Functional Analysis Exercise Class

Week 9 November – 13 November

Deadline to hand in the homeworks: your exercise class on week 16 November – 20 November

Exercises

(1) Show that if T ∈ B(X, Y ) and S ∈ B(Y, Z) are bounded operators then

∀x ∈ X : kT xk ≤ kT kkxk,

and
kST k ≤ kSkkT k.
Solution: Recall the definition of the operator norm,

kT xk
kT k := sup .
x∈X\{0} kxk

Then obviously for all x ∈ X \ {0}, we have

kT xk kT xk
≤ sup = kT k ,
kxk x∈X\{0} kxk

and hence kT xk ≤ kT k kxk, as claimed. The inequality kT 0k ≤ kT k k0k is

trivial, as both sides are equal to 0.
For the second part, clearly,

kST (x)k = kS(T x)k ≤ kSkkT xk ≤ kSkkT kkxk.

(2) Show that for a bounded linear map T : X → Y the following expressions for
the norm are equivalent:

a) kT k = inf{C ≥ 0 : kT xk ≤ Ckxk for all x ∈ X},

kT xk
b) kT k = supx6=0 kxk
,
c) kT k = supkxk≤1 kT xk,

1
 d) kT k = supkxk=1 kT xk.

Solution: Let

I = inf{C ≥ 0 : kT xk ≤ Ckxk for all x ∈ X},

kT xk
I1 = sup ,
x6=0 kxk
I2 = sup kT xk,
kxk≤1

I3 = sup kT xk.
kxk=1

kT xk x
Straight from definition we have that I3 ≤ I2 , and since kxk
= kT ( kxk )k we
have I1 ≤ I3 .
kT xk
Now if kxk ≤ 1 we have kT xk ≤ kxk
. Then I2 ≤ I1 and therefore

I1 = I2 = I3 .

Note that kT xk ≤ I1 kxk for all x ∈ X. Then I ≤ I1 . Moreover by definition of

sup there exists a sequence {xn }n such that

kT xn k/kxn k ≥ I1 − 1/n for all n.

By definition of I we have that for all n, I ≥ kT xn k/kxn k. Therefore I = I1 .

(3) a) Let T ∈ B(X, Y ) be a bounded linear operator. Show that if λ is an

eigenvalue of T then

|λ| ≤ kT k . (0.1)

b) A matrix T ∈ Rn×n is called stochastic if all its entries are non-negative,

and the sum of each row is 1. Show that all eigenvalues of a stochastic
matrix have absolute value at most 1.
Solution:

a) Let λ be an eigenvalue with eigenvector x ∈ X \ {0}. Then

|λ| kxk = kλxk = kT xk ≤ kT k kxk ,

from which (0.1) follows.

2
 b) Equip Rn with the ∞-norm kxk∞ := max1≤i≤n |x(i)|. We have seen in Pnthe
lecture that in this case the induced operator norm is kT k = max1≤i≤n j=1 |Tij |.
Hence, for a stochastic matrix kT k = 1, and by the previous point, every
eigenvalue λ of T satisfies |λ| ≤ kT k = 1.

(4) Let δ : C([0, 1]) → R be the linear functional that evaluates a function at the
origin: δ(f ) = f (0).

a) Show that if C([0, 1]) is equipped with the sup-norm

kf k∞ = sup |f (x)|
0≤x≤1

then δ is bounded, and compute its norm.

b) Show that if C([0, 1]) is equipped with the one-norm
Z 1
kf k1 = |f (x)|dx
0

then δ is unbounded.

Solution:

a) By one of the equivalent definitions of the norm we have

kδk = sup |δ(f )| = sup |f (0)|.

kf k∞ =1 kf k∞ =1

Since kf k∞ = 1, it means that ∀0 ≤ x ≤ 1: |f (x)| ≤ 1. In particular,

|f (0)| ≤ 1. Therefore kδk ≤ 1. Taking the constant 1 function f (x) = 1, x ∈
[0, 1], we get that kf k∞ = 1, and hence kδk ≥ |δ(f )| = 1. Combining the
two inequalities we get kδk = 1.
b) We are going to construct a sequence of functions fn , n ≥ 2, such that
kfn k1 = 1 for all n, and |δ(fn )| → ∞, proving the unboundedness of δ. For
every n ∈ N, n ≥ 2, define
 n2
− 2 x + n 0 ≤ x ≤ 2/n
fn (x) =
0 2/n ≤ x ≤ 1.
R1
Clearly, kfn k1 = 0 |f (x)|dx = 1 for all n, and f (0) = n. Therefore
|δ(fn )| = n → ∞ as n → ∞. So δ is unbounded.

3
(5) Define the linear functional

ϕ(x) := x(1) − 4x(2) + 5x(2015), x ∈ lp := lp (N, K).

Show that ϕ is a bounded linear functional on lp for all 1 ≤ p ≤ +∞, and

compute its norm.
Solution: Linearity is obvious, and ϕ is of the form


 1, i = 1,
+∞ 
X  −4, i = 2,
ϕ(x) = y(i)x(i), with y(i) =
i=1


 5, i = 2015,

0, otherwise.

Clearly, y ∈ lq (N, K) for any 1 ≤ q ≤ +∞. Assume first that 1 < p < +∞. As
we have seen in the lecture, ϕ defines a bounded linear functional on lp , and its
norm is
1 1
kϕk = kykq = (1 + 4q + 5q )1/q for =1− .
q p
For p = 1, we have
+∞
X
|ϕ(x)| ≤ |x(1)| + 4|x(2)| + 5|x(2015)| ≤ 5 |x(i)| = 5 kxk1 , x ∈ l1 ,
i=1

and hence kϕk ≤ 5 = kyk∞ . On the other hand, choosing x ∈ l1 such that
x(2015) = 1 and x(i) = 0 for i 6= 2015, we get kxk1 = 1 and ϕ(x) = 5, and
hence kϕk ≥ 5. Combining the two inequalities, we get kϕk = 5.
For p = +∞, we have for every x ∈ l∞

|ϕ(x)| ≤ |x(1)| + 4|x(2)| + 5|x(2015)| ≤ kxk∞ (1 + 4 + 5) = kxk∞ kyk1 ,

and hence kϕk ≤ kyk1 = 10. On the other hand, choosing


1,
 i = 1 or i = 2015,
x(i) = −1, i = 2,

0, otherwise,


we see that kxk∞ = 1, and ϕ(x) = 10, and thus kϕk ≥ 10. Combining the two
inequalities, we get kϕk = 10.

4
(6) Let

c00 := x ∈ KN : #{i : x(i) 6= 0} < +∞

be the space of K-valued sequences that are non-zero only at finitely many places.

a) Show that c00 is dense in lp (N, K) for any 1 ≤ p < +∞.

p 1/p
P+∞

b) Equip c00 with the p-norm kxkp := i=1 |x(i)| for some 1 < p < +∞,
and define the linear functional ϕ : c00 → K as
+∞
X 1
ϕ(x) := √ x(i).
i=1
i

For which values of p is ϕ bounded?

Solution:

a) Let 1 ≤ p < +∞. We have to show that for every x ∈ lp := lp (N, K)

and every ε > 0, there exists an xε ∈ c00 such that kx − xε kp < ε. If
P+∞ p
x ∈ lp then, by definition,
P+∞ i=1 |x(i)| < +∞, and hence there exists an
Nε ∈ N such that i=Nε +1 |x(i)|p < εp . Define xε (i) := x(i), i ≤ Nε , and
xε (i) := 0, i > Nε . Then xε ∈ c00 , and

+∞
!1/p +∞
!1/p
X X
kx − xε kp = |x(i) − xε (i)|p = |x(i)|p < ε.
i=1 i=Nε +1

b) Let y(i) := √1i such that ϕ(x) = +∞

P
i=1 y(i)x(i), and let q be the Hölder
1 1
conjugate of p, i.e., q + p = 1. Then

+∞ +∞
X
q
X 1
|y(i)| = q/2
,
i=1 i=1
i

and hence y ∈ lq (N, K) ⇐⇒ q > 2. If q > 2 then x 7→ +∞

P
i=1 y(i)x(i) defines
a bounded linear functional on lp , and ϕ is its restriction to c00 , and hence
ϕ is bounded, too.
On the other hand, if q ≤ 2 then kykq = +∞, and hence for any K > 0,
there exists an NK ∈ N such that N q
P K
i=1 |y(i)| ≥ K. Let yK (i) := y(i), 1 ≤
i ≤ NK , and yK (i) := 0, i > NK . Then yK can be considered as an
element of lq ([NK ], K), and we have seen (week 2) that there exists an

5
 xK ∈ lp ([NK ], K) such that kxK kp = 1 and N
P K
i=1 yK (i)xK (i) = kyK kq . This
xK can be naturally considered as an element in c00 with kxK kp = 1 and
NK
X
ϕ(xK ) = y(i)xK (i) = kyK kq ≥ K 1/q .
i=1

This shows that kϕk ≥ supK>0 K 1/q = +∞.

(7) a) Show that a linear operator from a normed space X to a normed space Y
is bounded if and only if the image of the unit ball around 0 is a bounded
set in Y .
b) Show that if X is finite-dimensional then every linear operator to any
normed space Y is bounded.
c) Is it true that if Y is finite-dimensional then any linear operator from any
normed space X is bounded?
d) The rank of a linear operator T : X → Y is the dimension of its range,
i.e., rk T := dim{T x : x ∈ X}. Is it true that every finite-rank operator is
bounded?
Solution:
a) Let T : X → Y be a linear operator. Note that T is bounded if and only
if its norm

kT xk x
kT k = sup = sup T kxk =
sup kT xk
x∈X\{0} kxk x∈X\{0} x∈B 1 (0)\{0}

is finite, from which the assertion follows.

b) Let k kX denote the norm on X. Let e1 , . . . , ed be a basis in X, and for
every X 3 x = di=1 xi ei , define
P

d
!1/2
X
kxk2 := |xi |2 .
i=1

As we have seen in the lecture, all norms on a finite-dimensional space

are equivalent, and hence there exists a constant C > 0 such that kxk2 ≤
C kxkX , x ∈ X. Then

d

d d
!1/2 d
!1/2
X X X X
kT xk = T xi ei ≤ |xi | kT ei k ≤ |xi |2 kT ei k2


i=1 i=1 i=1 i=1
d
!1/2
X
≤ C kxkX kT ei k2 ,
i=1

6
 where the first inequality is due to the triangle inequality, and the second
one follows from the Cauchy-Schwarz inequality. Hence, T is bounded, and
P 1/2
d 2
kT k ≤ C i=1 kT e i k .
c) No. Consider the examples in (4)b) and (6)b), where Y is one-dimensional,
and the given linear functional is unbounded.
d) No, for the same reason as in the previous point.

(8) In the lecture it was proved that for p, q ∈ (1, ∞) with 1/p + 1/q = 1 the space
lq := lq (N, K) is isometrically isomorphic to lp (N, K)∗ , i.e. lq ∼
= lp∗ . Prove the
∗ ∼
same for p = 1, i.e. prove that l1 = l∞ .
∗
P∞ the same argument as in the lecture. Define T : l∞ → l1 ,
Solution: We follow
by (T x)(y) := k=1 xk yk for x ∈ l∞ and y ∈ l1 . Then
∞
X
|(T x)(y)| ≤ |xk yk | ≤ kxk∞ kyk1 ≤ ∞.
k=1

This means that for all x ∈ l∞ we have T x ∈ l1∗ and

X |(T x)(y)|
kT xk = ≤ kxk∞ . (0.2)
kyk1
y∈l1 \{0}

On the other hand, for every k such that xk 6= 0, let y (k) (k) := |xk |/xk , and let

y (k) (i) := 0, i 6= k. Then y (k) 1 = 1, and (T x)(y (k) ) = |xk | implies that kT xk ≥
|xk |. Since this holds for every k ∈ N, we obtain kT xk ≥ supk∈N |xk | = kxk∞ .
Combining the two inequalities, we get kT xk = kxk∞ , i.e., T is an isometry.
This immediately implies that T is injective, and hence the only thing we are
left to show is the surjectivity of T .
To this end, let ϕ ∈ l1∗ and define x = (x1 , x2 , . . . ) by xk := ϕ(e(k) ), where
e(k) = (0, . . . , 0, 1, 0, . . . ) has 1 at the k-th place and 0 everywhere else. We need
to show that T x = ϕ, i.e. for all y ∈ l1 : (T x)(y) = ϕ(y). Since cc := {y ∈
KN : #{i : y(i) 6= 0} < +∞} is dense in l1 , it suffices to consider y ∈ cc . By
construction, for all y ∈ cc we have (T x)(y) = ϕ(y).

(9) Prove that equivalent norms on a normed linear space X lead to equivalent
norms on the space B(X) of bounded linear operators on X.
Solution: Let k · k1 and k · k2 be the two equivalent norms on a normed linear
space X, i.e. there exist constants c and C such that for all x ∈ X

ckxk1 ≤ kxk2 ≤ Ckxk1 .

7
Let T ∈ B(X). Then

kT (x)k1 c−1 kT xk2 C

kT k1 = sup ≤ sup −1 = kT k2 .
x6=0 kxk1 x6=0 C kxk2 c

Similarly,
kT (x)k2 CkT xk1 C
kT k2 = sup ≤ sup = kT k1 .
x6=0 kxk2 x6=0 ckxk1 c
In other words,
c C
kT k1 ≤ kT k2 ≤ kT k1 .
C c

8
 Homework with solutions

(1) Define the right shift operator S ∈ B(lp ) on lp := lp (N, K) by

Sx := (0, x1 , x2 , . . . ).
Define the left shift operator T ∈ B(lp ) by
T x := (x2 , x3 , . . . ).
a) Show that S is an injection but not a surjection, and T is a surjection, but
not an injection.
b) Show that kSk = kT k = 1.
c) Show that T S = I (the identity operator), but ST 6= I.
Solution:
a) If Sx = 0 then for all n we have xn = 0, therefore x = 0. So S is injection.
Since the sequence (1, 0, . . . ) has no pre-image, S is not a surjection.
Since for the sequence x = (1, 0, . . . ), T x = 0, T is not injective.
For every y ∈ lp define x = (0, y1 , y2 , . . . ), then T x = y, and so T is
surjective.
P 1/p
∞ p
b) For every x ∈ lp we have kSxkp = j=1 |x j | = kxkp . Therefore
kSk = supx∈lp \{0} kSxkp /kxkp = 1.
P 1/p
∞ p
For every x ∈ lp , kT xkp = j=2 |xj | ≤ kxkp . Therefore kT k ≤ 1.
But for every x = (0, x1 , x2 , . . . ) we have kT xkp = kxkp . Thus kT k = 1.
c) For x = (x1 , x2 , . . . ), we have T S(x) = T ((0, x1 , x2 , . . . )) = (x1 , x2 , . . . ) =
x. But ST (x) = S((x2 , x3 , . . . )) = (0, x2 , x3 , . . . ) 6= x, so ST 6= I.
(2) Let X = (C([0, 1]), k k∞ ), and define a linear operator K on X by
Z 1
(Kf )(x) := k(x, y)f (y)dy,
0
where k : [0, 1] × [0, 1] → R is continuous. Prove that K is a bounded operator.
Solution: Note that k is continuous on the compact set [0, 1] × [0, 1], and hence
it is bounded, i.e., kkk∞ := maxx,y∈[0,1] |k(x, y)| < +∞. Consider
Z 1 
kKf k∞ = sup |(Kf )(x)| = sup  k(x, y)f (y)dy 
 
0≤x≤1 0≤x≤1 0
Z 1
≤ sup |k(x, y)||f (y)|dy ≤ kf k∞ kkk∞ .
0≤x≤1 0

9
 Therefore K is bounded, and kKk ≤ kkk∞ .
Score: 3 points.

(3) Let c0 := {x ∈ KN : limn→+∞ x(n) = 0}. Show that c∗0 ∼

= l1 .
Solution: We follow the same argument as in the lecture. Define T : l1 → c∗0 ,
by
∞
X
(T x)(y) := xk yk (0.3)
k=1

for x ∈ P
l1 and y ∈ c0 . The above sum is well-defined, as it is absolute convergent
due to ∞ k=1 |xk yk | ≤ kxk1 kyk∞ ≤ ∞. Then

∞
X
|(T x)(y)| ≤ |xk yk | ≤ kxk1 kyk∞ ≤ ∞.
k=1

This means that for all x ∈ l1 we have T x ∈ c∗0 , and

|(T x)(y)|
kT xk = sup ≤ kxk1 . (0.4)
y∈c0 \{0} kyk∞

On the other hand, for every k such that xk 6= 0, let yk := |xk |/xk , and let yi := 0
(N ) (N )
otherwise. For every N ∈ N, let yi := yi , 1 ≤ i ≤ N , and yi := 0, otherwise,
(N ) (N ) (N )
i.e., y = (y1 , . . . , yN , 0, 0, . . .). For all N ∈ N, y ∈ c0 and y

∞
≤ 1, and
hence
N
X
(N )
kT xk ≥ |(T x)y |= |xi |. (0.5)
i=1

Taking the limit N → +∞, we get kT xk ≥ kxk1 , and combining it with (0.4)
yields kT xk = kxk1 . That is, T is an isometry. This immediately implies that
T is injective. Hence, we are left to prove that T is also surjective.
To this end, let ϕ ∈ c∗0 and define x = (x1 , x2 , . . . ) by xk := ϕ(e(k) ), where
e(k) = (0, . . . , 0, 1,
P0, . . . ) has 1 at the k-th place and 0 everywhere else. Note
+∞
that (T x)(y) := k=1 xk yk defines a linear functional on cc := {y ∈ KN : #{i :
y(i) 6= 0} < +∞} such that T x = ϕ|cc , and the same argument as in (0.5)
shows that kxk1 ≤ kT xk = kϕ|cc k ≤ kϕk < +∞, i.e., x ∈ l1 . Hence, T x can
be extended to a bounded linear functional on l1 by the formula given in (0.3).
Since T x and ϕ are both bounded linear functionals on l1 , and they cocincide
on the dense subspace cc , we finally obtain that T x = ϕ.

10
 Score: Continuous embedding of l1 into c∗0 : 3 points. The embedding is isome-
try+injectivity: 4 points. Surjectivity: 3 points.
Total score: 10 points.

Total score for HW: 13 points.

11