You are on page 1of 11

University of Bergen

General Functional Analysis


Problems with solutions 6



1) Prove that 0 is unique in any normed space.

Solution of 1) Let us suppose that there are 2 zeros 01 and 02 . Then


01 = 01 + 02 = 02 + 01 = 02 .

2) Let X be a compact space. Prove that the space C(X) of continuous real-valued functions is a complete
metric space.

Solution of 2) Let {xn } be a Cauchy sequence in C(X); that is for any ε > 0 there exists N0 ∈ N such
that
(1) n, m ≥ N0 =⇒ |xn (t) − xm (t)| ≤ kxn − xm k∞ < ε ∀ t ∈ X.
It implies that the sequence {xn } converges uniformly on X. Since X is compact the limit function x(t)
will be continuous. Letting m → ∞ in (1) we get
|xn (t) − x(t)| < ε ∀ t ∈ X and ∀ n ≥ N0 .
This means that {xn } converges uniformly, or in other words converges in the metric of C(X).

3) Let A be a subset of vector space V . Construct a smallest linear subspace (linear manifold) contain-
ing A.

4) Show that the set P of all polynomials on [0, 1] is a linear manifold in C[0, 1]. Is it closed? Give an
example of a closed linear manifold in C[0, 1].

5) Let X be a vector space. Let B = {x ∈ X : kxk < 1}. Prove that B is open and that
B = {x ∈ X : kxk ≤ 1}.

6) Suppose that A is a linear operator between two normed vector spaces X and Y . Show that the
following are equivalent:
(1) A is continuous at one point,
(2) A is continuous at all points,
(3) A is uniformly continuous,
(4) A is bounded.

Solutions of 6)
(4) =⇒ (3): If A is bounded, then kAxk ≤ M kxk for all x ∈ X. Then kAx − Ayk ≤ M kx − yk,
so A is uniformly continuous.
(3) =⇒ (2): This is true by the definitions of continuity and uniform continuity.
(2) =⇒ (1): This is true by definition.
(1) =⇒ (4): Suppose that A is continuous at x0 . Then for all ε > 0 there is some δ > 0 such that
kx − x0 k =⇒ kAx − Ax0 k < ε.
Therefore if kyk < δ then
kAyk = kA(y + x0 ) − Ax0 k < ε,
so we can assume x0 = 0.
1
2

If we let ε = 1 we have that


kyk < δ =⇒ kAyk < 1.
δ
Thus, if we take an arbitrary x 6= 0 (A is linear so A0 = 0) and we consider y = 2kxk x then

2kxk 2kxk 2
kAxk = kA( y)k = kAyk ≤ kxk.
δ δ δ
We conclude that A is bounded.

7) Let X be a normed vector space and Z its subspace (linear manifold). Prove that if y ∈ X has distance
d from Z, then there exists a linear functional Λ : X → R such that
kΛk ≤ 1, Λ(y) = d and Λ(z) = 0, ∀z ∈ Z.
(The distance referred is d = inf z∈Z kz − yk)

Solutions of 7) If y ∈ Z, then d(y, Z) = 0 and the zero functional works, so we may assume that
y ∈ X \ Z. Consider the subspace Y = {λy + Z : λ ∈ R} ⊂ X. Since y ∈
/ Z for every x ∈ Y , there is a
unique α ∈ R and z ∈ Z such that x = αy + z. Define Λ(x) = Λ(αy + z) = ad. Observe that Λ(z) = 0
for all z ∈ Z and that Λ(y) = d. The functional Λ is linear and
1
|Λ(x)| = αd ≤ |α|ky + zk = kxk
α
by the definition of d. Thus Λ is linear bounded functional with kΛk ≤ 1. Applying the Hahn - Banach
theorem we extend Λ on all X with the same estimate for the norm.
³ ´
8) Let C[−1, 1], k · k∞ be a space of continuous functions on [−1, 1] with kxk∞ = sup |x(t)|. Show
t∈[−1,1]
that the functional
Z 0 Z 1
f (x) = x(t) dt − x(t) dt
−1 0
is linear, continuous and bounded. Find its norm.

Solutions of 8) Let us show that it is linear. Let α, β ∈ R1 , x, y ∈ C[−1, 1].


Z0 Z1
f (αx + βy) = (αx + βy)(t) dt − (αx + βy)(t) dt
−1 0
Z0 Z0 Z1 Z1
= α x(t) dt + β y(t) dt − α x(t) dt − β y(t) dt
−1 −1 0 0
= αf (x) + βf (y).
Now we show that the functional is bounded and thus is continuous. We have
¯ Z0 ¯ ¯ Z1 ¯
¯ ¯ ¯ ¯
|f (x)| ≤ ¯ x(t) dt¯ + ¯ x(t) dt¯ ≤ 2kxk∞
−1 0

for any x ∈ C[−1, 1]. If kxk∞ ≤ ε/2 then kf k∞ ≤ ε ∀ε > 0.


We conclude that kf k ≤ 2. Let us show that kf k = 2. Consider the following sequence of functions

 1 if x ∈ [−1, − n1 ),
xn (t) = −1 if x ∈ ( n1 , 1],

nx if x ∈ [− n1 . n1 ],
3

1
We have |f (xn )| = 2 − n and kxn k∞ = 1. Thus
|f (xn )|
sup =2
n∈N kxn k∞

that shows kf k = 2

9) Verify, that the function


m
X
kxk = max |x(k) (t)|
t∈[a,b]
k=0
define a norm in the space C [a, b] of the continuous functions that have up to the m derivatives x(k)
m

and the derivatives are also continuous on [a, b]. What does it mean that a sequence converge in this
space?

10) Let (X, k · k) be a normed space. Let {xn } and {yn } be two Cauchy sequences in X. Show that the
sequence λn = kxn − yn k converges.

Solution of 10) Let ε > 0 and N0 ∈ N is such that


ε ε
if n, m ≥ N0 =⇒ kxn − xm k < and kyn − ym k < .
2 2
Since
kxn − yn k ≤ kxn − xm k + kxm − ym k + kym − yn k
and
kxm − ym k ≤ kxm − xn k + kxn − yn k + kyn − ym k,
then
|λn − λm | = | kxn − yn k − kxm − ym k | ≤ kxn − xm k + kym − yn k < ε.
Since (R, | · |) is complete the Cauchy sequence {λn } has a limit Λ ∈ R.

11) Let c0 be a space of real sequences x = {xn }∞


n=0 converging to 0. Let l∞ be a set of real sequences
w = {wk }∞
k=0 furnished with the norm kwk ∞ = sup |wn |. Prove that c0 is closed in l∞ .
1≤n<∞

Solution of 11) Let x(k) ∈ c0 be a sequence converging to ω ∈ l∞ . Take ε > 0 and N0 ∈ N such that
(k) (k)
ρsup (x(k) , ω) = sup |xn − ωn | < ε/2 for all k > N0 . For each k choose N1 ∈ N such that |xn | ≤ ε/2
1≤n≤∞
for all n > N1 . Thus,
|ωn | ≤ |ωn − x(k) (k)
n | + |xn | ≤ ε
for n > N1 and k > N0 , that means that the sequence ωn converges to 0 and ω ∈ c0 .

12) Show the inequality


|f + g|p ≤ 2p (|f |p + |g|p ).

13) Let X = (X, A, µ) be a measure space with σ-algebra A and positive measure µ. Let f ∈ Lp . Show
that ³Z ´p
kf kp = |f |p dµ
X
is a norm. (Extend the Minkowskii and Hölder inequality from a sum to an integral.)

14) Let us denote by L∞ (X) a set of measurable functions f on X = (X, A, µ) endowed with the norm
© ª
kf k∞ = ess sup f (t) = inf M : µ{t : f (t) > M } = 0 < ∞.
Show that the space L∞ (X) is complete.
4

Solution of 14) Let {fn }∞


n=1 be a Cauchy sequence in L∞ (X). Let also

Ak = {x ∈ X : |fk (x)| > kfk k∞ }


and
Bm,n = {x ∈ X : |fn (x) − fm (x)| > kfn − fm k∞ }.
By E we denote the union of Ak and Bm,n over all k, m, n ∈ N. Measure of E is zero because of the
zero measure of all Ak and Bm,n . On the complement of E the sequence {fn } converges uniformly to a
bounded function f . (Why?) Define f (x) = 0 for x ∈ E. Then f (x) ∈ L∞ (X) and we get
kfn − f k∞ → 0 as n → ∞.
R
15) If f ∈ L1 (X) and g ∈ L∞ (X) then show that X
|f g| dµ ≤ kf k1 · kgk∞ .

Solution of 15) Notice that


|f (x)g(x)| ≤ kgk∞ |f (x)|
for almost all x ∈ X. Then
Z Z
|f (x)g(x)| dµ ≤ kgk∞ |f (x)| dµ = kf k1 · kgk∞ .
X

16) Let E be a normed space and f ∈ E ∗ . Define kernel of f by Ker(f ) = {x ∈ E : f (x) = 0}. Show
that
|f (x)|
dist(x, Ker(f )) = , ∀ x ∈ E.
kf k

Solution for 16) Remember that ρ(x, L) = inf kx − yk. If y ∈ L, then


y∈L

|f (x)| = |f (x − y)| ≤ kf k · kx − yk.


Thus
|f (x)|
ρ(x, L) = inf kx − yk ≥ .
y∈L kf k
Let us show the reciprocal inequality. We take a sequence {xn } ∈ E such that kxn k = 1 for all n ∈ N
and
n
kf k < f (xn ).
n+1
Consider the element y = x − ff(x
(x)
n)
xn It is easy to see that y ∈ L. Moreover,
¯ f (x) ¯ n + 1 |f (x)|
¯ ¯
kx − yk ≤ ¯ ¯ · kxn k ≤ .
f (xn ) n kf k
Taking firstly the limit as n → ∞, then taking the infimum over all y ∈ L, we get the result
|f (x)|
ρ(x, L) ≤ .
kf k

17) Show that all norms on Rn are equivalent.

Solution of 17) Let k · k0 be the Euclidean norm (with respect to a basis e1 , . . . , en ) and let k · k be any
other norm on Rn .
Pn
We claim that x 7→ kxk is continuous with respect to the Euclidean norm. Write x = k=1 ck ek .
Then by the triangle inequality,
n
X ¡ ¢
kxk ≤ |ck | kek k ≤ n max kek k max |ck | ≤ M kxk0 ,
k=1
5

ε
where M = n max kek k. Thus if kx − yk0 ≤ M then kx − yk ≤ ε and we conclude that the map is
continuous.
Now we consider the region K = {x : kxk0 = 1}. This is just the unit sphere in Rn with respect to
the Euclidean norm, which is compact. The map x 7→ kxk is continuous, so it attains a minimum m and
x
a maximum M on K. Thus, for x ∈ K, m ≤ kxk ≤ M . Now for x ∈ \{0}, kxk 0
∈ K and then
kxk
m≤ ≤M =⇒ mkxk0 ≤ kxk ≤ M kxk0 .
kxk0
We see that the norm are equivalent.
R1 √ R1
18) Consider the linear functionals f (x) = 0
x( t) dt and f (x) = 0 x(t2 ) dt over C[0, 1]. Are they
bounded?

19) Let V be a normed space and {fn }∞ ∗


n=1 ∈ V . Prove that fn converges if and only if {fn } converges
uniformly on the unit ball B(0, 1) = {x ∈ V : kxk ≤ 1}.

20) Let V be a normed space and x ∈ V . Show that


kxk = sup |f (x)|.
f ∈V ∗ ,kf k=1

21) Given f ∈ Lp (), 1 ≤ p < ∞ and ε > 0 there is a bounded measurable function b with |b| < M and
kf − bkp ≤ ε.

21) We define the cut-function



 n for n ≤ f (x),
fn (x) = f (x) for −n ≤ f (x) ≤ n,

−n for f (x) ≤ −n.
Then |fn | ≤ n, and the sequence {fn } converges to f almost everywhere. Thus |f − fn | → 0 also almost
everywhere. Since |fn (x)| ≤ |f (x)| we have |f − fn |p ≤ 2p |f |p . Moreover f ∈ Lp and we get that
Z
kf − fn kpp = |f − fn |p → 0 as n → ∞

by the Lebesgue Convergence Theorem. We conclude that given ε > 0 there is b = fn such that
|b| ≤ M (= n) (function is bounded) and kf − bkp ≤ ε.
¡ ¢
22) Let g be an integrable function on [0, 1] g ∈ L1 ([0, 1]) . Show that there is bounded measurable
function f (f ∈ L∞ ([0, 1])) such that kf k∞ 6= 0 and
Z
f g = kgk1 · kf k∞ .
[0,1]

Solution of 22) Let us suppose that we defined the linear functional F as in the case of p ∈ [1, ∞)
for the characteristic function, for the step function ψ, and for the bounded measurable function f . Let
g ∈ L1 ([0, 1]) and ε > 0. Then there is a bounded measurable function f such that kf − gk1 < ε. Since
R
f is bounded then F (f ) = f g by the previous considerations. Then
Z
|F (g) − f g| = |F (g) − F (f )| ≤ kF k kf − gk1 ≤ kF kε.
R
thus F (g) = f g since ε is arbitrary.

23) Let c0 be a space of sequences x = (x1 , . . . , xn , . . .) converging to 0 with the norm kxk = supn |xn |.
Show that the dual space is isomorphic to the l1 of all absolutely summable sequences f = (f1 , . . . , fn , . . .)
P∞
with the norm kf k = n=1 |fn |.
6

Solution of 23) Any element f ∈ l1 defines the bounded linear functional F over c0 by

X
F (x) = fn xn .
n=1
P∞
Since |F (x)| ≤ kxk n=1 |fn | we have

X
kF k ≤ |fn | = kf k.
n=1

Now we prove the reciprocal inequality. Let us consider the unit elements in c0 :
e1 = (1, 0, 0, . . . , 0, 0, . . .),
e2 = (0, 1, 0, . . . , 0, 0, . . .),
..................,
en = (0, 0, 0, . . . , 1, 0, . . .),
...................

Set ( P
N fn
n=1 |fn | en for fn =
6 0,
x(N ) =
0 for fn = 0.
(N ) (N )
Then x ∈ c0 , kx k ≤ 1 and
XN XN
fn
F (x(N ) ) = F (en ) = |fn |.
n=1
|fn | n=1
P∞ |F (x(N ) )| P∞
We conclude that limN →∞ F (x(N ) ) = n=1 |fn | = kf k. Thus kF k ≥ kx(N ) k ≥ n=1 |fn | = kf k.
Finally, we obtain kF k = kf k.
We constructed the linear isometry f → F of the space l1 into the space c∗0 . Let us show that it is the
P∞
map onto; that is any functional F ∈ c∗0 has the representation F (x) = fn xn , where f = {fn } ∈ l1 .
P∞ n=1
For any x = {xn } ∈ c0 we have x = n=1 xn en . The series on the right hand side converges to x in
c0 because of
° N
X °
° °
°x − xn en ° = sup |xn | → 0 as N → ∞.
n=1 n>N

The functional F ∈ c∗0 is continuous, thus



X
F (x) = xn F (en ).
n=1
P∞ PN F (en )
So we have to verify that n=1 F (en ) < ∞. Setting x(N ) = n=1 |F (en )| en we deduce
N
X XN
F (en )
|F (en )| = F (en ) = F (x(N ) ) ≤ kF k
n=1 n=1
|F (e n )|
P∞
because of x(N ) (N )
∈ c0 , kx k ≤ 1. Since N was arbitrary we conclude that n=1 F (en ) < ∞.

24) Given f ∈ Lp , 1 < p ≤ ∞ and ε > 0, there is a step function ψ such that
kf − ψkp < ε.

Solution of 24) By the problem 1 we may choose a bounded function b such that kf − bkp < ε/2 and
b < M . Any bounded measurable function can be approximated by the step functions, so we can find a
7

step function ψ such that |b − ψ| ≤ ε/4 except on a set E of the measure µ(E) < δ (we presice δ later).
We set ψ(x) = 0 on E. Then
R R R
kb − ψkpp = [0,1] |b − ψ|p = [0,1]\E |b − ψ|p + E |b − ψ|p
εp p εp
< + M δ ≤
4p 2p
εp
if we choose δ = M p 4p . Consequently kb − ψkp < ε/2. Applying the Minkowskii inequality we get
kf − ψkp ≤ kf − bkp + kb − ψkp < ε.

25) Study the analogue of Hahn-Banach theorem for complex vector space. Let X be a complex
vector space, S a linear subspace, p a real-valued function on X such that p(x + y) ≤
p(x) + p(y), and p(αx) = |α|p(x). Let f be a complex linear functional on S such that
|f (s)| ≤ p(s) for all s ∈ S. Then there is a linear functional F defined on X such that
F (s) = f (s) for s ∈ S and |F (x)| ≤ p(x) for all x ∈ X.

26) Let l∞ be a space of bounded real-valued sequences x = (x1 , . . . , xn , . . .) with the norm kxk =
sup |xn |. Prove that l∞ = l1∗ , but l∞

6= l1 by using the Hahn - Banach theorem. For the last part, note
n=1,... P

that l1 can be identified with a subset of l∞ by a(b) = an bn for a = {an }∞ ∞
n=1 ∈ l1 , b = {bn }n=1 ∈ l∞ .
Show that not every element of l∞ is of this form.

Solution of 26) First note that given any b = {bn }∞ n=1 ∈ l∞ we obtain a bounded linear functional on
P P
l1 by a(b) = an bn for a = {an }∞
n=1 ∈ l1 . The sum is well-defined because {b n } is bounded and a n is
absolutely convergent. It is clearly linear in {an } and ka(b)k ≤ kbk∞ kakl1 . So kak ≤ kbk∞ . We showed
that any element from l∞ defines a linear functional on l1
We need to show that any bounded linear functional F ∈ l1∗ is defined by unique elements from l∞ .
Suppose we have F ∈ l1∗ . Let bi = F (ei ), where ei has 1 in the i-th place, and zeros everywhere else. F
is a bounded linear functional, so we have that |bi | = |F (ei )| ≤ kF k, so if we let b = {bi }, then b ∈ l∞ .
P
We now claim that F ({an }) = an bn . Indeed, consider the following sequence am ∈ l1
½
an if n ≤ m,
am
n =
0 if n > m.
The functional F is linear, so
m m m
¡X ¢ X X
F (am ) = F a n en = an F (en ) = an bn .
n=1 n=1 n=1
P
The series an is absolutely convergent and {bn } is bounded, so the righthand side above converges to
P
an bn as m → ∞. F is continuous on l1 , and am → an in l1 , so the lefthand side converges to F ({an }),
giving us that F comes from the element b ∈ l∞ .
The last step is to show that the constructed map is an isometry. We already know that kbkl1∗ ≤ kbk∞ .
The norm kbk∞ = sup |bn | implies that given ε > 0, there is an N ∈ N such that |bN | > kbk∞ − ε. then
kbkl1∗ ≥ |b(eN )| = |bN | > |bk∞ − ε =⇒ kbkl1∗ ≥ kbk∞ .

Now we show that l1 6= l∞ . First note that the first paragraph above also shows that we may identify
∗ ∗
l1 with a subspace of l∞ , so it is enough to show that subspace in not all of l∞ . Consider the closed
subspace c of l∞ consisting of convergent sequences. If {an } ∈ c, define f ({an }) = limn→∞ an . This
defines a bounded linear functional on c, which is closed. The Hahn - Banach theorem says that we can
extend it to a bounded linear functional F on l∞ . Observe, that f 6= 0 and that F |c0 = 0. Suppose that

P
F ∈ l1 considered as a subspace of l∞ , then F ({an }) = an bn for some {bn } ∈ l1 . The functional f
vanishes on c0 , so bn = f (en ) = 0 because en ∈ c0 . This tells us that F should be the zero functional,

but it is not so in the subspace of l∞ corresponding to l1 .
8

27) Let D0 ([0, 1]) denote the topological dual of C ∞ ([0, 1]) with the corresponding topology. One calls
elements of D0 ([0, 1]) distributions or generalised functions. Let ι : L1 ([0, 1]) → D0 ([0, 1]) be the inclusion
map defined by
ι(φ)[ψ], φ ∈ L1 ([0, 1]), ψ ∈ C ∞ ([0, 1]).
(1) Show that ι indeed maps into D0 ([0, 1]), is injective and continuous. One regards L1 ([0, 1]) as a
subset of D0 ([0, 1]).
(2) Let τ = ι|C ∞ ([0,1]) . Show that τ : C ∞ ([0, 1]) → D0 ([0, 1]) is continuous.
d d
(3) Show that dx : C ∞ ([0, 1]) → C ∞ ([0, 1]) has a continuous extension to a map dx : D0 ([0, 1]) →
D0 ([0, 1]), given by
d dψ
u[ψ] = −u , u ∈ D0 ([0, 1]), ψ ∈ C ∞ ([0, 1]).
dx dx
Thus every distribution, in particular every L1 function, can be differentiated arbitrary many times, in
the sense of distributions.

Solution of 27) (1) Suppose f ∈ L1 ([0, 1]), ψ ∈ C ∞ ([0, 1]), then


Z
¯ ¯
|ι(f )[ψ]| = ¯ f ψ ¯ ≤ kf k1 kψk∞ ≤ Ckψk∞ .
[0,1]
0
R
So ι(f ) ∈ D ([0, 1]). The functional f → f ψ is continuous that follows from the above inequality.
Now if f ∈ C([0, 1]) and ι(f ) = 0, then we claim that f = 0. Indeed, if f 6= 0, then {f > 0} and
{f < 0} are both open and at least one of them is non-empty. Without loss of generality, we may assume
there is a nonempty open set on which {f > 0}. We may find ψ ∈ C ∞ ([0, 1]) such that the support
supp ψ ⊂ {f > 0}, ψ ≥ 0, and ψ(x) > 0 for some x ∈ {f > 0}. (Construct such function!). Then
Z Z
0 = ι(f )[ψ] = f ψ = fψ > 0
{f >0}

a contradiction, so that f = 0. Thus ι|C([0,1]) is injective.


For f ∈ L1 ([0, 1]), ι(f ) = 0, we consider the functions
Z 1 Z 1
k x+ k k k
fk (x) = f (y) dy = f (x + y) dy.
2 x− k1 2 − k1
The functions fk are continuous because if xn → x, we have
Z Z x+ k1
k ³ x− k ´
1

|fk (xn ) − fk (x)| ≤ |f (y)| dy + |f (y)| dy → 0


2 xn − k1 1
xn + k

because f ∈ L1 ([0, 1]), with a similar statement holding for xn → x.


Lebesgue’s theorem tells us that fk → f almost everywhere. Now, for ψ ∈ C ∞ ([0, 1]) we have that
Z Z 1
k 1 k
ι(fk )[ψ] = f (x + y)ψ(x) dydx =
2 0 − k1
Z 1 Z 1 Z 1
k k k k
= f (x)ψ(x − y) dxdy = (ι(f ))[ψ(· − y)]dy = 0
2 − k1 0 2 − k1
by Fubini’s theorem (since f ψ is integrable on the product space). Then ι(fk ) = 0, so fk = 0. The
sequence fk → f almost everywhere, so f = 0 almost everywhere and thus f = 0, so ι is injective.
(2) We just need to show that the inclusion j : C ∞ ([0, 1]) → L1 ([0, 1]) is continuous, but we have for
ψ ∈ C ∞ ([0, 1])
Z 1
kψk1 = |ψ| ≤ 2πkψk∞
0
so it is continuous, so then τ = ι ◦ j is continuous.
d
(3) In order to show that the map dx : D0 ([0, 1]) → D0 ([0, 1]) is continuous then it is enough to show
that for each ψ, the map u 7→ ( dx u)[ψ] is continuous. But this functional is just u 7→ u(− dψ
d
dx ), and
9

d
so it is continuous. We now wish to show that dx indeed extends the map from C ∞ ([0, 1]). Indeed, if
ψφ ∈ C ∞ ([0, 1]), we have
Z Z
d dψ dψ dφ dφ
( (τ φ))[ψ] = −(τ φ)[ ] = − φ = ψ = τ ( )[ψ]
dx dx dx dx dx

by integration by part.

28) Suppose V is an inner product space (positive definite scalar product) either over R or over C.
(1) Prove that the inner product can be recovered from the norm by the polarisation identity:

1
(x, y) = (kx + yk2 − kx − yk2 )
4

if the field if R, and

1 i
(x, y) = (kx + yk2 − kx − yk2 ) − (kx + iyk2 − kx − iyk2 )
4 4

if the field is C.
(2) Prove that a normed vector space (V, k · k) is a n inner product space (with the induced norm begin
k · k) if and only if the norm satisfies the parallelogram law:

kx + yk2 + kx − yk2 = 2kxk2 + 2kyk2 ∀ x, y ∈ V.

(3) Conclude that the standard norms on lp and Lp (R), p 6= 2, do not arise from inner products.

Solution of 28) (1) For the R case, we just expand the right hand side and use the symmetry of the
inner product:

1 ¡ ¢
(kx + yk2 − kx − yk2 ) = 1
4 (x, x) + (y, y) + (x, y) + (y, x) − (x, x) − (y, y) + (x, y) + (y, x)
4 ¡ ¢
= 21 (x, y) + (y, x) = (x, y).

For the other case, we again expand the right side, using the relation we just proved:

1
4 (kx+ yk2 − kx − yk2 ) − 4i (kx + iyk2 − kx − iyk2 )
¡ ¢ ¡ ¢
= 21 (x, y) + (y, x) − 2i (x, iy) + (iy, x))
i2 i2
= (x, y) − 2 (x, y) + 2 (y, x) = (x, y).

(2) If the norm comes from the inner product, then we have

kx + yk2 + kx − yk2 = 2(x, x) + 2(y, y) + (x, y) + (y, x) − (x, y) − (y, x) = 2(kxk2 + kyk2 ).

Now we suppose that the norm satisfies the parallelogram law. Assume the field is C (the argument is
similar to R),and define the inner product via the polarisation identity from part (1). If x, y, z ∈ V , we
10

y+z y−z
have that (writing x + y = x + 2 + 2 and similar for x + z):
1¡ ¢
(x, y) + (x, z) = kx + yk2 + kx + zk2 − kx − yk2 − kx − zk2
4
i¡ ¢
− kx + iyk2 + kx + izk2 − kx − iyk2 − kx − izk2
4
1³ y+z 2 y−z 2 y+z 2 y − z 2´
= kx + k +k k − kx − k −k k
2 2 2 2 2
i ³ y+z 2 y−z 2 y+z 2 y − z 2´
− kx + i k + ki k − kx − i k − ki k
2 2 2 2 2
1 ³ y+z 2 y+z 2 y+z 2 y+z 2 ´
= kx + k +k k − kx − k −k k
2 2 2 2 2
i ³ y+z 2 y+z 2 y+z 2 y + z 2´
− kx + i k + ki k − kx − i k − ki k
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 = (x, y + z).
4
p
This holds for all x, y, z ∈ V , so, in particular, (x, ny) = n(x, y), and so if r = q ∈ Q, then (x, ry) = r(x, y).
Moreover, by the polarisation identity,
1¡ ¢ i¡ ¢
(x, iy) = kx + iyk2 − kx − iyk2 − kx − yk2 − kx + yk2 = i(x, y).
4 4
Combining these two results gives that if α ∈ Q + iQ, then (x, αy) = α(x, y). If α ∈ C, then there are
αn ∈ Q + iQ, αn → α, so
(x, αn y) → (x, αy)
because all the norms above must converge. This gives us that

α(x, y) = lim αn (x, y) = lim (x, αn y) = (x, αy),


n→∞ n→∞

so (x, y) is linear.
Observe that because ki(x − iy)k = kx − iyk, we have that (y, x) = (x, y), and that
1 i
(x, x) = (2kxk2 ) = (|1 + i|2 kxk2 − |1 − i|2 kxk2 ) = kxk2 ,
4 4
so this shows that the norm is induced by (·, ·) and that it is also positive definite, so it is an inner
product.
(3) Let f = χ[0,1/2] , g = χ[1/2,1] ∈ Lp (R), so then

kf + gk2 = kf − gk2 = 1,

and
³ 1 ´2/p
kf k2 = kgk2 = ,
2
so the parallelogram law does not hold unless p = 2. (For p = ∞, the same examples work, but the norms
are slightly different.)
³ ´2/p
Let a = (1/2, 1/2, 0, 0, . . .), b = (1/2 − 1/2, 0, 0, . . .) ∈ lp , then kak2 = kbk2 = ( 21 )p + ( 12 )p , while
ka + bk2 = ka − bk2 = 1. So the parallelogram law does not hold unless p = 2.

29) Let Tt be the operator


Tt (φ)(x) = φ(x + t)
on L2 (R). What is the norm of Tt ? What does it converge to as t → ∞ and in what topology?
11

Solution of 29) First we claim that for each phi ∈ C ∞ (R) ⊂ L2 (R), Tt φ → 0 weakly as t ∈ ∞. Indeed,
let ψ ∈ L∗2 (R) = L2 (R). Fix ε > 0. Since ψ ∈ L2 (R) there is some K such that
³Z ´1/2
|ψ(x)|2 dx < ε.
|x|>K

We let M be such that supp φ ⊂ [−M, M ]. Then if t > M + K we have that if φ(x) 6= 0, then x − t < −K,
and so we have that
Z Z
|(Tt φ, ψ)| = | φ(x + t)ψ(x) dx| = | φ(x)ψ(x − t) dx|
Z M ³ ´1/2
= | φ(x)ψ(x − t) dx| ≤ |ψ(x − t)| dx kφk2 ≤ Cφ ε,
−M
so |(Tt φ, ψ)| → 0 as t → ∞.
Now for f ∈ L2 (R), we choose φ ∈ C ∞ (R) with compact support such that kφ − f k2 ≤ ε, so then we
have
|(Tt f, ψ)| ≤ |(Tt (f − φ), ψ)| + |(Tt φ, ψ)| ≤ 2ε
for large enough t. Thus Tt → 0 in the weak operator topology, assuming that it is bounded, which we
show now.
Foe any φ ∈ L2 (R), we have that
Z Z
|φ(x + t)|2 dx = |φ(x)|2 dx =⇒ kTt φk = kφk2 =⇒ kTt k = 1.
R R
This also shows that Tt does not converge to anything in the strong operator topology (and hence not in
norm, either) because it must converge to 0, but kTt φk = kφk.

30) Let f ∈ L1 ([0, 1]). Let g be a bounded measurable function on [0, 1]. Then gf is in L1 ([0, 1]).

Solution of 30) Let ϕn be a sequence of step functions converging both L1 and almost everywhere to
f . Let ψn be a sequence of a step functions converging pointwise to g. Then ϕn ψn is a sequence of step
functions and as n → ∞ this sequence converges almost everywhere to f g. Changing f and g on a set of
measure 0 (e.g. giving them the value 0), we can assume that this convergence is pointwise everywhere.
If C is a bound for g, i.e. |g(x)| ≤ C for all x, then
|f g| ≤ C|f |.
We conclude that f g ∈ L1 ([0, 1]) by the Dominated Convergence Theorem.

You might also like