Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
0% found this document useful (0 votes)
100 views

Solved Problem Stuffs

This document contains Bob Lutz's solutions to various problems from Problem Set II in MATH 138 dated February 17, 2012. The problems addressed include proving that L^p spaces are metric linear spaces for 0<p<1, applying the generalized dominated convergence theorem, and using Egoroff's theorem for σ-finite measures. The document provides detailed proofs for each part of the problems addressed.

Uploaded by

Molina
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
100 views

Solved Problem Stuffs

This document contains Bob Lutz's solutions to various problems from Problem Set II in MATH 138 dated February 17, 2012. The problems addressed include proving that L^p spaces are metric linear spaces for 0<p<1, applying the generalized dominated convergence theorem, and using Egoroff's theorem for σ-finite measures. The document provides detailed proofs for each part of the problems addressed.

Uploaded by

Molina
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 6

Bob Lutz

MATH 138
Problem Set II
02/17/2012

Problems 1(a)(b),4(i)(ii)(iii),5,6(a)(b),9(a)(b)
Problem 1.
(a) Suppose that X is a vector space with an additive norm, prove that X is a metric
linear space, that is scalar multiplication is jointly continuous, provided:
i. for each scalar , the map f 7 f is continuous at f = 0;
ii. for each vector f , the map 7 f is continuous at = 0;
iii. the scalar multiplication map (, f ) 7 f is continuous at (0, 0).
(b) Prove that L p is a metric linear space for all 0 < p < 1.
(a) Suppose that (i) - (iii) hold. Fix > 0, and let 0 R and f 0 X . By (i) there
exists (0 ) > 0 such that if k f k < (0 ), then k0 f k < 3 . Likewise, by (ii) there exists
( f 0 ) > 0 such that if || < ( f 0 ), then k f 0 k < 3 . By (iii), there exists 0 > 0 such
that if || < 0 and k f k < 0 , then k f k < 3 . Thus if | 0 | < min{( f 0 ), 0 } and
k f f 0 k < min{(0 ), 0 }, then

k0 f 0 f k = k( 0 )( f 0 f ) + f 0 (0 ) + 0 ( f 0 f )k
k( 0 )( f f 0 )k + k f 0 ( 0 )k + k0 ( f f 0 )k
< 3 + 3 + 3 = ,
proving that (, f ) 7 f is continuous.

(b) Suppose that p (0, 1). Given f L p which is nonzero on a set of positive mea1/p
sure, if || < k f k p
, then k f k p = || p k f k p < , so (a)(i) holds. Given 6= 0, if
k f k p || p , then k f k p = || p k f k p < , so (a)(ii) holds as well. Finally, if ||, k f k p <
min{1/2 , 1/(2p) }, then k f k = || p k f k p < 1/2 1/2 = , so (a)(iii) holds. Hence L p is a
metric linear space by (a).

Problem 2. Suppose 1 p < . If f n , f L p and f n f a.e., then k f n f k p 0 iff
k f n k p k f k p . (Use the generalized dominated convergence
which
R theorem
R
R statesR that
1
if f n , gn , f , g L , f n f and gn g a.e., | f n | gn , and gn g, then f n f .)
Suppose that k f n f k p 0, so that k f n f k p |k f n k p k f k p | 0 as well, whence
k f n k p k f k p . Suppose on the other hand that k f n k p k f k p . Note that

| f n f | p [2 max{| f n |, | f |}] p 2 p (| f n | p + | f | p ) = gn ,
1

and that | f n f | p 0 and gn 2 p+1 | f | p = g a.e. Since | f n f | p , gn , 0, g L1 and


Z
Z


1/p
1/p
1/p
2 p+1 k f k p = g,
gn = 2 p k f n k p + k f k p
it follows from the generalized dominated convergence theorem that
whence k f f n k p 0 as well.

| f f n | p 0,


Problem 5.
(a) If is -finite and f n f a.e., there exist measurable E1 , E2 , . . . X such that
S
(( 1 Ej )c ) = 0 and f n f uniformly on each Ej .
(b) Let be counting measure on N. Then f n f in measure iff f n f uniformly.
(a) Since is -finite, there exists a cover { Xk }1 of X such that ( X j ) < for all j N.
Since subsets of null sets are null, for all j N we have f n f for a.e. x X j . By
Egoroffs theorem, there exists a measurable1 set Fk (`) Xk such that ( Fj (k )c ) k1
S
and f n f uniformly on Fj (k). Let Ej (`) = `k=1 Fj (k ). Observe that { Ej (`)c }
`=1 is nested
downward, f n f uniformly on Ej (`), and
( Ej (`)c ) =

`
\

!
Fj (k)c

k =1

Thus Ej =

`=1

1
( Fj (`)c ) .
`

Ej (`) is measurable, f n f uniformly on Ej , and


( Ecj ) =

!
Ej (`)c

= lim ( Ej (`)c ) = 0.
`

`=1

by continuity from above. By countable subadditivty, ((


which concludes the proof.

S
1

Ej )c ) = (

T c
1 E j ) = 0,


(b) If f n f in measure, then given > 0 we have ({ x : | f n ( x ) f ( x )| > }) 0


as n . Since assumes natural values, there exists N N such that if n N, then
({ x : | f n ( x ) f ( x )| } = 0. But then { x : | f n ( x ) f ( x )| } = , whence f n f
uniformly. Suppose, on the other hand, that f n f uniformly. Given > 0 there exists
N N such that if n N, then { x : | f n ( x ) f ( x )| } = . But then for all n N we
have ({ x : | f n ( x ) f ( x )| > }) = 0, whence f n f in measure.


1 Folland

omits this distinction [p. 62], but Rudin does not [p. 73].

Problem 6.
(a) Suppose | f n | g L1 and f n f in measure.
R
R
i.
f = lim f n .
ii. f n f in L1 .
(b) In Egoroffs theorem, the hypothesis ( X ) < can be replaced by | f n | g for
all n, where g L1 ().

R
R
R
R
(a) Note that (i) follows from (ii), since | f f n | = | ( f f n )| | f f n |. Thus
1
it suffices to prove (ii). Let > 0 be
that
R given. Since g L , there exists E X such

( E) < , supxE | g( x )| < , and Ec g < 4 . Let An = { x : | f n ( x ) f ( x )| 3(E) }, and


note that lim ( An ) = 0. Since | f n f | | f n | + | f | 2g, we have
lim

| fn f |

| f n f | + lim

Ec

Z
E An

| f n f | + lim

Z
E\ An

| fn f |

2g + lim ( An ) sup 2g( x ) + ( E) sup | f n ( x ) f ( x )|

Ec

xE

+ ( E)
2

3( E)

x E\ An

< .

Hence f n f in L1 , as desired.

(b) In the proof of Egoroffs theorem [Folland, p. 62], we needed the former hypothesis
to invoke continuity from above on the sequence { En (k )}. Thus it suffices to show that if
| f n | g for all n N, then ( E1 (k)) < .
First, recall that
n
o
[
En (k) =
x : | f m ( x ) f ( x )| k1 .
m=n

Since | f m f | | f m | + | f | 2g for all m N, we have


n
o n
o
x : | f m ( x ) f ( x )| k1 x : 2g( x ) k1 .
Thus
E1 (k ) =

n
[

x : | f m ( x ) f ( x )| k1

n
o
x : 2g( x ) k1 .

m =1

Since g L1 (), we have


n

x : 2g( x ) k

Hence
( E1 (k ))

n

o

2k

x : 2g( x ) k

as desired.

g < .
o

< ,


Problem 8.
(a) Suppose ( X ) < . If f and g are complex-valued measurable functions on X,
define
Z
| f g|
( f , g) =
d.
1 + | f g|
Then is a metric on the space of measurable functions if we identify functions
that are equal a.e., and f n f with respect to this metric iff f n f in measure.
Prove this by showing that

k f k0 =

|f|
d
1 + |f|

is an additive norm. Also show that a sequence is Cauchy in measure if and only
if it is Cauchy in the metric.
(b) Show that the space of measurable functions is a metric linear space under the
additive norm k f k0 .
(a) Since | f | = | f |, we have k f k0 = k f k0 . To prove the triangle inequality for k k0 ,
let f , g be measurable. Since integration is linear, it suffices to show that

k f + g k0 =

|f|
| g|
| f + g|

+
.
1 + | f + g|
1 + | f | 1 + | g|

After multiplying through and canceling terms, this simplifies to the inequality

| f + g| | f | + | g| + | f || g|(2 + | f + g|),
which holds due to the absolute value triangle inequality. Thus k k0 is an additive norm.
Let h be a measurable function, and notice that
( f , h) = k f hk0 k f gk0 + k g hk0 = ( f , g) + ( g, h),
proving the triangle inequality for . In addition, since the integrand in the definition of
is nonnegative, ( f , g) = 0 if and only if the numerator of the integrand is zero a.e.,
which occurs if and only if f = g a.e. Hence if we identify functions that are equal a.e.,
is nondegenerate and therefore a metric.
Suppose that f n f in measure. Given > 0, let E = { x : | f n ( x ) f ( x )| > }. Since
the map 7 1+ is increasing and bounded above by 1, we have

| f n ( x ) f ( x )|
1
1 + | f n ( x ) f ( x )|
and

| f n ( x ) f ( x )|

<
1 + | f n ( x ) f ( x )|
1+
4

for all x E,

for all x Ec .

Thus
( f n , f ) =

Z
E

| f n ( x ) f ( x )|
d +
1 + | f n ( x ) f ( x )|

Z
E

d +

Z
Ec

Z
Ec

| f n ( x ) f ( x )|
d
1 + | f n ( x ) f ( x )|

= ( E) + ( X ).
But f n f in measure, so ( E) 0 as n . Since ( X ) < , it follows that
( f n , f ) 0.
Suppose on the other hand that ( f n , f ) 0. Then we have


Z
Z

| f n ( x ) f ( x )|
d
d =
( E ),
( f n , f ) =
1 + | f n ( x ) f ( x )|
1+
E 1+
whence ( E) 0, proving that f n f in measure. Hence f n f with respect to if and
only if f n f in measure. If instead we set E = { x : | f n ( x ) f m ( x )| > } and replace f
with f m , the preceding argument yields the Cauchy result as well.

(b) Let > 0 and R be given. Suppose that || 1, and let f be measurable such
that k f k0 < . Then

k f k0 =

||| f |
d
1 + ||| f |

|f|
d = k f k0 < .
1 + |f|

(1)

Suppose that || > 1, and let f be such that k f k0 < ||1 . Then

k f k0 =

||| f |
d
1 + ||| f |

||| f |
d = ||k f k0 < ,
1 + |f|

(2)

proving hypothesis (i) of Problem 1(a). Next, let f be a given measurable function. If
k f k0 < , then let || 1, so that (1) follows. Otherwise, let || < k f k01 , so that (2)
follows. Thus 1(a)(ii) holds as well. Finally, if || < 1 and k f k0 < , then (1) follows, so
1(a)(iii) holds. Hence the space of measurable functions is a metric linear space.

Problem 10. On Rk find a sequence with f n 0 in L1 and a.e. but not almost uniformly.
Let An = ik=1 (n, n + n1/k ] for all n N, and consider the sequence { f n } defined by
f n = En . Note that for each n N, since An is a half-open rectangle, we have
k

k f n k1 = ( A n ) = m
i =1



1
n, n +
k
n



k
n

k

1
0,
n

as n ,

whence f n 0 in L1 . Moreover, given x = ( x1 , x2 , . . . , xk ), if n > max{ xi : i = 1, . . . , k },


then f n (x) = 0. Thus f n 0 pointwise, proving that f n 0 a.e. Note that if f n 0
5

uniformly on E, then there exists N N such that n N implies An E = . Since the


An are mutually disjoint, monotonicity and countable additivity imply
!

[
1
= .
( Ec ) m
An =
n
n= N
n= N
Hence f n does not converge almost uniformly.

You might also like