 3

Problem 29.
Consider the Hilbert space H = L2 [−1, 1] equipped with the usual scalar product:
Z 1
hf, gi = f (t)g(t)dt, f, g ∈ H.
−1

Let E = {x ∈ H : f (−t) = f (t), t ∈ [−1, 1]}.


(a) Show that E is closed in H. Find E ⊥ .
(b) Calculate the distance from h to E for h(t) = et .

Solution.
(a) Define f˜(t) := f (−t). Define the map
S : H → H defined by Sf = f˜.
Clearly, S is linear. S is bounded, so continuous. Indeed,
µZ 1 ¶ 12 µZ 1 ¶ 21
kSf k2 = kf˜k2 = |f˜(t)|2 dt = |f (−t)| dt2
= kf k2 .
−1 −1

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
32 CHAPTER 3. HILBERT SPACES

In fact, S is an isometry. It follows that I − S is continuous. By definition, E =


Ker(I − S), so E is closed.
By definition, E consists of all even functions, so E ⊥ is the linear subspace of all
odd functions. In fact, we have
1 1
f (t) = [f (t) + f (−t)] + [f (t) − f (−t)] ≡ ϕ(t) + ψ(t),
|2 {z } |2 {z }
ϕ ψ

with ϕ is even, ψ is odd, and


Z 1
hϕ, ψi = ϕ(t)ψ(t)dt = 0, so that ϕ⊥ψ.
−1

(b) The distance from h ∈ H to E is the length of the projection vector of h on E ⊥ .


By the above expression,
1 1
ProjE ⊥ (h) = [h(t) − h(−t)] = (et − e−t ).
2 2
Therefore,
° °
¡ ¢2 °1 t °2
dist(h, E) = ° −t
° 2 (e − e )°
°
Z
1 1 t
= |e − e−t |2 dt
4 −1
1 2
= (e − e−2 − 4).
4
Thus,
1√ 2
dist(h, E) = e − e−2 − 4. ¥
2

Problem 30.
Let H be a Hilbert space and M be a closed subspace of H. Denoting by P :
H → M the orthogonal projection of H onto M, prove that, for any x, y ∈ H,

hP x, yi = hx, P yi.

(This is telling us that P is self-adjoint).

Solution.
We know that if M is a closed subspace of H, then

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
3.1. HILBERT SPACES 33

• For all u ∈ H, there exist unique uM ∈ M and uM ⊥ ∈ M ⊥ such that

u = uM + uM ⊥ .

• If P : H → M the orthogonal projection of H onto M, then

P u = uM .

Now for arbitrary x, y ∈ H, we have

x = xM + xM ⊥ y = yM + yM ⊥
P x = xM P y = yM .

With these, can have

hP x, yi = hxM , yM + yM ⊥ i = hxM , yM i,

since hxM , yM ⊥ i = 0, and

hx, P yi = hxM + xM ⊥ , yM i = hxM , yM i,

since hxM ⊥ , yM i = 0. Thus,

hP x, yi = hx, P yi. ¥

Problem 31.
Let H be a Hilbert space and A ⊂ H a closed convex non-empty set. Prove that
PA : H → H is non-expansive, i.e.,

kPA (x) − PA (y)k ≤ kx − yk, ∀x, y ∈ H.

(PA is the orthogonal projection on A).

Solution.
We claim:
D E
(∗) Re x − PA (x), PA (x) − a ≥ 0, ∀x ∈ H, a ∈ A.

Let xA = PA (x). Then x can be decomposed uniquely as

x = xA + x0A , ∀xA ∈ A, x0A ∈ A⊥ .

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
34 CHAPTER 3. HILBERT SPACES

We have
D E
2Re x − PA (x), PA (x) − a = hx0A , xA − ai + hxA − a, x0A i
= hx0A , x0A i + hxA − a, xA − ai
= kx0A k2 + kxA − ak2 ≥ 0.
Hence (∗) is proved.
Replacing a with PA (y) in (∗), we obtain
D E
(3.1) Re x − PA (x), PA (x) − PA (y) ≥ 0.

Analogously, D E
Re y − PA (y), PA (y) − PA (x) ≥ 0.
And therefore,
D E
(3.2) Re PA (y) − y, PA (x) − PA (y) ≥ 0.

Adding 3.1 and 3.2 we get


D E
Re x − y − [PA (x) − PA (y)], PA (x) − PA (y) ≥ 0, i.e.,
D E
(3.3) Re x − y, PA (x) − PA (y) ≥ kPA (x) − PA (y)k2 .

Form the Cauchy-Schwarz inequality, we have


D E
Re x − y, PA (x) − PA (y) ≤ |hx − y, PA (x) − PA (y)i|
≤ kx − yk kPA (x) − PA (y)k
and from here, replacing in 3.3, we obtain that
kPA (x) − PA (y)k2 ≤ kx − yk kPA (x) − PA (y)k.
Thus,
kPA (x) − PA (y)k ≤ kx − yk. ¥

Problem 32.
Let X be a Hibert space, and G1 ⊂ G2 ⊂ ... ⊂ Gn ⊂ ... be a sequence of closed
linear subspaces of X. Let
à !
[
G = Span Gn .
n∈N

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
3.1. HILBERT SPACES 35

(a) Prove that d(x, G) = limn→∞ d(x, Gn ), ∀x ∈ X.


(b) Prove that PG (x) = limn→∞ PGn (x), ∀x ∈ X. Note: PG is the orthogonal
projection of X on G.

Solution.
(a) Let à !
[
A = Span Gn .
n∈N

Then we have
d(x, G) = d(x, Ā) = d(x, A).
For any ε > 0 we have

d(x, G) + ε = d(x, A) + ε > d(x, A) = inf kx − ak.


a∈A

From this, we deduce that there is some aε ∈ A such that

d(x, G) + ε > kx − aε k.
¡S ¢ S
Since aε ∈ Span n∈N Gn , we can find λ1 , ..., λk ∈ K and x1 , ..., xk ∈ n∈N Gn such
that
aε = λ1 x1 + ... + λk xk .
Then there are n1 , ..., nk ∈ N such that x1 ∈ Gn1 , ..., xk ∈ Gnk . Let n = max{n1 , ..., nk }.
By hypothesis, the sequence (Gn ) is increasing , so we have Gn1 , ..., Gnk ⊂ Gn , so
x1 , ..., xk ∈ Gn . And since Gn is a linear space,

aε = λ1 x1 + ... + λk xk ∈ Gn .

That is,

(∗) ∀ε > 0, ∃n ∈ N : d(x, G) + ε > kx − aε k ≥ d(x, Gn ).

From Gn ⊂ G it follows that

(∗∗) d(x, G) ≤ d(x, Gn ), ∀n ∈ N.

(∗) and (∗∗) imply that

d(x, G) ≤ d(x, Gn ) < d(x, G) + ε.

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
36 CHAPTER 3. HILBERT SPACES

Hence,
d(x, G) = inf d(x, Gn ).
n∈N

From Gn ³⊂ Gn+1 ,´ it follows that d(x, Gn+1 ) ≤ d(x, Gn ). The sequence of real
numbers d(x, Gn ) , which is decreasing and bounded below, must converge to its
infimum. Thus
lim d(x, Gn ) = d(x, G).
n→∞

(b) Let an = PGn (x), a = PG (x). Then by part (a) we have

lim kx − an k = lim d(x, Gn ) = d(x, G).


n→∞ n→∞

Since Gn ⊂ G, ∀n ∈ N, then (an ) ⊂ G and so an → PG (x) in norm, that is,

lim kan − ak = 0,
n→∞

which gives that


lim PGn (x) = PG (x). ¥
n→∞

Problem 33.
Let H be a Hilbert space.
(a) Prove that for any two subspaces M, N of H we have

(M + N )⊥ = M ⊥ ∩ N ⊥ .

(b) Prove that for any two closed subspaces E, F of H we have

(E ∩ F )⊥ = E ⊥ + F ⊥ .

Solution.
(a) If x ∈ (M + N )⊥ , then for every m ∈ M and n ∈ N we have

hm + n, xi = 0

since m + n ∈ M + N . For n = 0 we have hm, xi = 0. This holds for all m ∈ M , so


x ∈ M ⊥ . Similarly x ∈ N ⊥ . Thus x ∈ M ⊥ ∩ N ⊥ , and hence (M + N )⊥ ⊂ M ⊥ ∩ N ⊥ .
If x ∈ M ⊥ ∩ N ⊥ , then we have

hm, xi = 0 and hn, xi = 0, ∀m ∈ M, n ∈ N.

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
3.1. HILBERT SPACES 37

Hence
hm + n, xi = 0.
This means that x ∈ (M + N )⊥ . Hence M ⊥ ∩ N ⊥ ⊂ (M + N )⊥ .
(b) From part (a) it follows that
M + N = (M ⊥ ∩ N ⊥ )⊥ .
Setting E ⊥ in the place of M and F ⊥ in the place of N , we obtain
E ⊥ + F ⊥ = (E ∩ F )⊥ . ¥
Why in question (b) E and F must be closed?

Problem 34.
A systemPn {xi }i∈N in a normed space X is called a complete system if
Span { i=1 αi xi : ∀n ∈ N, αi ∈ F} is dense on X.
If {xi }i∈N is a complete system in a Hilbert space H and x⊥xi for every i, show
that x = 0.

Solution. P P
Given x ∈ X, if x⊥xi for every i, then x⊥ Span { ni=1 αi xi }. Let D := Span { ni=1 αi xi }.
By definition, D = X. Then there exists a sequence (xn ) in D converging to x and
x⊥xn for every n. Hence
0 = hx, xn i → hx, xi = kxk2 as n → ∞.
Thus x = 0. ¥

Problem 35.
Let H be a Hilbert space and {ϕi }∞
i=1 an orthonormal system in H. Show that

kϕn − ϕm k = 2 for m 6= n.

Solution.
Using orthogonality and the fact that kϕk k = 1, ∀k ∈ N, we get for n 6= m,
kϕn − ϕm k2 = hϕn − ϕm , ϕn − ϕm i
= hϕn , ϕn i − hϕn , ϕm i − hϕm , ϕn i + hϕm , ϕm i
= kϕn k2 + kϕm k2 = 2. ¥

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
38 CHAPTER 3. HILBERT SPACES

Problem 36.
Let {ei }i∈N be an orthonormal set in a Hilbert space H. Prove that

X
|hx, ei ihy, ei i| ≤ kxk kyk, ∀x, y ∈ H.
i=1

Solution.
Using the Cauchy-Schwarz inequality for `2 and the Bessel inequality for H, we have


à ∞
! 12 Ã ∞ ! 21
X X X
|hx, ei ihy, ei i| ≤ |hx, ei i|2 |hx, ei i|2
i=1 i=1 i=1
¡ ¢1 ¡ ¢1
≤ kxk 2 kyk2 2 = kxk kyk.
2

Problem 37.
Let H be a Hilbert space and A and B be any two subsets of H. Show that
(a) A⊥ is a closed subspace of H.
(b) A ⊂ (A⊥ )⊥ := A⊥⊥ .
(c) A ⊂ B ⇒ B ⊥ ⊂ A⊥ .

(d) A⊥ = A = A⊥ .

Solution.
(a) Let x, y ∈ A⊥ and α, β ∈ F. Then for any a ∈ A,

hαx + βy, ai = hαx, ai + hβy, ai


= αhx, ai + βhy, ai = 0.

Hence αx + βy ∈ A⊥ , so A⊥ is a subspace of H.
We show now that A⊥ is closed. Suppose that the sequence (xn ) in A⊥ converges
to some x ∈ H. Since hxn , ai = 0 for all a ∈ A, by continuity of the inner product,
we have
hx, ai = h lim xn , ai = lim hxn , ai = 0.
n→∞ n→∞

So x ∈ A⊥ , and A⊥ is closed.

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
3.1. HILBERT SPACES 39

(b) For any x ∈ A, we have x⊥A⊥ . This implies that x ∈ (A⊥ )⊥ . Thus

A ⊂ (A⊥ )⊥ := A⊥⊥ .

(c) Take any x ∈ B ⊥ . Then x⊥y for all y ∈ B. But A ⊂ B, so x⊥y for all y ∈ A.
Hence x ∈ A⊥ .
(d) From (a) we have
A⊥ = A⊥ ,
which is the second equality in (d). Now pick any x⊥A, that is, x ∈ A⊥ . If a ∈ A,
then there exists a sequence (an ) is A such that an → a. Since x⊥A, we have x⊥an
for all n. Hence
hx, ai = hx, lim an i = lim hx, an i = 0.
n→∞ n→∞

Thus we have x⊥A, so x⊥A. Therefore,



A⊥ ⊂ A . (i)

On the other hand, by (b) we have



A ⊂ A =⇒ A ⊂ A⊥ . (ii)

(i) and (ii) complete the proof. ¥

Problem 38. © ª
(a) Show that M := x = (xn ) ∈ `2 : x2n = 0, ∀n ∈ N is a closed subspace of
the Hilbert space `2 .
(b) Find M ⊥ .

Solution.
(a) Take any x = (xn ), y = (yn ) ∈ M . It is clear that for any scalars α, β,

(αx + βy)2n = αx2n + βy2n = 0.

That gives that αx + βy ∈ M . Hence M is a linear subspace of `2 .


(k)
Let us prove that it is closed. Take x ∈ M . There exists a sequence x(k) = (xn ) ∈ M
(k)
converging to x as k → ∞. Since x2n = 0, we obtain
(k)
x2n = lim x2n = 0,
k→∞

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
40 CHAPTER 3. HILBERT SPACES

that is, x ∈ M . Hence M is closed.


(b) Now

z ∈ M ⊥ ⇐⇒ hz, xi = 0, ∀x ∈ M
X∞ ∞
X
⇐⇒ z2n+1 x2n+1 = 0 for all scalars x2n+1 such that |x2n+1 |2 < ∞
n=0 n=0
⇐⇒ z2n+1 = 0, ∀n = 0, 1, 2, ...

Therefore

M ⊥ = {z = (zn ) ∈ `2 : z2n+1 = 0, ∀n = 0, 1, 2, ...}. ¥

Problem 39.
Let V be a subspace of a Hilbert space.

(a) Show that V = V ⊥ .
(b) Show that V is dense in H if and only if V ⊥ = {0}.

Solution.
(a) From Problem 37d we get (a).
(b) If V is dense, then V = H. Hence

V⊥ =V = H ⊥ = {0}.

Conversely, suppose V ⊥ = {0}. If V is not dense in H, that is, V ( H, pick


x ∈ H \ V . Let x0 = PV x. Then

x − x0 ∈ V = V ⊥ = {0}.

Thus x = x0 ∈ V . This is a contradiction. Thus V is dense in H. ¥

3.2 Weak convergence

Problem 40.
Prove that in any finite dimensional vector space, strong convergence and weak
convergence are equivalent.

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
3.2. WEAK CONVERGENCE 41

Solution.
Consider first the case that X = Fd under the Euclidian norm k.k2 . Suppose that
the sequence (xn ) converges weakly to x in Fd . Then for each standard basis vector
ek , k = 1, 1, .., d, we have

hxn , ek i → hx, ek i as n → ∞.

That is, weak convergence implies componentwise convergence. But since there are
only finitely many components, this implies norm convergence, since
d
X
kxn − xk22 = |hxn , ek i − hx, ek i|2 → 0 as n → ∞.
k=1

For the general case, choose any basis {e1 , e2 , ..., ed } for X, and then use the fact that
all norms on X are equivalent to define an isomorphism between X and Fd . ¥

Problem 41.
Show that if the sequence (xn ) in a normed space X is weakly convergent to
x0 ∈ X, then
lim inf kxn k ≥ kx0 k.
n→∞

Solution.
If x0 = 0 then kx0 k = 0 and the statement is obviously true. Now assume kx0 k 6= 0.
By a well known theorem 1 , there is some f ∈ X ∗ such that

kf k = 1, f (x0 ) = kx0 k.

Since (xn ) converges weakly to x0 and f is continuous, we have

lim f (xn ) = f (x0 ) = kx0 k.


n→∞

But
f (xn ) ≤ |f (xn )| ≤ kf k kxn k = kxn k.
1
(Kreyszig, p 223) Let X be a normed space and x0 6= 0 be an element in X. Then there exists
a bounded linear functional f ∈ X ∗ such that

kf k = 1, f (x0 ) = kx0 k.

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
42 CHAPTER 3. HILBERT SPACES

Hence,
lim inf kxn k ≥ lim f (xn ) = kx0 k. ¥
n→∞ n→∞

Note.
If X = H is a Hilbert space, using the definition of weak convergence we can have
different solution.
kx0 k2 = hx0 , x0 i = lim hx0 , xn i.
n→∞

Since hx, xn i ≤ kx0 k kxn k, so we have

kx0 k2 = lim hx, xn i ≤ kx0 | lim inf kxn k.


n→∞ n→∞

Thus
kx0 k ≤ lim inf kxn k.
n→∞

Problem 42.
Let X and Y be normed spaces, T ∈ B(X, Y ) and (xn ) a sequence in X. Show
w w
that if xn −
→ x, then T xn −
→ Tx

Solution.

Recall: Definition of weak convergence in a normed space:


w
→ x ⇐⇒ f (xn ) → f (x), ∀f ∈ X ∗ .
xn −

We must show that


ϕ(T xn ) → ϕ(T x), ∀ϕ ∈ Y ∗ .
That is,
(ϕ ◦ T )xn → (ϕ ◦ T )x, ∀ϕ ∈ Y ∗ .
w
But ϕ ◦ T ∈ X ∗ , so our hypothesis xn −
→ x guarantees our desired conclusion. ¥

Problem 43.
w
Let H be a Hilbert space and (xn ) be a sequence in H. Suppose xn −
→ x. Show
that
lim kxn − xk = 0 ⇐⇒ kxk ≥ lim sup kxn k (1).
n→∞ n→∞

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
3.2. WEAK CONVERGENCE 43

Solution.
By Problem 41 we have
kxk ≤ lim inf kxn k.
n→∞

So the right hand side of (1) is equivalent to kxk = limn→∞ kxn k. Now note that

(i) kxn − xk2 = kxn k2 − 2Rehxn , xi + kxk2


¯ ¯
(ii) ¯kxn k − kxk¯ ≤ kxn − xk.

If limn→∞ kxn − xk = 0 then by (ii) we get

lim kxn k = kxk.


n→∞

If limn→∞ kxn k = kxk then (i) give that

lim kxn − xk = 0. ¥.
n→∞

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
44 CHAPTER 3. HILBERT SPACES

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D

Chapter 4

Linear Operators - Linear


Functionals

4.1 Linear bounded operators

Problem 44.
Let (X, k.k1 ) and (Y, k.k2 ) be norm spaces, and T ∈ B(X, Y ). We define kT k as

kT k := inf {M : kT xk2 ≤ M kxk1 }.


x∈X

(a) Show that

kT xk2
kT k = sup kT xk2 = sup kT xk2 = sup .
kxk1 =1 kxk1 ≤1 x∈X, x6=0 kxk1

(b) Show that


kT k = sup kT xk2 .
kxk1 <1

Solution.
(a) We have
kT xk2
≤ M, for all x 6= 0, x ∈ X.
kxk1

45
www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
46 CHAPTER 4. LINEAR OPERATORS - LINEAR FUNCTIONALS

By definition,

kT xk2
kT k := inf {M : kT xk2 ≤ M kxk1 } = sup (i).
x∈X x∈X, x6=0 kxk1

x
Now, let y = kxk1
for x ∈ X, x 6= 0. Then y ∈ X and kyk = 1. By (i) we have
° µ ¶°
kT xk2 ° kxk y °
= sup ° ° = sup kT yk2 = sup kT xk2
1
kT k = sup ° T (ii).
x∈X, x6=0 kxk1 kyk=1 kxk1 °2 kyk=1 kxk=1

From (ii) it follows that

kT xk2 kT xk2
kT k = sup kT xk2 ≤ sup kT xk2 ≤ sup ≤ sup = kT k.
kxk1 =1 kxk1 ≤1 x∈X, kxk1 ≤1 kxk1 x∈X, x6=0 kxk1

Thus,
kT k = sup kT xk2 .
kxk1 ≤1

(b) Let
B := {x ∈ X : kxk1 ≤ 1} and B o := {x ∈ X : kxk1 < 1}.
Since kT k = supx∈B kT xk2 , there exists a sequence (xn ) in B such that

kT k = lim kT xn k2 .
n→∞

Consider the sequence (yn ) defined by yn = (1 − 21n )xn . It is clear that yn ∈ B o for
all n ∈ N. Moreover,
° µ ¶ ° µ ¶
° 1 ° 1
° °
lim kT yn k2 = lim °T 1 − n xn ° = lim 1 − n kT xn k2
n→∞ n→∞ 2 n→∞ 2
1
= lim (1 − n ) lim kT xn k2 = kT k.
n→∞ 2 n→∞
Thus
sup kT xk2 ≥ kT k.
x∈B o

On the other hand,


sup kT xk2 ≤ sup kT xk2 = kT k.
x∈B o x∈B

The proof is complete. ¥

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
4.1. LINEAR BOUNDED OPERATORS 47

Problem 46.
Let H = `2 be the well known Hilbert space. Consider the left shift defined by

L : `2 → `2 , x = (x1 , x2 , x3 , ...) 7→ Lx = (x2 , x3 , ...).

(a) Show that L is a linear bounded operator. Find kLk.


(b) Define the right shift and answer the same question as in part (a).

Solution.
We answer only the first part.
• It is easy to check the linearity of L.
• For any x = (x1 , x2 , x3 , ...) ∈ `2 , we have

X
2 2
kLxk = k(x2 , x3 , ...)k = |xn |2
n=2

X
≤ |xn |2 = kxk2 .
n=1

Hence
kLxk ≤ kxk, ∀x ∈ `2 .
This implies that T is bounded, and
kLk ≤ 1. (∗)
On the other hand, consider the sequence e = (0, 1, 0, 0, ...) ∈ `2 . We have
kek = 1 and kLek = k(1, 0, 0, ...)k = 1.
So (see Problem 39)
kLk = sup kLxk ≥ 1. (∗∗)
kxk=1

Combining (∗) and (∗∗) we obtain kLk = 1. ¥

Problem 47.
Let X = C[0, 1] with the max-norm (the uniform norm). We define the integral
operator Z x
K : X → X by Kf (x) = f (y)dy.
0

Show that K is bounded. Find kKk.

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
48 CHAPTER 4. LINEAR OPERATORS - LINEAR FUNCTIONALS

Solution.
The operator K is bounded. Indeed,
Z x Z 1
kKf k ≤ sup |f (y)|dy ≤ |f (y)|dy ≤ kf k.
x∈[0,1] 0 0

Hence kKk ≤ 1.
In fact, kKk = 1, since 1 ∈ X, K1 = x and so kK1k = kxk = 1. ¥

**Problem 48.
Let X = L2 [0, 1] with the norm k.k2 . We define the integral operator
Z x
A : X → X by Af (x) = f (y)dy.
0

Show that A is bounded. Find kAk.

Solution.
Warning: It’s completely different from the previous problem!!!

First solution: Using Cauchy-Schwarz inequality, we get


Z 1 ¯Z t ¯2 Z 1 ¯¯Z t r ¯2
¯
¯ ¯ ¯ πs f (s) ¯
kAf k22 = ¯ f (s)ds¯ dt = ¯ cos . p ds ¯ dt
¯ ¯ ¯ 2 cos πs ¯
0 0 0 0 2
Z 1 µZ t Z t ¶
πs |f (s)|2
≤ cos ds πs ds dt
0 0 2 0 cos 2
Z µZ t ¶
2 1 πt |f (s)|2
= sin . ds dt
π 0 0 2 cos πs 2
Z µZ 1 ¶
2 t πt |f (s)|2
= sin dt ds
π 0 0 2 cos πs 2
Z µZ 1 ¶
2 t πt |f (s)|2
= sin dt ds
π 0 0 2 cos πs
2
µ ¶2 Z t
2 |f (s)|2
= πs ds.
π 0 cos 2

Equality holds for f (s) = cos πs


2
. Thus
2
kAk = . ¥
π

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
4.1. LINEAR BOUNDED OPERATORS 49
p
Second solution: We find the norm of T = A∗ A, then kAk = kT k.
Since A is compact (we will see this somewhere later) and A∗ A is self adjoint, we
have that T is a compact, normal operator. Hence, its spectrum is countable with
0 as the only possible cluster point and kT k is equal to the spectral radius of T .
These two facts together imply that

kT k = max{|λ| : T f = λf for some f ∈ X},

We have that A∗ is given by

Z 1

A f (x) = f (y)dy.
x

Hence

Z 1 Z y
(∗) T f (x) = f (z)dz dy.
x 0

By the Fundamental Theorem of Calculus, T f is twice differentiable for all f ∈ X,


so any eigenvector of T must satisfy the differential equation

Z 1 Z y µ Z x ¶
d2 d2 d
λ 2 f (x) = 2 f (z)dz dy = − f (z)dz = −f (x), λ ∈ C \ {0}.
dx dx x 0 dx 0

By a theorem from differential equations, we know that there are only two linearly
independent solutions to the above differential equation over C, namely, eiωx and
e−iωx , where ω 2 = λ, λ, ω 6= 0. Hence the general solution to the above equation is

f (x) = αeiωx + βe−iωx , α, β ∈ C.

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
50 CHAPTER 4. LINEAR OPERATORS - LINEAR FUNCTIONALS

When we apply T , we get


Z 1 Z y
iωx −iωx
T (αe + βe ) = (αeiωz + βe−iωz )dz dy
Zx 1 µ0 iωz ¶
αe βe−iωz ¯¯y
= − ¯ dy
x iω iω 0
Z 1 ·µ iωy −iωy
¶ µ ¶¸
αe βe α β
= − − − dy
x iω iω iω iω
µ iωy ¶ µ ¶ ¯
αe βe−iωy α β ¯1
= 2
− 2
− − y¯
−ω −ω iω iω x
µ iω −iω
¶ µ ¶
αe βe α β
= 2
− 2
− −
−ω −ω iω iω
·µ iωx −iωx
¶ µ ¶ ¸
αe βe α β
− − − − x
−ω 2 −ω 2 iω iω
µ ¶
1 iωx −iωx α β
= (αe + βe )+ − x
ω2 iω iω
µ iω ¶ µ ¶
αe βe−iω α β
− − − − .
ω2 ω2 iω iω

Now since αeiωx + βe−iωx is an eigenvector, we must have that


µ ¶ µ iω ¶ µ ¶
α β αe βe−iω α β
− x− − − − = 0, ∀x ∈ [0, 1],
iω iω ω2 ω2 iω iω

which implies α = β. Hence

αeiωx + βe−iωx = 2α cos ωx.

Moreover, we must have that T f (1) = 0 for any eigenvector f by (∗), so 2α cos ω = 0.
Since α 6= 0, we must have that

(2n + 1)π
ω= , n ∈ Z.
2
Hence the eigenvalues are of the form

1 22
λn = 2 = , n ∈ Z.
ωn (2n + 1)2 π 2

Thus
4
max{λn : n ∈ Z} = (take n = 0, −1).
π2

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
4.1. LINEAR BOUNDED OPERATORS 51

We conclude that
4 2
kT k = 2
and kAk = . ¥
π π

Problem 49.
Let a, b be real numbers such that a < b. Consider the Hilbert space L2 [a, b] over
R and the operator T : L2 [a, b] → R defined by
Z b
Tf = f (x)dx, f ∈ L2 [a, b].
a

(a) Show that T is bounded. Compute kT k.


(b) According to the Riesz’s Theorem, there exists a function g ∈ L2 [a, b] such
that
T f = hf, gi for all f ∈ L2 [a, b].
Find such a function g and verify that kgkL2 = kT k.

Solution.
(a) By Hölder’s inequality, we have
¯Z b ¯
¯ ¯
|T f | = ¯¯ f (x)dx¯¯
a
Z b
≤ |1.f (x)|dx
a
µZ b ¶1/2 µZ b ¶1/2
2 2
≤ 1 dx |f (x)|
a a

= b − a kf kL2 .
Hence, T is bounded, i.e., T ∈ B(L2 , R) = (L2 )∗ , the dual space of L2 .
From the above we get √
kT k(L2 )∗ ≤ b − a.
1
Now consider the function h(x) = √b−a , x ∈ (a, b). It is obviously that h ∈ L2 (a, b)
and
µZ b ¶1/2 ¯Z b ¯
¯ 1 ¯ √
khkL2 = (b − a)−1
= 1 and |T h| = ¯¯ √ dx¯¯ = b − a.
a a b−a
Hence, √
b − a = |T h| ≤ kT k(L2 )∗ khkL2 = kT k(L2 )∗ .

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
52 CHAPTER 4. LINEAR OPERATORS - LINEAR FUNCTIONALS

Therefore, √
kT k(L2 )∗ = b − a.
(b) By the Riesz’s Theorem, there exists a function g ∈ L2 [a, b] such that T f =
hf, gi for all f ∈ L2 [a, b]. Here, functions are real-valued, so we can write
Z b Z b
T f = hf, gi ⇔ f (x)dx = f (x)g(x)dx.
a a

It is evident that the above equation is satisfied for all f ∈ L2 [a, b] if we choose
g(x) = 1 on [a, b]. By the uniqueness of this representation guaranteed by Riesz’s
Theorem, we can definitely conclude that

g(x) = 1, x ∈ [a, b].

Also, we can verify that


µZ ¶1/2
b
2

kgkL2 = 1 dx = b − a = kT k(L2 )∗ . ¥
a

Problem 50.
Let X, Y be normed spaces and T ∈ B(X, Y ). Consider the following statement:

T is an isometry ⇔ kT k = 1.

Do you agree with it? Why?

Solution.
• The direct way (⇒) is correct. Indeed, suppose that T is an isometry; then

kT kB(X,Y ) = sup kT xkY


x∈X
kxkX =1

= sup kxkX (since kT xkY = kxkX )


x∈X
kxkX =1
= 1.

• The other way (⇐) is false. The left shift on `2 is a counter-example (see Problem
40). ¥

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
4.1. LINEAR BOUNDED OPERATORS 53

Problem 51.
Define T : C[0, 1] → C[0, 1] by
Z t
(T x)(t) = t x(s)ds.
0

(a) Prove that T is a bounded linear operator. Compute kT k.


(b) Prove that the inverse T −1 : Image(T ) → C[0, 1] exists but not bounded.

Solution.
(a) For all x1 , x2 ∈ C[0, 1] and all α1 , α2 ∈ R and all t ∈ [0, 1] we have
Z t
T (α1 x1 + α2 x2 )(t) = t (α1 x1 + α2 x2 )(s)ds
0
Z t Z t
= α1 t x1 (s)ds + α2 t x2 (s)ds
0 0
= α1 T (x1 )(t) + α2 T (x2 )(t)
= [α1 T (x1 ) + α2 T (x2 )] (t).

Hence,
T (α1 x1 + α2 x2 ) = α1 T (x1 ) + α2 T (x2 ).
This shows that T is linear.
For each x ∈ C[0, 1] we have
¯ Z t ¯ Z t
¯ ¯
kT xk = max ¯¯t x(s)ds¯¯ ≤ max |t| |x(s)|ds
t∈[0,1] 0 t∈[0,1] 0
Z t
≤ max t kxkds = max t2 kxk = kxk.
t∈[0,1] 0 t∈[0,1]

Hence T is a bounded
Rt with kT k ≤ 1. Moreover, if x(t) = 1, t ∈ [0, 1] then kxk = 1
and (T x)(t) = t 0 ds = t2 , therefore, kT xk = maxt∈[0,1] t2 = 1. Thus,

1 = kT xk ≤ kT kkxk and so kT k ≥ 1.

Hence we have proved that kT k ≤ 1 and kT k ≥ 1, so kT k = 1.


(b) Suppose that x ∈ C[0, 1] satisfies
Z t
T x = 0, i.e., t x(s)ds = 0, ∀t ∈ [0, 1].
0

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
54 CHAPTER 4. LINEAR OPERATORS - LINEAR FUNCTIONALS
Rt
It follows that 0 x(s)ds = 0, ∀t ∈ (0, 1]. By differentiation w.r.t. t we get x(t) =
0, ∀t ∈ (0, 1]. Since x(t) is continuous, we then also have x(0) = 0. Hence, x = 0.
We have thus proved

∀x ∈ C[0, 1], T (x) = 0 ⇒ x = 0.

Hence, T −1 exists.
Now we show that T −1 is not bounded. Given any n ∈ N, we let xn (t) = tn . Then

xn ∈ C[0, 1] and kxn k = max |tn | = 1.


t∈[0,1]

We let Z t
1
yn (t) = T xn (t) = t sn ds = tn+2 .
0 n+1
Then ¯ ¯
¯ 1 ¯ 1
kyn k = max ¯¯ t ¯¯ =
n+2
.
t∈[0,1] n + 1 n+1
Also by construction, yn Image(T ) and T −1 yn = xn ; thus

kT −1 yn k = kxn k = 1.

This shows that T −1 cannot be bounded. (For if T −1 were bounded, then we would
have
1
kT −1 yn k ≤ kT −1 kkyn k, i.e., 1 ≤ kT −1 k , ∀n ∈ N.
n+1
This is impossible. ¥

Problem 53.
Let 1 < p < ∞ and q be its exponent conjugate. For f ∈ Lp (0, ∞), let
Z Z 1
1 x
(T f )(x) = f (t)dt = f (tx)dt.
x 0 0

Show that
(a) T f is well-defined on (0, ∞), and that T f is continuous with respect to x.
(b) T f ∈ Lp (0, ∞).
(c) T is a bounded linear operator from Lp (0, ∞) to itself. Calculate its norm.
(Hint: Use fn = x−1/p χ{1≤x≤n} .)

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
4.1. LINEAR BOUNDED OPERATORS 55

Solution. Rx
(a) For f ∈ Lp (0, ∞), we need to show ψ : x 7→ 0 f (t)dt is well-defined and
continuous w.r.t. x. By Holder’s inequality we have
Z x
|f (t)|dt ≤ |x|1/q kf kp < ∞,
0

which shows that ψ is well-defined. Also, for x, y ∈ (0, ∞), we have


¯Z y ¯ Z y
¯ ¯
ψ(x) − ψ(y)| = ¯¯ f (t)dt¯¯ ≤ |f (t)|dt ≤ |x − y|1/q kf kp .
x x

This shows that ψ is continuous.


Since (T f )(x) = ψ(x)
x
, T f is also continuous on (0, ∞). Moreover we have

|(T f )(x)| ≤ x−1/p kf kp → 0 as x → ∞.

So (T f )(x) → 0 as x → ∞.
(b) To prove this part (b) we will use the following theorem:
Given a σ-finite measure space (X, RA, µ). Let 1 < p < ∞ and q be its exponent
conjugate. If |ϕ| < ∞ a.e on X and if X ϕψdµ exists in C for every ψ ∈ Lq (X),
then f ∈ Lp (X).
Now, for any g ∈ Lq (0, ∞), by Fubini’s theorem we have
Z Z Z 1
|(T f )(x)g(x)|dx ≤ |(T f )(x)||g(x)|dtdx
(0,∞) (0,∞) 0
Z Z 1
1
= |f (x)||g(x/t)|dtdx
(0,∞) 0 t
Z 1
dt
≤ kf kp kg(./t)kq .
0 t
But µZ ¶1/q
q
kg(./t)kq = |g(x/t)| dx = t1/q kgkq .
(0,∞)

Thus,
Z Z 1
|(T f )(x)g(x)|dx ≤ kf kp kgkq t1/q−1 dt = qkf kp kgkq < ∞.
(0,∞) 0

Hence, T f ∈ Lp (0, ∞).


(c) By the last inequality we have

kT f kp ≤ qkf kp .

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
56 CHAPTER 4. LINEAR OPERATORS - LINEAR FUNCTIONALS

This shows that T is an bounded operator on Lp (0, ∞) with kT kp,p ≤ q. We would


like to establish the equality. Consider the function fn , n ∈ N defined by fn =
x−1/p χ{1≤x≤n} . We have
Z n ¡ −1/p ¢p
kfn kpp = x dx = ln n.
1

For x > 1 we have


Z min{x,n}
1 min{x, n}1/q − 1
(T fn )(x) = t−1/p dt = q .
x 1 x

For 0 < ε < 1, since (1 − ε)p ≥ 1 − pε we have


Z µZ n Z ∞ ¶
p p 1 −1/q p
[(T fn )(x)] dx = q (1 − x ) dx + (n1/q − 1)p −p
x dx
(0,∞) 1 x n
µ ¶
p p 1
≥ q ln n − +
q p−1
µ ¶
p p p 1
≥ q kfn kp − + .
q p−1

This implies that


kT fn kp
lim ≥ q.
n→∞ kfn kp

Hence, kT kp,p ≥ q. Finally, we get

kT kp,p = q. ¥

Problem 54.
Let (cj )∞ 2
j=1 be a sequence of complex numbers. Define an operator D on ` by

Dx = (c1 x1 , c2 x2 , ...) for x = (x1 , x2 , ...) ∈ `2 .

Prove that D is bounded if and only if (cj )∞


j=1 is bounded, and in this case kDk =
supj |cj |.

Solution.

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
4.1. LINEAR BOUNDED OPERATORS 57

• Suppose (cj )∞
j=1 is bounded. Let M = supj |cj | < ∞. Then
Ã∞ !1/2 à ∞ !1/2
X X
kDxk = |cj xj |2 = |cj |2 |xj |2
j=1 j=1
Ã∞ !1/2 à ∞
!1/2
X X
≤ M 2 |xj |2 =M |xj |2
j=1 j=1
= M kxk.
Hence, D is bounded and kDk ≤ M .
• Suppose D is bounded. We want to show that (cj )∞ j=1 is bounded. Consider the
vector ej = (0, 0, .., 0, 1, 0, ...) where the number 1 appears at the j-th coordinate.
Clearly kej k = 1 and kDej k = |cj | for all j = 1, 2, .. Since D is bounded,
|cj | = kDej k ≤ kDk for any j = 1, 2, ...
Hence (cj )∞
j=1 is bounded and M = supj |cj | ≤ kDk. Finally

kDk = M. ¥

Problem 55.
Prove that B(F, Y ) is not a Banach space if Y is not complete.

Hint: Take a Cauchy sequence (yn ) in Y which does not converge and consider the sequence
of operators (Bn ) defined by :
Bn λ := λyn ; λ ∈ F.

Solution.
We follow the suggestion above. It is easy to see that
Bn ∈ B(F, Y ) and kBn k = kyn k, ∀n ∈ N.
Since (Bn − Bm )λ = λ(yn − ym ), we have
kBn − Bm k = kyn − ym k, ∀n ∈ N.
Therefore (Bn ) is a Cauchy sequence in B(F, Y ). Suppose there exists B ∈ B(F, Y )
such that kBn − Bk → 0 as n → ∞. Let y := B1 ∈ Y , where 1 is the unit element
in F. Then
kyn − yk = kBn 1 − B1k ≤ kBn − Bk → 0 as n → ∞,

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
58 CHAPTER 4. LINEAR OPERATORS - LINEAR FUNCTIONALS

i.e., the sequence (yn ) converges to y. This contradiction proves that (Bn ) cannot
be convergent. Hence, B(F, Y ) is not a Banach space. ¥

Problem 56.
Let T1 , T2 , ... be the following bounded linear operators `1 → `∞ :

T1 (x1 , x2 , x3 , ...) = (x1 , x1 , x1 , x1 , ...)


T2 (x1 , x2 , x3 , ...) = (x1 , x2 , x2 , x2 , ...)
T3 (x1 , x2 , x3 , ...) = (x1 , x2 , x3 , x3 , ...)...etc

Prove that the sequence (Tn ) is strongly operator convergent. Also prove that
(Tn ) is not uniformly operator convergent.

Solution.
Let T : `1 → `∞ be the bounded linear operator given by

T (x1 , x2 , x3 , ...) = (x1 , x2 , x3 , ...).

T is obviously linear. It is bounded with norm kT k ≤ 1. Indeed, if x = (x1 , x2 , x3 , ...) ∈


`1 then
X∞
|xk | = kxk1 < ∞, so |xk | ≤ kxk1 , ∀k = 1, 2, ...
k=1

Hence,
kxk∞ ≤ kxk1 .
We claim that (Tn ) strongly operator converges to T .
For any x = (x1 , x2 , x3 , ...) ∈ `1 ,

kTn x − T xk∞ = k(0, ..0, xn − xn+1 , xn − xn+2 , ...)k∞


= sup |xn − xn+j |.
j≥1

P∞
Since x = (x1 , x2 , x3 , ...) ∈ `1 , k=1 |xk | < ∞, so we have limk→∞ |xk | = 0. Hence,
given any ε > 0, there is some K ∈ N such that

|xk | ≤ ε, for all k ≥ K.

Then if n ≥ K we have
n + j ≥ K, ∀j ∈ N.

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
4.1. LINEAR BOUNDED OPERATORS 59

Hence,
|xn − xn+j | ≤ |xn | + |xn+j | ≤ 2ε.

Thus, for n ≥ K,
kTn x − T xk∞ = sup |xn − xn+j | ≤ 2ε.
j≥1

In other words,
lim kTn x − T xk∞ = 0.
n→∞

This is true for every x ∈ `1 . Hence, the sequence (Tn ) is strongly operator conver-
gent to T .
It follows from this that if (Tn ) would be uniformly operator convergent, then the
limit must be equal to T . We show that this is not true. Now we have

(Tn − T )(x) = (0, .., 0, xn − xn+1 , xn − xn+2 , ...).

In particular, if x = (0, .., 0, 1, 0, ..) with 1 at the n-th position, then

(Tn − T )(x) = (0, .., 0, 1, 1, ...).

Here kxk1 = 1 and k(0, .., 0, 1, 1, ...)k∞ = 1. Hence,

kTn − T k ≥ 1, for all n.

Thus, (Tn ) is not uniformly operator convergent. ¥

Problem 57.
Let H1 and H2 be two Hilbert spaces. Let {a1 , ..., an } be an orthonormal system of
H1 and {b1 , ...,n } be an orthonormal system of H2 , and λ1 , ..., λn ∈ K. Consider
the operator
X
U : H1 → H2 defined by U (x) = λi bi hx, ai i.
i=1

Calculate kU k.

Solution.

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
60 CHAPTER 4. LINEAR OPERATORS - LINEAR FUNCTIONALS

From the Pythagoras theorem and the Bessel inequality we have


n
X
kU (x)k2 = |λi |2 kbi k2 |hx, ai i|2
i=1
n
X
= |λi |2 |hx, ai i|2
i=1
n
X
2
≤ M |hx, ai i|2 ≤ M 2 kxk2 ,
i=1

where M = max1≤i≤n |λi |. Hence, we have that


kU (x)k ≤ M kxk, ∀x ∈ H.
Therefore, kU k ≤ max1≤i≤n |λi |.
On the other hand, we have
kU (ai )k ≤ kU k kai k = kU k,
U (ai ) = λi bi , ∀i ∈ {1, ..., n}.
It follows that |λi | ≤ kU k, ∀i ∈ {1, ..., n}. This implies max1≤i≤n |λi | ≤ kU k. Thus,
kU k = max |λi |. ¥
1≤i≤n

Problem 58.
(a) Let X be a Hilbert space. Let a, b ∈ H be two non-zero orthogonal elements.
Consider the operator

U : H → H, U (x) = ahx, bi + bhx, ai.

Calculate kU k.
(b) Consider the operator T : L2 [0, π] → L2 [0, π] defined by
Z π Z π
(T f )(x) = sin x f (t) cos tdt + cos x f (t) sin tdt.
0 0

Calculate kT k.

Solution.
(a) Note that ahx, bi and bhx, ai are two orthogonal vectors. Using the Pythagoras
theorem we have
kU (x)k2 = kak2 |hx, bi|2 + kbk2 |hx, ai|2 .

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
4.1. LINEAR BOUNDED OPERATORS 61

From the Bessel inequality we deduce that


ï¿ À¯2 ¯¿ À¯2 !
¯ b ¯ ¯ a ¯
kU (x)k2 = kak2 kbk2 ¯¯ x, ¯ + ¯ x, ¯ ≤ kak2 kbk2 kxk2 .
kbk ¯ ¯ kak ¯

Hence, kU k ≤ kak kbk.


But
U (a) = aha, bi + bha, ai = kak2 b.
Therefore,
kU (a)k = kak2 kbk ≤ kU k kak.
Hence, kU k ≥ kak kbk. Thus,

kU k = kak kbk.

(b) Let H = L2 [0, π]. Then H, with the usual inner product, is a Hilbert space.
The two vectors a = sin x and b = cos x are orthogonal. Indeed,
Z π
ha, bi = hsin x, cos xi = sin x cos xdx = 0.
0

By (a) we get
kT k = k sin xk k cos xk.
But Z π Z π
2 π π
k sin xk = 2
(sin x) dx = ; k cos xk2 = (cos x)2 dx = .
0 2 0 2
Thus,
π
kT k = . ¥
2

Problem 59.
(a) Let H be a Hilbert space and {e1 , e2 } ⊂ H an orthonormal system. Let
µ ¶
a b
A=
c d

be scalar square matrix. Consider the operators U, V : H → H defined by

U (x) = ahx, e1 ie1 + bhx, e2 ie2 ,


V (x) = chx, e1 ie1 + dhx, e2 ie2 .

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
62 CHAPTER 4. LINEAR OPERATORS - LINEAR FUNCTIONALS

Prove that
kU + V k2 + kU − V k2 = 2(kU k2 + kV k2 )
if and only if

(max{|a+c|,|b+d|})2 +(max{|a−c|,|b−d|})2 =2(max{|a|,|b|}2 +max{|c|,|d|}2 ).

(b) Prove that if dim H ≥ 2 then B(H) is not a Hilbert space

Solution.
(a) Using Problem 57 we have

kU k = max{|a|, |b|}; kV k = max{|c|, |d|}.


kU + V k = max{|a + c|, |b + d|}; kU − V k = max{|a − c|, |b − d|}.

From here we obtain the statement.


(b) Since dim H ≥ 2 we can find x, y ∈ H, two linearly independent vectors. Using
the Gram-Schmidt procedure on {x, y} we can construct an orthonormal system
{e1 , e2 } ⊂ H. Now construct U, V using the matrix
µ ¶
2 1
A= .
0 2

That is U, V : H → H defined by

U (x) = 2hx, e1 ie1 + hx, e2 ie2 , V (x) = 2hx, e2 ie2 .

If B(H) were a Hilbert space, then by the parallelogram law we must have

kU + V k2 + kU − V k2 = 2(kU k2 + kV k2 ),

i,e., by (a)
¡ ¢
(max{2, 3|})2 + (max{2, 1})2 = 2 max{2, 1}2 + max{0, 2}2 .

This is not true. ¥

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
4.2. LINEAR FUNCTIONALS 63

4.2 Linear Functionals


Review:

Definition 1 Let X be a linear space. A map p : X → R is called sublinear if it is subadditive


and positive homogeneous, i.e.,

p(x + y) ≤ p(x) + p(y), ∀x, y ∈ X.


p(λx) = λp(x), ∀x ∈ X, ∀λ ≥ 0.

Definition 2 Let X be a linear space. A map p : X → K is called seminorm if it is subadditive


and absolutely homogeneous, i.e.,

p(x + y) ≤ p(x) + p(y), ∀x, y ∈ X.


p(λx) = |λ|p(x), ∀x ∈ X, ∀λ ∈ K.

The Hahn-Banach theorems

Theorem 1 (The case K = R). Let X be a real linear space, G ⊂ X a linear subspace, p : X → R
a sublinear functional and f : G → R a linear functional such that f (x) ≤ p(x), ∀x ∈ G. Then
there is a linear functional f¯ : X → R which extends f , i.e.,

f¯(x) = f (x), ∀x ∈ G and f¯(x) ≤ p(x), ∀x ∈ X.

Theorem 2 (The case K = R or C). Let X be a linear space, G ⊂ X a linear subspace,


p : X → R a seminorm and f : G → K a linear functional such that |f (x)| ≤ p(x), ∀x ∈ G. Then
there is a linear functional f¯ : X → K which extends f such that

|f¯(x)| ≤ p(x), ∀x ∈ X.

Theorem 3 (The normed space case). Let X be a linear space, G ⊂ X a linear subspace and
f : G → K a linear and continuous functional. Then there is f¯ : X → K, a linear and continuous
functional which extends f such that kf¯k = kf k. Such an f¯ is called a Hahn-Banach extension for
f.

Theorem 4 Let X be a normed space, x0 ∈ X and G ⊂ X a linear subspace such that δ =


d(x0 , G) > 0. Then there is f : X → K, a linear and continuous functional, such that
1
f = 0 on G, f (x0 ) = 1 and kf k = .
δ

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
64 CHAPTER 4. LINEAR OPERATORS - LINEAR FUNCTIONALS

Three classic problems.

Problem 60.
The dual space of `1 is `∞ , that is, (`1 )∗ = `∞ .

Solution.
Let (ek ) be the standard basis for `1 , where ek = δkj . Every x ∈ `1 has unique
representation

X
x= ξk ek .
k=1
1 ∞
Norms on ` and on ` are respectively:

X
kxk1 = |ξk |, kxk∞ = sup |ξk |.
k∈N
k=1

Take f ∈ (`1 )∗ , that is, f ∈ B(`1 , F). For x ∈ `1 ,


Ã∞ ! ∞
X X
f (x) = f ξk ek = ξk γk where γk = f (ek ).
k=1 k=1

Then for all k ∈ N,


|γk | = |f (ek )| ≤ kf k kek k = kf k.
So,
(∗) sup |γk | ≤ kf k.
k∈N

On the other hand, for every b = (βk ) ∈ `∞ we can obtain a corresponding bounded
linear functional g on `1 . In fact, we may define g on `1 by

X
g(x) = ξk βk where x = (ξk ) ∈ `1 .
k=1

Then g : `1 → F is linear, and the boundedness follows from



X ∞
X
|g(x)| ≤ |ξk βk | ≤ sup |βk | |ξk | = sup |βk | kxk1 .
k∈N k∈N
k=1 k=1

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
4.2. LINEAR FUNCTIONALS 65

Hence g ∈ (`1 )∗ .
We finally show that the norm of f is the norm on `∞ . We have
¯ ¯
¯X∞ ¯ ∞
X
¯ ¯
|f (x)| = ¯ ξk γk ¯ ≤ sup |γk | |ξk | = kxk sup |γk |.
¯ ¯ k∈N k∈N
k=1 k=1

Taking the supremum over x such that kxk = 1, we see that

kf k ≤ sup |γk | = k(γk )k∞ .


k∈N

From(∗) and this, we obtain


kf k = k(γk )k∞ .
Thus there is an isometric isomorphism between (`1 )∗ and `∞ , so that we can write
(`1 )∗ = `∞ . ¥

Problem 61.
1 1
The dual space of `p is `q , that is, (`p )∗ = `q , here 1 < p < ∞ and p
+ q
= 1.

Solution.
The basis for `p is (ek ), where ek = δkj as in the previous problem. Every x ∈ `p
has unique representation
X∞
x= ξk ek .
k=1

Take f ∈ (` ) , that is, f ∈ B(` , F). For x ∈ `p ,


p ∗ p


X
(1) f (x) = ξk γk where γk = f (ek ).
k=1

(n)
Let q be the conjugate of p. Consider xn = (ξk ) with
( q
|γk |
(n) γk
if 1 ≤ k ≤ n and γk 6= 0
ξk =
0 if k > n or γk = 0.

By substituting this into (1) we obtain



X n
X
(n)
f (xn ) = ξ k γk = |γk |q .
k=1 k=1

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
66 CHAPTER 4. LINEAR OPERATORS - LINEAR FUNCTIONALS

We also have
à n ! p1
X (n)
f (xn ) ≤ kf k kxn k = kf k |ξk |p
k=1
à n ! p1
X
= kf k |γk |(q−1)p
k=1
à n ! p1
X
= kf k |γk |q .
k=1

Together,
n
à n ! p1
X X
f (xn ) = |γk |q ≤ kf k |γk |q .
k=1 k=1
1 1
Dividing the last factor and using 1 − p
= q
, we get
à n !1− p1 à n ! 1q
X X
|γk |q = |γk |q ≤ kf k.
k=1 k=1

Since n is arbitrary, letting n → ∞, we obtain


̰ ! 1q
X
(2) |γk |q ≤ kf k.
k=1

Conversely, for any b = (βk ) ∈ `q we can get a corresponding bounded linear func-
tional g on `p . In fact, we may define g on `p by setting

X
g(x) = ξk βk where x = (ξk ) ∈ `p .
k=1

Then g is linear. The boundedness follows from the Hölder inequality. Hence g ∈
(`p )∗ . We finally prove that the norm of f is the norm of (γk ) in `q . From (1) and
the Hölder inequality we have
¯ ¯ Ã∞ ! p1 à ∞ ! 1q
¯X ∞ ¯ X X
¯ ¯
|f (x)| ≤ ¯ ξk γk ¯ ≤ |ξk |p |γk |q
¯ ¯
k=1 k=1 k=1
̰ ! 1q
X
= kxk |γk |q .
k=1

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
4.2. LINEAR FUNCTIONALS 67

Hence ̰ ! 1q
X
kf k ≤ |γk |q .
k=1

From (2) we obtain


̰ ! 1q
X
kf k = |γk |q .
k=1

This can be written


kf k = k(γk )kq where γk = f (ek ).
Thus there is an isometric isomorphism between (`p )∗ and `q . ¥

Problem 62.
The dual space of c0 is `1 , that is, (c0 )∗ = `1 .

Solution.
Recall that c0 ⊂ `∞ , and if x = (λn ) ∈ c0 the norm of x in c0 is kxk = supn∈N |λn |.
Let x = (λn ) ∈ c0 and (ek ), where ek = δkj , be the basis for `∞ as in preceding
examples. Then x has unique representation

X
x= λk ek , and lim λk = 0.
k→∞
k=1

Consider any f ∈ (c0 )∗ . Since f is linear,



X
f (x) = λk γk where γk = f (ek ).
k=1

(0)
For a given N ∈ N, take a special sequence xN0 = (λk ) ∈ c0 where
(
γk
(0) if 1 ≤ k ≤ N and γk 6= 0
λk = |γk |
0 if k > N or γk = 0.

Note that kxN


0 k = 1, and we have

N
X
|f (xN
0 )| = |f (ek )| ≤ kf k kxN
0 k = kf k < ∞.
k=1

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
68 CHAPTER 4. LINEAR OPERATORS - LINEAR FUNCTIONALS

Since N is arbitrary,

X ∞
X
|f (ek )| = |γk | = kf k < ∞.
k=1 k=1

This shows that (γk ) ∈ `1 and

(i) k(γk )k ≤ kf k.

On the other hand,


¯ ¯
¯X∞ ¯ ∞
X
¯ ¯
|f (x)| = ¯ λk γk ¯ ≤ sup |λk | |γk | = kxk k(γk )k < ∞.
¯ ¯ k∈N
k=1 k=1

It follows that
(ii) sup |f (x)| = kf k ≤ k(γk )k.
kxk=1

From (i) and (ii) we obtain kf k = k(γk )k.


Now given (ξk ) ∈ `1 , we want to construct a linear bounded functional g on c0 . Let
x = (λk ) ∈ c0 . Consider the function g : c0 → F defined by

X
g(x) = λk ξk .
k=1

It is clear that g is linear. Its boundedness follows from



X ∞
X
|g(x)| ≤ sup |λk | |ξk | = kxk |ξk | < ∞.
k∈N
k=1 k=1

Hence g ∈ (c0 )∗ . Thus there is an isometric isomorphism between (c0 )∗ and `1 , so


we can write (c0 )∗ = `1 . ¥

∗ ∗ ∗∗

Problem 63.
Let X be a normed space and f, g are nonzero linear functionals on X. Show
that
ker(f ) = ker(g) ⇐⇒ f = cg for some nonzero scalar c.

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
4.2. LINEAR FUNCTIONALS 69

Solution.
The reverse way (⇐) is trivial.
We show the direct way. Suppose ker(f ) = ker(g). Since f 6= 0, there exists some
x0 ∈ X such that f (x0 ) 6= 0, and by rescaling, we can assume that f (x0 ) = 1. Since
x0 ∈
/ ker(f ) = ker(g), we have g(x0 ) 6= 0. Given any y ∈ X, we have
¡ ¢
f y − f (y)x0 = f (y) − f (y)f (x0 ) = 0.

Therefore, y − f (y)x0 ∈ ker(f ) = ker(g). Hence,


¡ ¢
g(y) − f (y)g(x0 ) = g y − f (y)x0 = 0.

This implies that


g(y) = g(x0 )f (y), ∀y ∈ X.
Hence g = cf with c = g(x0 ) 6= 0. ¥

Problem 64.
Let X be a normed space and f are nonzero linear functional on X. Show that
f is continuous if and only if ker(f ) is closed.

Solution.
Suppose that f is continuous. Since ker(f ) = f −1 ({0}), so ker(f ) is closed.
Conversely, suppose that ker(f ) is closed. Pick an x0 ∈ X such that f (x0 ) = 1.
Assume that f is not continuous, that is, f is not bounded. Then there exists a
sequence (xn ) in X such that

kxn k = 1 and f (xn ) ≥ n, ∀n ∈ N.


xn
Define yn = x0 − f (xn )
. Then

yn ∈ ker(f ) for all n and lim yn = x0 .


n→∞

By hypothesis, ker(f ) is closed, so x0 ∈ ker(f ), that is, f (x0 ) = 0. This contradicts


our assumption that f (x0 ) = 1. Thus f is continuous. ¥

Problem 65
Let Z be a subspace of a normed space X, and y ∈ X. Let d = d(y, Z). Prove
that there exists Λ ∈ X ∗ such that kΛk ≤ 1, Λ(y) = d and Λ(z) = 0 for all
z ∈ Z.

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
70 CHAPTER 4. LINEAR OPERATORS - LINEAR FUNCTIONALS

Solution.
If y ∈ Z, then d = 0 and so the zero functional works, so we may assume that y ∈ / Z.
Consider the subspace Y = Cy + Z ⊂ X. Since y ∈ / Z, so for every x ∈ Y , there is
a unique α ∈ C and z ∈ Z such that x = αy + z. Define

λ(x) = αd.

We observe that λ(z) = 0 since

x = z ⇒ αy = 0
⇒ α = 0 (otherwise y = 0 so y ∈ Z).

Also we have λ(y) = d since y = 1.y + 0. It is clearly that λ is linear in x and

1
kxk = |α|ky + zk ≥ αd = |λ(x)|.
α
Thus, λ is continuous, that is, λ ∈ Y ∗ and kλk ≤ 1. By the Hahn-Banach theorem
we may extend λ to an element Λ ∈ X ∗ with the same norm. ¥

Problem 66
Let X be a normed space and (xn ) be a sequence in X. Set V := Span {x1 , x2 , ...}.
Let W be the set of all continuous f ∈ X ∗ such that f (xn ) = 0, ∀n ∈ N. Prove
that
V = {x ∈ X : f (x) = 0, ∀f ∈ W }.

Solution.
Let Y := {x ∈ X : f (x) = 0, ∀f ∈ W }.
• We show that V ⊂ Y . Take any x0 ∈ V . There exists a sequence (uk ) in V such
that uk → x0 as k → ∞. Since V := Span {x1 , x2 , ...}, for each k ∈ N, there are
(k) (k) P k (k)
scalars c1 , ..., cnk such that uk = nn=1 cn xn . By linearity of f and by definition
of W , we have
nk
X
f (uk ) = c(k)
n f (xn ) = 0, ∀f ∈ W.
n=1

By continuity of f , we have

f (x0 ) = 0, ∀f ∈ W.

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
4.2. LINEAR FUNCTIONALS 71

Hence x0 ∈ Y , and so V ⊂ Y .
• Now given any x0 ∈ Y , we want to show x0 ∈ V . Assume that x0 ∈ / V . Then

d(x0 , V ) = d > 0. Using the result in Problem 65, there is an F ∈ X such that

F (x0 ) = d and F (x) = 0, ∀x ∈ V .

It follows that
F (x0 ) = d and F (xn ) = 0, ∀n ∈ N.
Hence F ∈ W but F (x0 ) 6= 0. It means that x0 ∈
/ Y : a contradiction. Thus x0 ∈ V ,
and so Y ⊂ V .
The proof is complete. ¥

Problem 67.
Let X be a normed space and {x1 , ..., xn } ⊂ X a linearly independent system.
Prove that for any α1 , ..., αn ∈ K there exists x∗ ∈ X ∗ such that

x∗ (xi ) = αi , ∀i ∈ {1, ..., n}.

Solution.
Let Y = Span{x1 , ..., xn }. Since the system {x1 , ..., xn } ⊂ X a linearly independent,
it follows that {x1 , ..., xn } is an algebraic basis for Y . Define f : Y → K linear with
f (xi ) = αi , ∀i ∈ {1, ..., n}. Since Y is of finite dimension, f is continuous. By the
Hahn-Banach theorem there is an extension x∗ ∈ X ∗ for f . Then

x∗ (xi ) = f (xi ) = αi , ∀i ∈ {1, ..., n}. ¥

Problem 68.
Let X be a normed space and Y ⊂ X a linear subspace. For x0 ∈ X \ Y we
define
f : Span{Y, x0 } → K, f (y + λx0 ) = λ, ∀y ∈ Y, ∀λ ∈ K.
(a) Prove that f is well defined and linear.
(b) Prove that f is continuous
T / Y.
if and only if x0 ∈
(c) Prove that Y = {ker x : x ∈ X , Y ⊂ ker x∗ }.
∗ ∗ ∗

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
72 CHAPTER 4. LINEAR OPERATORS - LINEAR FUNCTIONALS

Solution.
(a) Let us first observe that
Span{Y, x0 } = {y + λx0 : y ∈ Y, λ ∈ K}.
To show that f is well defined, we show that
(y + λx0 = y 0 + λ0 x0 ) ⇒ (y = y 0 , λ = λ0 ).
Assume that λ 6= λ0 . Then
y − y0
x0 = ∈Y
λ − λ0
since y, y 0 ∈ Y and Y is a linear subspace. This is a contradiction. Thus, λ = λ0
implies that y = y 0 . The linearity of f is obvious.
(b) Suppose that f is continuous. We will prove that x0 ∈ / Y . Assume, for a
contradiction, that x0 ∈ Y . Then there is a sequence (yn ) ⊂ Y such that yn → x0 .
Since yn , x0 ∈ Y , yn → x0 and f is continuous, f (yn ) → f (x0 ). By definition of
f, f (yn ) = 0, ∀n ∈ N. It follows that f (x0 ) = 0, which is false, since by definition
of f , we have f (x0 ) = 1.
Suppose now that that x0 ∈ / Y , i.e., x0 ∈ (Y )c , which is open. Then
∃ε > 0 : B(x0 ; ε) ⊂ (Y )c .
The inclusion is equivalent to Y ∩ B(x0 ; ε) = ∅. Let x = y + λx0 ∈ Span{Y, x0 }. If
λ = 0, then
kxk
(∗) |f (x)| = 0 ≤ .
° ε °
y
If λ 6= 0, then −λ y
∈ Y so −λ ∈/ B(x0 ; ε), i.e., °− λy − x0 ° ≥ ε. Therefore,
y
(∗∗) kxk = |λ| k − − x0 k ≥ |λ|ε = ε|f (x)|.
λ
Both cases (*) and (**) show that f is continuous.
(c) If x∗ ∈ X ∗ and Y ⊂ ker x∗ then Y ⊂ ker x∗ (since ker x∗ is closed) and therefore
\
Y ⊂ {ker x∗ : x∗ ∈ X ∗ , Y ⊂ ker x∗ }.
T
Conversely, let x0 ∈ {ker x∗ : x∗ ∈ X ∗ , Y ⊂ ker x∗ }, and assume that x0 ∈ / Y.
By (b), there is an f : Span{Y, x0 } → K linear and continuous such that f |Y =
0, f (x0 ) = 1. Let x∗ ∈ X ∗ be a Hahn-Banach extension for f given by the Hahn-
Banach theorem. Then x∗ = f on Span{Y, x0 }. In particular, x∗ = 0 on Y . Hence
Y ⊂ ker x∗ and x∗ (x) = f (x0 ) = 1, that is x ∈
/ ker x∗ and so
\
/ {ker x∗ : x∗ ∈ X ∗ , Y ⊂ ker x∗ },
x0 ∈

which is a contradiction. ¥

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D

Chapter 5

Fundamental Theorems

Review some main points:


1. The Baire category theorem:

Definition 3 Let X be a metric space and E be a subset of X.


(a) E is called a set of the first category if E is a countable union of nowhere dense (non-dense)
sets in X.
(b) If E is not a set of the first category, then E is called a set of the second category.

Theorem 5 (Baire category theorem - form I)


If X is a complete metric space, then X is a set of the second category.
S∞
Let X be a complete metric space. The Baire category theorem tells us that if X = n=1 An , then
some of the set An must have non-empty interior.

Theorem 6 (Baire category theorem - form II)


Let X be a complete metric space. If (Un )n∈N is a sequence of open dense subsets of X, then
T
n∈N Un is dense in X.

It can be shown that these two forms of Baire category theorem are equivalent (try it!).

2. The principle of uniform boundedness (P.U.B.)

Theorem 7 Let X be a Banach space and Y be a normed space. Let F be a family of bounded
linear operators from X to Y . Suppose that for each x ∈ X, {kT xk : T ∈ F} is bounded, then
{kT k : T ∈ F} is bounded.

One of the P.U.B. consequences is:

Theorem 8 (Banach-Steinhaus theorem)


Let X be a Banach space and Y be a normed space and (Tn ) be a sequence in B(X, Y ). Suppose
for every x ∈ X, Tn x → T x as n → ∞. Then the family {kTn k, n ∈ N} is bounded, i.e.,
supn∈N kTn k < ∞, and T is linear and bounded, i.e., T ∈ B(X, Y ).

73
www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
74 CHAPTER 5. FUNDAMENTAL THEOREMS

3. The open mapping theorem

Theorem 9 Let X and Y be Banach spaces and T ∈ B(X, Y ). If T is onto (surjective), then T
is an open mapping, (that is, if U is open in X, then T (U ) is open in Y ).

Consequence:

Theorem 10 (The inverse mapping theorem)


Let X and Y be Banach spaces. If T ∈ B(X, Y ) is bijective, then T −1 exists and T −1 ∈ B(Y, X).

5. The closed graph theorem

Theorem 11 Let X and Y be Banach spaces and T be a linear map from X to Y . If we define
the graph of T by
Γ(T ) := {(x, y) ∈ X × Y : y = T x},
then T is bounded (continuous) if and only if Γ(T ) is closed in X × Y .

Problem 69.
(a) If X is a normed space, prove that any proper closed linear subspace of X is
a nowhere dense set.
(b) Prove that c0 is a nowhere dense set of c.

Solution.
(a) Let A be a proper closed linear subspace of X. If A is not nowhere dense, then
˚ = Å 6= ∅ since A is closed. Therefore, A contain an open ball B(a; ε). Take

x ∈ B(0; ε) then we have

kxk < ε ⇒ a + x ∈ B(a; ε) ⊂ A.

Since a, a + x ∈ A and A is a linear space, x must be in A. Therefore B(0; ε) ⊂ A.


Now take any y ∈ X with y 6= 0, then we have
ε
y ∈ B(0; ε).
2kyk
ε
This implies that 2kyk y ∈ A, and so y ∈ A. Thus X = A. This is a contradiction.
(b) We know that c0 is a proper closed linear subspace of c. So c0 is nowhere dense
in c. ¥

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
75

Problem 70.
Show that L2 [0, 1] is a subset of the first category in L1 [0, 1].

Solution.
For short, we write L1 := L1 [0, 1] and L2 := L2 [0, 1]. We know that L2 ⊂ L1 . For
each n ∈ N, let
An = {f : kf k2 ≤ n}.
Every function in L2 will be in some An , Thus
[
L2 = An .
n∈N

We first prove that An is closed. Fix n. Let (fk ) be a sequence in An such that
kfk − f k1 → 0. We show that f ∈ An . Since kfk − f k1 → 0, (fk ) converges
in measure to f . Hence there exists a subsequence (fkj ) converging to f almost
everywhere. By definition of An we have
Z 1
fk2j (t)dt ≤ n2 for j = 1, 2, ...
0

Applying Fatou’s lemma, we obtain


Z 1
f (t)dt ≤ n2 .
0

Thus f ∈ An , and An is closed.


¡ ¢0
Now we show that An is nowhere dense, i.e. An = (An )0 = ∅. To do this, it
suffices to show that for any open ball Bε (f ) in L1 , there exists a point g ∈ Bε (f )
but g ∈/ An . Take
ε 1
h(t) = √ .
4 t
/ L2 and
Then h ∈ Z 1
ε
h(t)dt = ,
0 2
so that
ε
h ∈ L1 and khk1 = .
2
2
Let g = f + h. Then g ∈
/ L so g ∈
/ An , and
ε
kf − gk1 = khk1 = .
2

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
76 CHAPTER 5. FUNDAMENTAL THEOREMS

Hence g ∈ Bε (f ). Thus An is nowhere dense, and L2 is of the first category in


L1 . ¥

Problem 71.
Let X 6= {0} be a normed space, and A ⊂ X which is not nowhere dense. We
denote by A0 the derivative set of A, i.e., the set of all accumulation points of
A. Prove that

∃x ∈ X and ε > 0 such that B(x; ε) ⊂ A0 .

(Note: Ā = A ∪ A0 ).

Solution.
˚ 6= ∅,
Since Ā
∃x ∈ X and ε > 0 such that B(x; ε) ⊂ Ā.
We will prove B(x; ε) ⊂ A0 . Let y ∈ B(x; ε). If y ∈
/ A0 , then there is r > 0 such that

B(y; r) ∩ (A \ {y}) = ∅.

Let δ = min{r, 12 (ε − ky − xk)} > 0. We claim B(y; δ) ⊂ B(x; ε). Indeed, if


z ∈ B(y; δ) then

kz − xk ≤ kz − yk + ky − xk
< δ + ky − xk
1
≤ (ε − ky − xk) + ky − xk
2
1
= (ε + ky − xk) < ε.
2

Hence z ∈ B(x; ε).


Obviously, B(y; δ) ⊂ B(y; r). So

B(y; δ) ∩ (A \ {y}) ⊂ B(y; r) ∩ (A \ {y}) = ∅.

That is B(y; δ) ∩ A ⊂ {y}. Since y ∈ B(x; ε) ⊂ Ā, so y ∈ Ā, from whence


B(y; δ) ∩ A 6= ∅, and therefore

B(y; δ) ∩ A = {y}.

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
77

Since B(y; δ) ∩ Ā ⊂ B(y; δ) ∩ A and B(y; δ) ⊂ B(x; ε) ⊂ Ā, i.e., B(y; δ) ∩ Ā =


B(y; δ). Hence, B(y; δ) = {y}, which is false, since X 6= {0}. (Indeed, ∃a ∈ X, a 6=
a
0. Take b = kak ∈ X. Then kbk = 1 and
δ
y+ b ∈ B(y; δ) = {y},
2
which implies that b = 0 : a contradiction). ¥

Problem 72.
Let X is a Banach space and A ⊂ X a dense set. Can we find a function
f : X → R such that, for every x ∈ A, we have limt→x |f (t)| = ∞?

Solution.
Suppose that such a function exists. Since f takes finite values, for every x ∈ X,
there is k ∈ N such that |f (x)| ≤ k, i.e.,
[
X= Ak where Ak = {x ∈ X : |f (x)| ≤ k}.
k∈N

Since X is a Banach space, X is of the second Baire category. So there is k ∈ N


˚ 6= ∅. Using Problem 74, it follows that there is x ∈ X and ε > 0 such
such that Ā k
that B(x; ε) ⊂ A0k . Since A is dense, we have
∅ 6= B(x; ε) ∩ A ⊂ A0k ∩ A.
It follows that there is an element a ∈ A which is in A0k . By definition of accumula-
tion points, there is a sequence (xn ) ⊂ Ak with xn 6= a, ∀n ∈ N such that xn → a.
Since a ∈ A, by hypothesis we have limt→a |f (t)| = ∞. Hence, limn→∞ |f (xn )| = ∞.
This is not possible since (xn ) ⊂ Ak implies that |f (xn )| ≤ k, ∀k ∈ N. ¥

Problem 73.
Show that Q is a subset of the first category of R.

Solution.
Since Q is countable, we have
[
Q= {xn } and ({xn })0 = ({xn })0 = ∅. ¥
n∈N

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
78 CHAPTER 5. FUNDAMENTAL THEOREMS

Problem 74.
Show that the set of piecewise linear functions on R is of the first category.

Solution.
Let P denote the set of piecewise linear functions on R. Let Pn be the set of piecewise
linear functions having n intervals of linearity. We have

[
P = Pn .
n=1

We need to prove that Pn is nowhere dense for every n. Fix an arbitrary n. Let
f ∈ Pn . Consider the ball Br (f ), r > 0. Let
r
g(t) = sin(2π4nt) + f (t).
2
Then
r
|f (t) − g(t)| = | sin(2π4nt)|.
2
Hence d(f, g) = 2r , and so g ∈ Br (f ). We claim that the ball B r2 (g) contains no
element from Pn . Pick h ∈ Br (f ) ∩ Pn and suppose d(g, h) < 2r . Then
¯r ¯ r
¯ ¯
(∗) d(g, h) = sup |g(t) − h(t)| = sup ¯ sin(2π4nt) + f (t) − h(t)¯ < .
2 2
Observing that the term 2r sin(2π4nt) oscillates between − 2r and 2r 4n times on
[0, 1]. Thus the term f (t) − h(t) must also oscillate between negative and positive
values 4n times for (*) to hold. But this is impossible since the term f (t) − h(t) is
a piecewise linear function with at most 2n intervals of linearity. So, the open ball
B r2 (g) contains no element from Pn . Since n is arbitrary, we see that Pn is nowhere
dense, and hence P is of the first category. ¥

Problem 75. (Inverse mapping theorem)


Let X and Y be Banach spaces and T ∈ B(X, Y ). Suppose T is bijective. Show
that there exist real numbers a, b > 0 such that

akxk ≤ kT xk ≤ bkxk, ∀x ∈ X.

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
79

Solution.
Since T is linear, bijective and bounded, T −1 exists, is linear and bounded by the
inverse mapping theorem. Let kT −1 k = a1 and kT k = b. Note that T 6= 0, a, b > 0.
Now since T is bounded,

(i) kT xk ≤ kT k kxk = bkxk, ∀x ∈ X.

Also, since T −1 is bounded,


1
(ii) kxk = kT −1 (T x)k ≤ kT −1 k kT xk = kT xk, ∀x ∈ X.
a
(i) and (ii) imply that

akxk ≤ kT xk ≤ bkxk, ∀x ∈ X. ¥

Problem 76. Let X = C 1 [0, 1] be the space of continuously differentiable func-


tions on [0, 1] and Y = C[0, 1] . The norm on C[0, 1] and C 1 [0, 1] is the sup-
norm. Consider the map

dx(t)
T : C 1 [0, 1] → C[0, 1] defined by T x(t) = , t ∈ [0, 1].
dt
Show that the the graph of T is closed but T is not bounded. Does this contradict
the closed graph theorem ?

Solution.
• It is clear that T is linear.
• We show that Γ(T ) is closed. Suppose xn → x in X = C 1 [0, 1] and T xn → y in
Y = C[0, 1]. We must show that y = T x. For any t ∈ [0, 1], we have
Z t Z t
dxn
y(s)ds = lim ds, y ∈ C[0, 1]
0 0 n→∞ ds
Z t
dxn
= lim ds (uniform convergence)
n→∞ 0 ds
¡ ¢
= lim xn (t) − xn (0) = x(t) − x(0).
n→∞

Thus, with y ∈ C[0, 1], we have


Z t
x(t) = x(0) + y(s)ds, t ∈ [0, 1].
0

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
80 CHAPTER 5. FUNDAMENTAL THEOREMS

Hence
dx
x ∈ C 1 [0, 1] and = y on [0, 1].
ds

That is T x = y, and so (x, y) ∈ Γ(T ), and Γ(T ) is closed.


• We show that T is not bounded. Take fn (t) = tn , n ∈ N, t ∈ [0, 1]. Then
fn ∈ C 1 [0, 1] and T fn = nfn−1 for n > 1. But we have

kfn k = 1 and kT fn k = n,

which shows that T is unbounded. The reason? Is C 1 [0, 1] with the sup-norm a
Banach space? ¥

Problem 77. (P.U.B.)


Let H be a separable Hilbert space with an orthonormal basis {en }∞
n=1 . Let {xn }
be a sequence in H. Prove that the following two statements are equivalent:
(1) limn→∞ hx, xn i = 0, ∀x ∈ H.
(2) limn→∞ hem , xn i = 0, ∀m ∈ N and {kxn k} is bounded.

Solution.
• (1) ⇒ (2)
Assume that (1) is true . Then the first part of (2) is automatically true. We
have only to show that {kxn k} is bounded. Consider, for each n, the functional
fn (x) = hx, xn i. This is a bounded functional by Schwarz inequality, and, because
fn (x) → 0 for each x, we have that the set {kfn (x)k : n ∈ N} is bounded. The
principle of uniform boundedness then gives us that the set {kfn k} is bounded. But
kfn k = kxn k, so {kxn k} is bounded.
• (2) ⇒ (1)
Assume
P that (2) holds. Let B be the bound of {kxn k} and let x ∈ X. We write
x = n hen , xien . For every ε > 0, let K be such that

X ε
|hem , xi|2 < .
m>K
B

We know that, for m fixed, hem , xn i → 0 as n → ∞. So we may find N such that if


n > N then |hem , xn i| < K sup ε|her ,xi| (the denominator is finite because the sequence

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
81

her , xi is an `2 -sequence). Then, if n > N , we have


¯ ¯
¯X∞ ¯
¯ ¯
|hx, yi| = ¯ hem , xihem , xn i¯
¯m=1 ¯
K
X X
ε
≤ P |hem , xi| + hem , xihem , xn i
m=1
K |he r , xi| m>K
à !1/2 à !1/2
X X
≤ ε+ |hem , xi|2 + |hem , xn i|2
m>K m>K
ε
≤ ε+ kxn k ≤ 2ε. ¥
B

Problem 78. (Closed graph theorem)


Let H be a Hilbert space and T : H → H a linear operator which is symmetric,
i.e.,
hT x, yi = hx, T yi, ∀x, y ∈ H.
Prove that T is continuous.

Solution.
Since H is a complete space and T is a linear operator, it is sufficient to prove that
T is a closed graph operator. Let (xn ) be a sequence in H such that xn → x ∈ H
and T xn → y ∈ H. We will show that y = T x .
By hypothesis, we have
hT (xn − x), yi = hxn − x, T yi, ∀y ∈ H.
Hence
hT (xn − x), yi → 0, ∀y ∈ H.
That is
w w
T (xn − x) −
→ 0, or, equivalently T xn −
→ T x.
w
But T xn → y, so T xn −→ y. Since the limit is unique, we have T x = y. Thus, T is
a closed graph operator. ¥

Problem 79. (Banach-Steinhaus theorem)


Let a = (a1 , a2 , ...) = (an )n∈N be a sequence of scalars such that the sequence
(an xn )n∈N ∈ c0 for all sequences x = (xn )n∈N ∈ c0 . Prove that a ∈ `∞ .

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
82 CHAPTER 5. FUNDAMENTAL THEOREMS

Solution.
For every n ∈ N, consider the operator

Un : c0 → c0 defined by Un (x) = Un (x1 , x2 , ...) = (a1 x1 , a2 x2 , ..., an xn , 0, ...).

Then Un is linear, and

kUn (x1 , x2 , ...)k = k(a1 x1 , a2 x2 , ..., an xn , 0, ...)k


= max(|a1 x1 |, ..., |an xn |)
≤ kxk max(|a1 |, ..., |an |), ∀x ∈ c0 .

So Un is continuous, and kUn k ≤ max(|a1 |, ..., |an |).


For 1 ≤ k ≤ n, we have

kUn k ≥ kUn (ek )k = k(0, ..., 0, ak , 0, ...)k = |ak |,

where ek = (0, .., 0, 1, 0, ...) ∈ c0 . Therefore max(|a1 |, ..., |an |) ≤ kUn k. Hence,
| {z }
k−1

kUn k = max(|a1 |, ..., |an |), ∀n ∈ N.

Now consider the operator

U : c0 → c0 defined by U (x) = (a1 x1 , a1 x2 , ...) = (an xn )n∈N .

We have that Un (x) → U (x), ∀x ∈ c0 because

kUn (x) − U (x)k = sup |ak xk | → 0.


k≥n+1

Now from the Banach-Steinhaus theorem we get that supn∈N kUn k < ∞, i.e., supn∈N |an | <
∞. In other words, a ∈ `∞ . ¥

Problem 80. (Similar problem)


Let a = (a1 , a2 , ...) = (an )n∈N be a sequence of scalars such that the sequence (an xn )n∈N ∈ c0 for
all sequences x = (xn )n∈N ∈ `∞ . Prove that a ∈ c0 .

Problem 81. (P.U.B.)



P∞ x = (x1 , x2 , ...) = (xi )i=1 be a sequence of scalars such that1 the series
Let
i=1 xi yi is convergent for all y = (y1 , y2 , ...) ∈ c0 . Prove that x ∈ ` .

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
83

Solution.
For every n ∈ N, we define the linear operator
n
X
Tn : c0 → C, Tn (y) = xi y i .
i=1

Then we have à !
n
X n
X
|Tn (y)| ≤ |xi yi | ≤ |xi | kyk∞ .
i=1 i=1

This shows that Tn is bounded with


n
X
kTn k ≤ |xi |.
i=1

P∞ ³ ´
By hypothesis, i=1 x i yi < ∞, the sequence T n (y) converges for every y ∈ co ,
n∈N
and c0 is a Banach space, the principle of uniform boundedness implies that

∃M > 0 : kTn k ≤ M, ∀n ∈ N.
¡ (n) (n) (n) ¢
Now let x(n) = x1 , x2 , ..., xn , 0, 0, ... be a truncated version of x (so that x(n) ∈
`1 ). Let
X n
(n)
Tn (y) = xi yi , y = (y1 , y2 , ...) ∈ c0 .
i=1
¡ (n) (n) ¢
Define y (n) = y1 , y2 , ..., by
 (n)
 xk (n)
if xk 6= 0,
(n) (n)
yk = |xk |
0 (n)
if xk = 0.

Then
n
X
(n) (n)
|Tn (y )| = |xk | = kx(n) k1 = kx(n) k1 ky (n) k∞ .
k=1

Hence
kTn k ≥ kx(n) k1 ,
which in turn implies that

kx(n) k1 ≤ M, ∀n ∈ N.

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
84 CHAPTER 5. FUNDAMENTAL THEOREMS
¡ ¢
But it is clear from the definition of x(n) that kx(n) k1 is an increasing sequence of
real numbers. Being bounded above by M , it must converge. Hence

X
|xi | < ∞,
i=1

and so x ∈ `1 . ¥

Problem 82. (Very similar problem) P∞


Let c = (c1 , c2 , ...) = (ci )∞
i=1 be a sequence of scalars such that the series i=1 ci ai is convergent
for all a = (a1 , a2 , ...) ∈ `1 . Prove that x ∈ `∞ .

Problem 83. (Closed graph theorem)


Let X, Y and Z be Banach spaces. Suppose that T : X → Y is linear, that
J : Y → Z is linear, bounded and injective, and that JT ≡ J ◦ T : X → Z is
bounded.
Show that T is also bounded.

Solution.
We will show that the graph Γ(T ) is closed. Then by the closed graph theorem, this
implies
¡ that ¢T is bounded (continuous).
Let (xn , yn ) n∈N be a convergent sequence in X × Y , that is,

xn → x in X, yn → y in Y and T xn = yn .

Since J and JT are continuous,


£ ¤ £ ¤
yn → y ⇒ Jyn → Jy and xn → x ⇒ JT xn → JT x .

Since T xn = yn , we have Jyn → JT x. Since the limit is unique, this gives that
Jy = JT x. But by hypothesis J is injective, so we have

Jy = JT x ⇒ y = T x.

This shows that (x, y) ∈ Γ(T ), and Γ(T ) is closed. ¥

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
85

Problem 84. (Closed graph theorem)


Let X be a Banach space and E, F two closed subspaces of X such that X =
E ⊕ F . Consider the projections on E and on F defined by

PE : X → E, PE (u) = x,
PF : X → F, PF (u) = y, where u = x + y, x ∈ E, y ∈ F.

Use the closed graph theorem to show that PE ∈ B(X, E) and PF ∈ B(X, F ).

Solution.
The linearity of these two maps are easy to check. Let us prove that they are
bounded by using the closed graph theorem. Denote the graph of PE by ΓE . We
can write
ΓE = {(x, y) ∈ X × E : x − y ∈ F }.
Let (xn , yn ) ∈ ΓE for every n ∈ N. Suppose (xn , yn ) → (x, y) as n → ∞. Since
xn − yn ∈ F for every n ∈ N, and F is a closed subspace of X, limn→∞ (xn − yn ) =
x − y ∈ F . It follows that (x, y) ∈ ΓE . Thus, ΓE is closed, and so PE is bounded.
The proof for PF is the same. ¥

Problem 85. (Inverse mapping theorem)


Let (X, k.k1 ) and (X, k.k2 ) be Banach spaces. Suppose that

∃C ≥ 0 : kxk2 ≤ Ckxk1 , ∀x ∈ X.

Show that the two norms k.k1 and k.k2 are equivalent.

Solution.
Consider the identity map
id : (X, k.k1 ) → (X, k.k2 ), id(x) = x.
It is clear that the identity map is linear and bijective. It is continuous since by
hypothesis we have
(∗) kid(x)k2 = kxk2 ≤ Ckxk1 , ∀x ∈ X.
By the inverse mapping theorem, the inverse map id−1 exists and continuous. That
is
(∗∗) ∃C 0 ≥ 0 : kxk1 ≤ C 0 kxk2 , ∀x ∈ X.

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
86 CHAPTER 5. FUNDAMENTAL THEOREMS

(∗) and (∗∗) together imply that k.k1 and k.k2 are equivalent. ¥

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D

Chapter 6

Linear Operators on Hilbert


Spaces

Review.
1. Definition and elementary properties
Let T : H → K be a linear operator between Hilbert spaces H and K.
• The following statements are equivalent:
1. T is continuous at 0,
2. T is continuous,
3. T is bounded on H.
• An isomorphism between H and K is a linear surjection U : H → K such that

hU x, U yi = hx, yi, ∀x, y ∈ H.

An isomorphism is an isometry and so preserves completeness, but an isometry need not to be an


isomorphism.

Proposition 1 Two Hilbert spaces are isomorphic if and only if they have the same dimension.

2. Adjoint of an Operator
• Let A ∈ B(H). Then A∗ is called the adjoint operator of A if

hAx, xi = hx, A∗ xi, ∀x ∈ H.

Proposition 2 If A, B ∈ B(H) and α ∈ F, then

(a) (αA + B)∗ = ᾱA∗ + B ∗ .


(b) (AB)∗ = B ∗ A∗ .
(c) A∗∗ := (A∗ )∗ = A.
(d) If A is invertible in B(H) and A−1 is its inverse, then A∗ is invertible and (A∗ )−1 = (A−1 )∗ .

87
www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
88 CHAPTER 6. LINEAR OPERATORS ON HILBERT SPACES

Proposition 3 It A ∈ B(H) then


kAk = kA∗ k = kA∗ Ak1/2 .

3. Self-adjoint, normal, unitary operators

Definition 4 If A ∈ B(H), then

(a) A is Hermitian or self-adjoint if A∗ = A.


(b) A is normal if AA∗ = A∗ A.
(c) A is unitary if it is a surjective isometry.

Proposition 4 Let A ∈ B(H). The following statements are equivalent.

(a) A∗ A = AA∗ = I.
(b) A is unitary.
(c) A is a normal isometry.

4. Positive operators

Definition 5 Let H be a Hilbert space. An operator A ∈ B(X) is called positive if


hAx, xi ≥ 0, ∀x ∈ X.
We write A ≥ 0. If A, B ∈ B(X) and A − B ≥ 0 then we write B ≤ A.

Proposition 5 If A ∈ B(X) then A∗ A ≥ 0. In addition, if A ≥ 0, then


1. A is self adjoint,
2. there exists a unique B ∈ B(X) such that B ≥ 0 and B 2 = A. Furthermore, B is also
√ self
adjoint and commutes with every bounded operator which commutes with A. We write B = A.

We define |A| = A∗ A.
5. Projection, Orthogonal projection

Definition 6
If P ∈ B(H) and P 2 = P, then P is called a projection.
If P ∈ B(H), P = P 2 and P ∗ = P, then P is called an orthogonal projection.

Proposition 6
If P : H → H is a projection then H = Image P ⊕ ker P .
If H = M ⊕ N , where M, N are subspaces of H, then there is a projection P : H → H with
Image P = M and ker P = N.

∗∗∗∗∗

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
89

Problem 86. Let P be an orthogonal projection defined on a Hilbert space H.


Show that kP k = 1.

Solution.
If x ∈ H and P x 6= 0, then the use of the Cauchy-Schwarz inequality implies that
hP x, P xi
kP xk =
kP xk
hx, P 2 xi
= (since P ∗ = P )
kP xk
hx, P xi kxk kP xk
= ≤
kP xk kP xk
≤ kxk.

Therefore kP k ≤ 1.
Now, if P 6= 0, then there is an x0 ∈ H such that

P x0 6= 0 and kP (P x0 )k = kP x0 k.

This implies that kP k ≥ 1. Thus kP k = 1. ¥

Problem 87. Given a function φ : [0, 1] → C, consider the operator

P : L2 [0, 1] → L2 [0, 1] defined by P f (x) = φ(x)f (x).

Find necessary and sufficient conditions on the function φ for P to be an or-


thogonal projection.

Solution.
First, in order for P to be a well-defined operator acting on L2 [0, 1], the function
φf needs to be in L2 [0, 1] for all f ∈ L2 [0, 1]. In particular φf is measurable, and
taking f ≡ 1, it follows that φ is a measurable function on [0, 1].
Secondly, P is an orthogonal projection if and only if P ∗ = P and P 2 = P . The
last equality is equivalent to φ2 (x)f (x) = φ(x)f (x), ∀f ∈ L2 [0, 1]. Again by taking
f ≡ 1, we have a2 (x) = a(x) for almost every x ∈ [0, 1]. Thus

a(x) = 0 or a(x) = 1 for almost all x ∈ [0, 1].

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
90 CHAPTER 6. LINEAR OPERATORS ON HILBERT SPACES

In particular φ takes real values. Then


Z 1
hP f, gi = P f (x)g(x)dx
0
Z 1
= φ(x)f (x)g(x)dx
0
Z 1
= f (x)φ(x)g(x)dx
0
Z 1
= f (x)P g(x)dx
0
= hf, P gi,
which proves that P is self-adjoint. Since 0 ≤ φ(x) ≤ 1 for a.e. on [0, 1], we have
that
µZ 1 ¶ 21 µZ 1 ¶ 12
2 2 2
kP f kL2 = φ(x) |f (x)| dx ≤ |f (x)| dx = kf kL2 .
0 0
Thus, P is bounded.
In conclusion, the necessary and sufficient conditions for P to be an orthogonal
projection is φ a measurable satisfying φ(x) = 0 or φ(x) = 1 for almost all x ∈
[0, 1]. ¥

Problem 88. Consider the right-shift on the Hilbert space `2 :

S : `2 → `2 , S(α1 , α2 , ...) = (0, α1 , α2 , ...).


Define its adjoint operator.

Solution.
For (αn ) = (α1 , α2 , ...) and (βn ) = (β1 , β2 , ...) in `2 ,
­ ∗ ® ­ ®
S (αn ), (βn ) = (αn ), S(βn )
= h(α1 , α2 , ...), (0, β1 , β2 , ...)i
= α2 β̄1 + α3 β̄2 + ...
= h(α2 , α3 , ...), (β1 , β2 , ...)i.
Thus
S ∗ (α1 , α2 , ...) = (α2 , α3 , ...).
Hence, the adjoint of the right-shift is the left-shift. ¥

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
91

Problem 89. Let A : `2 → `2 be defined by

Ax = A(x1 , x2 , x3 , ...) = (0, 0, x3 , x4 , ...).



Prove A is linear, continuous, self-adjoint and positive. Find A.

Solution.
With similar argument as the previous problem, we can show that A is linear. We
have ∞ ∞
X X
kAxk2 = |xk |2 ≤ |xk |2 = kxk2 , ∀x = (x1 , x2 , x3 , ...) ∈ `2 .
k=3 k=1
Therefore,
kAxk ≤ kxk, ∀x ∈ `2 .
This shows that A is continuous. Also, for all x, y ∈ `2 ,
X∞
hAx, yi = xk y¯k = hx, Ayi.
k=3

Hence, A is self-adjoint. And,



X
hAx, xi = |xk |2 ≥ 0, ∀x ∈ `2 .
k=3

So A is positive. Then there exists the square root A : `2 → `2 . We have
A2 x = A(Ax) = A(0, 0, x3 , x4 , ...) = (0, 0, x3 , x4 , ...), ∀x ∈ `2 .

It follows that A2 = A. Hence, A = A. ¥

Problem 90.(Multiplication operator)


Let (X, Ω, µ) be a σ-finite measure space. Consider the Hilbert space H =
L2 (X, Ω, µ) =: L2 (µ). If φ ∈ L∞ (µ), define

Mφ : L2 (µ) → L2 (µ) by Mφ f = φf.

(a) Show that


Mφ ∈ B(H) and kMφ k = kφk∞ .
Here kφk∞ is the µ-essential supremum norm.
(b) Show that Mφ∗ = Mφ̄ .
(c) Show that Mφ is normal. When Mφ is self adjoint? unitary?

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
92 CHAPTER 6. LINEAR OPERATORS ON HILBERT SPACES

Solution.
(a) The linearity of the operator Mφ is evident. We show that Mφ is bounded and
calculate its norm.
By definition, kφk∞ is the infimum of all c > 0 such that |φ(x)| ≤ c a.e. [µ], and so
|φ(x)| ≤ kφk∞ a.e. [µ]. Thus we can assume that φ is a bounded measurable and
|φ(x)| ≤ kφk∞ for all x. If f ∈ L2 (µ), then
Z Z
|φf | dµ ≤ kφk∞ |f |2 dµ.
2 2

That is, ¡ ¢
Mφ ∈ B L2 (µ) and kMφ k ≤ kφk∞ (∗).
If ε > 0, the σ-finiteness of the measure space implies that

∃∆ ∈ Ω such that 0 < µ(∆) < ∞ and |φ(x)| ≥ kφk∞ − ε, ∀x ∈ ∆.

If we take f = √ 1 χ∆ , then we have


µ(∆)

f ∈ L2 (µ) and kf k2 = 1.

So Z
2 12
kMφ k ≥ kφf k = |φ|2 dµ ≥ (kφk∞ − ε)2 .
µ(∆) ∆
Letting ε → 0, we get that

kMφ k ≥ kφk∞ (∗∗).

(*) and (**) give that kMφ k = kφk∞ .


(b) For f, g ∈ L2 (µ), we have
Z
¡ ¢
hf, Mφ gi = f Mφ g dµ
Z
= f (φg)dµ
Z
= (φ̄f )ḡdµ
= hMφ̄ f, gi.

This shows that


Mφ∗ = Mφ̄ .
(c) Every multiplication operator Mφ is normal. Indeed,

Mφ Mφ∗ = Mφ Mφ̄ = Mφ̄ Mφ = Mφ∗ Mφ .

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
93

Mφ is self-adjoint if and only if φ = φ̄, that is, φ is real-valued.


Mφ is unitary if and only if |φ| = 1 a.e. [µ]. ¥

Problem 91. Let H be a Hilbert space and A ∈ B(H). Show that


(a) Image A = (ker A∗ )⊥ .
(b) ker A = (Image A∗ )⊥ .

Solution.
(a) Take any x ∈ Image A. Then there is a y ∈ H such that x = Ay. For any
z ∈ ker A∗ , we have

hx, zi = hAy, zi = hy, A∗ zi = hy, 0i = 0.

hence x ∈ (ker A∗ )⊥ . This proves that Image A ⊂ (ker A∗ )⊥ . Since (ker A∗ )⊥ is


closed, it follows that
(i) Image A ⊂ (ker A∗ )⊥ .
On the other hand, if x ∈ (Image A)⊥ , then for all y ∈ H, we have

0 = hAy, xi = hy, A∗ xi.

Therefore A∗ x = 0, that is, x ∈ ker A∗ . This prove that (Image A)⊥ ⊂ ker A∗ .
Taking orthogonal complements both sides, we obtain

(ii) (ker A∗ )⊥ ⊂ Image A ⊂ Image A.

From (i) and (ii) it follows that

Image A = (ker A∗ )⊥ .

(b) Replacing A by A∗ in (a), we get

(ker A)⊥ = Image A∗ .

Taking orthogonal complements both sides and using a result in Problem 39, we
obtain
ker A = (Image A∗ )⊥ = (Image A∗ )⊥ . ¥

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
94 CHAPTER 6. LINEAR OPERATORS ON HILBERT SPACES

Problem 92.(Integral operator)


Let (X, Ω, µ) be a measure space. Let k : X × X → F be an Ω × Ω-measurable
function for which there are constants c1 and c2 such that
Z
|k(x, y)|dµ(y) ≤ c1 a.e.[µ],
Z X

|k(x, y)|dµ(x) ≤ c2 a.e.[µ].


X

Consider the operator K : L2 (µ) → L2 (µ) defined by


Z
(Kf )(x) = k(x, y)f (y)dµ(y).

The function k is called the kernel of the operator K.



(a) Show that K is a bounded linear operator and kKk ≤ c1 c2 .
(b) Show that K ∗ is the integral operator with kernel k ∗ (x, y) = k(x, y).

Solution. R
(a) Linearity of K comes from linearity of the integral . It suffices to show that K
is bounded. Actually it must be shown first that Kf ∈ L2 (µ), but this will follow
from the argument that demonstrates the boundedness of K. If f ∈ L2 (µ),
Z
|Kf (x)| ≤ |k(x, y)| |f (y)|dµ(y)
Z
= |k(x, y)|1/2 |k(x, y)|1/2 |f (y)|dµ(y)
·Z ¸1/2 ·Z ¸1/2
2
≤ |k(x, y)|dµ(y) |k(x, y)| |f (y)| dµ(y)
·Z ¸1/2
√ 2
≤ c1 |k(x, y)| |f (y)| dµ(y) .

Hence
Z Z Z
2
|Kf (x)| dµ(y) ≤ c1 |k(x, y)| |f (y)|2 dµ(y)dµ(x)
Z Z
2
= c1 |f (y)| |k(x, y)|dµ(x)dµ(y)

≤ c1 c2 kf k2 .

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
95

Now this shows that the formula used to define Kf is finite a.e. [µ], and so

Kf ∈ L2 (µ) and kKf k2 ≤ c1 c2 kf k2 .

(b) By definition,
Z
hKf, gi = k(x, y)f (y)g(y)dµ(y)
Z
= f (y)k(x, y)g(y)dµ(y)
Z
∗ ∗
= hf, K gi, where K g(y) = k(x, y)g(y)dµ(y).

Hence, the kernel of K ∗ is k ∗ (x, y) = k(x, y). ¥

Problem 93. Let H = H ⊕ H where H be a Hilbert space. Let A ∈ B(H) and


B be the operator defined on H by
µ ¶
0 iA
B=
−iA∗ 0

Prove that kAk = kBk and that B is self-adjoint.

Solution. µ ¶ µ ¶
x1 y1
For any x = , y= with x1 , x2 , y1 , y2 ∈ H we have
x2 y2
¿µ ¶µ ¶ µ ¶À
0 iA x1 y1
hBx, yi = ,
−iA∗ 0 x2 y2

= hiAx2 , y1 i + h−iA x1 , y2 i
= hx2 , −iA∗ y1 i + hx1 , iAy2 i
¿µ ¶ µ ¶µ ¶À
x1 0 iA y1
= ,
x2 −iA∗ 0 y2
= hx, Byi.

Moreover,

kBxk2 = kiAx2 k2 + k − iA∗ x1 k2


≤ (max{kAk, kA∗ k})2 kxk2
= kAk2 kxk2 .

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
96 CHAPTER 6. LINEAR OPERATORS ON HILBERT SPACES
µ ¶
0
Hence, kBk ≤ kAk. Conversely, one can take x̃ = and obtain
x2

kB x̃k = kAx2 k ≤ kBkkx̃k = kBkkx2 k.

Therefore, kAk ≤ kBk. Finally, we obtain kAk ≤ kBk. ¥


————————–
Remark:
Note that norm on H ⊕ H is
k(a, b)k = kak + kbk, a, b ∈ H.

Problem 94. Let T be a self-adjoint operator on a Hilbert space H. Show that


its norm is given by
kT k = sup |hT x, xi|.
kxk=1

Solution.
For kxk = 1 we have

|hT x, xi| ≤ kT xkkxk = kT xk ≤ kT k.

Therefore
(i) sup |hT x, xi| ≤ kT k.
kxk=1

In order to establish the inverse inequality, we consider the case:


1 p
z ∈ H, kzk = 1, T z 6= 0 and u = T z where λ = kT zk.
λ
If we denote by α := supkxk=1 |hT x, xi|, then we have
­ ®
kT zk2 = T (λz), u
1 £­ ® ­ ®¤
= T (λz + u), λz + u − T (λz − u), λz − u
4
α£ ¤
≤ kλz + uk2 + kλz − uk2
4
α£ ¤
= kλzk2 + kuk2
2
α£ ¤
= kλk2 + kT zk2 = αkT zk.
2

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
97

This implies that, for any z ∈ H with kzk = 1, we have kT zk ≤ α, and hence

(ii) kT k ≤ α = sup |hT x, xi|.


kxk=1

(i) and (ii) completes the proof. ¥

Problem 95
Let H be a Hilbert space and A a positive self-adjoint operator on H. Prove that
the following assertions are equivalent:
(i) A(H) is dense in H.
(ii) Ker A = {0}.
(iii) A is positive definite, i.e., hAx, xi > 0, ∀x ∈ H \ {0}.

Solution.
• (i) ⇒ (ii)
Suppose Ax = 0. Then, for any y ∈ H,

hAx, yi = hx, Ayi = 0 ( since A is self-adjoint)


⇒ x⊥A(H)
⇒ x⊥A(H) = H ( since A(H) is dense)
⇒ x = 0.

• (ii) ⇒ (iii) √
Since A is positive, A = B exists. It is also a self-adjoint operator on H. To show
A is positive definite, we show hAx, xi = 0 ⇒ x = 0. Now

0 = hAx, xi = hB 2 x, xi = hB(Bx), xi = hBx, Bxi = kBxk2 .

This implies that Bx = 0. Therefore,

Ax = B(Bx) = 0.

Since Ker A = {0}, we have x = 0.


• (iii) ⇒ (i)
Assume that A(H) is not dense in H. Then there is x ∈ H \ {0} such that x⊥A(H).
In particular, x⊥Ax, i.e., hAx, xi = 0. But A is positive definite, so x = 0, a
contradiction. ¥

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
98 CHAPTER 6. LINEAR OPERATORS ON HILBERT SPACES

Problem 96
Let H be a Hilbert space. If A, B : H → H are self-adjoint operators with
0 ≤ A ≤ B and B is compact, prove that A is compact.

Solution.
Let (xn ) be a sequence in the closed unit ball BH . Since B is compact, there is a
subsequence (xnk ) such that (Bxnk ) converges. From the Cauchy-Schwarz inequality
we have
hBx, xi ≤ kBxk kxk, ∀x ∈ H.
It follows that
hBxnk − Bxmk , xmk − xnk i ≤ kBxmk − Bxnk k kxmk − xnk k
≤ kBxmk − Bxnk k (kxmk k + kxnk k)
≤ 2kBxmk − Bxnk k.
From 0 ≤ A ≤ B we get
hA(xnk − xmk ), xmk − xnk i ≤ hB(xnk − xmk ), xmk − xnk i
≤ 2kBxmk − Bxnk k.
On the other hand, we get
√ √ √
k A(x)k2 = h A(x), A(x)i
√ 2
= h A (x), xi
= hAx, xi
≤ kAxk kxk, ∀x ∈ H.
Then
√ √
k Axmk − Axnk k2 ≤ kA(xmk − xnk )k kxmk − xnk k
≤ kA(xmk − xnk )k(kxmk k + kxnk k)
≤ 2kA(xmk − xnk )k.
Therefore, √ √
k Axmk − Axnk k2 ≤ hA(xmk − xnk ), xmk − xnk i.
Hence √ √
k Axmk − Axnk k2 ≤ 2kBxmk − Bxnk k.

From this we see that the sequence ( Axnk ) is Cauchy, hence converges. The
√ √ 2
operator A is compact. And so is the operator A = A . ¥

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D

Chapter 7

Compact Operators

In this chapter we study general properties of compact operators on Banach and Hilbert spaces.
Spectral properties of these operators will be discussed later.

Definition 7 Let X and Y be Banach spaces. An operator T ∈ B(X, Y ) is called compact op-
erator if the image of every bounded set in X has compact closure in Y (relatively compact set).
Equivalently, T ∈ B(X, Y ) is compact if and only if for every bounded sequence (xn ) in X, (T xn )
has a convergent subsequence in Y .
The set of all compact operators is denoted by B0 (X, Y ).

Proposition 7 Let X and Y be Banach spaces. Then B0 (X, Y ) is a closed subspace of B(X, Y ).
That is, if (Tn ) is a sequence of compact operators and T ∈ B(X, Y ) such that kTn − T k → 0, then
T ∈ B0 (X, Y ).

Definition 8 (Finite rank operators)


An operator T : X → Y has finite rank if Image T := T (X) is finite dimensional.

Proposition 8 Let X and Y be Banach spaces. Every finite rank operator from X to Y is
compact.

w
Proposition 9 Let X and Y be Banach spaces, and T : X → Y be a compact operator. If xn → x
then T xn → T x.

Proposition 10 Let H and K be Hilbert spaces. Then T is compact if and only if T ∗ is compact.

Proposition 11 Let H and K be Hilbert spaces and T ∈ B(H, K). Then T is compact if and only
if for any sequence (xn ) ⊂ H converging weakly to x, the sequence (T xn ) converges (strongly) to
T x in K.

99
www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
100 CHAPTER 7. COMPACT OPERATORS

Problem 97. Let X be a Banach space. Prove that if T ∈ B(X) is arbitrary


and A ∈ B0 (X), then AT and T A are compact operators. (This is called the two
sides ideal property for compact operators).

Solution.
Suppose (xn ) is a sequence in H such that kxn k ≤ 1 for every n ∈ N. Since T is
continuous,
kT xn k ≤ kT k kxn k ≤ kT k, ∀n ∈ N.
If we set yn = TkTxnk , and then we have kyn k ≤ 1 for every n ∈ N. Since A is compact,
the sequence (Ayn ) has a convergent subsequence. Now we have

kT kAT xn
kT kAyn = = AT xn , ∀n ∈ N.
kT k

It follows that the sequence (AT xn ) also has a convergent subsequence. Thus AT is
compact. The similar argument for T A ¥

Problem 98.
(a) Let X be a Banach space. Show that the identity I : X → X is compact if
and only if X has finite dimensional.
(b) Let X, Y be Banach spaces and A ∈ B(X, Y ). Suppose that A has the
property:
∃c > 0 : kAxk ≥ ckxk, ∀x ∈ X,
Find condition(s) for X so that A can be a compact operator.

Solution.
(a) See Problem 16.
(b) First we note that A is injective. Indeed,

Ax = 0 ⇒ cx = 0 ⇒ x = 0.

Let Z = A(X), then U : X → Z defined by U (x) = A(x) is bijective. Let us


consider U −1 : Z → X. Clearly U −1 is linear. we claim:
1
(∗) kU −1 (y)k ≤ kyk, ∀y ∈ Z
c

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
101

Proof: Since y ∈ A(X), there is an x ∈ X such that A(x) = U (x) = y. This implies
that x = U −1 (y). By our hypothesis,

kAxk = kU (x)k = kyk ≥ ckxk.

Thus,
kyk ≥ ckU −1 (y)k.
Hence, (*) is proved. It follows that U −1 is linear and continuous. If U is a compact
operator, then by the ideal property for the compact operators (Problem 97), it
follows that I = U −1 U : X → X is compact, which means that X is finite dimen-
sional. Conversely, if X is finite dimensional, then every A ∈ B(X, Y ) is compact
(in a finite dimensional normed space, a set is compact if and only if it is closed and
bounded). Hence A is compact if and only if X is finite dimensional. ¥

Problem 99.
Let H and K be Hilbert spaces and A ∈ B(H, K). Show that A is compact if
and only if A∗ A is compact.

Solution.
• Suppose A ∈ B(H, K) is compact. Let (xn ) be a sequence in X converging weakly
to 0. We have
kA∗ Axn k ≤ kA∗ k kAxn k.
Since A is compact, Axn → 0 (strongly) in Y . Thus A∗ Axn → 0, and so A∗ A is
compact.
w
• Reciprocally, suppose A∗ A is compact. For any sequence (xn ) such that xn → 0,
we have
kAxn k2 = hAxn , Axn i = hxn , A∗ Axn i ≤ kA∗ Axn k kxn k.
Since kxk is uniformly bounded and A∗ A is compact, A∗ Axn → 0. Therefore,
Axn → 0, and hence A is compact. ¥

Problem 100.
Let X be c0 or `p , 1 ≤ p ≤ ∞. Consider the operator

U : X → X, U (x) = U (x1 , x2 , ...) = (0, x1 , 0, x3 , 0, x5 , ...).

Prove that U is not compact but U 2 is compact.

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
102 CHAPTER 7. COMPACT OPERATORS

Solution.
We first note that c0 and `p , 1 ≤ p ≤ ∞ (with appropriate norms) are Banach
spaces (see Problems 18,19). We have

U 2 (x) = U (U (x)) = U (0, x1 , 0, x3 , 0, x5 , ...) = (0, 0, ...).

Thus, U 2 = 0, therefore U 2 is compact.


On the other hand, if
en = (0, ..., 0, 1, 0, ...) ∈ X
| {z }
n

then
U (e2n−1 ) = e2n , ∀n ∈ N.
Now, we have explicitly

e2n = (0, ..., 0, 1, 0, ...), e2(n+k) = (0, ..., 0, 0, ..., 0, 1, 0, ...),


| {z } | {z } | {z }
2n 2n 2k

so that
e2(n+k) − e2n = (0, ..., 0, −1, 0, ..., 0, 1, 0, ...).
| {z } | {z }
2n 2k

For X = c0 or X = ` we have

ke2(n+k) − e2n k∞ = 1.

For X = `p , 1 ≤ p < ∞ we have

ke2(n+k) − e2n kp = 21/p .


³ ´
It follows that, in both cases, the sequence U (e2n−1 ) cannot have any convergent
subsequence. Thus, U is not a compact operator. ¥

Problem 101
Let 1 ≤ p < ∞, and λ = (λn )n∈N ⊂ K with supn∈N |λn | < ∞. We define the
multiplication operator

Mλ : `p → `p , Mλ (x) = (λ1 x1 , λ2 x2 , ...), x = (x1 , x2 , ...) ∈ `p .

Prove that:
(a) Mλ is continuous and kMλ k = supn∈N |λn |.
(b) Mλ is a compact operator if and only if λ ∈ c0 .

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
103

Solution.
(a) We have
|λn xn |p ≤ kλkp∞ |xn |p , ∀n ∈ N.
P∞ p
Since
P∞ the series n=1 |xn | converges and kλk∞ < ∞ by hypothesis, the series
p
n=1 |λn xn | converges. Moreover,
̰ !1/p
X
kMλ (x)k = |λn xn |p
n=1
à ∞
!1/p
X
≤ kλk∞ |xn |p
n=1
= kλk∞ kxk, ∀x ∈ `p .

This shows that Mλ is continuous and kMλ k ≤ kλk∞ . Also, for any n ∈ N,

|λn | = |Mλ (en )| ≤ kMλ k ken k = kMλ k.

Here en = (0, ..., 0, 1, 0, ...) ∈ `p . Therefore, kλk∞ = supn∈N |λn | ≤ kMλ k. Thus,
| {z }
n−1
kMλ k = supn∈N |λn |.
(b) Suppose Mλ is a compact operator. i.e., Mλ (B`p ) is relatively compact (B`p the
closed unit ball in `p ). Then

X
∀ε > 0, ∃nε ∈ N : |Mλ (x)|p ≤ εp , ∀x ∈ B`p .
k=nε

Let n ≥ nε . Then for en ∈ B`p we have



X
|Mλ (en )|p ≤ εp ,
k=nε

that is
n ≥ nε ⇒ |λn | < ε.
So λn → 0, that is, λ ∈ c0 .
Conversely, if λn → 0, then

∀ε > 0, ∃nε ∈ N : n ≥ nε ⇒ |λn | < ε.

Let x = (x1 , x2 , ...) ∈ B`p , then



X ∞
X ∞
X ∞
X
p p p p p
|Mλ (x)| = |λk xk | ≤ ε |xk | ≤ ε |xk |p ≤ εp .
k=nε k=nε k=nε k=1

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
104 CHAPTER 7. COMPACT OPERATORS

Therefore Mλ (B`p ) is relatively compact. ¥

Problem 102
Consider the linear operator defined by
µ ¶
2 2 ξ1 ξ2 ξ3
T : ` → ` , x = (ξ1 , ξ2 , ξ3 , ...) 7→ T x = , , , ... .
1 2 3

Show that T is compact.

Solution.
It is clear that T is linear. To show that it is compact, we will show that it is the
norm limit of a sequence of compact operators. Let
µ ¶
2 2 ξ1 ξ2 ξn
Tn : ` → ` , T n x = , , ..., , 0, 0, ... .
1 2 n
Then Tn is linear, bounded, and of finite rank so compact. Furthermore,
X∞
2 1 2
k(T − Tn )xk = |ξi |
i=n+1
i2

X
1
≤ 2
|ξi |2
(n + 1) i=1
kxk2
= .
(n + 1)2
Taking the supremum over all x of norm 1, we see that
1
kT − Tn k ≤ .
n+1
Hence, Tn → T in norm. Thus, T is compact. ¥

Problem 103
Let (cj )∞ 2
j=1 be a sequence of complex numbers. Define an operator D on ` by

Dx = (c1 x1 , c2 x2 , ...), x = (x1 , x2 , ...) ∈ `2 .

Prove that D is compact if and only if limj→∞ cj = 0.

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
105

Solution.
• We note that D is linear. To show that it is compact, we will show that it is the
norm limit of a sequence of compact operators. Suppose limj→∞ cj = 0. Define Dn
by
Dn = (c1 x1 , ..., cn xn , 0, 0, ...).
We obtain that
(D − Dn ) = (0, ..., 0.cn+1 xn+1 , cn+2 xn+2 , ...)
and moreover,
kD − Dn k = sup |cj | → 0 as n → ∞.
j≥n+1

Since each Dn has finite rank and hence is compact, the operator D is compact.
• Assume that (cj ) does not converge to zero as j → ∞. Then, for a given ε > 0,
there exists a subsequence (cjk ) such that |cjk | ≥ ε. Consider the sequence of vectors
(ej ) of the standard basis. We have kejk k = 1 and for any indices m, k we have

kDejm − Dejk k2 = kcjm ejm − cjk ejk k2 = |cjm |2 + |cjk |2 ≥ 2ε2 > 0.

We conclude that the sequence (Dejk ) does not contain a convergent subsequence
and thus the operator D is not compact. ¥

Trick used in problems 102 and 103 is called ”cut off ” method: From the sequence x = (x1 , x2 , ...)
we get the sequence (x1 , ..., xn , 0, 0, ...) by cutting off the tail of x.

Problem 104
Let g ∈ C[0, 1] be a fixed function. Consider the operator A ∈ B(C[0, 1]) defined
by
(Au)(s) := g(s)u(s),
i.e., the operator of multiplication by g. Is this operator compact?

Solution.
Note first that C[0, 1], equipped with the sup-norm, is a Banach space.
It is clear that if g ≡ 0 then A is compact. Let us prove that if g is not identically
zero then A is not compact. Indeed, since g is not identically zero, there exists a
subinterval [a, b] ⊂ [0, 1] such that

m := min |g(s)| > 0.


s∈[a,b]

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
106 CHAPTER 7. COMPACT OPERATORS

Consider the sequence (un ):


µ ¶
−a ns
un ∈ C[0, 1]; un (s) := sin 2 π ; s ∈ [0, 1], n ∈ N.
b−a
It is clear that (un ) is a bounded sequence. On the other hand, (Aun ) does not have
Cauchy subsequences. Indeed, take arbitrary k, n ∈ N with k > n. Let
1
sn := a + (b − a)
2n+1
Then sn ∈ [a, b] and
kAuk − Aun k = max |g(s)(uk (s) − un (s))|
s∈[a,b]

≥ m max |uk (s) − un (s)|


s∈[a,b]

≥ m|uk (sn ) − un (sn )|


= m| sin(2k−n−1 π) − sin(π/2)|
= m|0 − 1| = m > 0.
Hence (Aun ) cannot have any convergent subsequence, A is not compact. ¥

Problem 105
Given k ∈ L2 ([0, 1] × [0, 1]), define the operator A : L2 ([0, 1]) → L2 ([0, 1]) by
Z 1
(Af )(x) = k(x, y)f (y)dy.
0

(a) Show that A is bounded.


(b) Under what condition on k, the operator A is self-adjoint.
(c) Show that A is compact.

Solution.
(Look at Problem 92! They are different!)
(a) We estimate kAk to see A is bounded.
Z 1 ¯Z 1 ¯2
¯ ¯
2
kAf k = ¯ k(x, y)f (y)dy ¯ dx
¯ ¯
0 0
Z 1 µZ 1 ¶ Z 1
2
≤ |k(x, y)| dy dx. |f (y)|2 dy (Cauchy-Schwarz)
0 0 0
Z 1Z 1
≤ kf k2 . |k(x, y)|2 dydx.
0 0

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
107
R1R1
Since k ∈ L2 ([0, 1] × [0, 1]), 0 0
|k(x, y)|2 dydx < ∞; hence,
µZ 1 Z 1 ¶1/2
2
kAk ≤ |k(x, y)| dydx < ∞.
0 0

Thus A is bounded.
(b) We have
Z 1
(Af )(x) = k(x, y)f (y)dy.
0
Z 1

(A g)(x) = k(x, y)g(y)dy.
0

Therefore,
Z 1 Z 1
hAf, gi = k(x, y)f (y)dy g(x) dx
0 0
Z 1 Z 1
= f (y) k(x, y) g(x) dxdy
0 0
= hf, A∗ gi.

Hence, A is self-adjoint if
k(x, y) = k(y, x).
(c) Let (uj )∞ 2
j=1 be an orthonormal basis in L [0, 1]. Then


X Z 1
k(x, y) = kj (y)uj (x), where kj (y) = k(x, y)uj (x) dx,
j=1 0

for almost all y. Due to the Parseval identity, we have, for almost all y
Z 1 ∞
X
|k(x, y)|2 dx = |kj (y)|2 ,
0 j=1

and Z Z ∞ Z
1 1 X 1
2
(1) |kj (y)| dxdy = |kj (y)|2 dy.
0 0 j=1 0

We now define the following operator of rank N


Z 1
kN f (x) = kN (x, y)f (y)dy,
0

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
108 CHAPTER 7. COMPACT OPERATORS
PN
where kN (x, y) = kj (y)uj (x). By Cauchy-Schwarz inequality we obtain
j=1
Z 1 ¯Z 1 ¯2
¯ ¯
k(A − kN )f k2 = ¯ (k(x, y) − k (x, y))f (y)dy ¯ dx
¯ N ¯
µ0Z 1 Z0 1 ¶ µZ 1 ¶
2 2
≤ |k(x, y) − kN (x, y)| dxdy |f (y)| dy .
0 0 0
µZ 1 Z 1 ¶
2 2
≤ kf k |k(x, y) − kN (x, y)| dxdy .
0 0

Thus by using that the right hand side in (1) is absolutely convergent, we find
Z 1Z 1
2
k(A − kN )k ≤ |k(x, y) − kN (x, y)|2 dxdy
0 0
Z 1 Z 1 Z 1 Z 1 N
X
2
= |k(x, y)| dxdy − k(x, y) kj (y)uj (x) dxdy
0 0 0 0 j=1
Z 1 Z 1 N
X N Z
X 1
− k(x, y) kj (y)uj (x)dxdy + |kj (y)|2 dy
0 0 j=1 j=1 0
Z 1 Z 1 N
X
2
= |k(x, y)| dxdy − |kj (y)|2 dy → 0 as N → ∞. ¥
0 0 j=1

Problem 106
Part I
Consider the operator
Z x
U : C[0, 1] → C[0, 1] defined by (U f )(x) = et f (t)dt, x ∈ [0, 1],
0

and the sequence of operators


Z Ã n !
x X tk
Un : C[0, 1] → C[0, 1] defined by (Un f )(x) = f (t)dt, x ∈ [0, 1].
0 k=0
k!

Prove that limn→∞ kUn − U k = 0.


Part II
1. Let M be a set of C 1 -functions f on [0, 1]. Prove that M is relatively compact
in C[0, 1] if f satisfies following conditions
Z 1
|f (0)| ≤ k1 and |f 0 (x)|2 dx ≤ k2
0

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
109

where k1 , k2 are positive constants. (Hint: Use Arzela-Ascoli theorem).


2. Show that the operator U in Part I is compact.

Solution.
Part I Rx
From Calculus we know that if φ ∈ C[0, 1] then x 7→ 0 φ(t)dt is continuous.
P∞ Hence,
tn
U, Un take their values in C[0, 1]. Using the Taylor expansion et = n=0 n! we
obtain
Z xà X∞ k
!
t
(U f − Un f )(x) = f (t)dt, ∀x ∈ [0, 1].
0 k=n+1
k!

Then
Z ∞
à !
Xx
tk
|(U f − Un f )(x)| ≤ |f (t)|dt
0 k=n+1
k!
Z 1Ã X ∞
!
tk
≤ kf k∞ dt, ∀x ∈ [0, 1].
0 k=n+1
k!

Thus,

Z

à !
X 1
tk
kUn − U k ≤ kf k∞ dt, ∀f ∈ C[0, 1],
0 k=n+1
k!
Z 1Ã X ∞
!
tk
≤ dt.
0 k=n+1
k!

P∞ tk
But if un (t) = k=n+1 k! then un : [0, 1] → R and

X∞
1
|un (t)| ≤ → 0,
k=n+1
k!
R1
that is, un → 0 uniformly on [0, 1]. Hence, 0
un (t)dt → 0 as n → ∞. Thus,

lim kUn − U k = 0.
n→∞

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
110 CHAPTER 7. COMPACT OPERATORS

Part II
1. For all x, y ∈ [0, 1] with x < y, and all f ∈ M , we have
¯Z y ¯
¯ ¯
|f (x) − f (y)| = ¯¯ |f (x)|¯¯ dt
0
x
µZ y ¶1/2 µZ y ¶1/2
0 2
≤ |f (x)| dt 1dt
x x
µZ y ¶1/2
√ 0 2
≤ y−x |f (x)| dt
x
p √
≤ k2 y − x.

This shows that M is qui-continuous.


Now for all x ∈ M , and all f ∈ M , we have

|f (x)| ≤ |f (x) − f (0)| + |f (0)|


p
≤ k1 + k2 .

So M is uniformly bounded. Thus, by Arzela-Ascoli Theorem1 , M is relatively


compact in C[0, 1].
R1
2. For kf k ≤ 1, let g(x) = 0 etx f (t)dt, ∀x ∈ [0, 1]. Using the differentiation
theorem for the Riemann integral with parameter, we have
Z 1 Z 1
0 ∂ tx
g (x) = (e f (t))dt = tetx f (t)dt,
0 ∂x
Z 1 Z 10
|g 0 (x)| ≤ tetx |f (t)|dt ≤ tetx dt ≤ e, ∀x ∈ [0, 1].
0 0

It follows that the set of all functions g is uniformly Lipschitz. We also have
¯Z 1 ¯ Z 1
¯ ¯
|g(0)| = ¯¯ f (t)dt¯¯ ≤ |f (t)|dt ≤ kf k ≤ 1, ∀g.
0 0

From previous question, it follows that A = {U f : kf k ≤ 1} is relatively compact.


Thus the operator U is compact. ¥

1
Arzela-Ascoli Theorem: Let (X, d) be a compact metric space and A ⊂ C(X). Then the
following assertions are equivalent:
1. A is relatively compact.
2. A is uniformly bounded and equi-continuous,
3. Any sequence (fn ) ⊂ A contains a uniformly convergent subsequence.

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
111

Problem 107
(a) Let X be an infinite dimensional Banach space, and A be a compact operator
on X. Prove that there is y ∈ X such that the equation A(x) = y has no solution,
i.e., A is not surjective.
(b) Let 1 ≤ p < ∞, and the operator
³ x x ´
2 3
U : ` → ` , U (x) = x1 , , , ... , x = (x1 , x2 , ...) ∈ `p .
p p
2 3
Find an element y ∈ `p for which the equation U (x) = y has no solution.
(c) Consider the operator
³ x x ´
2 3
A : c00 → c00 defined by A(x) = A(x1 , x2 , ...) = x1 , , , ... .
2 3
Prove that A is compact and bijective. On c00 we have the `∞ - norm.

Solution.
(a) Let us suppose, for a contradiction, that A is surjective. From the open mapping
theorem it follows that A is an open operator, in particular A(B(0; 1)) ⊂ X is an
open set, i.e.,

∃ε > 0 : B(0; ε) = εB(0; 1) ⊂ A(B(0; 1)),

where B(0; 1) = {x ∈ X : kxk < 1}. Since A is compact it follows that A(B(0; 1))
is relatively compact. Hence, B(0; ε) is relatively compact, from whence B(0; 1) is
relatively compact, therefore compact. But if B(0; 1) is compact, then, by Problem
16, X must be finite dimensional, which is a contradiction. Thus, A is not surjective.
¡ ¢
(b) Choose y = 1, 21α , 31α , ... ∈ `p . If x = (x1 , x2 , ...) ∈ `p has the property that
U (x) = y, then

1 1
xn = α , ∀n ∈ N,
n n

that is,

1
xn = , ∀n ∈ N.
nα−1

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
112 CHAPTER 7. COMPACT OPERATORS
P
Since x = (x1 , x2 , ...) ∈ `p , the generalized harmonic series ∞ 1 2
n=1 n(α−1)p converges ,
1
therefore, (α − 1)p > 1 ⇔ α > 1 + p . From here, it follows that for
µ ¶
1 1
y= 1, 1 , 1 , ... ,
2 p+1 3 p+1
the equation U (x) = y has no solution.
(c) For n ∈ N, consider
³ x xn ´
2
An : c00 → c00 , An (x) = An (x1 , x2 , ...) = x1 , , .., , 0, ... .
2 n
Then
1
kA − An k = , ∀n ∈ N.
n+1
Therefore An → A in norm. Since every An is a finite rank operator, so An is
compact. The sequence of compact operators (An ) converges to A in norm, so A
must be compact. The fact that A is bijective is obvious from the expression which
defines A. ¥

2
P∞ 1
The generalized harmonic series n=1 nαp converges if and only if

1
αp > 1 ⇔ α > .
p

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D

Chapter 8

Bounded Operators on Banach


Spaces and Their Spectra

Review:
1. Definitions

Let X be a Banach space and T ∈ B(X), λ ∈ C.


• Resolvent and spectrum of T :
Set Tλ = T − λI. The set ρ(T ) of all λ such that Tλ has an inverse Rλ (T ) = (T − λI)−1 is called
the resolvent of T . The set σ(T ) = C \ ρ(T ) is called the spectrum of T .
• Eigenvalues and eigenvectors of T :
An x 6= 0 which satisfies T x = λx for some λ is called an eigenvector of T . The corresponding λ
is an eigenvalue of T . It is evident that λ ∈ σ(T ).
2. Basic properties

Theorem 12 (Spectrum)
If T is a bounded linear operator on a Banach space X, then its spectrum σ(T ) is compact and lies
in the disk given by
|λ| ≤ kT k.

Theorem 13 (Resolvent equation)


Let T ∈ B(X, X) where X is a Banach space. Then

1. The resolvent Rλ (T ) satisfies the following equation called the resolvent equation

Rµ − Rλ = (µ − λ)Rµ Rλ for µ, λ ∈ ρ(T ).

2. Rλ commutes with any S ∈ B(X, X) which commutes with T .


3. We have
Rµ Rλ = Rλ Rµ for µ, λ ∈ ρ(T ).

113
www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
114CHAPTER 8. BOUNDED OPERATORS ON BANACH SPACES AND THEIR SPECTRA

3. Classification of spectrum

• λ ∈ C is called a regular point of A ∈ B(X) iff (A − λI)−1 exists and is bounded.


• If λ is not a regular point, it is called a spectrum point. All such points form the spectrum
σ(A).
• Every λ ∈ C with |λ| > kAk is a regular point.
• Classification of spectrum:
1. The point spectrum: σp (A) is the set of eigenvalues of A.
2. The continuous spectrum: λ ∈ σc (A) iff λ ∈ σ(A) \ σp (A) and Image(A − λI) is dense
in X.
3. The residual spectrum: σr (A) = σ(A) \ (σp (A) ∪ σc (A)). If λ ∈ σr (A) then

Image(A − λI) 6= X and ker(A − λI) = 0

4. Spectral radius

Definition 9 Let T ∈ B(X, X) where X is a Banach space. The spectral radius rσ (T ) of T is the
radius of the smallest closed disk centered at the orgin and containing σ(T ).

rσ (T ) := sup |λ|.
λ∈σ(T )

Formula for spectral radius:


It can be shown that p
rσ (T ) = lim n
kT n k.
n→∞

5. Spectral mapping theorem

Theorem 14 (Spectral theorem for polynomials)


Let T ∈ B(X, X) where X is a Banach space, and

p(λ) = αn λn + αn−1 λn−1 + ... + α0 , αn 6= 0.

Then
σ(p(T )) = p(σ(T )),
that is, the spectrum of the operator

p(T ) = αn T n + αn−1 T n−1 + ... + α0 I

consists precisely of all those values which the polynomial p assumes on the spectrum σ(T ) of T .

∗∗∗∗∗

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
115

Problem 108
Let X be a Banach space. Suppose that A ∈ B(X) is an invertible operator.
Show that
σ(A−1 ) = {λ−1 : λ =
6 0, λ ∈ σ(A)}.

Solution.
For λ 6= 0, we can write

A−1 − λ−1 I = (λI − A)λ−1 A−1 .

From this equality we conclude that A−1 − λ−1 I is invertible if and only if A − λI
is invertible. Hence, we have

σ(A−1 ) = {λ−1 : λ ∈ σ(A)}. ¥

Problem 109
Let X be a Banach space, let A ∈ B(X) and λ ∈ C. Assume that there exists a
sequence (xn ) in X such that

kxn k = 1, ∀n ∈ N and Axn − λxn → 0 as n → ∞.

Prove that λ ∈ σ(A).

Solution.
Assume that A − λI is invertible. Then, there exists a number c > 0 such that (see
problem 75)
k(A − λI)xk ≥ ckxk, ∀x ∈ X.
Replace x by xn with kxn k = 1 for all n, we have

kAxn − λxn k = k(A − λI)xn k ≥ ckxn k = c.

This contradicts the condition in the statement of the problem. ¥

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
116CHAPTER 8. BOUNDED OPERATORS ON BANACH SPACES AND THEIR SPECTRA

Problem 110
Let (an ) and (bn ) be complex sequences such that
(n→∞) (n→∞)
|an−1 | > |an | −−−−→ 0 and |bn−1 | > |bn | −−−−→ 0.

Consider the operator T : `2 → `2 defined by

T x = (a1 x1 , a2 x2 + b1 x1 , a3 x3 + b2 x2 , ...), x = (xj ) ∈ `2 .

(a) Show that T is compact.


(b) Find all eigenvalues and eigenvectors of T .

Solution.
(a) Let the sequence of operators Tn : `2 → `2 , n = 1, 2, ... be defined by for any
x ∈ `2 : (
(T x)j if 1 ≤ j ≤ n
(Tn x)j =
0 if j > n.
All the Tn ’s are operators of finite rank and hence compact. Moreover, we have
kTn x − T xk ≤ (|an+1 | + |bn |)kxk,
which implies that
kTn − T k → 0 as n → ∞.
But a uniform limit of a sequence of compact operators is compact. Hence, T is
compact.
(b) Suppose λ ∈ C is an eigenvalue of T and that x 6= 0 is the corresponding
eigenvector. Then
¡ ¢
0 = T x − λx = (a1 − λ)x1 , (a2 − λ)x2 + b1 x1 , ..., (an − λ)xn + bn−1 xn−1 , ... .
If λ coincides with none of the an ’s, then x = 0: impossible. So it is necessary that
λ = an for some n. In this case, xn can be chosen arbitrary, and x1 = ... = xn−1 = 0,
and for k = 1, 2, ... we have 0 = (an+k −λ)xn+k +bn+k−1 xn+k−1 . If we choose xn = 1
then we get
bn+k−1 bn+k−2 bn
xn+k = ... , k = 1, 2, ...
λ − an+k λ − an+k−1 λ − an+1
Thus, for any n, λ = an is a simple eigenvalue. The corresponding eigenvector is
x = (0, ..0, xn , xn+1 , ...) defined as above. ¥

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
117

Problem 111
Let X be a Banach space and let A ∈ B(X) such that An = 0 for some n ∈ N
(A is nilpotent). Find σ(A).

Solution.
The spectral mapping theorem implies
{λn : λ ∈ σ(A)} = σ(An ) = σ(0) = {0}.
Therefore,
λ ∈ σ(A) ⇔ λn = 0 ⇔ λ = 0.
Thus, σ(A) = {0}. ¥

Problem 112
Let P ∈ B(X) be a projection, i.e., a linear operator on X such that P 2 = P .
Construct the resolvent R(P ; λ) of P

Solution.
If P = 0, then obviously
P − λI = −λI; σ(P ) = {0}; R(P ; λ) := (P − λI)−1 = −λ−1 I.
If P = I, then
P − λI = (1 − λ)I; σ(P ) = {1}; R(P ; λ) := (P − λI)−1 = (1 − λ)−1 I.
Suppose P is non- trivial, i.e., P 6= 0, I. Take any λ 6= 0, 1. Then using the equalities
P 2 = P ; Q2 = Q and QP = P Q where Q = I − P , we obtain
³ ´ ³ ´
(1 − λ)−1 P − λ−1 Q (P − λI) = (1 − λ)−1 P − λ−1 Q ((1 − λ)P − λQ)
= P + Q = I.
Similarly, we have ³ ´
(P − λI) (1 − λ)−1 P − λ−1 Q = I.
Thus, ³ ´
R(P ; λ) = (1 − λ)−1 P − λ−1 Q = λ−1 (1 − λ)−1 P − I . ¥

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
118CHAPTER 8. BOUNDED OPERATORS ON BANACH SPACES AND THEIR SPECTRA

Problem 113
Let CR be the space of all continuous and bounded functions x(t) on R with norm
kxk = supR |x(t)|. On the space CR we define the operator A by

(Ax)(t) = x(t + c),

where c ∈ R is a constant. Prove that

σ(A) = {λ ∈ C : |λ| = 1}.

Solution.
Notice that
kAxk = sup |x(t + c)| = sup |x(τ )| = kxk.
t∈R τ ∈R

It follows that kAk = 1 and therefore all of the point of {λ ∈ C : |λ| > 1} are
regular points of A. The operator A is invertible since the operator defined by

(A−1 x)(t) = x(t − c)

is bounded and is the inverse of A. Next kA−1 k = 1 and hence all of the point of
{λ ∈ C : |λ| > 1} are regular points of A−1 . From Problem 65 we deduce that all
of the points {λ ∈ C : |λ| < 1} are regular points of A.
Consider |λ| = 1. This means that

λ = eiϕ , 0 ≤ ϕ ≤ 2π.

Set a = c
and xa (t) = eat . We obtain xa ∈ CR and

(Axa )(t) = ea(t+c) = eat eac = λxa .

This means that λ is an eigenvalue of A. Thus,

σ(A) = σp (A) = {λ ∈ C : |λ| = 1}. ¥

Problem 114
(a) Let H be a Hilbert space and a, b ∈ H. Consider the rank-one operator

U : H → H, U (x) = hx, aib, x ∈ H.

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
119

Calculate kU k and the spectral radius r(U ). Show that

r(U ) = kU k ⇔ a, b are linearly independent.

(b) Let X be a Banach space, x∗ ∈ X ∗ and y ∈ X. Consider the rank-one


operator
V : X → X, V (x) = x∗ (x)y, x ∈ X.
Calculate kV k and the spectral radius r(V ). Show that

r(V ) = kV k ⇔ |x∗ (y)| = kx∗ k kyk.

Solution.
(a) We have
kU (x)k = |hx, ai| kbk ≤ kak kbk kxk, ∀x ∈ H.
Hence, kU k ≤ kak kbk. We also have

kU k kak ≥ kU (a)k = |ha, ai| kbk = kak2 kbk.

Therefore, kU k ≥ kak kbk. Thus,

kU k = kak kbk.

For x ∈ H, let y = U (x). Then

U 2 (x) = U (y) = hy, aib, where hy, ai = hU x, ai = hx, aiha, bi.

Denoting λ = ha, bi, we have

U 2 (x) = λhx, aib = λU (x), ∀x ∈ H.

Therefore
U 2 = λU.
From here, by induction, we get

U n = λn−1 U, ∀n ∈ N.

Now ³ ´
1 n−1 1
r(U ) = lim (kU kn ) n = lim |λ| n kU k n = |λ| = |ha, bi|.
n→∞ n→∞

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
120CHAPTER 8. BOUNDED OPERATORS ON BANACH SPACES AND THEIR SPECTRA

The last assertion is a consequence of the fact proved above and the fact that in
the Cauchy-Schwarz inequality we have equality if and only if {a, b} are linearly
independent.
r(U ) = kU k ⇔ kak kbk = |ha, bi| ⇔ a, b are linearly independent.
(b) We have
kV (x)k = |x∗ (x)| kyk ≤ kx∗ k kyk kxk, ∀x ∈ X.
Therefore, kV k ≤ kx∗ k kyk.
If y = 0 then kV k = 0. Suppose y 6= 0. For x ∈ X, we have
|x∗ (x)| kyk = kV (x)k ≤ kV k kxk.
Then
kV k kV k
|x∗ (x)| ≤ kxk ⇒ kx∗ k ≤
kyk kyk
⇒ kV k ≥ kx∗ k kyk.
Thus,
kV k = kx∗ k kyk.
To calculate r(V ) we use the same procedure as in (a). For x ∈ H, let z = V (x).
Then
V 2 (x) = V (z) = x∗ (z)y
z = V (x) = x∗ (x)y,
x∗ (z) = x∗ (x)x∗ (y).
Therefore,
V 2 (x) = x∗ (y)x∗ (x)y = λx∗ (x)y, λ = x∗ (y).
Thus, V 2 = λV . And from here, by induction, we get
V n = λn−1 V, ∀n ∈ N.
And as above we obtain
r(V ) = |λ| = |x∗ (y)|. ¥

Problem 115
Let k ∈ C([0, 1] × [0, 1]) be a given function. Consider the operator
Z s
B ∈ B(C[0, 1]) defined by (Bu)(s) = k(s, t)u(t)dt.
0

Find σ(B) and r(B).

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
121

Solution.
Let us prove by induction that

Mn n
(∗) |(B n u)(s)| ≤ s kuk∞ , ∀s ∈ [0, 1], ∀n ∈ {0, 1, 2, ...}
n!
where
M := max |k(s, t)|.
(s,t)∈[0,1]2

For n = 0, then B 0 = I, and (*) is trivial. Suppose (*) holds for n = k. Then for
n = k + 1 we have
¯Z s ¯
¯ k+1 ¯ ¯ ¯
¯(B u)(s)¯ = ¯ k(s, t)(B u)(t)dt¯¯
k
¯
Z 0s
¯ ¯
≤ |k(s, t)| ¯(B k u)(t)¯ dt
0
Z s ¯ k ¯
≤ M ¯(B u)(t)¯ dt
Z0 s
Mk k
≤ M t kuk∞ dt
0 k!
k+1 Z s
M
= kuk∞ tk dt
k! 0
k+1
M
= sk+1 kuk∞ , ∀s ∈ [0, 1].
(k + 1)!

Hence, (*) is proved by induction.


It follows from (*) that

Mn
kB n uk∞ ≤ kuk∞ , ∀u ∈ C[0, 1],
n!
i.e.,
Mn
kB n k ≤ , ∀n ∈ {0, 1, 2, ...}.
n!
Therefore
M
r(B) = lim kB n k1/n ≤ lim = 0.
n→∞ n→∞ (n!)1/n
Since r(B) = 0, σ(B) cannot contain nonzero elements. Taking into account that
σ(B) is nonempty, we conclude that σ(B) = {0}. ¥

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
122CHAPTER 8. BOUNDED OPERATORS ON BANACH SPACES AND THEIR SPECTRA

Problem 116
Let X be a Banach space and A, B ∈ B(X). Suppose AB = BA. Prove that

r(A + B) ≤ r(A) + r(B),

where r(T ) is the spectral radius of an operator T ∈ B(X).

Solution.
Recall
r(T ) := sup{|λ| : λ ∈ σ(T )} and r(T ) = lim kT n k1/n .
n→∞

Take an arbitrary ε > 0. The spectral radius formula implies that


¡ ¢n ¡ ¢n
kAn k ≤ r(A) + ε , kB n k ≤ r(B) + ε

for sufficiently large n ∈ N. Therefore there exists a constant M ≥ 1 such that


¡ ¢n ¡ ¢n
kAn k ≤ M r(A) + ε , kB n k ≤ M r(B) + ε , ∀n ∈ N.

Since AB = BA, we have


n
X
n n!
(A + B) = An−k B k .
k=0
k!(n − k)!

Hence
n
X
n n!
k(A + B) k ≤ kAn−k k kB k k
k=0
k!(n − k)!
n
X
2 n! ¡ ¢k
≤ M r(A) + ε)n−k (r(B) + ε
k=0
k!(n − k)!
¡ ¢n
= M 2 r(A) + r(B) + 2ε , ∀n ∈ N.

Consequently,

r(A + B) = lim k(A + B)n k1/n ≤ r(A) + r(B) + 2ε, ∀ε > 0.


n→∞

This implies that


r(A + B) ≤ r(A) + r(B). ¥

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
123

Problem 117
Let K ⊂ C be an arbitrary non-empty compact set. Construct an operator
B ∈ B(`p ), 1 ≤ p ≤ ∞, such that σ(B) = K.

Solution.
Let {λk }k∈N be a dense subset of K. (Recall that every metric compact space is
separable, that is, it contains a countable dense subset.) Consider the operator
B : `p → `p defined by

Bx = (λ1 x1 , λ2 x2 , ...), ∀x = (x1 , x2 , ...) ∈ `p .

Then B is a bounded operator and λk ’s are its eigenvalues. Consequently,

{λk }k∈N ⊂ σ(B).

Since σ(B) is closed and {λk }k∈N is dense in K,

K ⊂ σ(B).

On the other hand, let λ ∈ C \ K. Then d := inf k∈N |λk − λ| > 0 and B − λI has a
bounded inverse (B − λI)−1 : `p → `p defined by
µ ¶
−1 1 1
(B − λI) x = x1 , ..., xk , ... , ∀x = (x1 , x2 , ...) ∈ `p .
λ1 − λ λk − λ

Hence, λ ∈
/ σ(B). Therefore, σ(B) ⊂ K. Finally,

σ(B) = K. ¥

Problem 118
Let g ∈ C[0, 1] be a fixed function and A ∈ B(C[0, 1]) be defined by

(Af )(t) = g(t)f (t), t ∈ [0, 1].

Find σ(A) and construct effectively the resolvent R(A; λ). Find the eigenvalues
and eigenvectors of A.

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
124CHAPTER 8. BOUNDED OPERATORS ON BANACH SPACES AND THEIR SPECTRA

Solution.
Let λ ∈ C, λ ∈
/ g([0, 1]) := {g(t); t ∈ [0, 1]}. Then, since g ∈ C[0, 1],

1
∈ C[0, 1]
g−λ

and A − λI has an inverse

R(A; λ) = (A − λI)−1 ∈ B(C[0, 1])

defined by
R(A; λ)f (t) = (g(t) − λ)−1 f (t), t ∈ [0, 1].

Hence, σ(A) ⊂ g[0, 1].


Suppose now λ ∈ g([0, 1]), i.e., λ = g(t0 ) for some t0 ∈ [0, 1]. Then

(A − λI)f (t0 ) = (g(t0 ) − λ)f (t0 ) = 0,

i.e., Image(A − λI) consists of functions vanishing at t0 .


Consequently, Image(A − λI) 6= C[0, 1] and A − λI is not invertible. Therefore,
g([0, 1]) ⊂ σ(A). Finally,
σ(A) = g([0, 1]).

Take an arbitrary λ ∈ g([0, 1]). Let g −1 (λ) := {τ ∈ [0, 1] : g(τ ) = λ}. The equation
Af = λf , i.e., (g(t) − λ)f (t) = 0 is equivalent to f (t) = 0, ∀t ∈ [0, 1] \ g −1 (λ) . If
g −1 (λ) contains an interval of positive length, then it is easy to see that the set

{f ∈ C[0, 1] \ {0} : f (t) = 0, ∀t ∈ [0, 1] \ g −1 (λ)}

is non-empty and coincides with the set of all eigenvectors corresponding to the
eigenvalues λ. If g −1 (λ) does not contain an interval of positive length, then [0, 1] \
g −1 (λ) is dense in [0, 1] and f (t) = 0, ∀t ∈ [0, 1] \ g −1 (λ) implies by continuity that
f ≡ 0. In this case λ is not an eigenvalue. ¥

Problem 119
Let X be a Banach space and A, B ∈ B(X). Show that for any λ ∈ ρ(A) ∩ ρ(B),

R(B; λ) − R(A; λ) = R(B; λ)(A − B)R(A; λ).

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
125

Solution.

£ ¤
R(B; λ)(A − B)R(A; λ) = R(B; λ) (A − λI) − (B − λI) R(A; λ)
£ ¤
= R(B; λ)(A − λI) − R(B; λ)(B − λI) R(A; λ)
= R(B; λ)(A − λI)R(A; λ) − R(B; λ)(B − λI)R(A; λ)
= R(B; λ) − R(A; λ). ¥

Problem 120
Let k ∈ C([0, 1] × [0, 1]) be given. Consider the operator B ∈ B(C[0, 1]) defined
by Z s
(Bu)(s) = k(s, t)u(t)dt.
0

Find the spectral radius of B. What is the spectrum of B. ( Hint: Prove by


induction that
Mn n
|(B n u)(s)| ≤ s kuk∞ , ∀n ∈ N,
n!
for some constant M > 0).

Solution.
Let us first prove by induction that

Mn n
|(B n u)(s)| ≤ s kuk∞ , ∀s ∈ [0, 1], ∀n ∈ N, (1)
n!

where
M := max |k(s, t)|.
(s,t)∈[0,1]2

(1) is true for n = 1. Indeed,

Z s Z s
M1 1
|(Bu)(s)| ≤ M |u(t)|dt ≤ M kuk∞ dt = s kuk∞ .
0 0 1!

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
126CHAPTER 8. BOUNDED OPERATORS ON BANACH SPACES AND THEIR SPECTRA

Suppose (1) holds for n = k. Then for n = k + 1 we have


¯Z s ¯
¯ ¯
|(B u)(s)| = ¯¯
k+1
k(s, t)(B u)(t)dt¯¯
k

Z 0s
≤ |k(s, t)||(B k u)(t)|dt
0
Z s
≤ M |(B k u)(t)|dt
Z0 s k
M k
≤ M t kuk∞ dt
0 k!
Z s
M k+1
= kuk∞ tk dt
k! 0
M k+1 k+1
= s kuk∞ , ∀s ∈ [0, 1].
(k + 1)!
Hence, (1) is true for n = k + 1. Thus (1) is proved by induction.
It follows from (1) that
Mn
kB n uk∞ ≤ kuk∞ , ∀u ∈ C[0, 1].
n!
It follows that
Mn
kB n k ≤ , ∀n ∈ N.
n!
Therefore,
M
r(B) = lim kB n k1/n ≤ lim = 0.
n→∞ n→∞ (n!)1/n

Since r(B) = 0, σ(B) cannot contain non-zero elements. Taking into account that
σ(B) is not empty, we conclude that σ(B) = {0}. ¥

Problem 121
Determine the spectra of the left and the right shift operators on `2 :

R(x1 , x2 , x3 , ....) = (0, x1 , x2 , ...),


L(x1 , x2 , x3 , ....) = (x2 , x3 , x4 , ....).

Classify them into point, continuous, and residual spectrum.

Solution.
We have shown that kRk = kLk = 1 (problem 46). It follows that
{λ ∈ C : |λ| > 1} ⊂ ρ(R) and {λ ∈ C : |λ| > 1} ⊂ ρ(L).

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
127

Now I prove the following four claims, and then I will state the conclusion.
Claim(1): R − λI is injective (one-to-one) for all λ ∈ C such that |λ| ≤ 1.
Proof.
If λ = 0, then
Rx = 0 ⇒ xi = 0, ∀i ∈ N.
Hence x = 0, and so R − λI is injective.
Suppose 0 < |λ| ≤ 1. Then

(R − λI)x = 0 ⇒ (0, x1 , x2 , ...) = (λx1 , λx2 , λx3 , ....).

Since λ 6= 0, this implies that x1 = x2 = ... = 0. Hence x = 0, and R − λI is


injective. ¤
Claim(2): R − λI is not surjective (onto) for all λ ∈ C such that |λ| ≤ 1.
Proof.
Note that if λ = 0, then e1 = (1, 0, ...) ∈
/ Image(R − λI) = Image R.
Suppose 0 < |λ| ≤ 1. Then
1
(R − λI)x = e1 ⇒ xn = − , ∀n ∈ N
λn
X∞ X∞ µ ¶n
2 2 1
⇒ kxk = |xn | = 2
.
n=1 n=1
|λ|
1
The above series cannot be convergent because 0 < |λ| ≤ 1 implies that |λ|2
≥ 1.
And hence e1 ∈
/ Image(R − λI). Therefore, R − λI is not surjective. ¤
Claim(3): L − λI injective (one-to-one) for all λ ∈ C such that |λ| = 1.
Proof.
Suppose it was not injective. There would be some nonzero x ∈ `2 such that
(L − λI)x = 0. Then

(x2 , x3 , x4 , ....) = (λx1 , λx2 , λx3 , ....).

Hence
xn = λn−1 x1 , ∀n ∈ N.
Since x 6= 0, x1 6= 0. Since |λ| = 1, we have

X ∞
X
2 2n
kxk = |x1 | |λ| = |x1 |2 .
n=0 n=0

This sum cannot be finite since x1 6= 0, but this is impossible. So x must be zero,
and hence L − λI injective. ¤

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
128CHAPTER 8. BOUNDED OPERATORS ON BANACH SPACES AND THEIR SPECTRA

Claim(4): L − λI is not injective (one-to-one) for all λ ∈ C such that |λ| < 1.
Proof.
By a similar argument as above, we see that any nonzero x that satisfies the equation
(L − λI)x = 0 is of the form

x = (x1 , λx1 , λ2 x1 , λ3 x1 , ....).

Choose x1 = 1, then

X
kxk = (|λ|2 )n .
n=0

The series is convergent since |λ| < 1, so x ∈ `2 is nonzero and satisfies the equation
(L − λI)x = 0. Thus L − λI is not injective. ¤
Conclusion:
• Claims (1) and (2) show that

σ(R) = {λ ∈ C : |λ| ≤ 1}.

Recall that Image(R − λI) is dense if and only if ker(R − λI)∗ = ker(L − λ̄I) = {0}.
Since |λ̄| = |λ|, claims (3) and (4) show that Image(R − λI) is dense if and only if
|λ| = 1. Therefore

σc (R) = {λ ∈ C : |λ| = 1} and σr (R) = {λ ∈ C : |λ| < 1}.

Also from the above results we get

σp (R) = ∅.

• Note that claim (4) shows us that

σp (L) = {λ ∈ C : |λ| < 1}.

Since σ(L) is a closed set, we get

σ(L) = {λ ∈ C : |λ| ≤ 1}.

(As from the above we know that |λ| > 1 implies that λ ∈ ρ(L).)
For |λ| = 1 we know that ker(L − λI)∗ = ker(R − λ̄I) = {0} by claim (1). Hence
Image(L − λI) is dense. Therefore

σc (L) = {λ ∈ C : |λ| = 1} and σr (L) = ∅. ¥

∗ ∗ ∗∗

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
129

Alternate solution.

Consider R, L : `2 → `2 defined by

Rx = (0, x1 , x2 , ...); Lx = (x2 , x3 , ...); x = (x1 , x2 , x3 , ...) ∈ `2 .

It is clear that kRk = kLk = 1. So, every λ ∈ C with |λ| > 1 is a regular point for
both of the operators R and L. Concerning the eigenvalues of these operators, we
obtain the following:

Lx = λx (x 6= 0) ⇒ x2 = λx1 ; x3 = λx2 ; ...


⇒ x = (1, λ, λ2 , λ3 , ...)x1 .

Such a vector belongs to `2 iff |λ| < 1. Hence,

σp (L) = {λ ∈ C : |λ| < 1}.

From the above result we also have dim ker(L − λI) = 1.


For R we have

Rx = λx (x 6= 0) ⇒ 0 = λx1 ; x1 = λx2 ; x2 = λx3 ; ...


⇒ x1 = x2 = x3 = ... = 0...
⇒ x = 0 : a contradiction.

Hence, σp (R) = ∅.
Next, since L∗ = R and R∗ = L, we obtain

(1) Image(R − λI)⊥ = ker(L − λ̄I),


(2) Image(L − λI)⊥ = ker(R − λ̄I).

For |λ| < 1 the relation (1) yields

codim Image(R − λI) = dim ker(L − λ̄I) = 1.

Hence, {λ ∈ C : |λ| < 1} ⊂ σr (R). Since the spectrum of an operator is closed, we


conclude that

{λ ∈ C : |λ| = 1} ⊂ σ(L) and {λ ∈ C : |λ| = 1} ⊂ σ(R).

Moreover, for |λ| = 1, form (1) and (2) we have

Image(R − λI) = Image(L − λI) = `2 .

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
130CHAPTER 8. BOUNDED OPERATORS ON BANACH SPACES AND THEIR SPECTRA

Hence, σc (R) = σc (L) = {λ ∈ C : |λ| = 1}.


Conclusion:

σ(R) = σ(L) = {λ ∈ C : |λ| ≤ 1}


σp (R) = σp (L) = {λ ∈ C : |λ| < 1}
σr (L) = ∅, σr (R) = {λ ∈ C : |λ| < 1}
σc (L) = σc (R) = {λ ∈ C : |λ| = 1}. ¥

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D

Chapter 9

Compact Operators and Their


Spectra

As bounded linear operators, compact operators share spectral properties of bounded linear oper-
ators. Besides, compact operators have some more particular spectral properties.
Let T ∈ B(X) be a compact operator on a Banach space X. Suppose dim X = ∞.
1. 0 ∈ σ(T ). Every spectral value λ 6= 0 is an eigenvalue.
2. For λ 6= 0, dim ker(Tλ ) ≡ dim ker(T − λI) < ∞.
3. For λ 6= 0, the range of Tλ ≡ T − λI is closed.
4. The set of eigenvalues of T , namely σp (T ), is at most countable. The value λ = 0 is the only
possible point of accumulation of that set.

Problem 122
Let T ∈ B(X) be a compact operator on a Banach space X. Suppose dim X = ∞.
Show that

(a) dim ker(Tλn ) < ∞ ∀n ∈ N,


(b) {0} = ker(Tλ0 ) ⊂ ker(Tλ1 ) ⊂ ker(Tλ2 ) ⊂ ...

Solution.
Since Tλ is linear, Tλ 0 = 0. By induction we get

Tλn x = 0 ⇒ Tλn+1 x = 0, ∀n ∈ N,

131
www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
132 CHAPTER 9. COMPACT OPERATORS AND THEIR SPECTRA

and so (b) follows.


We now prove (a). By the binomial formula,
n µ ¶
X n
Tλn = (T − λI) n
= T k (−λ)n−k
k=0
k
n µ ¶
X
n n
= (−λ) I + T T k−1 (−λ)n−k .
k
|k=1 {z }
S

This can be written

Tλn = W − µI, where µ = −(−λ)n .

Note that T is compact and S is bounded, so W = T S = ST is compact. The


property 2 above gives that dim ker(Tλn ) < ∞. ¥

Problem 123
Let T ∈ B(X) be a compact operator on a Banach space X. Suppose dim X = ∞.
Show that
0 ∈ σ(T ).

Solution.
/ σ(T ) then T is invertible, and we have T T −1 = I. But T and T −1 are compact,
If 0 ∈
so I is compact. This requires that the dimension of X is finite (problems 16, 98):
a contradiction. Thus 0 ∈ σ(T ). ¥

Problem 124
Let T : X → X be a compact operator on a normed space X and let λ 6= 0. Then
there exists a smallest integer r (depending on λ) such that from n = r on, the
kernels ker(Tλn ) are equal, and if r > 0, the inclusions

ker(Tλ0 ) ⊂ ker(Tλ1 ) ⊂ ... ⊂ ker(Tλr )

are all proper (strict).

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
133

Solution.
For simplicity, we let Nn := ker(Tλn ).
• We know that Nm ⊂ Nm+1 (Problem 122). Suppose that Nm = Nm+1 for no m.
Then Nn is a proper subspace of Nn+1 for every n. Since these kernels are closed,
by Riesz lemma, there is a sequence (yn ) in Nn such that
1
kyn k = 1 and kyn − xk ≥ ∀x ∈ Nn−1 .
2
We show that
1
(∗) kT yn − T ym k ≥ |λ| for m < n,
2
so that the sequence (T yn ) has no convergence subsequences. This contradicts the
compactness of T .
From Tλ = T − λI we have T = Tλ + λI and
T yn − T ym = λyn − x̃ where x̃ = Tλ ym + λym − Tλ yn .
Let m < n. We show that x̃ ∈ Nn−1 . Since m ≤ n − 1, we clearly have λym ∈ Nm ⊂
Nn−1 . Also ym ∈ Nm implies
0 = Tλm ym = Tλm−1 (Tλ ym ),
that is, Tλ ym ∈ Nm−1 ⊂ Nn−1 . Similarly, yn ∈ Nn implies Tλ yn ∈ Nn−1 . Together,
x̃ ∈ Nn−1 . Also x = λ1 x̃ ∈ Nn−1 . Hence
1
kλyn − x̃k = |λ| kyn − xk ≥ |λ|.
2
Thus we have (∗). Therefore, we must have Nm = Nm+1 for some m.
• We now prove that
(∗∗) Nm = Nm+1 =⇒ Nn = Nn+1 for all n > m.
Assume that this does not hold. Then Nn is a proper subspace of Nn+1 for some
n > m. We consider an x ∈ Nn+1 \ Nn . By definition,
Tλn+1 x = 0 but Tλn x 6= 0.
Set z = Tλn−m x. Then
Tλm+1 z = Tλn+1 x = 0 but Tλm z = Tλn x 6= 0.
Hence
z ∈ Nm+1 but z ∈
/ Nm .
So Nm is a proper subspace of Nm+1 . This contradicts (∗∗). The first statement is
proved, where r is the smallest n such that Nn = Nn+1 . Consequently, if r > 0, the
inclusions in the theorem are strict. ¥

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
134 CHAPTER 9. COMPACT OPERATORS AND THEIR SPECTRA

Problem 125
Let T : X → X be a compact operator on a Banach space X and let λ 6= 0.
Then there exists a smallest integer q (depending on λ) such that from n = q
on, the ranges Tλn (X) are equal, and if q > 0, the inclusions

Tλ0 (X) ⊃ Tλ1 (X) ⊃ ... ⊃ Tλq (X)

are all proper (strict).

Solution.
For simplicity, we let Rn := Tλn (X). Suppose that Rs = Rs+1 for no s. Then Rn+1 is
a proper subspace of Rn for every n. Since these ranges are closed, by Riesz lemma,
there is a sequence (xn ) in Rn such that
1
kxn k = 1 and kxn − xk ≥ ∀x ∈ Rn+1 .
2
Let m < n. Since T = Tλ + λI, we can write

(i) T xm − T xn = λxm − (−Tλ xm + Tλ xn + λxn ).

On the right hand side, λxm ∈ Rm , xm ∈ Rm , so that Tλ xm ∈ Rm+1 . Since n > m,


also Tλ xn + λxn ∈ Rn ⊂ Rn+1 . Hence (i) is of the form

T xm − T xn = λ(xm − x) with x ∈ Rm+1 .

Consequently,
1
kT xm − T xn k = |λ| kxm − xk ≥ |λ|.
2
This contradicts the fact that (T xn ) has a convergent subsequence since (xn ) is
bounded and T is compact. Thus, Rs = Rs+1 for some s. Let q be the smallest s
such that Rs = Rs+1 . Then, if q > 0, the inclusions in the theorem are proper. Fur-
thermore, Rq+1 = Rq means that Tλ maps Rq onto itself. Hence repeated application
of Tλ gives Rn+1 = Rn for every n > q. ¥

Problem 126
Let A be an invertible operator, and let K be a compact operator in a Banach
space. Prove that
(a) dim(ker(A + K)) < ∞.
(b) codim(Image(A + K)) < ∞.

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
135

Solution.
(a) Since A is invertible, we can write
A + K = A(I + A−1 K).
The operator A−1 K is compact (see note above), so dim(ker(I + A−1 K)) < ∞. This
implies that
A invertible ⇒ ker(AB) = ker(B), ∀B ∈ B(X).
Indeed,
x ∈ ker(B) ⇒ ABx = A0 = 0
⇒ x ∈ ker(AB).
And
x ∈ ker(AB) ⇒ ABx = 0
⇒ Bx = 0
⇒ x ∈ ker(B).
It follows that
dim(ker(A + K)) = dim(A(I + A−1 K))
= dim(I + A−1 K) < ∞.
(b) One can write
A + K = (I + KA−1 )A.
The operator KA−1 is compact, so codim(I + KA−1 ) < ∞. This implies that if A
is invertible, and then Image(BA) = Image(B). Indeed,
x ∈ Image(B) ⇒ x = By
⇒ x = (BA)A−1 y ∈ Image(BA).
And
x ∈ Image(BA) ⇒ x = BAy
⇒ x = B(Ay) ∈ Image(B).
Thus, we obtain
codim(Image(A + K)) = codim(Image((I + KA−1 )A))
= codim(Image(I + KA−1 )) < ∞. ¥

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
136 CHAPTER 9. COMPACT OPERATORS AND THEIR SPECTRA

Problem 127
Consider the operator K : L2 ([0, 1]) → L2 ([0, 1]) defined by
Z 1
(Kf )(t) = k(t, s)f (s)ds
0

where k(s, t) = min{t, s : t, s ∈ [0, 1]}.


(a) Prove that K is a compact self-adjoint operator.
(b) Find the spectrum σ(K) and the norm kKk.

Solution.
(a) Since k(t, s) = k(s, t) and k(t, s) is a continuous function, the operator is self-
adjoint and compact (see Problem 105).
(b) Since K is compact and self-adjoint, the spectrum of K consists of zero and real
eigenvalues1 . Assume that λy = Ky. This means that
Z t Z 1
(1) λy(t) = min{t, s}y(s)ds + min{t, s}y(s)ds
0 t
Z t Z 1
= sy(s)ds + ty(s)ds
0 t
Z t Z 1
= sy(s)ds + t y(s)ds.
0 t

Taking the derivative twice, we obtain


Z 1 Z 1
0
(2) λy (t) = ty(t) + y(s)ds − ty(t) = y(s)ds
t t
λy 00 (t) = −y(t).

Clearly, λ 6= 0; otherwise, y = 0 so ker(K) = 0. We have the differential equation

(3) λy 00 + y = 0 with b.v.c. y 0 (0) = y(0) = 0

because of (1). Let us prove that λ > 0, which means that the operator K is positive.
Multiplying (3) by ȳ and integrating we obtain
Z 1
λ y 00 (t)ȳ(t)dt + kyk2 = 0.
0

1
If T ∈ B(H) is self adjoint, all its eigenvalues are real. (We will see this in the next chapter.)

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
137

Integrating by parts we obtain


µ Z 1 ¶
0 1
λ y ȳ|0 − |y | dt + kyk2 = 0.
0 2
0

The b.v.c. yield Z 1


−λ |y 0 |2 dt + kyk2 = 0.
0
Hence λ > 0.
The solution of the differential equation is
t t
y = C1 cos √ + C2 sin √ .
λ λ
C2
From the b.v.c. it follows that C1 = 0 and √ λ
cos √1λ = 0. Therefore, the eigenvalues
of K are
4
λk = 2 , k = 1, 2, ...
π (2k − 1)2
Since K is seft-adjoint, we obtain
4
kKk = max |λk | = |λ1 | = . ¥
k∈N π2

Problem 128(Similar problem)


Consider the operator K : L2 ([0, 1]) → L2 ([0, 1]) defined by
Z 1
(Kf )(t) = k(t, s)f (s)ds
0

where k(s, t) = max{t, s : t, s ∈ [0, 1]}.


(a) Prove that K is a compact self-adjoint operator.
(b) Find the spectrum σ(K) and the norm kKk.
(c) Is K a positive operator?

Problem 129
Let S be the operator defined on C[0, 1] by
Z x
(Sf )(x) = f (y)dy.
0

(a) Compute the spectrum of S.


(b) Show that S is compact.

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
138 CHAPTER 9. COMPACT OPERATORS AND THEIR SPECTRA

Solution.
(a) First we show that S is continuous. For f, g ∈ C[0, 1] we have
¯Z x ¯
¯ ¯
kSf − Sgk = kS(f − g)k = ¯ (f (y) − g(y))dy ¯¯
¯
Z 0x
≤ |f (y) − g(y)|dy
0
Z x
≤ sup |f (y) − g(y)|dy
x∈[0,1] 0
≤ |[0, 1]|.kf − gk = kf − gk.

Hence, S is Lipschitz continuous with constant 1. Thus, kSk ≤ 1. Next, we show


that kS n k1/n → 0 as n → ∞. Observe that
Z x
x1
|(Sf )(x)| ≤ |f (t)|dt ≤ kf kx = kf k , x ∈ [0, 1].
0 1!

By induction
Z x
n+1
|(S f )(x)| ≤ |S n f (t)|dt
0
Z x n
t
≤ kf k dt
0 n!
xn+1
= kf k , x ∈ [0, 1], n = 1, 2, ...
(n + 1)!

Thus,
1
kS n f k ≤ kf k , n = 1, 2, ...
n!
So µ ¶1/n
n 1/n 1
kS k ≤ → 0 as n → ∞.
n!
Recall that the spectral radius of S is given by

r(S) = lim kS n k1/n .


n→∞

This implies that r(S) = 0. Thus σ(S) = {0}.


(b) Suppose F ⊂ C[0, 1] is a bounded subset. Put

kF k := sup kf k.
f ∈F

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
139

Then SF is equi-continuous by the Fundamental Theorem of the Calculus:

∀f ∈ F, |(Sf )0 (x)| ≤ kf k ≤ kF k.

SF is bounded since:

∀f ∈ F, |(Sf )(x)| ≤ kf k ≤ kF k.

By Ascoli-Arzela theorem SF is relatively compact. Thus S is compact. ¥

Problem 130
Set Z 1−x
(T f )(x) = f (y)dy, f ∈ C[0, 1], x ∈ [0, 1].
0

(a) Prove that T is a linear bounded and compact operator on C[0, 1].
(b) Calculate σ(T ) and the eigenvalues of T .

Solution.
(a)

• Linearity of T : trivial.

• Boundedness of T :
¯Z ¯
¯ 1−x ¯
|(T f )(x)| = ¯¯ f (y)dy ¯¯ ≤ kf k,
0

where kf k = maxx∈[0,1] |f (x)|. Hence,

kT k ≤ 1.

• Compactness of T : let (fn )∞


n=1 be a bounded sequence in C[0, 1]. Hence,

kfn k ≤ M ∀n ∈ N

for some M ≥ 0. By Arzela-Ascoli theorem, it suffices to show that A :=


{T fn ; n ∈ N} is bounded and equicontinuous. We have

kT fn k ≤ kT kkfn k ≤ kfn k ≤ M ∀n ∈ N.

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
140 CHAPTER 9. COMPACT OPERATORS AND THEIR SPECTRA

Given any ε > 0, without lost of generality, we can assume x < y, then
¯Z 1−x ¯
¯ ¯
¯
|(T fn )(x) − (T fn )(y)| = ¯ f (y)dy ¯¯ ≤ M |x − y|.
1−y

Thus,
ε
|(T fn )(x) − (T fn )(y)| < ε ∀n ∈ N provided |x − y| < .
M
(b) First we see that λ = 0 is an eigenvalue. Assume that λ 6= 0 is an eigenvalue,
i.e., Z 1−x
λg(x) = (T g)(x) = g(y)dy, x ∈ [0, 1]
0
for some 0 6= g ∈ C[0, 1]. This implies that
T g ∈ C 1 [0, 1] and λg 0 (x) = −g(1 − x), x ∈ [0, 1].
Moreover, we have g(1) = 0. But g ∈ C 1 [0, 1] implies that g ∈ C 2 [0, 1]. By
differentiating once more we get
g(x)
λg 00 (x) = , x ∈ [0, 1] and g(1) = g 0 (0) = 0.
λ
Hence
g(x) = A cos(x/λ) with g(1) = 0.
This gives that
1
λk = π , k ∈ Z.
2
+ kπ
Check if all these λ’s are eigenvalues. We calculate
Z 1−x
(T gk )(x) = cos(t/λk )dt
0
= λk [sin(t/λk )]1−x
h³ π 0 ´ i
= λk sin + kπ (1 − x)
2
= λk (−1)k gk (x).
Hence, λ = λ2l , l ∈ Z are the eigenvalues of T , i.e.,
σp (T ) = {λ2l , l ∈ Z}.
We know that σ(T ) is closed and σ(T ) \ {0} ⊂ σp (T ). This yields
σ(T ) = {0} ∪ σp (T ). ¥

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
141

Problem 131
Let T be a compact operator on a Hilbert space H and (λn ) be a sequence of
complex numbers. Suppose there exists a nested sequence of distinct subspaces
(Mn ) such that for all n ∈ N

(i) Mn ( Mn+1
(ii) (T − λn I)Mn+1 ⊂ Mn .

Prove that limn→∞ λn = 0.

Solution.
Since Mn is a subspace of Mn+1 , we can write Mn+1 = Mn ⊕(Mn⊥ )M , where (Mn⊥ )M is
the orthogonal complement of Mn in Mn+1 . For short we write (Mn⊥ )M = Mn+1 ªMn .
Let {en } be a sequence of unit vectors defined by

e1 ∈ M1 , en+1 ∈ Mn+1 ª Mn , ∀n ∈ N.

Clearly, that is an orthonormal system. Moreover,


­ ®
(T − λn I)en , en = 0, for all n ≥ 2,

which implies that


¯­ ®¯
kT en k ≥ |hT en , en i| = ¯ (T − λn I)en , en ¯ + |hλn en , en i|
= 0 + |hλn en , en i| = |λn |.
w
Since T is compact and en → 0, it follows that limn→∞ T en = 0. Thus

lim λn = 0. ¥
n→∞

Problem 132
Let T be a compact operator on a Hilbert space H and any C > 0. Prove
that there is a finite number of linearly independent eigenvectors x1 , ..., xn of T
corresponding to eigenvalues λ1 , ..., λn such that λi > C for all i = 1, ..., n.

Solution.
We can rescale to get kxi k = 1 for all i = 1, ..., n. Suppose to the contrary that

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
142 CHAPTER 9. COMPACT OPERATORS AND THEIR SPECTRA

there is an infinite sequence {xn } of unit vectors, and a sequence of eigenvalues {λn }
satisfying
λn > C and T xn = λn xn , ∀n ∈ N.
Let Mn = Span{x1 , ..., xn }, then {Mn } is a nested sequence of subspaces of H and
the inclusions
Pn Mn ⊂ Mn+1 are strict. Let x ∈ Mn , then there are c1 , ..., cn ∈ C such
that x = i=1 ci xi . So we have
n
X
(T − λn I)x = (T − λn I) c i xi
i=1
n
X
= ci (T − λn I)xi
i=1
n
X
= ci (T xi − λn xi )
i=1
Xn
= ci (λi − λn )xi ∈ Mn−1 .
i=1

This implies that


(T − λn I)Mn ⊂ Mn−1 , n ≥ 2.
From Problem 130, we obtain
lim λn = 0.
n→∞

This contradicts the assumption λn > C for all n ∈ N. ¥

Note: Argument in problems 130, 131 is the proof of Proposition 4 in the review at the beginning
of this chapter.

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D

Chapter 10

Bounded Self Adjoint Operators


and Their Spectra

Review some main points.


Bounded self adjoint operators on Hilbert spaces were defined and considered before. This chapter
is devoted to their spectral properties.
Definition:
Let T ∈ B(H) where H is a complex Hilbert space. The adjoint operator of T is the operator
T ∗ : H → H defined by
hT x, yi = hx, T ∗ yi, ∀x, y ∈ H.
T is said to be self adjoint if T = T ∗ . We can say that T is self adjoint if and only if

hT x, yi = hx, T yi, ∀x, y ∈ H.

Another equivalent condition is:


T is self adjoint if and only if hT x, xi is real for all x ∈ H.

Let T ∈ B(H) be a bounded self adjoint operator on the complex Hilbert space H.

Proposition 12 (Eigenvalues and eigenvectors)


1. All eigenvalues of T (if they exist) are real.
2. Eigenvectors corresponding to different eigenvalues of T are orthogonal.

Proposition 13 (Spectrum)
1. σ(T ) ⊂ R.
2. σ(T ) ⊂ [m, M ] where m = inf kxk=1 hT x, xi and M = supkxk=1 hT x, xi.
3. m, M ∈ σ(T ).

Proposition 14 (Norm)
kT k = max{|m|, |M |} = supkxk=1 |hT x, xi|.

143
www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
144CHAPTER 10. BOUNDED SELF ADJOINT OPERATORS AND THEIR SPECTRA

Problem 133
Let T : H → H be a bounded self-adjoint linear operator on a complex Hilbert
space H. Prove that the residual spectrum of T is empty, that is,

σr (T ) = ∅.

Solution.
Assume that σr (T ) 6= ∅. Take λ ∈ σr (T ). By the definition of σr (T ), Tλ−1 exists but
its domain is not dense in H. Hence, by the projection theorem, there is a y 6= 0 in
H such that y is perpendicular to the domain D(Tλ−1 ) of Tλ−1 . But D(Tλ−1 ) is the
range of Tλ , hence
hTλ x, yi = 0, ∀x ∈ H.
Since λ is real and T is self-adjoint, we obtain
hx, Tλ yi = 0, ∀x ∈ H.
Taking x = Tλ y, we get kTλ yk2 = 0, so that
Tλ y = T y − λy = 0.
Since y =6 0, this shows that λ is an eigenvalue of T . But this contradicts λ ∈
σr (T ). ¥

Second solution:
By Problem 91, noting that T is self adjoint, for any λ ∈ C we have
Image(T − λI)⊥ = ker(T ∗ − λ̄I) = ker(T − λ̄I).
And hence, if Image(T − λI) 6= H, then λ̄ is an eigenvalue of T . Since T is self
adjoint, λ is real. Thus λ = λ̄ is an eigenvalue of T . Therefore λ does not belong to
the residual spectrum of T. ¥

Problem 134
Let T : H → H be a bounded self-adjoint linear operator on a complex Hilbert
space H. Prove that

(∗) λ ∈ ρ(T ) ⇐⇒ ∃c > 0 : ∀x ∈ H, kTλ xk ≥ ckxk.

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
145

Solution.
• If λ ∈ ρ(T ) then Rλ := Tλ−1 : H → H exists and is bounded, say kRλ k = k > 0.
Now since I = Rλ Tλ , we have for every x ∈ H

kxk = kRλ Tλ xk ≤ kRλ k kTλ xk = kkTλ xk.

This gives
1
kTλ xk ≥ ckxk, where c = .
k
• Conversely, suppose (∗) holds. We shall show:

(a) Tλ : H → Tλ (H) is bijective;

(b) Tλ (H) is dense in H;

(c) Tλ (H) is closed in H;

so that Tλ (H) = H and Rλ := Tλ−1 is bounded by the bounded inverse theorem.1


(a) By (7.1), we have for x1 , x2 ∈ H

kTλ x1 − Tλ x2 k = kTλ (x1 − x2 )k ≥ ckx1 − x2 k.

Therefore,
Tλ x1 = Tλ x2 =⇒ x1 = x2 .
Thus Tλ : H → Tλ (H) is bijective.
(b) We show that x0 ⊥Tλ (H) implies x0 = 0, so that Tλ (H) = H by the projection
theorem.2 Let x0 ⊥Tλ (H) Then for all x ∈ H we have

0 = hTλ x, x0 i = hT x, x0 i − λhx, x0 i.

Since T is self-adjoint,
hT x, x0 i = hx, T x0 i.
It follows that
hx, T x0 i = λhx, x0 i = hx, λ̄x0 i.
Thus T x0 = λ̄x0 . So x0 = 0 since otherwise, λ̄ = λ would be an eigenvalue of T ,
and Tλ x0 = 0, which would imply

0 = kTλ x0 k ≥ ckx0 k > 0 : a contradiction.


1
A bounded linear operator T from a Banach space X onto a Banach space Y is an open
mapping. Hence if T is bijective, then T is continuous and thus bounded.
2
If Y is a closed subspace of H, then H = Y ⊕ Y ⊥ .

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
146CHAPTER 10. BOUNDED SELF ADJOINT OPERATORS AND THEIR SPECTRA

(c) To show Tλ (H) is closed we show

y ∈ Tλ (H) =⇒ y ∈ Tλ (H).

Let y ∈ Tλ (H). There is a sequence (yn ) in Tλ (H) which converges to y. For every
n we have yn = Tλ xn for some xn ∈ H. By (7.1),

1 1
kxn − xm k ≤ kTλ (xn − xm )k = kyn − ym k.
c c
Hence (xn ) is Cauchy. Since H is complete, (xn ) converges, say, xn → x. Since
Tλ is continuous, yn = Tλ xn → Tλ x. Since the limit is unique, Tλ x = y. Hence
y ∈ Tλ (H). Thus Tλ (H) is closed in H. ¥

Problem 135
(a) Let A ∈ B(X) where X is a Banach space. Suppose there exists m > 0 such
that
kAxk ≥ mkxk, ∀x ∈ X.
Show that Image A is closed in X.
(b) Let A ∈ B(H) be self adjoint, where H is a Hilbert space. Let λ ∈ C such
that Imλ 6= 0. Prove that

kAx − λxk ≥ |Imλ| kxk, ∀x ∈ H.

Prove that λ is a regular point of A.

Solution.
(a) Let (xn ) be a sequence in X. Suppose Axn → y as n → ∞. From the hypothesis
we get
kAxn − Axm k ≥ mkxn − xm k for n 6= m.
Since (Axn ) is a Cauchy sequence in X, it follows that (xn ) is also a Cauchy sequence
in X. Hence x → x as n → ∞. But A is continuous, so Axn → Ax as n → ∞. By
uniqueness of the limit, we obtain y = Ax. This shows that Image A is closed.
(b) Let λ = a + ib with a, b ∈ R and b = Imλ 6= 0. We have for all x ∈ H

kAx − λxk2 = hAx − (a + ib)x, Ax − (a + ib)xi


= h(A − aI)x − ibx, (A − aI)x − ibxi
= k(A − aI)xk2 + b2 kxk2 ,

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
147

which implies that


kAx − λxk ≥ |b|kxk, ∀x ∈ H.
By part (a), Image(A − λI) is closed and therefore λ ∈
/ σc (A). By Problem 91
σr (A) = ∅, thus λ is a regular point of A. ¥

Problem 136
Let A ∈ B(H) be self adjoint, where H is a Hilbert space.
(a) Prove that
|hAx, xi|
kAk = sup .
x6=0 kxk2
(b) Prove that at least one of kAk or −kAk is an element of σ(A).

Solution.
(a) Using the Cauchy-Schwarz inequality, we have

|hAx, xi| ≤ kAxk kxk ≤ kAk kxk2 , ∀x ∈ H.

Hence
|hAx, xi|
sup ≤ kAk. (i)
x6=0 kxk2
Now we establish the reverse. Notice first that for all x, y ∈ H we have
­ ® ­ ® £ ¤
A(x + y), x + y − A(x − y), x − y = 2 hAx, yi + hAy, xi .

Using the triangle inequality, we get


¯ ¯ ¯ ¯ ¯ ¯
2¯hAx, yi + hAy, xi¯ ≤ ¯hA(x + y), x + yi¯ + ¯hA(x − y), x − yi¯.
|hAx,xi|
If we let C = supx6=0 kxk2
, then the Parallelogram Law gives that

¯ ¯ ¡ ¢
¯hAx, yi + hAy, xi¯ ≤ 1 C kx + yk2 + kx − yk2
2¡ ¢
= C kxk2 + kyk2 . (∗)
Ax
Now let x be any vector with kxk = 1 and let y = kAxk ( the case Ax = 0 does not
give the supremum, hence we may assume that Ax 6= 0). Then kyk = 1. From (∗)
we get ¯ ¯
¯ hAx, Axi hAx, Axi ¯
¯ ¯
¯ kAxk + kAxk ¯ ≤ 2C.

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
148CHAPTER 10. BOUNDED SELF ADJOINT OPERATORS AND THEIR SPECTRA

Hence kAxk ≤ C. This holds for all x ∈ H with kxk = 1. Thus kAk ≤ C. (ii)
Combine (i) and (ii) we obtain
|hAx, xi|
kAk = sup .
x6=0 kxk2
(b) If we take x arbitrary with kxk = 1, then we get
kAk = sup |hAx, xi|. (∗∗)
kxk=1

Let (xn ) be a sequence in H with kxn k = 1 such that |hAxn , xn i| converges to


kAk (this is possible by (∗∗)). Let hAxn , xn i → λ (it may be necessary to pass to
subsequence). Clearly, λ = ±kAk. Now,
0 ≤ kAxn − λxn k2 = kAxn k2 − 2λhAxn , xn i + λ2 kxn k2
≤ 2λ2 − 2λhAxn , xn i,
which converges to 0 as n → ∞. Thus λ ∈ σ(A) (see Problem 109). ¥

∗ ∗ ∗∗
To close this chapter, we introduce a well known theorem relative to compact self adjoint operators
on Hilbert spaces:

The spectral theorem for compact self adjoint operators on Hilbert spaces.

Problem 137
Let T ∈ B(H) be a compact self adjoint operator on a Hilbert space H.
(a) There exists a system (finite or infinite) of orthonormal eigenvectors
{e1 , e2 , ...} of T and corresponding eigenvalues {λ1 , λ2 , ...} of T such that |λ1 | ≥
|λ2 | ≥ .... If the system is infinite then λn → 0 as n → ∞.
(b) Eigenvectors and eigenvalues mentioned above satisfy the following equation:

X
Tx = λk hx, ek iek , ∀x ∈ H.
k=1

Solution.
(a) We use Proposition 13-3 (that we proved in Problem 136b) repeatedly for con-
structing eigenvalues and eigenvectors.

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
149

Let H1 = H and T1 = T . Then by Proposition 13-3, there exists an eigenvalue λ1


of T1 and a corresponding eigenvector e1 such that

ke1 k = 1 and |λ1 | = kT1 k.

Now Span{e1 } is a closed subspace of H1 hence, by the projection theorem,

H1 = Span{e1 } ⊕ Span{e1 }⊥ .

Let H2 = Span{e1 }⊥ . Clearly H2 is a closed subspace of H1 and T (H2 ) ⊂ H2 .


Indeed, if x ∈ H2 then x⊥e1 , hence T x = λx ⇒ T x⊥e1 . Let T2 be the restriction
of T1 on H2 , that is, T2 = T1 |H2 = T |H2 . Then T2 is a compact and self adjoint
operator in B(H2 ). If T2 = 0, then there is nothing to prove. Assume that T2 6= 0.
Then by Proposition 13-3, there exists an eigenvalue λ2 of T2 and a corresponding
eigenvector e2 such that

ke2 k = 1 and |λ2 | = kT2 k.

Since T2 is a restriction of T1 ,

|λ2 | = kT2 k ≤ kT1 k = λ1 .

By construction e1 and e2 are orthonormal.


Now let H3 = Span{e2 , e2 }⊥ . Clearly H3 ⊂ H2 and T (H3 ) ⊂ H3 . The operator
T3 = T |H3 is compact and self adjoint....If we continue to proceed in this way, either
after some stage, say n, we get Tn = 0 or there exists an infinite sequence (λn ) of
eigenvalues of T and corresponding eigenvectors (en ) satisfying

ken k = 1, ∀n ∈ N and |λ1 | ≥ |λ2 | ≥ ...

If the sequence (λn ) is infinite, we show that λn → 0 as n → ∞. Suppose that


λn 9 0 as n → ∞. Then there exists ε > 0 such that |λn | > ε for infinitely many
n. For n 6= m, we have

kT en − T em k2 = kλn en − λm em k2 = λ2n + λ2m > ε2 .

This shows that the sequence (T en ) has no convergent subsequence, a contradiction


to the compactness of T . Hence λn → 0 as n → ∞.
(b) There are two cases to consider:
Case 1. Tn = 0 P for some n.
Let xn = x − nk=1 hx, ek iek for all x ∈ H. Then xn ⊥ek for 1 ≤ k ≤ n, since
{e1 , e2 , ...} is an orthonormal system and

hxn , ek i = hx, ek i − hx, ek i = 0.

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D
150CHAPTER 10. BOUNDED SELF ADJOINT OPERATORS AND THEIR SPECTRA

Hence n n
X X
0 = T n xn = T x − hx, ek iTk ek = λk hx, ek iek .
k=1 k=1

That is,
n
X ∞
X
Tx = λk hx, ek iek = λk hx, ek iek , ∀x ∈ H.
k=1 k=1

Case 2. Tn 6= 0 for infinitely many n.


For x ∈ H, by Case 1, we have
° °
° Xn °
° °
°T x − λk hx, ek iek ° = kTn xx k ≤ kTn k kxn k
° °
k=1
= |λn | kxn k → 0 as n → ∞.

Hence ∞
X
Tx = λk hx, ek iek , ∀x ∈ H. ¥
k=1

THANK YOU and GOOD LUCK!


Anh Le
leqanh36@gmail.com

www.MATHVN.com
www.MATHVN.com - Anh Quang Le, Ph.D

Bibliography

[1] Conway, J.B. A course in Functional Analysis. Second edition. Springer. New
York, 1990.

[2] Kreyszig. E. Introductory Functional Analysis with Applications. John Wiley


and Son. New York, 1989.

[3] Reed, M.; Simon, B. Methods of modern Mathematics Physics. I. Functional


Analysis. Second edition. Academic Press, Inc. New York, 1980.

[4] Royden. H.L. Real Analysis. Third edition. Prentice Hall. NJ, 1988.

[5] Rudin. W Functional Analysis. McGraw-Hill, Inc. New York, 1991.

[6] Yeh, J. Real Analysis. Theory of measure and integration. Second edition. World
Scientific. NJ, 2006.

151
www.MATHVN.com

You might also like