Vous êtes sur la page 1sur 3

MATH414 FUNCTIONAL ANALYSIS

HOMEWORK 3

Hw 3: 7/82 5/101 6/101 8/101 9/101

Problem (7/82)

If dimY < in Rieszs lemma 2.5-4, show that one can even choose = 1.

Solution

Step1 If Y is a finite dimensional subspace of a normed space X then for


any x X there exists y0 Y such that d(x, y0 ) = d(x, Y ), that is,

||x y0 || = inf ||x y||.


yY

Let a = d(x, Y ), then by definition for any n N there is an element yn in Y



such that ||x yn || a + n1 . So {yn }n=1 forms a bounded sequence in Y . Let
||yn || M for all n in N. Notice that A = {y Y : ||y|| M } is a compact
set since it is closed and bounded subset of a finite dimensional normed space

Y (see theorem 2.5-3). Since {yn }n=1 is a sequence in A, it has a convergent

subsequence, say {ynk }k=1 with limit y0 . We claim that d(x, y0 ) = ||x y0 || = a.
Clearly it is a. On the other hand
1
||x y0 || ||x ynk || + ||ynk y0 || a + + ||ynk y0 || for all k N.
nk
So as k 7 we find ||x y0 || a. This proves the claim.

Step2 If dimY < , then we can re-formulate Rieszs lemma as:

If Y is a proper finite dimensional subspace of normed space X, then x in X


such that ||x y|| 1 for all y Y .

Proof. Let v X Y and a = inf yY ||v y||. Clearly a > 0 since Y is closed
(see theorem 2.4-2). We know from step1 that y0 Y such that a = ||v y0 ||.
So let
v y0 v y0
x= = .
||v y0 || a
Evidently ||x|| = 1 and for any y Y ,
v y0 1 1
||x y|| = || y|| = ||v y0 ay || a = 1.
a a | {z } a
in Y

1
Problem (5/101)

Show that the operator T : l l , (j ) 7 (j /j) is linear and bounded.

Solution T is linear since if x = (j ), y = (j ) are vectors and , scalars then


j + j j j
T (x + y) = T (j + j ) = ( ) = ( ) + ( ) = T x + T y.
j j j

And boundedness of T follows from


j
||T x|| = sup | | sup |j | = ||x|| (c = 1).
j j j

Problem (6/101)

(Range) Show that the range R(T ) of a bounded linear operator T : X Y


need not be closed in Y . Hint. Use T in Prob. 5.

Solution Consider the bounded


linear operator T in previous question. Since
for any n N, xn = (1, 2, ..., n, 0, 0, ...) is in l , we obtain
1 1
T xn = yn = (1, , ..., , 0, 0, ...) is in R(T ) n N.
2 n

Clearly {yn }n=1 is a convergent sequence with limit y = ( 1n )n=1 . But y is not

in R(T ). If not
x = (n ) l such that T x = y, but the definition of T
requires n = n which is a contradiction.

Problem (8/101)

Show that the inverse T 1 : R(T ) X of a bounded linear operator need


not be bounded. Hint. Use T in Prob. 5.

Solution

Consider again previous question. T is a bounded linear operator and its inverse
T 1 exits (from
R(T ) l ) since T is one-to-one. But T
1
is not bounded.
Since xn = (1, 2, ..., n, 0, 0, ...) l for all n,
1 1
T xn = yn = (1, , ..., , 0, 0, ...) R(T ).
2 n

Clearly ||x|| = n and ||y|| = 1. Given c R+ choosing n such that n>c
gives us
||T 1 yn || = ||xn || = n > c = c||yn ||.

2
Problem (9/101)

Let T : C[0, 1] C[0, 1] be defined by


Z t
T x(t) = y(t) = x( ) d.
0

Find R(T ) and T 1 : R(T ) C[0, 1]. Is T 1 linear and bounded?

Solution

Definition f : [0, 1] R (or C) is said to be differentiable on [0,1], if f is


differentiable on (0,1) in usual sense and differentiable at t = 0 and t = 1 by
considering only one-sided limits.

Definition f : [0, 1] R (or C) is said to be continuously differentiable on


[0,1], if f is differentiable on [0,1] and its derivative f 0 is continuous on [0,1].
1
(Notice that f (t) = t2 sin for t 6= 0 and 0 for t = 0, is differentiable on [0,1] but its derivative
t
is continuous only on (0,1].)

R(T ) = continuously differentiable functions y on [0,1] with y(0) = 0.

Let y be a such function, so y 0 C[0, 1] and we have


Z t
T y 0 (t) = y 0 ( ) d |{z}
= y(t) y(0) = y(t) y R(T ).
0
FTC

Conversely if y R(T ) then x C[0, 1] such that


Z t
T x(t) = y(t) = x( ) d.
0

Again by FTC we obtain y 0 = x, so y is continuously differentiable on [0,1] and


clearly y(0) = 0.

It is easy to show T is one-to-one, so T 1 : R(T ) C[0, 1] exists and lin-


ear (see theorem 2.6-10). Since for any y R(T ) T y 0 = y, then T 1 y = y 0 . But
T 1 is not bounded. Notice that yn (t) = tn is in R(T ) for any n, and we have
||yn || = 1, ||T 1 yn || = ||yn0 || = n. So given c R+ choosing n > c gives us

||T 1 yn || = ||yn0 || = n > c = c||yn ||.

Vous aimerez peut-être aussi