Sei sulla pagina 1di 15

Assignment I

REAL ANALYSIS II


April 13, 2016

Submitted to

Dr. Legesse Lemecha

Name
ID No
1.
Miliyon Tilahun . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . GSR/1401/08

Problems from Roydens Real Analysis Chapter 4


1. R(#9) Let E have measure zero. Show that if f is a bounded function on E, then f is measurable and
f = 0.
E
Solution. Let F be a measurable subset of R. Then f 1 (F ) is a subset of E, and since subsets of
measure zero sets have measure zero, f 1 (F ) has measure zero. In particular f 1 (F ) is measurable, so
f is measurable. Since f is bounded on E, for some constants and we have f . Moreover,
by monotonicity of Lebesgue integral we have
Z
m(E)
f m(E)
E

But m(E) = 0, implying


Z
f = 0.
E

2. (#10) Let f be a bounded


R
Rmeasurable function on a set of finite measure E. For a measurable subset
A of E, show that A f = E f A .
Solution. For a bounded measurable function f on a set of finite measure E we have
Z
Z
Z
f=
f+
f.
AB

where A and B are disjoint measurable subsets of E.


Since A E, E = A (E \ A). Clearly, A and E \ A are disjoint. Since A is a measurable subset of
E we have E \ A is measurable because the set E = A (E \ A) is measurable. Therefore
Z
Z
Z
Z
Z
f A =
f A +
f A =
f A =
f
E

E\A

This is because f A = f on A, and f A = 0 on E \ A.


3. (#13) Does the Bounded Convergence Theorem hold if m(E) < but we drop the assumption that
the sequence (|fn |) is uniformly bounded on E?
Solution. No it does not hold! Take fn = n2 [0, n1 ] which is not uniformly bounded. Then we see
that fn f 0. But


Z
Z
1
1
lim
fn = lim
n2 [0, n1 ] = lim n2 m 0,
= lim n2 =
n
n
n
n
n
n
This concludes

Z
fn = =
6 0=

lim

4. (#17) Let E be a set of measure zero and define f on E. Show that

R
E

f = 0.

Solution. By definition we have


Z

Z
f = sup
hf

h
E

where f is a nonnegative measurable function on measurable set E and h is a bounded measurable


R
function such that m{x : h(x)
R 6= 0} is finite. If h is such a function and m(E) = 0, we have E h = 0
by problem 1(#9). Hence E f = 0.
2

5. (#21) Let the function f be nonnegative and integrable over


R E and > 0. Show there is a simple
function on E that has finite support, 0 f on E and E |f | < R . If E is a closed, bounded
interval, show there is a step function h on E that has finite support and E |f h| < .
Solution. Let > 0. We know by the definition of the integral for non-negative measurable functions
that there exists a bounded, measurable function of finite support h f , such that
Z
Z
h>
f /2.
E

We also know that for any given bounded, measurable function h with finite support, there exists a
simple function h with the same support as h such that
Z
Z
>
h /2.
E

Therefore we see that there is a simple function with finite support such that f , and
Z
(f ) <
E

In the case that E is a closed and bounded interval, let =


that
Z
(f ) < /2.

Pn

k=1 ck Ek

be a simple function such

E
k
For each k, since m(Ek ) < , there exists a finite collection of disjoint open intervals {Iik }ni=1
such
that



Z
nk
nk
X
[


k
E

=
m
E

I
k
k
Ii
i < /2n.
k

i=1

i=1

Moreover, since E is a closed bounded interval, we may assume that each Iik is contained inside E, but
may no longer be open. Define a function by
=

nk
n X
X

ck Iik .

k=1 i=1

Note that this function is a step function even though some of the intervals may overlap, since any
finite collection of intervals that cover E define a partition of E by ordering their endpoints. It follows
that
Z
Z
nk
n
X
X


| |
ck Ek
Iik /2.
E

k=1

i=1

Therefore is a step function such that,


Z
| f | <
E

6. (#22) Let {fn } be a sequence of nonnegative measurable functions on R that converges pointwise on
R to f and f be integrable over R. Show that
Z
Z
Z
Z
if
f = lim
fn , then
f = lim
fn for any measurable set E.
R

R
R
R
Solution. Let E be a measurable set. Using the fact that R f = E f + R\E f , Fatous lemma on
R \ E implies
Z
Z
Z
Z
f lim inf
fn =
f lim sup
fn .
R\E

Therefore

R\E

Z
f lim sup
n

fn

(1)

fn .

(2)

However Fatous Lemma on E gives


Z

Z
f lim inf
n

By using the inequalities (1) and (2) we conclude that


Z
Z
f = lim
fn
n

7. (#25) Let {fn } be a sequence of nonnegative measurable functions on E that converges pointwise on
E to f . Suppose fn f on E for each n. Show that
Z
Z
fn =
f.
lim
n

R
Solution. Since fn f on E for each n,we have E fn E f , therefore
Z
Z
lim sup
fn
f.
n

However, since {fn } are nonnegative, Fatous Lemma implies


Z
Z
f lim inf
fn .
n

(3)

(4)

Thus, using (3) and (4) we arrive at the conclusion


Z
Z
fn =
f
lim
n

8. (#26) Show that the Monotone Convergence Theorem may not hold for decreasing sequences of functions.
Solution. Let f : R R be the zero function, and consider the sequence {fn } denoted by
fn (x) = [n,) (x)
Note that fn is a simple function and [n, ) is measurable. By definition
Z
fn = m([n, )) =
for every n Z+ . Now we will show that fn is monotone decreasing. Let n Z+ and x R. If
x < n + 1, then x
/ [n + 1, ) so
fn+1 (x) = 0 fn (x).
4

If x n + 1, then x [n + 1, ) and x [n, ), so


fn+1 (x) = 1 1 = fn (x).
This establishes that fn+1 fn for all n Z+ . In other words {fn } is a monotone decreasing sequence
of functions. Hence
Z
Z
Z
fn
f = 0 = 0 6= = lim = lim
n

showing that the Monotone Convergence Theorem need not hold for decreasing sequences.
R n+1
9. (#29) For a measurable function f on [1, ) which is bounded on bounded sets, define an =
f
n
P
for each natural number n. Is it true that f is integrable over [1, ) if and only if P
the series n=1 an

converges? Is it true that f is integrable over [1, ) if and only if the series
n=1 an converges
absolutely?
Solution. NO!
Take

(1)n
, n x n + 1, n 1
n
This function is not Lebesgue integrable though the series is convergent.
For the second one take,
(
1,
n x n + 21 for some n 1.
f (x) =
1, otherwise
f (x) =

Then an = 0 for every n, hence

n=1

|an | converges. But f is not Lebesgue integrable.

10. (#34) Let f be a nonnegative measurable function on R. Show that


Z n
Z
lim
f=
f.
n

Solution. Let fn = f [n,n] , then fn is nonnegative, monotone and fn f pointwise. By monotone


convergence theorem we have
Z n
Z
Z
lim
f = lim
fn =
f.
n

11. (#35) Let f be a real-valued function of two variables (x, y) that is defined on the square Q =
{(x, y)|0 x 1, 0 y 1} and is a measurable function of x for each fixed value of y. Suppose for
each fixed value of x, limy0 f (x, y) = f (x) and that for all y, we have |f (x, y)| g(x), where g is
integrable over [0, 1]. Show that
1

Z
lim

y0

Z
f (x, y)dx =

f (x)dx.
0

Also show that if the function f (x, y) is continuous in y for each x, then
Z
h(y) =

f (x, y)dx
0

is a continuous function of y.

Solution. It suffices to consider a sequence {yn } [0, 1] such that yn 0. Define fn (x) = f (x, yn ).
Then since fn f pointwise and |fn | g, Lebesgue dominated convergence gives
Z 1
Z 1
Z 1
f (x)dx.
fn (x)dx =
f (x, yn )dx = lim
lim
n

Since this is true for every such sequence {yn }, we have


Z 1
Z
f (x, y)dx =
lim
y0

f (x)dx.

Now fix y [0, 1] and suppose that {yn } [0, 1] such that yn y. Define fn (x) = f (x, yn ). Since
f (x, y) is continuous, fn (x) f (x, y) for all x [0, 1]. Once again, since |fn (x)| g(x) dominated
convergence gives
Z 1
Z 1
fn (x)dx =
f (x, y)dx = h(y).
lim h(yn ) = lim
n

Therefore h is continuous.
12. (#36) Let f be a real-valued function of two variables (x, y) that is defined on the square Q =
{(x, y)|0 x 1, 0 y 1} and is a measurable function on x for each fixed value of y. For each
(x, y) Q let the partial derivative f /y exist. Suppose there is a function g that is integrable over
[0, 1] and such that


f

(x, y) g(y) for all (x, y) Q.
y

Prove that
d
dy

Z

 Z
f (x, y)dx =

f
(x, y)dx for all y [0, 1].
y

Solution. Fix y0 [0, 1) and define for n 1,


Dn f (x) = n(f (x, y0 + 1/n) f (x, y0 )).
f
y (x, y0 ) pointwise. Also
|Dn f (x)| = | f
y (x, )| g(x). We

Since the derivative exists, Dn f (x)

by the Mean Value Theorem, there is

a [y0 , y0 + 1/n] such that


conclude by the Lebesgue Dominated
Convergence Theorem that
Z 1

Z 1
Z 1
Z 1
f
d
(x, y0 )dx = lim
Dn f (x)dx = lim Dn
f (x, y0 )dx =
f (x, y0 )dx .
n 0
n
dy 0
0 y
0
Since this is true for all y0 [0, 1) we have the result.
13. (#44) Let f be integrable over R and > 0. Establish the following three approximation properties.
R
(i) There is a simple function on R which has finite support and R |f | < .
R
(ii) There is a step function s on R which vanishes outside a closed, bounded interval and R |f s| < .
(Hint: Apply part (i) and Problem 18 of Chapter 3.)
R
(iii) There is a continuous function g on R which vanishes outside a bounded set and R |f g| < .
Solution. (i) If f is non-negative we may find an increasing sequence of non-negative simple functions
{n } with finite support such that n f pointwise. It follows by monotone convergence that we can
find a such that
Z
Z
|f | =
R

f < .
R

For
f we write Rf = f + f , and find 1 and 2 simple and of finite support such that
R general
+
|f 1 | < /2 and R |f 2 | < /2. Since f + and f have disjoint support, we see that 1
R
6

and 2 must also have disjoint support, therefore = 1 2 is also simple with finite support and it
follows that
Z
Z
Z
+
|f |
|f 1 | + |f 2 | < .
R

(ii) By part (i), since we can approximate by simple functions, by the triangle inequality it suffices to
show that the characteristic function E of a bounded measurable set E can be approximated by step
functions. Note S
that since E is measurable, we can find a disjoint collection of open intervals {Ik }
k=1

such that O = k=1 Ik , and


Sm(O \ E) < /2. Since O must have finite measure we can find an N
large enough such that m( k=N +1 Ik ) < /2. Therefore
s=

N
X

In

k=1

is a step function and


Z
|E s|
R

N Z
X
k=1

|EIk Ik | +

k=N +1

EIk



 [
[

N
Ik E
Ik \ E +m
m
k=N +1

k=1

 [


m(O \ E) + m
Ik < .
k=N +1

(iii) Using part (ii), once again we see by the triangle inequality that it suffices to show that any
characteristic function of a bounded interval [a,b] can be approximated by a continuous function. Let
g be the continuous function which is 1 on [a + /2, b /2] and linearly interpolated to 0 outside of
[a, b], then
Z
|[a,b] g| < m([a, a + /2) (b /2, b]) = .
R

14. (#46) Let f be integrable over (, ). Show that


Z
lim
f (x) cos nxdx = 0.
n

we know there exists a step function s, vanishing


Solution. Using the result from problem 13(#44),
R
outside of a closed bounded interval such that R |f s| < /2. Let s be the step function which we
write in canonical form as
K
X
s=
sk (ak ,bk ) ,
k=1

where {(ak , bk )} are a disjoint collection of bounded open intervals and {sk } are distinct. Note that it
doesnt matter that we dont define s at the end points of the intervals since they are a set of measure
0. We see that
Z
X
Z bk

K








s(x)
cos
nxdx
|s
|
cos
nxdx
k


ak

k=1

K
X
|sk |
k=1

| sin nbk sin nak |

2K max{|si |}
n

Therefore if n > N 4K max{si }/, we conclude


Z
Z
Z





|f (x) s(x)|dx +
f (x) cos nxdx



s(x) cos nxdx

< /2 + /2 = .

15. (#47) Let f be integrable over (, ).


(i) Show that for each t,

f (x)dx =

f (x + t)dx.

(ii) Let g be a bounded measurable function on R. Show that


Z
g(x) [f (x) f (x + t)]dx = 0.
lim
t0

(Hint: First show this, using uniform continuity of f on R, if f is continuous and vanishes outside a
bounded set. Then use the approximation property (iii) of Problem #44.)
P
Solution. (i) Note that this result is true for simple functions. Let = k=1 ci Ek be simple, then
since E (x + t) = Et (x), we haven by the translation invariance of Lebesgue
Z
Z
n
n
X
X
ck m(Ek ) =
(x)dx.
ck m(Ek t) =
(x + t)dx =
R

k=1

k=1

For the case of integrable f , we may restrict ourselves to f non-negative, since we may always write
any integrable f as f = f + f , where f + (x) = max{f (x), 0} and f (x) = max{f (x), 0}. For such
a non-negative f , we know that there is a sequence of simple functions with finite support {n }
n=1
such that n f with n converging to f pointwise and monotonically. In particular (+t) f (+t)
monotonically for every fixed t. It follows from the monotone convergence theorem that
Z
Z
Z
Z
n (x)dx =
f (x)dx.
n (x + t)dx = lim
f (x + t)dx = lim
n

(ii) Let M < be a constant such that |g| M . Since f is integrable, we know byRProblem 13 (#44)
that for every > 0 there exists h, continuous and of bounded support such that R |f h| < /2M .
It follows that h uniformly continuous and bounded by some constant N . Let {tn } be any sequence
tn 0 and define hn (x) = |h(x) h(x + tn )|. Then hn 0 pointwise and 0 hn (x) 2N . It follows
by the bounded convergence theorem that
Z
lim
|f (x) f (x + tn )| = 0.
n

Therefore
Z

Z
Z
Z


g(x)[f (x)f (x+tn )]dx M
|f
(x)h(x)|dx+M
|f
(x+t
)h(x+t
)|dx+M
|f (x)f (x+tn )|dx
n
n


R

By part i) and the way that h was chosen, the sum of the first two terms on the righthand side above
are bounded by . Taking the lim supn of both sides, we obtain
Z

Z



lim sup g(x) [f (x) f (x + t)]dx < + M lim
|f (x) f (x + tn )|dx = .
n

Since this is true for every > 0 and every {tn }, tn 0, we conclude
Z
lim g(x) [f (x) f (x + t)]dx.
t0

16. (#49) Let f be integrable over R. Show that the following four assertions are equivalent:
(i) f = 0 a.e on R.
R
(ii) R f g = 0 for every bounded measurable function g on R.
R
(iii) A f = 0 for every measurable set A.
R
(iv) O f = 0 for every open set O.
Solution. (1) (2)
If f = 0 a.e. on R, then m(E0 ) = 0, where
E0 = {x R : f (x) 6= 0}.
R
Since f is integrable over R, we have R |f |d < and g : R R being bounded and measurable so
|g(x)| < M for all x, then the function |f g| is bounded from above by M |f |, so we have
Z
Z
Z
|f g|d
M |f | = M
|f | < .
R

Thus f g is integrable and


Z

fg =
R

fg +

f g,
E0

RE0

But since f = 0 a.e on R and m(E0 ) = 0, we have


Z
Z
f g = 0, and

f g = 0,

E0

RE0

Thus,
Z
f g = 0,
R

as required.
(2) (3)
For a measurable set A, let g = A . Clearly, g is measurable and bounded. Therefore,
Z
Z
Z
0=
fg =
f A =
f = 0,
R

as required.
(3) (4)
Since every open set is measurable, then if (3) holds, we have
Z
f = 0 for every open set O,
O

as required.
(4) (1)
Let {Un }nN be a countable collection of open sets. Then,
\
G=
On ,
nN

T
is a G set with On = kn Uk , where each Uk is an open set. Thus, {On }nN is a countable descending
collection of open sets, i.e.
i N : Oi Oi+1 .
9

Now, for On being open, we have

Z
f = 0,
On

and by the continuity of integration,


Z
Z
f= T
G

Z
f = lim

nN

On

f = 0.
On

Since every measurable set E R is of the form G E0 , where G is a G subset of R and m(E0 ) = 0,
so that
Z
f = 0 for each measurable set E R.
E

Now, define
E + = {x R : f (x) 0} and E = {x R : f (x) 0},
then E + and E are measurable subsets of R and thus,
Z
Z
+
f =
f = 0,
E+

and

(f ) =

Z
f = 0,
E

where f + = max{f, 0} and f = max{f, 0}. Finally we use the fact (proposition 9 chapter 4) that
Z
f = 0 if and only if f = 0 a.e
E

so that f + and f vanishes almost everywhere in R and so does f as required.

Problems from Roydens Real Analysis Chapter 6


17. (#16) Let g be integrable over [a, b]. Define the antiderivative of g to be the function f defined on
[a, b] by
Z x
f (x) =
g
for all x [a, b].
a

Show that f is differentiable almost everywhere on (a, b).


Solution. Let us write g as a positive and negative part that is
g = g+ g
where
g + = max{f (x), 0}
g = max{f (x), 0}
Both g + and g are non negative. Then
Z

f (x) =
a

g+

Since f is the difference of two monotone increasing functions, then it is of bounded variation and f is
differentiable a.e on (a, b) by the corollary of Jordans theorem.

10

18. (#24) Show that for f defined in the last remark of this section, f 0 is not integrable over [0, 1].
Solution. The function in the remark is
(
f (x) =

x2 sin(1/x2 )
0

for 0 < x 1,
for x = 0.

To show f 0 is not integrable on [0, 1] it suffices to show that f is not of bounded variation on [0, 1].
Since every absolutely continuous function is of bounded variation it follows that f is not absolutely
continuous on [0, 1]. Hence by the Fundamental Theorem of Calculus for Lebesgue integrals we conclude
that f 0 is not integrable.
We show that f is not of bounded variation by showing that the total variation of f is bounded below
by a divergent series.
To make a suitable partition of [0, 1], note first that sin(x) = 1 for all x = 2 + 2k with k N, and
(1+2k)
with k N. Hence
sin(x) = 1 for all x = 3
2 + 2k with k N. So | sin(x)| = 1 for all x =
2
q
1
2
| sin( x2 )| = 1 for all x = (1+2k) for k N, and the value is negative if k is odd and positive if k is
even. So for each N N we can take a partition {xk }N
k=0 of [0, 1] so that
s
2
for all k = 0, . . . , N.
xk =
(1 + 2k)
Now f (xk ) and f (xk1 ) always have the opposite sign, so it follows that
T01 (f )

N
X

|f (xk ) f (xk1 )|

k=1

N 
X
k=1

N
X
k=1

N
X
k=1

2
2

(2k + 1) (2k 1)

2
(2k + 1)
N
2
2 X1
=
.
(3k)
3
k
k=1

This inequality holds for all N N, so we get that


T01 (f )

2 X1
= .
3
k
k=1

So f is not of bounded variation on [0, 1].


19. (#29) (a) Define
(
x2 cos(1/x2 )
f (x) =
0

if x 6= 0, x [1, 1]
if x = 0.

Is f of bounded variation on [1, 1]?


(b) Define
(
g(x) =

x2 cos(1/x)
0

Is g of bounded variation on [1, 1]?

11

if x 6= 0, x [1, 1]
if x = 0.

Solution. (a) We show that f is not of bounded variation by showing that the total variation of f is
bounded below by a divergent series.
Since f (x) = f (x) for all x [1, 1] it suffices to show that the total variation of f in [0, 1] is not
finite. To make a suitable partition of [0, 1], note first that cos(x) = 1 for all x = 2k with k N,
and cos(x) = 1 for all xq= (2k + 1) with k N. So | cos(x)| = 1 for all x = k with k N. Hence
| sin( x12 )| = 1 for all x =

1
k

for k N, and the value is negative if k is odd and positive if k is even.

So for each N N we can take a partition {xk }N


k=0 of [0, 1] so that
r
1
xk =
for all k = 0, . . . , N.
k
Now f (xk ) and f (xk1 ) always have the opposite sign, so it follows that
T01 (f )

N
X

|f (xk ) f (xk1 )|

k=1


N 
X
1
1

=
k (k 1)
k=1

N
N
X
1X1
1
=
.
k

k
k=1

k=1

This inequality holds for all N N, so we get that


T01 (f )

1X1
= .

k
k=1

So f is not of bounded variation on [1, 1].


(b) Note that for x 6= 0, we have
g 0 (x) = 2x cos(1/x) sin(1/x)
and x = 0,
h2 cos( h1 )
cos h
g(h) g(0)
= lim
= lim h cos(1/h) = lim
=0
h0
h0
h h
h0
h
h

g 0 (0) = lim

Thus, the derivative of g exists for all x [1, 1] and it is bounded by


|g 0 (x)| = |2x cos(1/x) sin(1/x)|
= 2|x|| cos(1/x)| + | sin(1/x)|
2+1=3
Since g(x) 0 as x 0 then g is continuous so the standard trick to find the Lipschitz constant of
g is to use the Mean Value Theorem(MVT). For any x < y in [1, 1], by MVT there exists a point
(x, y) such that
|g(x) g(y)| = |g 0 ()||x y|
Now since the derivative of g is bounded from above by 3, this implies
g(x) g(y) 3|x y|,

for all x, y [1, 1]

Therefore, g is Lipschitz. Since every Lipschitz function is of bounded variation g is of bounded


variation.

12

20. (#33) Let {fn } be a sequence of real-valued functions on [a, b] that converges pointwise on [a, b] to the
real-valued function f . Show that
T V (f ) lim inf T V (fn ).
Solution. Let P be any partition of [a, b], then since V (fn , P ) only depends on fn at a finite number
of points,
V (f, P ) = lim V (fn , P ).
n

Furthermore V (fn , P ) T V (fn ) and therefore


V (f, P ) = lim V (fn , P ) lim inf T V (fn ).
n

Taking the sup over all partitions P gives


T V (f ) lim inf T V (fn ).
n

21. (#35) For and positive numbers, define the function f on [0, 1] by
(
x sin(1/x ) for 0 < x 1
f (x) =
0
for x = 0.
Show that if > , then f is of bounded variation on [0, 1], by showing that f 0 is integrable over
[0, 1]. Then show that if , then f is not of bounded variation on [0, 1].
Solution. If > , then
f 0 (x) = x1 sin(1/x ) x1 cos(1/x ).
Since f is C 1 and bounded on (0, 1) we can use the fundamental theorem of calculus for Riemann
integrals to conclude that for any partition P
V (f, P ) =

n
X

Z
|f (xk ) f (xk1 )|

Z
T V (f )

|f 0 |

|f 0 |

k=0

and therefore

x1 + x1 < ,

since > 0 and > 0.


If , choose a partition Pn = {0, an , an1 , . . . , a1 }, where

an =
then we see that
V (f, Pn ) =

n
2

1/

/
n 
X
k
k=1

Note that this series diverges as n since / 1. Therefore


lim V (f, Pn ) T V (f ) = .

13

22. (#49) Let f be continuous on [a, b] and differentiable almost everywhere on (a, b). Show that
Z b
f 0 = f (b) f (a)
a

if and only if

Z
Z b
lim Diff1/n f = lim
a


Diff1/n f

Solution. Since f is continuous and differentiable a.e. on (a, b), we have that



Z b
Diff1/n f = lim Av1/n f (b) Av1/n f (a) = f (b) f (a),
lim
n

and

Z b
a

 Z
lim Diff1/n f =

f0

for almost all x. The result then follows immediately.


23. (#52) Let f and g be absolutely continuous on [a, b]. Show that
Z b
Z
f g 0 = f (b)g(b) f (a)g(a)
a

f 0 g.

Solution. Since g is absolutely continuous {Diff1/n g} is uniformly integrable and therefore {(Diff1/n g)
f } is uniformly integrable. Thus
Z b
Z b
f g 0 = lim
f Diff1/n g.
n

however by direct computation we see that


Z b
Z b
Z
f Diff1/n g = n
f (x)g(x + 1/n)dx
a

b1/n


f (x + 1/n)g(x + 1/n)dx

a1/n

(f (x) f (x + 1/n))g(x + 1/n)dx + n

=n

f (x + 1/n)g(x + 1/n)dx

b1/n

Z
n

f (x + 1/n)g(x + 1/n)dx
a1/n

Z
= Av1/n (f g)(b) Av1/n (f g)(a)

Diff1/n f g.

(5)

Since f and g are continuous,


Av1/n (f g)(b) f (b)g(b), Av(f g)(a) f (a)g(a),

(6)

and f is absolutely continuous so {Diff1/n f g} is uniformly integrable, therefore


Z

Z
Diff1/n f g

f 0 g.

(7)

Hence using (5),(6) and (7) we conclude that


Z b
Z
f g 0 = f (b)g(b) f (a)g(a)
a

14

f 0 g.

24. (#53) Let the function f be absolutely continuous on [a, b]. Show that f is Lipschitz on [a, b] if and
only if there is a c > 0 for which |f 0 | c a.e. on [a, b].
Solution. Suppose f is Lipschitz, then there exists a constant c > 0 such that
|f (y) f (x)| c|y x|,

x, y [a, b].

|f (y) f (x)|
c
|y x|

Now
f 0 (x) = lim

yx

f (y) f (x)
yx




f (y) f (x)
|f (y) f (x)|
lim
|f 0 (x)| = lim
lim c = c
yx
yx
y x yx
|y x|
Thus, |f 0 | c on [a, b].
Conversely, suppose |f 0 | c on [a, b]. Since f is absolutely continuous on [a, b], then it is differentiable
a.e on (a, b). By MVT there exists (a, b) such that
|f (x) f (y)| |f 0 ()||x y| c|x y|
Hence, f is Lipschitz.

15

Potrebbero piacerti anche