You are on page 1of 51

The Riemann-Stieltjes Integral

Riemann-Stieltjes integrals
b
7.1 Prove that ; d)  )Ÿb  " )Ÿa , directly from Definition 7.1.
a

Proof: Let f  1 on ¡a, b¢, then given any partition P  £a  x 0 , . . . , x n  b¤, then we
have
n
SŸP, 1, )   ! fŸt k   ) k , where t k  ¡x k"1 , x k ¢
k1
n
 ! )k
k1
 )Ÿb  " )Ÿa .
b
So, we know that ; d)  )Ÿb  " )Ÿa .
a

7.2 If f  RŸ)  on ¡a, b¢ and if ; ab fd)  0 for every f which is monotonic on ¡a, b¢,
prove that ) must be constant on ¡a, b¢.
Proof: Use integration by parts, and thus we have
b
; a )df  fŸb )Ÿb  " fŸa )Ÿa 
Given any point c  ¡a, b , we may choose a monotonic function f defined as follows.
0 if x t c
f
1 if x  c.
So, we have
b
; a )df  )Ÿc   )Ÿb .
So, we know that ) is constant on ¡a, b¢.
7.3 The following definition of a Riemann-Stieltjes integral is often used in the
literature: We say that f is integrable with respect to ) if there exists a real number A
having the property that for every /  0, there exists a -  0 such that for every partition
P of ¡a, b¢ with norm PPP  - and for every choice of t k in ¡x k"1 , x k ¢, we have
|SŸP, f, )  " A|  /.
b
(a) Show that if ; fd) exists according to this definition, then it is also exists according
a
to Definition 7.1 and the two integrals are equal.
Proof: Since refinement will decrease the norm, we know that if there exists a real
number A having the property that for every /  0, there exists a -  0 such that for every
partition P of ¡a, b¢ with norm PPP  - and for every choice of t k in ¡x k"1 , x k ¢, we have
|SŸP, f, )  " A|  /. Then choosing a P / with PP / P  -, then for P – P / ´ PPP  -. So,
we have
|SŸP, f, )  " A|  /.
b
That is, ; fd) exists according to this definition, then it is also exists according to
a
Definition 7.1 and the two integrals are equal.
(b) Let fŸx   )Ÿx   0 for a t x  c, fŸx   )Ÿx   1 for c  x t b,
b
fŸc   0, )Ÿc   1. Show that ; fd) exists according to Definition 7.1 but does not exist
a
by this second definition.
b b
Proof: Note that ; fd) exists and equals 0 according to Definition 7.1If ; fd) exists
a a
according to this definition, then given /  1, there exists a -  0 such that for every
partition P of ¡a, b¢ with norm PPP  - and for every choice of t k in ¡x k"1 , x k ¢, we have
|SŸP, f, ) |  1. We may choose a partition P  £a  x 0 , . . . , x n  b¤ with PPP  - and
c  Ÿx j , x j1  , where j  0, . . . , n " 1. Then
SŸP, f, )   fŸx ¡)Ÿx j1   " )Ÿx j  ¢  1, where x  Ÿc, x j1  
which contradicts to |SŸP, f, ) |  1.
7.4 If f  R according to Definition 7.1, prove that ; ab fŸx dx also exists according to
definition of Exercise 7.3. [Contrast with Exercise 7.3 (b).]
b
Hint: Let I  ; fŸx dx, M  sup£|fŸx | : x  ¡a, b¢. ¤ Given /  0, choose P / so that
a
UŸP / , f   I  //2 (notation of section 7.11). Let N be the number of subdivision points in
P / and let -  //Ÿ2MN . If PPP  -, write
UŸP, f   ! M k Ÿf  xk  S1  S2,
where S 1 is the sum of terms arising from those subintervals of P containing no points of
P / and S 2 is the sum of remaining terms. Then
S 1 t UŸP / , f   I  //2 and S 2 t NMPPP  NM-  //2,
and hence UŸP / , f   I  /. Similarly,
LŸP, f   I " / if PPP  - U for some - U .
Hence |SŸP, f  " I|  / if PPP  minŸ-, - U  .
Proof: The hint has proved it.
Remark: There are some exercises related with Riemann integrals, we write thme as
references.
b
(1) Suppose that f u 0 and f is continuous on ¡a, b¢, and ; fŸx dx  0. Prove that
a
fŸx   0 on ¡a, b¢.
Proof: Assume that there is a point c  ¡a, b¢ such that fŸc   0. Then by continuity of
fŸc 
f, we know that given /  2  0, there is a -  0 such that as |x " c|  -, x  ¡a, b¢, we
have
fŸc 
|fŸx  " fŸc | 
2
which implies that
fŸc 
 fŸx  if x  Ÿc " -, c  -  9 ¡a, b¢ : I
2
So, we have
fŸc  b
0 |I| t ; fŸx dx t ; fŸx dx  0, where 0  |I|, the length of I
2 I a
which is absurb. Hence, we obtain that fŸx   0 on ¡a, b¢.
(2) Let f be a continuous function defined on ¡a, b¢. Suppose that for every continuous
function g defined on ¡a, b¢ which satisfies that
b
; a gŸx dx  0,
we always have
b
; a fŸx gŸx dx  0.
Show that f is a constant function on ¡a, b¢.
b
Proof: Let ; fŸx dx  I, and define gŸx   fŸx  " I
b"a
, then we have
a
b
; a gŸx dx  0,
which implies that, by hypothesis,
b
; a fŸx gŸx dx  0
which implies that
b
; a ŸfŸx  " c gŸx dx  0 for any real c.
So, we have
b
; a ŸgŸx   2 dx  0 if letting c 
b"a
I
I
which implies that gŸx   0 forall x  ¡a, b¢ by (1). That is, fŸx   b"a
on ¡a, b¢.
(3) Define
0 if x  ¡0, 1¢ " Q
hŸx   1
n if x is the rational number m/n (in lowest terms)
1 if x  0.
Then h  RŸ¡0, 1¢ .
Proof: Note that we have shown that h is continuous only at irrational numbers on
¡0, 1¢ " Q. We use it to show that h is Riemann integrable, i.e., h  RŸ¡0, 1¢ . Consider the
upper sum UŸP, f  as follows.
Given /  0, there exists finitely many points x such that fŸx  u //2. Consider a
partition P /  £x 0  a, . . . , x n  b¤ so that its subintervals I j  ¡x j"1 , x j ¢ for some j
containing those points and !|I j |  //2. So, we have
n
UŸP, f   ! Mk xk
k1

 ! !
1 2
 //2  //2
/
where ! 1  ! 1 M j I j , and ! 2 , is the sum of others.
So, we have shown that f satisfies the Riemann condition with respect to )Ÿx   x.
Note: (1) The reader can show this by Theorem 7.48 (Lebesgue’’s Criterion for
Riemann Integrability). Also, compare Exercise 7.32 and Exercise 4.16 with this.
(2) In Theorem 7.19, if we can make sure that there is a partition P / such that
UŸP / , f, )  " LŸP / , f, )   /,
then we automatically have, for any finer PŸ– P /  ,
UŸP, f, )  " LŸP, f, )   /
since the refinement makes U increase and L decrease.
(4) Assume that the function fŸx  is differentiable on ¡a, b¢, but not a constant and that
fŸa   fŸb   0. Then there exists at least one point 8 on Ÿa, b  for which
b
|f U Ÿ8 |  4 ; a fŸx dx.
Ÿb " a  2
Proof: Consider sup x¡a,b¢ |f U Ÿx | : M as follows.
(i) If M  ., then it is clear.
(ii) We may assume that M  ..
Let x  ¡a, ab
2
¢, then
fŸx   fŸx  " fŸa   f U Ÿy Ÿx " a  t MŸx " a , where y  Ÿa, x . *
ʳʳʳʳ
and let x  ¡ ab
2
, b¢, then
fŸx   fŸx  " fŸb   f U Ÿz Ÿx " b  t MŸb " x , where z  Ÿx, b . **
ʳʳʳʳ
So, by (*) and (**), we know that
b ab b
; a fŸx dx  ; a 2
fŸx dx  ; ab fŸx dx
2
ab b
t M; Ÿx " a dx  M ; ab Ÿb " x dx
2

a 2
2
 M a"b
2
which implies that
b
Mu 4 ; fŸx dx.
Ÿb " a  2 a
Note that by (*) and (**), the equality does NOT hold since if it was, then we had
f U Ÿx   M on ¡a, b¢ which implies that f is a constant function. So, we have
b
M 4 ; fŸx dx.
Ÿb " a  2 a
By definition of supremum, we know that there exists at least one point 8 on Ÿa, b  for
which
b
|f U Ÿ8 |  4 ; a fŸx dx.
Ÿb " a  2
(5) Gronwall Lemma: Let f and g be continuous non-negative function defined on
¡a, b¢, and c u 0. If
x
fŸx  t c  ; gŸt fŸt dt for all x  ¡a, b¢,
a
then
x
; a gŸt dt
fŸx  t ce .
In particular, as c  0, we have f  0 on ¡a, b¢.
Proof: Let c  0 and define
x
FŸx   c  ; gŸt fŸt dt,
a
then we have
(i). FŸa   c  0.
(ii). F U Ÿx   gŸx fŸx  u 0 ´ F is increasing on ¡a, b¢ by Mean Value Theorem
(iii). FŸx  u fŸx  on ¡a, b¢ ´ F U Ÿx  t gŸx FŸx  by (ii).
So, from (iii), we know that
x x
; a gŸt dt ; a gŸt dt
FŸx  t FŸa e  ce by (i).
For c  0, we choose c n  1/n v 0, then by preceding result,
x
; gŸt dt
fŸx  t 1
ne a v 0 as n v ..
So, we have proved all.
(6) Define
x1
fŸx   ;x sinŸt 2  dt.

(a) Prove that |fŸx |  1/x if x  0.


Proof: Let x  0, then we have, by change of variable(u  t 2 ), and integration by
parts,
Ÿx1  2
fŸx   1
2
;x 2
sin u du
u
Ÿx1  2 dŸcos u 
 "1
2
;x 2 u
Ÿx1  2 Ÿx1  2
 "1 cos u ; cos u du
2 u x2 x2 2u 3/2

cosŸx 2   cos Ÿx  1  2 Ÿx1  2


 " "; cos u du
2x 2Ÿx  1  x2 4u 3/2
which implies that,
cosŸx 2   cos Ÿx  1  2 Ÿx1  2
|fŸx | t
2x

2Ÿx  1 
 ;x 2
cos u du
4u 3/2
Ÿx1  2
 1  1  1 ; du
2x 2Ÿx  1  4 x2 u 3/2
 1  1 " 1  1
2x 2Ÿx  1  2Ÿx  1  2x
 1/x.
Note: There is another proof by Second Mean Value Theorem to show above as
follows. Since
Ÿx1  2
fŸx   1
2
;x 2
sin u du by (a),
u
we know that, by Second Mean Value Theorem,
y Ÿx1  2
fŸx   1
2
1
x ; x2 sin udu  x 1 1 ;y sin udu

 1 1 ¡cosŸx 2   " cos y¢  1


x cos y " cos Ÿx  1  2
2 x1
 1 " 1x  1 cosŸy   1x cosŸx 2   " 1 cos Ÿx  1  2
2 x1 x1
which implies that
cosŸx 2   cos Ÿx  1  2
|fŸx | t 1 1
"x  1 |cos y|  x 
2 x1 x1

cosŸx 2   cos Ÿx  1  2
t 1 1 " 1
x  x 
2 x1 x1

 1 1 " 1
x  1x  1
2 x1 x1
since no x makes |cosŸx  |  cos Ÿx  1  2
2 1
 1/x.
(b) Prove that 2xfŸx   cosŸx 2   " cos Ÿx  1  2  rŸx , where |rŸx |  c/x and c is a
constant.
Proof: By (a), we have
cosŸx 2   cos Ÿx  1  2 Ÿx1  2
fŸx   " "; cos u du
2x 2Ÿx  1  x 2 4u 3/2
which implies that
Ÿx1  2
2xfŸx   cosŸx 2   " x cos Ÿx  1  2  x ; cos u du
x1 x2 2u 3/2
Ÿx1  2
 cosŸx 2   " cos Ÿx  1  2
 1 cos Ÿx  1  2  x ; cos u du
x1 x2 2u 3/2
where
Ÿx1  2
rŸx   1 cos Ÿx  1  2  x
x1 2
;x 2
cos u du
u 3/2
which implies that
Ÿx1  2
|cos u|
|rŸx | t 1  x
x1 2
;x 2 u 3/2
du
Ÿx1  2
 1  x ; u "3/2 du
x1 2 x2
 1  1
x1 x1
2
 x.

Note: Of course, we can use the note in (a) to show it. We write it as follows.
Proof: Since
fŸx   1 " 1x  1 cosŸy   1x cosŸx 2   " 1 cos Ÿx  1  2
2 x1 x1
which implies that
2xfŸx   x " 1 cosŸy   cosŸx 2   " x cos Ÿx  1  2
x1 x1
2
 cosŸx   " cos Ÿx  1  
2 1 cos Ÿx  1  2  x " 1 cosŸy 
x1 x1
where
rŸx   1 cos Ÿx  1  2  x " 1 cosŸy 
x1 x1
which implies that
|rŸx | t 1 1" x
x1 x1
 2
x1
 2/x.
(c) Find the upper and lower limits of xfŸx , as x v ..
Proof: Claim that lim sup xv. cosŸx 2   " cos Ÿx  1  2  2 as follows. Taking
x  n 2= , where n  Z, then
cosŸx 2   " cos Ÿx  1  2  " cos n 8=  1 . 1
ʳʳʳʳ
If we can show that n 8= is dense in ¡0, 2=¢ modulus 2=. It is equivalent to show that
n =2 is dense in ¡0, 1¢ modulus 1. So, by lemma £ar : a  Z¤, where r  Q c is dense
in ¡0, 1¢ modulus 1, we have proved the claim. In other words, we have proved the claim.
Note: We use the lemma as follows. £ar  b : a  Z, b  Z¤, where r  Q c is dense in
R. It is equivalent to £ar : a  Z¤, where r  Q c is dense in ¡0, 1¢ modulus 1.
Proof: Say £ar  b : a  Z, b  Z¤  S, and since r  Q c , then by Exercise 1.16,
there are infinitely many rational numbers h/k with k  0 such that |kr " h|  1k . Consider
Ÿx " -, x  -  : I, where -  0, and thus choosing k 0 large enough so that 1/k 0  -.
Define L  |k 0 r " h 0 |, then we have sL  I for some s  Z. So,
sL  Ÿo ¡Ÿsk 0  r " Ÿsh 0  ¢  S. That is, we have proved that S is dense in R.
.
(d) Does ; sinŸt 2  dt converge?
0

Proof: Yes,
xU xU
; x sin 2 tdt  1
2
;x sin u du by the process of (a)
u
y xU
 1
2
1
x ; x sin udu  x1U ; y sin udu by Second Mean Value Theorem

 1 2  2
x
2 xU
 2x
which implies that the integral exists.
Note: (i) We can show it without Second Mean Value Theorem by the method of (a).
However Second Mean Value Theorem is more powerful for this exercise.
(ii) Here is the famous Integral named Dirichlet Integral used widely in the STUDY
of Fourier Series. We write it as follows. Show that the Dirichlet Integral
.
; 0 sinx x dx
converges but not absolutely converges. In other words, the Dirichlet Integral converges
conditionally.
Proof: Consider
xU y xU
;x sin x dx  1
x x ; x sin xdx  x1U ; y sin xdx by Second Mean Value Theorem;
we have
xU
;x
sin x dx t 2  2  4 .
x x xU x
So, we know that Dirichlet Integral converges.
Define I n  ¡ =4  2n=, =2  2n=¢, then
.
; 0 sinx x dx u ;In sinx x dx
2

u ;I =
2
 2n=
dx
n 4
2
. Ÿ =4  
u ! =
2
 2n=
v ..
n0 4

So, we know that Dirichlet Integral does NOT converges absolutely.


(7) Deal similarity with
x1
fŸx   ;x sinŸe t  dt.
Show that
e x |fŸx |  2
and that
e x fŸx   cosŸe x   " e "1 cosŸe x1    rŸx ,
where |rŸx |  minŸ1, Ce "x  , for all x and
Proof: Since
x1
fŸx   ;x sinŸe t  dt
e x1
 ;e x
sin u du by Change of Variable (let u  e t )
u
e x1

e x " e x1 " ; e x


 cos ex cos e x1 cos u du by Integration by parts *
ʳʳʳʳ
u2
which implies that
e x1
|fŸx |  cos e x  cos e x1  ; du since cos u is not constant 1
ex e x1
e x u2
t e1x  x1 1  1 1" 1
ex e
e
which implies that
e x |fŸx |  2.
In addition, by (*), we have
e x fŸx   cosŸe x   " e "1 cosŸe x1    rŸx ,
where
e x1
rŸx   "e x ;e x
cos u du
u2
which implies that
|rŸx |  1 " e "1  1 for all x **
ʳʳʳʳ
or which implies that, by Integration by parts,
e x1
|rŸx |  e x ;e x
cos u du
u2
e x1
 ex sin e x1 " sin e x  2 ; sin u du
e 2Ÿx1  ex ex u3
e x1
 ex 1  12x  2 ; du since sin u is not constant 1
e 2Ÿx1  e ex u3
 2e "x for all x. ***
ʳʳʳʳ
By (**) and (***), we have proved that |rŸx |  minŸ1, Ce "x   for all x, where C  2.
Note: We give another proof on (7) by Second Mean Value Theorem as follows.
Proof: Since
e x1
fŸx   ;e x
sin u du
u
y e x1
 e1x ; sin udu  x1 1 ; sin udu by Second Mean Value Theorem
ex e y

 e1x Ÿcos e x " cos y   x1 1 Ÿcos y " cos e x1   *


ʳʳʳʳ
e
which implies that
e x |fŸx |  |cos e x " cos y  e "1 Ÿcos y " cos e x1  |
 |Ÿcos e x " e "1 cos e x1    cos yŸ1 " e "1  |
t |cos e x " e "1 cos e x1 |  Ÿ1 " e "1  
 Ÿ1  e "1    Ÿ1 " e "1   since no x makes |cos e x |  |cos e x1 |  1.
 2.
In addition, by (*), we know that
e x fŸx   cos e x " e "1 cos e x1  rŸx 
where
rŸx   Ÿe "1 cos y " cos y 
which implies that
|rŸx | t 1 " e "1  1 for all x. **
ʳʳʳʳ
In addition, from the proof of the process in (7), we know that
e x1
|rŸx |  ex ;e x
cos u du
u2
y e x1
 ex 1 ; cos udu  1 ; cos udu
e 2x e x e 2Ÿx1  y
 e "x |sin y " sin e x  e "2 Ÿsin e x1 " sin y |
 e "x |sin yŸ1 " e "2    Ÿe "2 sin e x1 " sin e x  |
t e "x Ÿ1 " e "2    e "x |e "2 sin e x1 |  e "x |sin e x |
 e "x Ÿ1 " e "2    e "x Ÿ1  e "2   since no x makes |sin e x1 |  |sin e x |  1
 2e "x for all x. ***
ʳʳʳʳ
So, by (**) and (***), we have proved that |rŸx |  minŸ1, Ce "x  , where C  2.
(8) Suppose that f is real, continuously differentiable function on ¡a, b¢,
fŸa   fŸb   0, and
b
; a f 2 Ÿx dx  1.
Prove that
b
; a xfŸx f U Ÿx dx  "1
2
and that
b b
; a ¡f U Ÿx ¢ 2 dx  ; a x 2 f 2 Ÿx dx  1.
4
Proof: Consider
b b
; a xfŸx f U Ÿx dx  ; a xfŸx dfŸx 
b
 xf 2 Ÿx | ba " ; fŸx dŸxfŸx  
a
b b
 " ; f 2 Ÿx dx  ; xfŸx f U Ÿx dx since fŸa   fŸb   0,
a a
so we have
b
"1 .
2
; a xfŸx f U Ÿx dx 
In addition, by Cauchy-Schwarz Inequality, we know that
b b b 2
;a ¡f U Ÿx ¢ 2 dx  ; x 2 f 2 Ÿx dx u
a
 1.
4
;a xfŸx f U Ÿx dx

Note that the equality does NOT hold since if it was, then we have f U Ÿx   kxfŸx . It
implies that
"kx 2
¡f U Ÿx  " kxfŸx ¢e 2 0
which implies that
"kx 2 U
fe 2 0
which implies that
"kx 2
fŸx   Ce 2 , a constant
which implies that
C  0 since fŸa   0.
That is, fŸx   0 on ¡a, b¢ which is absurb.
7.5 Let £a n ¤ be a sequence of real numbers. For x u 0, define
¡x¢
AŸx   ! an  ! an,
ntx n1
where ¡x¢ is the largest integer in x and empty sums are interpreted as zero. Let f have a
continuous derivative in the interval 1 t x t a. Use Stieltjes integrals to derive the
following formula:
a
! a n fŸn   " ; 1 AŸx f U Ÿx dx  AŸa fŸa .
nta

Proof: Since
a a
; 1 AŸx f U Ÿx dx  ; 1 AŸx dfŸx  since f has a continous derivative on ¡1, a¢
a
 " ; fŸx dAŸx   AŸa fŸa  " AŸ1 fŸ1  by integration by parts
1
¡a¢
a
 " ! a n fŸn   AŸa fŸa  by ; 1 fŸx dAŸx   ! a n fŸn  and AŸ1   a 1 ,
nta n2
we know that
a
! a n fŸn   " ; 1 AŸx f U Ÿx dx  AŸa fŸa .
nta

7.6 Use Euler’’s summation formula, integration by parts in a Stieltjes integral, to


derive the following identities:
n n
(a) ! k1 k1s  n s"1
1
 s ; x¡x¢
s1 dx if s p 1.
1

Proof:
n n
! k1s  ; 1 x "s d¡x¢  1
k1
n
 " ; ¡x¢dx "s  n "s ¡n¢ " 1 "s ¡1¢  1
1
n
¡x¢
 s; s1
dx  n 1"s
1 x
n ¡x¢
 1s"1  s ; s1 dx if s p 1.
n 1 x
n n x"¡x¢
(b) ! k1 1
k
 log n " ; x2
 1.
1

Proof:
n n
! 1k  ;1 1 d¡x¢  1
x
k1
n
 " ; ¡x¢dx "1  n "1 ¡n¢ " 1 "1 ¡1¢  1
1
n n n ¡x¢
 ; 1 x "1 dx " ; 1 x "1 dx  ; 1 x2
dx  1
n x " ¡x¢
 log n " ;  1.
1 x2
7.7 Assume that f U is continuous on ¡1, 2n¢ and use Euler’’s summation formula or
integration by parts to prove that
2n 2n
!Ÿ"1  k fŸk   ; 1 f U Ÿx Ÿ¡x¢ " 2¡x/2¢ dx.
k1

Proof:
2n 2n n
!Ÿ"1  fŸk   " ! fŸk   2 ! fŸ2k 
k

k1 k1 k1


2n 2n
" ;1 fŸx d¡x¢  fŸ1   2 ;1 fŸx d¡x/2¢
2n 2n
 " " ; ¡x¢dfŸx   ¡2n¢fŸ2n   2 " ; ¡x/2¢dfŸx   fŸ2n ¡2n/2¢ " fŸ1 ¡1/2¢
1 1

since fU is continuous on ¡1, 2n¢


2n 2n
 ;1 f U Ÿx ¡x¢dx " 2nfŸ2n  " ; f U Ÿx ¡x/2¢dx  2nfŸ2n 
1
2n
 ;1 f U Ÿx Ÿ¡x¢ " 2¡x/2¢ dx.

7.8x Let C 1  x " ¡x¢ " 1


if x p integer, and let C 1  0 if x  integer. Also, let
2
C 2  ; C 1 Ÿt dt. If f UU is continuous on ¡1, n¢ prove that Euler’’s summation formula implies
0
that
n n n fŸ1   fŸn 
! fŸk   ; 1 fŸx dx " ; 1 C 2 Ÿx f UU Ÿx dx  2
.
k1
Proof: Using Theorem 7.13, then we have
n n n fŸ1   fŸn 
! fŸk   ; 1 fŸx dx  ; 1 f U Ÿx C 1 Ÿx dx  2
k1
n n fŸ1   fŸn 
 ; 1 fŸx dx  ; 1 f U Ÿx dC 2 Ÿx   2
n n fŸ1   fŸn 
 ; 1 fŸx dx  " ; C 2 Ÿx df U Ÿx   f U Ÿn C 2 Ÿn  " f U Ÿ1 C 2 Ÿ1  
1 2
n n fŸ1   fŸn 
 ; 1 fŸx dx " ; 1 C 2 Ÿx df U Ÿx   2
n n fŸ1   fŸn 
 ; fŸx dx " ; C 2 Ÿx f UU Ÿx dx  since f UU is continuous on ¡1, n¢.
1 1 2
7.9 Take fŸx   log x in Exercise 7.8 and prove that
n C Ÿt 
log n!  n  1 log n " n  1  ; 2 2 dt.
2 1 t
Proof: Let fŸx   log x, then by Exercise 7.8, it is clear. So, we omit the proof.
Remark: By Euler’’s summation formula, we can show that
n n
! log k  ; 1 log xdx  ; 1 x " ¡x¢ " 12 dxx  log2 n . *
ʳʳʳʳ
1ktn
Since
x " ¡x¢ " 1 t 1/2
2
and
a1
;a x " ¡x¢ " 1 dx  0 for all real a,
2
**
ʳʳʳʳ
we thus have the convergence of the improper integral
.
; 1 x " ¡x¢ " 12 dx by Second Mean Value Theorem.
So, by (*), we have
log n!  n  1 log n " n  C  + n
2
where
.
C  1  ; x " ¡x¢ " 1 dx ,
1 2 x
and
.
+n  " ; x " ¡x¢ " 1 dx .
x
n 2
So,
lim n!  e C : C 1 . ***
ʳʳʳʳ
nv. e "n n n1/2
Now, using Wallis formula, we have
2  2  4  4   Ÿ2n Ÿ2n 
lim
nv. 1  3  3  5  5   Ÿ2n " 1 Ÿ2n  1 
 =/2

which implies that


Ÿ2 n n!  4
Ÿ1  oŸ1    =/2
¡Ÿ2n !¢ 2 Ÿ2n  1 
which implies that, by (***),
4
C 41 Ÿ2 n n n1/2 e "n  
Ÿ1  oŸ1    =/2
C 21 Ÿ2n  2n1/2 e "2n Ÿ2n  1 
which implies that
C 21 n
Ÿ1  oŸ1    =/2.
2Ÿ2n  1 
.
Let n v ., we have C 1  2= , and ; Ÿx " ¡x¢ " 1
2
 dx  1
2
log 2= " 1.
1

Note: In (***), the formula is called Stirling formula. The reader should be noted that
Wallis formula is equivalent to Stirling formula.
7.10 If x u 1, let =Ÿx  denote the number of primes p t x, that is,
=Ÿx   ! 1,
ptx

where the sum is extended over all primes p t x. The prime number theorem states that
log x
lim
xv.
=Ÿx  x  1.
This is usually proved by studying a related function G given by
GŸx   ! log p,
ptx

where again the sum is extended over all primes p t x. Both function = and G are step
functions with jumps at the primes. This exercise shows how the Riemann-Stieltjes integral
can be used to relate these two functions.
(a) If x u 2, prove that =Ÿx  and GŸx  can be expressed as the following
Riemann-Stieltjes integrals:
x x
GŸx   ; log td=Ÿt , =Ÿx   ; 1 dGŸt .
3/2 3/2 log t
Note. The lower limit can be replaced by any number in the open interval Ÿ1, 2 .
Proof: Since GŸx   ! ptx log p, we know that by Theorem 7.9,
x
GŸx   ; 3/2 log td=Ÿt ,
and =Ÿx   ! ptx 1, we know that by Theorem 7.9,
x
=Ÿx   ; 3/2 1 dGŸt .
log t
(b) If x u 2, use integration by parts to show that
x =Ÿt 
GŸx   =Ÿx  log x " ; t dt,
2
GŸx  x GŸt 
;
=Ÿx   dt.
log x 2
2 t log t

These equations can be used to prove that the prime number theorem is equivalent to
GŸx 
the relation lim xv. x  1.
Proof: Use integration by parts, we know that
x x =Ÿt 
GŸx   ; log td=Ÿt   " ; t dt  log x=Ÿx  " logŸ3/2 =Ÿ3/2 
3/2 3/2
x =Ÿt 
 "; t dt  log x=Ÿx  since =Ÿ3/2   0
3/2
x =Ÿt  2 =Ÿt 
 =Ÿx  log x " ; t dt since ; 3/2 t dt  0 by =Ÿx   0 on ¡0, 2 ,
2
and
x x GŸt  GŸx  GŸ3/2 
=Ÿx   ; 3/2 1 dGŸt  
log t
; 3/2 2
t log t
dt 
log x
"
logŸ3/2 
x GŸt  GŸx 
 ; 3/2 2
t log t
dt 
log x
since GŸ3/2   0

GŸx  x GŸt  2 GŸt 


 ; dt since ; dt  0 by GŸx   0 on ¡0, 2 .
log x 2
2 t log t
2
3/2 t log t

7.11 If ) { on ¡a, b¢, prove that


b c b
(a) ; fd)  ; fd)  ; fd), (a  c  b)
a a c

Proof: Given /  0, there is a partition P such that


UŸP, f, )   IŸa, b   /. 1
ʳʳʳʳ
Let P U  £c¤ : P  P 1 : P 2 , where P 1  £a  x 0 , . . . , x n 1  c¤ and
P 2  £x n 1  c, . . . , x n 2  b¤then we have
IŸa, c   IŸc, b  t UŸP 1 , f, )   UŸP 2 , f, )   UŸP U , f, )  t UŸP, f, ) . 2
ʳʳʳʳ
So, by (1) and (2), we have
IŸa, c   IŸc, b  t IŸa, b  *
ʳʳʳʳ
since / is arbitrary.
On the other hand, given /  0, there is a partition P 1 and P 2 such that
UŸP 1 , f, )   UŸP 2 , f, )  t IŸa, c   IŸc, b   /
which implies that, let P  P 1 : P 2
UŸP, f, )   UŸP 1 , f, )   UŸP 2 , f, )  t IŸa, c   IŸc, b   /. 3
ʳʳʳʳ
Also,
IŸa, b  t UŸP, f, ) . 4
ʳʳʳʳ
By (3) and (4), we have
IŸa, b  t IŸa, c   IŸc, b  **
ʳʳʳʳ
since / is arbitrary.
So, by (*) and (**), we have proved it.
b b b
(b) ; Ÿf  g d) t ; fd)  ; gd).
a a a

Proof: In any compact interval J, we have


supŸf  g  t sup f  sup g. 1
ʳʳʳʳ
xJ xJ xJ
So, given /  0, there is a partition P f and P g such that
n1
b
! M k Ÿf  )k t ; a fd)  //2 2
ʳʳʳʳ
k1
and
n2
b
! M k Ÿg  )k t ; a gd)  //2. 3
ʳʳʳʳ
k1
So, consider P  P f : P g , then we have, by (1),
UŸP, f  g, )  t UŸP, f, )   UŸP, g, ) 
along with
n1 n2
UŸP, f, )  t ! M k Ÿf  ) k and UŸP, g, )  t ! M k Ÿg  )k
k1 k1
which implies that, by (2) and (3),
b b
UŸP, f  g, )  t ; a fd)  ; a gd)  /
which implies that
b b b
;a Ÿf  g d) t ;a fd)  ; gd)
a
since / is arbitrary.
b b b
(c) ; Ÿf  g d) u ; fd)  ; gd)
a " a a
" "

Proof: Similarly by (b), so we omit the proof.


7.12 Give an example of bounded function f and an increasing function ) defined on
b
¡a, b¢ such that |f|  RŸ)  but for which ; a fd) does not exist.
Solution: Let
1 if x  ¡0, 1¢ 9 Q
fŸx  
"1 if x  ¡0, 1¢ 9 Q c
and )Ÿx   x on ¡0, 1¢. Then it is clear that f  RŸ)  on ¡a, b¢ and |f|  RŸ)  on ¡a, b¢.
7.13 Let ) be a continuous function of bounded variation on ¡a, b¢. Assume that
x
g  RŸ)  on ¡a, b¢ and define *Ÿx   ; gŸt d)Ÿt  if x  ¡a, b¢. Show that:
a
(a) If f { on ¡a, b¢, there exists a point x 0 in ¡a, b¢ such that
b x b
; fd*  fŸa  ; 0 gd)  fŸb  ; gd).
a a x0

Proof: Since ) is a continuous function of bounded variation on ¡a, b¢, and g  RŸ) 
on ¡a, b¢, we know that *Ÿx  is a continuous function of bounded variation on ¡a, b¢, by
Theorem 7.32. Hence, by Second Mean-Value Theorem for Riemann-Stieltjes
integrals, we know that
b x0 b
; a fd*  fŸa  ; a d*Ÿx   fŸb  ; d*Ÿx 
x0
which implies that, by Theorem 7.26,
b x0 b
; a fd*  fŸa  ; a gd)  fŸb  ; gd).
x0

(b) If, in addition, f is continuous on ¡a, b¢, we also have


b x0 b
; a fŸx gŸx dx  fŸa  ; a gd)  fŸb  ; gd).
x0

Proof: Since
b b
; a fd*  ; a fŸx gŸx dx by Theorem 7.26,
we know that, by (a),
b x0 b
; a fŸx gŸx dx  fŸa  ; a gd)  fŸb  ; gd).
x0

Remark: We do NOT need the hypothesis that f is continuous on ¡a, b¢.


7.14 Assume that f  RŸ)  on ¡a, b¢, where ) is of bounded variation on ¡a, b¢. Let
VŸx  denote the total variation of ) on ¡a, x¢ for each x in Ÿa, b¢, and let VŸa   0. Show
that
b b
; a fd) t ; a |f|dV t MVŸb ,
where M is an upper bound for |f| on ¡a, b¢. In particular, when )Ÿx   x, the inequality
becomes
b
; a fd) t MŸb " a .

Proof: Given /  0, there is a partition P  £a  x 0 , . . . , x n  b¤ such that


b
; a fd) " /  SŸP, f, ) 
n
 ! fŸt k   ) k , where t k  ¡x k"1 , x k ¢
k1
n
t !|fŸt k  ||)Ÿx k   " )Ÿx k"1  |
k1
n
t !|fŸt k  |ŸVŸx k   " VŸx k"1   
k1
 SŸP, |f|, V 
t UŸP, |f|, V  since V is increasing on ¡a, b¢
which implies that, taking infimum,
b b
; a fd) " / t ; a |f|dV
since |f|  RŸV  on ¡a, b¢.
So, we have
b b
; a fd) t ; a |f|dV *
ʳʳʳʳ
b
since ; |f|dV is clear non-negative. If M is an upper bound for |f| on ¡a, b¢, then (*) implies
a
that
b b
; a fd) t ; a |f|dV t MVŸb 
which implies that
b
; a fd) t MŸb " a 

if )Ÿx   x.
7.15 Let £) n ¤ be a sequence of functions of bounded variation on ¡a, b¢. Suppose
there exists a function ) defined on ¡a, b¢ such that the total variation of ) " ) n on ¡a, b¢
tends to 0 as n v .. Assume also that )Ÿa   ) n Ÿa   0 for each n  1, 2, . . . . If f is
continuous on ¡a, b¢, prove that
b b
lim
nv.
; a fŸx d) n Ÿx   ; a fŸx d)Ÿx .
Proof: Use Exercise 7.14, we then have
b
; a fŸx dŸ) " ) n Ÿx   t MV n Ÿb  v 0 as n v .

where V n is the total variation of ) " ) n , and M  sup x¡a,b¢ |fŸx |.


So, we have
b b
lim
nv.
; a fŸx d) n Ÿx   ; a fŸx d)Ÿx .
Remark: We do NOT need the hypothesis )Ÿa   ) n Ÿa   0 for each n  1, 2, . . . .
7.16 If f  RŸ) , f 2  RŸ) , g  RŸ) , and g 2  RŸ)  on ¡a, b¢, prove that
2
b b fŸx  gŸx 
1
2
;a ;a fŸy  gŸy 
d)Ÿy  d)Ÿx 

b b b 2
 ;a f 2 Ÿx d)Ÿx  ;a g 2 Ÿx d)Ÿx  " ; a fŸx gŸx d)Ÿx  .

When ) { on ¡a, b¢, deduce the Cauchy-Schwarz inequality


b 2 b b
;a fŸx gŸx d)Ÿx  t ; a f 2 Ÿx d)Ÿx  ; a g 2 Ÿx d)Ÿx  .

(Compare with Exercise 1.23.)


Proof: Consider
2
b b fŸx  gŸx 
1
2
;a ;a fŸy  gŸy 
d)Ÿy  d)Ÿx 

b b
 1
2
; a ; a ŸfŸx gŸy  " fŸy gŸx   2 d)Ÿy  d)Ÿx 
b b
 1
2
; a ; a Ÿf 2 Ÿx g 2 Ÿy  " 2fŸx gŸy fŸy gŸx   f 2 Ÿy g 2 Ÿx  d)Ÿy  d)Ÿx 
b b
 1
2
; a f 2 Ÿx d)Ÿx  ; a g 2 Ÿy d)Ÿy 
b b
" ; fŸx gŸx d)Ÿx  ; fŸy gŸy d)Ÿy 
a a
b b
 ; g 2 Ÿx d)Ÿx  ; f 2 Ÿy d)Ÿy 
a a
b b b 2
 ; a f 2 Ÿx d)Ÿx  ; a g 2 Ÿy d)Ÿy  " ; a fŸx gŸx d)Ÿx  ,

if ) { on ¡a, b¢, then we have


2
b b fŸx  gŸx 
0t 1
2
;a ;a fŸy  gŸy 
d)Ÿy  d)Ÿx 

b b b 2
 ;a f 2 Ÿx d)Ÿx  ;a g 2 Ÿy d)Ÿy  " ; a fŸx gŸx d)Ÿx 
which implies that
b 2 b b
; a fŸx gŸx d)Ÿx  t ; a f 2 Ÿx d)Ÿx  ; a g 2 Ÿx d)Ÿx  .

b b
Remark: (1) Here is another proof: Let A  ; f 2 Ÿx d)Ÿx , B  ; fŸx gŸx d)Ÿx , and
a a
b
C  ; g 2 Ÿx d)Ÿx . From the fact,
a
b
0t ; a ¡fŸx z  gŸx ¢ 2 dx for any real z
 Az 2  2Bz  C.
It implies that
B 2 t AC.
That is,
b 2 b b
; a fŸx gŸx d)Ÿx  t ; a f 2 Ÿx d)Ÿx  ; a g 2 Ÿx d)Ÿx  .

Note: (1) The reader may recall the inner product in Linear Algebra. We often
consider Riemann Integral by defining
b
 f, g : ; a fŸx gŸx dx
where f and g are real continuous functions defined on ¡a, b¢. This definition is a real case.
For complex case, we need to preserve its positive definite. So, we define
b
 f, g : ; a fŸx g Ÿx dx
where f and g are complex continuous functions defined on ¡a, b¢, and g means its
conjugate. In addition, in this sense, we have the triangular inequality:
Pf " gP t Pf " hP  Pf " hP, where PfP   f, f  .

(2) Suppose that f  RŸ)  on ¡a, b¢ where ) { on ¡a, b¢ and given /  0, then there
exists a continuous function g on ¡a, b¢ such that
Pf " gP  /.
Proof: Let K  sup x¡a,b¢ |fŸx |, and given /  0, we want to show that
Pf " gP  /.
Since f  RŸ)  on ¡a, b¢ where ) { on ¡a, b¢, given Ÿ1  / U  0, there is a partition
P  £x 0  a, . . . , x n  b¤ such that
n
UŸP, f, )  " LŸP, f, )   !¡M j Ÿf  " m j Ÿf ¢ ) j  Ÿ/ U   2 . 1
ʳʳʳʳ
j1

Write P  A : B, where A  £x j : M j Ÿf  " m j Ÿf   / U ¤ and


B  £x j : M j Ÿf  " m j Ÿf  u / U ¤, then
/ ! )j t !¡M j Ÿf  " m j Ÿf ¢ ) j  Ÿ/ U   2 by (1)
B B
which implies that
! )j  /U. 2
ʳʳʳʳ
B
For this partition P, we define the function g as follows.
x "t t"x
gŸt   x j "j x j"1 fŸx j"1    x j " xj"1
j"1
fŸx j  , where x j"1 t t t x j .
So, it is clear that g is continuous on ¡a, b¢. In every subinterval ¡x j"1 , x j ¢
xj " t t " x j"1
|fŸt  " gŸt |  x j " x j"1 ¡fŸt  " fŸx j"1  ¢  x j " x j"1 ¡fŸt  " fŸx j  ¢
t |fŸt  " fŸx j"1  |  |fŸt  " fŸx j  |
t 2¡M j Ÿf  " m j Ÿf ¢ 3
ʳʳʳʳ
Consider
! ;¡x j"1 ,x j ¢
2
|fŸt  " gŸt | d) t ! 4¡M j Ÿf  " m j Ÿf ¢ 2 ) j by (3)
A A

t 4 ! / U ¡M j Ÿf  " m j Ÿf ¢ ) j by definition of A


A

t 4/ U ! ) j by / U  1
A
 U
4/ ¡)Ÿb  " )Ÿa ¢
and
! ;¡x j"1 ,x j ¢
2
|fŸt  " gŸt | d) t ! 4K 2 )j
B B

t 4K 2 / U by (2).
Hence,
b
; a |fŸt  " gŸt | 2 d) t 4/ U ¡)Ÿb  " )Ÿa ¢  4K 2 / U
 /2
if we choose / U is small enough so that 4/ U ¡)Ÿb  " )Ÿa ¢  4K 2 / U  / 2 . That is, we have
proved that
Pf " gP  /.
P.S.: The exercise tells us a Riemann-Stieltjes integrable function can be approximated
(approached) by continuous functions.
(3)There is another important result called Holder’’s inequality. It is useful in Analysis
and more general than Cauchy-Schwarz inequality. In fact, it is the case p  q  2 in
Holder’’s inequality. We consider the following results.
Let p and q be positive real numbers such that
1  1  1.
p q
Prove that the following statements.
(a) If u u 0 and v u 0, then
p p
uv t up  vq .
Equality holds if and only if u p  v q .
p p
Proof: Let fŸu   up  vq " uv be a function defined on ¡0, . , where 1
p  1
q  1,
p  0, q  0 and v u 0, then f U Ÿu   u p"1 " v. So, we know that
1 1
f U Ÿu   0 if u  0, v p"1 and f U Ÿu   0 if u  v p"1 , .
1
which implies that, by f v p"1  0, fŸu  u 0. Hence, we know that fŸu  u 0 for all u u 0.
1
up vp
That is, uv t p  q . In addition, fŸu   0 if and only if u  v p"1 if and only if u p  v q .
So, Equality holds if and only if u p  v q .
Note: (1) Here is another good proof by using Young’’s Inequality, let fŸx  be an
strictly increasing and continuous function defined on £x : x u 0¤, with fŸ0   0. Then
we have, let a  0 and b  0,
a b
ab t ; 0 fŸx dx  ; 0 f "1 Ÿx dx, where f "1 is the inverse function of f.
And the equality holds if and only if fŸa   b.
Proof: The proof is easy by drawing the function f on x " y plane. So, we omit it.
So, by Young’’s Inequality, let fŸx   x ) , where )  0, we have the Holder’’s
inequality.
(2) The reader should be noted that there are many proofs of (a), for example, using the
concept of convex function, or using A. P. u G. P. along with continuity.
(b) If f, g  RŸ)  on ¡a, b¢ where ) { on ¡a, b¢, f, g u 0 on ¡a, b¢, and
b b
; a f p d)  1  ; a g q d),
then
b
; a fgd) t 1.
Proof: By Holder’’s inequality, we have
fp gq
fg t p  q
b b
which implies that, by ) { on ¡a, b¢, and ; f p d)  1  ; g q d),
a a
b b b
fp gq
; a fgd) t ; a p d)  ; a
1 1
q d)  p  q  1.

(c) If f and g are complex functions in RŸ) , where ) { on ¡a, b¢, then
b b 1/p b 1/q
;a fgd) t ;a |f| p d) ;a |g| q d) . *
ʳʳʳʳ

Proof: First, we note that


b b
; a fgd) t ; a |fg|d).
Also,
b b p
|f|
; a |f| p d)  M p ´ ; a M
d)  1
and
b b q
|g|
; a |g| q d)  N q ´ ; a N
d)  1.
Then we have by (b),
b
|f| |g|
;a M N
d) t 1
which implies that, by (*)
b b 1/p b 1/q
; a fgd) t MN  ; a |f| p d) ; a |g| q d) .

(d) Show that Holder’’s inequality is also true for the ””improper”” integrals.
Proof: It is clear by (c), so we omit the proof.
7.17 Assume that f  RŸ) , g  RŸ) , and f  g  RŸ)  on ¡a, b¢. Show that
b b
1
2
; a ; a ŸfŸy  " fŸx  ŸgŸy  " gŸx  d)Ÿy  d)Ÿx 
b b b
 Ÿ)Ÿb  " )Ÿa   ; fŸx gŸx d)Ÿx  " ; a fŸx d)Ÿx  ; a gŸx d)Ÿx  .
a
If ) { on ¡a, b¢, deduce the inequality
b b b
; a fŸx d)Ÿx  ; a gŸx d)Ÿx  t Ÿ)Ÿb  " )Ÿa   ; fŸx gŸx d)Ÿx 
a
when both f and g are increasing (or both are decreasing) on ¡a, b¢. Show that the reverse
inequality holds if f increases and g decreases on ¡a, b¢.
Proof: Since
b b
1
2
; a ; a ŸfŸy  " fŸx  ŸgŸy  " gŸx  d)Ÿy  d)Ÿx 
b b
 1
2
; a ; a fŸy gŸy  " fŸy gŸx  " fŸx gŸy   fŸx gŸx d)Ÿy  d)Ÿx 
b b b
 Ÿ)Ÿb  " )Ÿa   ; fŸy gŸy d)Ÿy  " ; a fŸx d)Ÿx  ; a gŸx d)Ÿx 
a
which implies that, (let ), f, and g { on ¡a, b¢),
b b
0 t 1 ; ; ŸfŸy  " fŸx  ŸgŸy  " gŸx  d)Ÿy  d)Ÿx 
2 a a
and (let ), and f { on ¡a, b¢, g | on ¡a, b¢),
b b
0 u 1 ; ; ŸfŸy  " fŸx  ŸgŸy  " gŸx  d)Ÿy  d)Ÿx ,
2 a a
we know that, (let ), f, and g { on ¡a, b¢)
b b b
; a fŸx d)Ÿx  ; a gŸx d)Ÿx  t Ÿ)Ÿb  " )Ÿa   ; fŸx gŸx d)Ÿx 
a
and (let ), and f { on ¡a, b¢, g | on ¡a, b¢)
b b b
; a fŸx d)Ÿx  ; a gŸx d)Ÿx  u Ÿ)Ÿb  " )Ÿa   ; fŸx gŸx d)Ÿx .
a

Riemann integrals
7.18 Assume f  RŸ)  on ¡a, b¢. Use Exercise 7.4 to prove that the limit
n
lim b"a !f akb" a
nv. n n
k1
b
exists and has the value ; fŸx dx. Deduce that
a
n n
lim ! n  = , lim !Ÿn 2  k 2   "1/2  log 1 2 .
nv. k2  n2 4 nv.
k1 k1
Proof: Since f  RŸ)  on ¡a, b¢, given /  0, there exists a -  0 such that as
PPP  -, we have
b
SŸP, f  " ; fŸx dx  /.
a
b"a b"a
For this -, we choose n large enough so that n  -, that is, as n u N, we have n  -.
So,
b
SŸP, f  " ; fŸx dx  /
a
which implies that
n b
b"a
n !f akb"
n
a " ; fŸx dx  /.
a
k1
That is,
n
lim b"a !f akb" a
nv. n n
k1
b
exists and has the value ; fŸx dx.
n a n
Since ! k1 k 2 n
n
2  n ! k1
1
k
1
2 , we know that by above result,
n 1
n n
lim
nv. ! n
k2  n2
 lim 1
nv. n ! k
1
2
k1 k1 Ÿ   1
n
1
 ;0dx
1  x2
 arctan 1 " arctan 0
 =/4.
n n
Since ! k1 Ÿn 2  k 2   "1/2  1
n ! k1 1
2 1/2
, we know that by above result,
1 k
n
n n
lim
nv. !Ÿn 2  k 2   "1/2  lim 1
nv. n ! 1
2 1/2
k1 k1 1  Ÿ nk  
1
 ;0 dx
1/2
Ÿ1  x 2  
=/4
 ;0 sec 2d2, let x  tan 2
=/4
 ;0 sec 2 sec 2  tan 2 d2
sec 2  tan 2
1 2
 ;1 du , let sec 2  tan 2  u
u
 log 1  2 .

7.19 Define
x 2 1
fŸx   ; 0 e "t dt
2
, gŸx   ;0 e "x 2 Ÿt 2 1  dt.
t2  1
(a) Show that g U Ÿx   f U Ÿx   0 for all x and deduce that fŸx   gŸx   =/4.
Proof: Since
x
f U Ÿx   2 ; 0 e "t dt2
e "x
2

"x 2 t 2 1
and note that if hŸx, t   e t 2 1 , we know that h is continuous on ¡0, a¢ • ¡0, 1¢ for any
real a  0, and h x  "2xe "x 2 Ÿt 2 1  is continuous on ¡0, a¢ • ¡0, 1¢ for any real a  0,
1
g U Ÿx   ; 0 h x dt
1
 ; 0 "2xe "x Ÿt 1  dt 2 2

1
 "2e "x
2
; 0 xe "Ÿxt  dt 2

x
 "2e "x 2 ; e "u 2 du,
0
we know that 
g U Ÿx   0 for all x. Hence, we have fŸx   gŸx   C for all x,
f U Ÿx 
1 dt
constant. Since C  fŸ0   gŸ0   ; 1t 2  =/4, fŸx   gŸx   =/4.
0

Remark: The reader should think it twice on how to find the auxiliary function g.
(b) Use (a) to prove that
x
lim
xv.
; 0 e "t dt 
2 1 =.
2
Proof: Note that
"x Ÿt 1  2 2
hŸx, t   e 2 t |e "x 2 Ÿt 2 1  | t 1 for all x  0;
t 1 x 2 Ÿt 2  1 
we know that
1 1
;0 e "x 2 Ÿt 2 1  dt t
t2  1
1
x2
;0 dt
1  t2
v 0 as x v ..

So, by (a), we get


lim
xv.
fŸx   =/4
which implies that
x
lim
xv.
; 0 e "t dt 
2 1 =
2
x x x
since lim xv. ; e "t dt exists by ; e "t dt t ; dt
2 2
 arctan x v =/2 as x v ..
0 0 0 1t 2

Remark: (1) There are many methods to show this. But here is an elementary proof
with help of Taylor series and Wallis formula. We prove it as follows. In addition, the
reader will learn some beautiful and useful methods in the future. For example, use the
application of Gamma function, and so on.
Proof: Note that two inequalities,
.
! xk!
2k
1 x2 t e x2  for all x
k0
and
. .
! x 2k t
k!
! x 2k  1 if |x|  1
1 " x2
k0 k0
which implies that
1 " x 2 t e "x 2 if 0 t x t 1 ´ Ÿ1 " x 2   n t e "nx 2 1
ʳʳʳʳ
and
n
2
e "x t 1 if x t 0 ´ e "nx 2 t 1 . 2
ʳʳʳʳ
1  x2 1  x2
So, we have, by (1) and (2),
1 1 . . n
; 0 Ÿ1 " x 2   n dx t ; 0 e "nx dx t ; 0 e "nx dx t ; 0
2 2 1
1  x2
dx. 3
ʳʳʳʳ
Note that
. .
; 0 e "nx dx 
2 1
n
; 0 e "x dx :
2 K .
n
Also,
1 =/2 2  4  6   Ÿ2n " 2 Ÿ2n 
; 0 Ÿ1 " x 2   n dx  ; 0 sin 2n1 tdt 
1  3  5   Ÿ2n  1 
and
. n =/2 1  3  5   Ÿ2n " 3  =
;0 1
1  x2
dx  ;0 sin 2n"2 tdt 
2  4  6   Ÿ2n " 2  2
,

so
2  4  6   Ÿ2n " 2 Ÿ2n  1  3  5   Ÿ2n " 3  =
n tKt n
1  3  5   Ÿ2n  1  2  4  6   Ÿ2n " 2  2
which implies that
n ¡2  4  6   Ÿ2n " 2 Ÿ2n ¢ 2 n ¡1  3  5   Ÿ2n " 3 ¢ 2 Ÿ2n " 1  = 2
t K 2 t 4
ʳʳʳʳ
2n  1 ¡1  3  5   Ÿ2n " 1 ¢ 2 Ÿ2n  1  2n " 1 ¡2  4  6   Ÿ2n " 2 ¢ 2 2
By Wallis formula, we know that, by (4)
=
K .
2
That is, we have proved that Euler-Possion Integral
. =
; 0 e "x dx 
2

2
.

Note: (Wallis formula)


¡2  4  6   Ÿ2n " 2 Ÿ2n ¢ 2
lim  =.
nv. 2
¡1  3  5   Ÿ2n " 1 ¢ Ÿ2n  1  2

Proof: As 0 t x t =/2, we have


sin 2n1 t t sin 2n t t sin 2n"1 t, where n  N.
So, we know that
=/2 =/2 =/2
;0 sin 2n1 tdt t ;0 sin 2n tdt t ;0 sin 2n"1 tdt
which implies that
Ÿ2n Ÿ2n " 2    4  2 Ÿ2n " 1 Ÿ2n " 3    3  1 = Ÿ2n " 2 Ÿ2n " 4    4  2
t t .
Ÿ2n  1 Ÿ2n " 1    3  1 Ÿ2n Ÿ2n " 2    4  2 2 Ÿ2n " 1 Ÿ2n " 3    3  1
So,
2 2
Ÿ2n Ÿ2n " 2    4  2 Ÿ2n Ÿ2n " 2    4  2
1 t = t 1 .
Ÿ2n " 1 Ÿ2n " 3    3  1 2n  1 2 Ÿ2n " 1 Ÿ2n " 3    3  1 2n
Hence, from
2
Ÿ2n Ÿ2n " 2    4  2 1 " 1 t 1 = v 0,
Ÿ2n " 1 Ÿ2n " 3    3  1 2n 2n  1 2n 2
we know that
¡2  4  6   Ÿ2n " 2 Ÿ2n ¢ 2
lim  =.
nv. 2
¡1  3  5   Ÿ2n " 1 ¢ Ÿ2n  1  2

(2) Here is another exercise from Hadamard’’s result. We Write it as follows. Let
f  C k ŸR  with fŸ0   0. Prove that there exists an unique function g  C k"1 ŸR  such that
f  xgŸx  on R.
Proof: Consider
fŸx   fŸx  " fŸ0 
1
 ; 0 dfŸxt 
1
 ; 0 xf U Ÿxt dt
1
 x ; f U Ÿxt dt;
0
1
we know that if gŸx  : ; f U Ÿxt dt, then we have prove it.
0

Note: In fact, we can do this job by rountine work. Define


fŸx 
x if x p 0
gŸx  
0 if x  0.
However, it is too long to write. The trouble is to make sure that g  C k"1 ŸR .
x
7.20 Assume g  R on ¡a, b¢ and define fŸx   ; gŸt dt if t  ¡a, b¢. Prove that the
a
x
integral ; |gŸt |dt gives the total variation of f on ¡a, x¢.
a
x
Proof: Since ; |gŸt |dt exists, given /  0, there exists a partition
a
P 1  £x 0  a, . . . , x n  x¤ such that
x
LŸP, |g|   ; a |gŸt |dt " /. 1
ʳʳʳʳ
So, for this P 1 , we have
n n xk n
!|fŸx k   " fŸx k"1  |  ! ;x k"1
gŸt dt  !|c k Ÿx k " x k"1  | by Mean Value Theorem 2
ʳʳʳʳ
k1 k1 k1
where inf x¡x k"1 ,x k ¢ |gŸx | t c k t sup x¡x k"1 ,x k ¢ |gŸx |.
Hence, we know that, by (1) and (2),
n x
!|fŸx k   " fŸx k"1  |  ; a |gŸt |dt " /
k1
which implies that
x
V f Ÿa, b  u ; a |gŸt |dt
since / is arbitrary.
x
Conversely, since ; |gŸt |dt exists, given /  0, there exists a partition P 2 such that
a
x
UŸP 2 , |g|   ; a |gŸt |dt  //2. 3
ʳʳʳʳ
Also, for the same /, there exists a partition P 3  £t 0  a, . . . , t m  x¤ such that
m
V f Ÿa, b  " //2  !|fŸt k   " fŸt k"1  |. 4
ʳʳʳʳ
k1
Let P  P 2 : P 3  £s 0  a, . . . , s p  x¤, then by (3) and (4), we have
x
UŸP, |g|   ; a |gŸt |dt  //2
and
p
V f Ÿa, b  " //2  !|fŸs k   " fŸs k"1  |
k1
p
sk
 ! ;s gŸt dt
k1 k"1

p
 !|c k Ÿx k " x k"1  |
k1
t UŸP, |g| 
which imply that
x
V f Ÿa, b  t ; a |gŸt |dt
since / is arbitrary.
Therefore, from above discussion, we have proved that
x
V f Ÿa, b   ; a |gŸt |dt.
7.21 If f  Ÿf 1 , . . . , f n   be a vector-valued function with a continuous derivative f U on
¡a, b¢. Prove that the curve described by f has length
b
" f Ÿa, b   ; a Pf U Ÿt Pdt.
Proof: Since f U  Ÿf U1 , . . . , f Un   is continuous on ¡a, b¢, we know that
n 2 1/2
! j1 Ÿf Uj   Ÿt   Pf U Ÿt P is uniformly continuous on ¡a, b¢. So, given /  0, there exists
a - 1  0 such that as |x " y|  - 1 , where x, y  ¡a, b¢, we have
|Pf U Ÿx P " Pf U Ÿy P|  / . 1
ʳʳʳʳ
3Ÿb " a 
Since Pf U Ÿt P  R on ¡a, b¢, for the same /, there exists - 2  0 such that as PP 1 P  - 2 ,
where P 1  £x 0  a, . . . , x n  b¤we have
b n
SŸP 1 , Pf U P  " ; Pf U Ÿt Pdt  //3, where SŸP 1 , Pf U P   !Pf U Ÿt j  P xj 2
ʳʳʳʳ
a
j1

and " f Ÿa, b  exists by Theorem 6.17, for the same /, there exists a partition
P 2  £s 0  a, . . . , s m  b¤ such that
m
" f Ÿa, b  " //3  !PfŸs k   " fŸs k"1  P
k1
m n 1/2

 ! !¡Ÿf j  Ÿs k   " Ÿf j  Ÿs k"1  ¢ 2


. 3
ʳʳʳʳ
k1 j1

Let -  minŸ- 1 , - 2   and P – P 2 so that PPP  -, where P  £y 0  a, . . . , y q  b¤then by


(1)-(3), we have
(i) As |x " y|  -, where x, y  ¡a, b¢, we have
|Pf U Ÿx P " Pf U Ÿy P|  / . 4
ʳʳʳʳ
3Ÿb " a 
(ii) As PPP  -, we have
q
b
SŸP, Pf U P  "; Pf U Ÿt Pdt  //3, where SŸP, Pf U P   !Pf U Ÿt j  P yj 5
ʳʳʳʳ
a
j1

(iii) As PPP  -, we have


m n 1/2

" f Ÿa, b  " //3 t ! !¡Ÿf j  Ÿs k   " Ÿf j  Ÿs k"1  ¢ 2

k1 j1

q n 1/2

t ! !¡Ÿf j  Ÿy k   " Ÿf j  Ÿy k"1  ¢ 2


k1 j1
q n
 ! !¡f Uj Ÿz k  ¢ 2 y j , by Mean Value Theorem
k1 j1
q
 !Pf U Ÿz k  P yj
k1
t " f Ÿa, b  6
ʳʳʳʳ
By (ii) and (iii), we have
q q q

! gŸz k   y j " SŸP, g   ! gŸz k   y j " ! gŸt j   y j


k1 k1 j1
q
t !|gŸz k   " gŸt j  | yj
k1
q
 ! 3Ÿb /" a  yj
k1
 //3. 7
ʳʳʳʳ
Hence, (5)-(7) implies that
b
; a Pf U Ÿt Pdt " " f Ÿa, b   /.

Since / is arbitrary, we have proved that


b
" f Ÿa, b   ; a Pf U Ÿt Pdt.
7.22 If f Ÿn1  is continuous on ¡a, x¢, define
x
I n Ÿx   1 ; Ÿx " t  n f Ÿn1  Ÿt dt.
n! a
(a) Show that
f Ÿk  Ÿa Ÿx " a  k
I k"1 Ÿx  " I k Ÿx   , k  1, 2, . . . , n.
k!
Proof: Since, for k  1, 2, . . . , n,
x
I k Ÿx   1 ; Ÿx " t  k f Ÿk1  Ÿt dt
k! a
x
 1 ; Ÿx " t  k df Ÿk  Ÿt 
k! a
x
 1 Ÿx " t  k f Ÿk  Ÿt  a  k ; Ÿx " t  k"1 f Ÿk  Ÿt dt
x
k! a
k
f Ÿa Ÿx " a 
Ÿk  x
"
k!
 1 ;
Ÿk " 1 ! a
Ÿx " t  k"1 f Ÿk  Ÿt dt

f Ÿk  Ÿa Ÿx " a  k


"  I k"1 Ÿx ,
k!
we know that
f Ÿk  Ÿa Ÿx " a  k
I k"1 Ÿx  " I k Ÿx   , for k  1, 2, . . . , n.
k!
(b) Use (a) to express the remainder in Taylor’’s formula (Theorem 5.19) as an integral.
Proof: Since fŸx  " fŸa   I 0 Ÿx , we know that
fŸx   fŸa   I 0 Ÿx 
n
 fŸa   !¡I k"1 Ÿx  " I k Ÿx ¢  I n Ÿx 
k1
n
f Ÿk  Ÿa Ÿx " a  k x
 ! k!
 1
n!
; a Ÿx " t  n f Ÿn1  Ÿt dt by (a).
k0
So, by Taylor’’s formula, we know that
f Ÿn1  Ÿc Ÿx " a  n1 x
R n Ÿx  
Ÿn  1 !
 1
n!
; a Ÿx " t  n f Ÿn1  Ÿt dt, for some c  Ÿa, x .
where R n Ÿx  is the remainder term.
Remark: 1. The reader should be noted that with help of Mean Value Theorem, we
have
x Ÿn1  Ÿc Ÿx " a  n1
1 ; Ÿx " t  n f Ÿn1  Ÿt dt  f . *
ʳʳʳʳ
n! a Ÿn  1 !
2. Use Integration by parts repeatedly; we can show (*). Of course, there is other
proofs such as Mathematical Induction.
Proof: Since
; uv Ÿn1  dt  uv Ÿn  " u U v Ÿn"1   u UU v Ÿn"2  ". . . Ÿ"1  n u Ÿn  v  Ÿ"1  Ÿn1  ; u Ÿn1  vdt,
letting vŸt   Ÿx " t  n and uŸt   fŸt , then
n x
Ÿk  Ÿa 
fŸx   !f k!
Ÿx " a  k  1
n!
; a Ÿx " t  n f Ÿn1  Ÿt dt.
k0

Note: The reader should give it a try to show it. Since it is not hard, we omit the detail.
3. The remainder term as an integral is useful; the reader should see the textbook in
Ch9, pp242-244.
4. There is a good exercise related with an application of Taylor’’s Remainder. We write
it as a reference.
Let u UU Ÿt   fŸt uŸt   0, where fŸt  is continuous and non-negative on ¡0, c¢ If u is
defined and not a zero function on ¡0, c¢ and
b
; a Ÿb " t Ÿa " t fŸt   b " a for all a, b  ¡0, c¢, where a  b. *
ʳʳʳʳ
Then u at most has one zero on ¡0, c¢.
Proof: First, we note that u has at most finitely many zeros in the interval ¡0, c¢ by
uniqueness theorem on O.D.E. So, let uŸa   uŸb   0, where a, b  ¡0, c¢ with a  b,
and no point y  Ÿa, b  such that uŸy   0. Consider ¡a, b¢ and by Taylor’’s Theorem with
Remainder Term as an integral, we have
x
uŸx   uŸa   u U Ÿa Ÿx " a   ; Ÿx " t u UU Ÿt dt
a
x
 u U Ÿa Ÿx " a   ; Ÿx " t u UU Ÿt dt
a
x
 u U Ÿa Ÿx " a  " ; Ÿx " t uŸt fŸt dt. **
ʳʳʳʳ
a
Note that uŸx  is positive on Ÿa, b  ( Or, uŸx  is negative on Ÿa, b  ) So, we have
|uŸx | t |u U Ÿa |Ÿx " a . ***
ʳʳʳʳ
By (**),
b
0  uŸb   u U Ÿa Ÿb " a  " ; Ÿb " t uŸt fŸt dt
a
which implies that
b
u U Ÿa Ÿb " a   ; a Ÿb " t uŸt fŸt 
which implies that by (***), and note that u U Ÿa  p 0,
b
b"a t ; a Ÿb " t Ÿt " a fŸt dt
which contradicts to (*). So, u at most has one zero on ¡0, c¢.
Note: (i) In particular, let fŸt   e "t , we have (*) holds.
Proof: Since
b
; a Ÿb " t Ÿt " a e "t dt  e "a Ÿ"2  b " a   e "b Ÿ2  b " a 
by integration by parts twice, we have, (let b " a  x),
e "a Ÿ"2  b " a   e "b Ÿ2  b " a  " Ÿb " a 
 e "a Ÿ"2  x   e "x"a Ÿ2  x  " x
 xŸe "a " 1   e "a"x Ÿ"2e x  Ÿx  2  
 0 since a  b and e x  1  x.
(ii) In the proof of exercise, we use the uniqueness theorem: If pŸx  and qŸx  are
continuous on ¡0, a¢, then
y UU  pŸx y U  qŸx y  0, where yŸ0   y 0 , and y U Ÿ0   y U0
has one and only one solution. In particular, if yŸ0   y U Ÿ0   0, then y  0 on ¡0, a¢ is the
only solution. We do NOT give a proof; the reader can see the book, Theory of Ordinary
Differential Equation by Ince, section 3.32, or Theory of Ordinary Differential
Equation by Coddington and Levison, Chapter 6.
However, we need use the uniqueness theorem to show that u (in the exercise) has at
most finitely many zeros in ¡0, c¢.
Proof: Let S  £x : uŸx   0, x  ¡0, c¢¤. If #ŸS   ., then by Bolzano-Weierstrass
Theorem, S has an accumulation point p in ¡0, c¢. Then uŸp   0 by continuity of u. In
addition, let r n v p, and uŸr n    0, then
uŸx  " uŸp  uŸr n   " uŸp 
u U Ÿp   lim
xvp x"p  lim
nv. rn " p  0.
(Note that if p is the endpoint of ¡0, c¢, we may consider x v p  or x v p " ). So, by
uniqueness theorem, we then have u  0 on ¡0, c¢ which contradicts to the hypothesis, u
is not a zero function on ¡0, c¢. So, #ŸS   ..
7.23 Let f be continuous on ¡0, a¢. If x  ¡0, a¢, define f 0 Ÿx   fŸx  and let
x
f n1 Ÿx   1 ; Ÿx " t  n fŸt dt, n  0, 1, 2, . . .
n! 0
(a) Show that the nth derivative of f n exists and equals f.
Proof: Consider, by Chain Rule,
x
f Un  1
Ÿn " 1 !
; 0 Ÿx " t  n"1 fŸt dt  f n"1 for all n  N,
we have
f Ÿn 
n  f.
That is, nth derivative of f n exists and equals f.
Remark: (1) There is another proof by Mathematical Induction and Integration by
parts. It is not hard; we omit the proof.
(2) The reader should note that the exercise tells us that given any continuous function f
on ¡a, b¢, there exists a function g n on ¡a, b¢ such that g Ÿn 
n  f, where n  N. In fact, the
function
x
g n  1 ; Ÿx " t  n fŸt dt, n  0, 1, 2, . . .
n! a
(3) The reader should compare the exercise with 7.22. At the same time, look at two
integrands in both exercises.
(b) Prove the following theorem of M. Fekete: The number of changes in sign of f in
¡0, a¢ is not less than the number of changes in sign in the ordered set of numbers
fŸa , f 1 Ÿa , . . . , f n Ÿa .
Hint: Use mathematical induction.
Proof: Let TŸf  denote the number of changes in sign of f on ¡0, a¢ and S n Ÿf  the
number of changes in sign in the ordered set of numbers
fŸa , f 1 Ÿa , . . . , f n Ÿa .
We prove TŸf  u S n Ÿf  for each n by Mathematical Induction as follows. Note that
S n Ÿf  t n.
As n  1, if S 1 Ÿf   0, then there is nothing to prove it. If S 1 Ÿf   1, it means that
fŸa f 1 Ÿa   0. Without loss of generality, we may assume that fŸa   0, so f 1 Ÿa   0
which implies that
a
0  f 1 Ÿa   1 ; fŸt dt
0! 0
which implies that there exists a point y  ¡0, a  such that fŸy   0. Hence, TŸf  u S 1 Ÿf 
holds for any continuous functions defined on ¡0, a¢.
Assume that n  k holds for any continuous functions defined on ¡0, a¢, As n  k  1, .
we consider the ordered set of numbers
fŸa , f 1 Ÿa , . . . , f k Ÿa , f k1 Ÿa .
Note that
f n1 Ÿa   Ÿf 1   n Ÿa  for all n  N,
so by induction hypothesis,
TŸf 1   u S k Ÿf 1  
Suppose S k Ÿf 1    p, and f 1 Ÿ0   0, then f U1  f at least has p zeros by Rolle’’s
Theorem. Hence,
TŸf  u TŸf 1   u S k Ÿf 1    p *
ʳʳʳʳ
We consider two cases as follows.
(i) fŸa f 1 Ÿa  u 0 :With help of (*),
TŸf  u S k Ÿf 1    S k1 Ÿf .
(ii) fŸa f 1 Ÿa   0 :Claim that
TŸf   S k Ÿf 1    p
as follows. Suppose NOT, it means that TŸf   TŸf 1    p by (*). Say
fŸa 1    fŸa 2   . . .  fŸa p    0, where 0  a 1  a 2 . . .  a p  1.
and
f 1 Ÿb 1    f 1 Ÿb 2   . . .  f 1 Ÿb p    0, where 0  b 1  b 2 . . .  b p  1.
By fŸa f 1 Ÿa   0, we know that
fŸx f 1 Ÿx   0 where x  Ÿ0, c , c  minŸa 1 , b 1  
which is impossible since
fŸx f 1 Ÿx   fŸx ¡f 1 Ÿx  " f 1 Ÿ0 ¢ by f 1 Ÿ0   0
 fŸx f U1 Ÿy , where y  Ÿ0, x  – Ÿ0, c 
 fŸx fŸy 
 0 since fŸx  and fŸy  both positive or negative.
So, we obtain that TŸf   S k Ÿf 1    p. That is, TŸf  u S k Ÿf 1    1  S k1 Ÿf .
From above results, we have proved it by Mathmatical Induction.
(c) Use (b) to prove the following theorem of Feje’’r: The number of changes in sign of
f in ¡0, a¢ is not less than the number of changes in sign in the ordered set
a a a
fŸ0 , ; fŸt dt, ; tfŸt dt, . . . , ; t n fŸt dt.
0 0 0

Proof: Let gŸx   fŸa " x , then, define g 0 Ÿx   gŸx , and for n  0, 1, 2, . . . ,
a
g n1 Ÿa   1 ; Ÿa " t  n gŸt dt
n! 0
a
 1 ; u n fŸu du by change of variable (u  a " t).
n! 0
So, by (b), the number of changes in sign of g in ¡0, a¢ is not less than the number of
changes in sign in the ordered set
gŸa , g 1 Ÿa , . . . , g n1 Ÿa .
That is, the number of changes in sign of g in ¡0, a¢ is not less than the number of changes
in sign in the ordered set
a a a
fŸ0 , ; fŸt dt, ; tfŸt dt, . . . , ; t n fŸt dt.
0 0 0
Note that the number of changes in sign of g in ¡0, a¢ equals the number of changes in
sign of f in ¡0, a¢, so we have proved the Feje’’r Theorem.
7.24 Let f be a positive continuous function in ¡a, b¢. Let M denote the maximum
value of f on ¡a, b¢. Show that
b 1/n
lim
nv.
;a fŸx  n dx  M.

Proof: Since f is a positive continuous function in ¡a, b¢, there exists a point c  ¡a, b¢
such that fŸc   M  sup x¡a,b¢ fŸx   0. Then given ŸM  /  0, there is a -  0 such
that as x  BŸc, -  9 ¡a, b¢ : I, we have
Ÿ0  M " /  fŸx   M  /.
Hence, we have
1/n b 1/n
|I| 1/n ŸM " /  t ;I f n Ÿx dx t ; a fŸx  n dx t Ÿb " a  1/n M

which implies that


b 1/n
M " / t lim
nv.
inf ; a fŸx  dx n
t M.
b 1/n
So, lim nv. inf ; fŸx  n dx  M since / is arbitrary. Similarly, we can show that
a
b 1/n b 1/n
lim nv. sup ; fŸx  n dx  M. So, we have proved that lim nv. ; fŸx  n dx  M.
a a

Remark: There is good exercise; we write it as a reference. Let fŸx  and gŸx  are
continuous and non-negative function defined on ¡a, b¢. Then
b 1/n
lim
nv.
; a fŸx  gŸx dx
n
 max fŸx .
x¡a,b¢
x
Since the proof is similar, we omit it. (The reader may let )Ÿx   ; gŸt dt).
a

7.25 A function f of two real variables is defined for each point Ÿx, y  in the unit square
0 t x t 1, 0 t y t 1 as follows:
1 if x is rational,
fŸx, y  
2y if x is irrational.
1 1
(a) Compute ; fŸx, y dx and ; fŸx, y dx in terms of y.
0 0_

Proof: Consider two cases for upper and lower Riemann-Stieltjes integrals as
follows.
(i) As y  ¡0, 1/2¢ : Given any partition P  £x 0  0, . . . , x n  1¤, we have
sup fŸx, y   1, and inf fŸx, y   2y.
x¡x j"1 ,x j ¢ x¡x j"1 ,x j ¢
1 1
Hence, ; fŸx, y dx  1, and ; fŸx, y dx  2y.
0 0_
(ii) As y  Ÿ1/2, 1¢ : Given any partition P  £x 0  0, . . . , x n  1¤, we have
sup fŸx, y   2y, and inf fŸx, y   1.
x¡x j"1 ,x j ¢ x¡x j"1 ,x j ¢
1 1
Hence, ; fŸx, y dx  2y, and ; fŸx, y dx  1.
0 0_
1 t
(b) Show that ; fŸx, y dy exists for each fixed x and compute ; fŸx, y dy in terms of x
0 0
and t for 0 t x t 1, 0 t t t 1.
Proof: If x  Q 9 ¡0, 1¢, then fŸx, y   1. And if x  Q c 9 ¡0, 1¢, then fŸx, y   2y. So,
for each fixed x, we have
1 1
; 0 fŸx, y dy  ; 0 1dy  1 if x  Q 9 ¡0, 1¢
and
1 1
; 0 fŸx, y dy  ; 0 2ydy  1 if x  Q c 9 ¡0, 1¢.
In addition,
t t
; 0 fŸx, y dy  ; 0 1dy  t if x  Q 9 ¡0, 1¢
and
t t
; 0 fŸx, y dy  ; 0 2ydy  t 2 if x  Q c 9 ¡0, 1¢.
1 1
(c) Let FŸx   ; fŸx, y dy. Show that ; FŸx dx exists and find its value.
0 0

Proof: By (b), we have


FŸx   1 on ¡0, 1¢.
1
So, ; FŸx dx exists and
0
1
; 0 FŸx dx  1.
7.26 Let f be defined on ¡0, 1¢ as follows: fŸ0   0; if 2 "n"1  x t 2 "n , then
fŸx   2 "n for n  0, 1, 2, . . .
1
(a) Give two reasons why ; fŸx dx exists.
0

Proof: (i) fŸx  is monotonic decreasing on ¡0, 1¢. (ii) x : f is discontinuous at x has
measure zero.
Remark: We compute the value of the integral as follows.
Solution: Consider the interval I n  ¡2 "n , 1¢ where n  N, then we have f  R on I n for
each n, and
1 n 2 "k1
;2 "n
fŸx dx  ! ;2 "k
fŸx dx
k1
n 2 "k1
 ! 2 "k1 ;2 "k
dx
k1
n
 !Ÿ2 "k1  Ÿ2 "k  
k1
1 n
4
1 " Ÿ 14  
2
1 " 14
n
 2 1" 1 v 2 as n v ..
3 4 3
1
So, th integral ; fŸx dx  2
3
.
0

Note: In the remark, we use the following fact. If f  R on ¡a, b¢, then
b b
; a fŸx dx  lim ;
nv. a
n
fŸx dx
where £a n ¤ is a sequence with a n v a, and a n u a for all n.
Proof: Since a n v a, given /  0, there is a positive integer N such that as n u N, we
have
|a n " a|  //M, where M  sup |fŸx |
x¡a,b¢

So,
b b an
; a fŸx dx " ; a n
fŸx dx  ;a fŸx dx

t M|a n " a|
 /.
b b
That is, ; fŸx dx  lim nv. ; fŸx dx.
a an
x
(b) Let FŸx   ; fŸt dt. Show that for 0  x t 1 we have
0

FŸx   xAŸx  " 1 AŸx  2 ,


3
where AŸx   2 "¡"log x/ log 2¢ and where ¡y¢ is the greatest integer in y.
1 1 1
Proof: First, we note that FŸx   ; fŸt dt " ; fŸt dt  2
3
" ; fŸt dt. So, it suffices to
0 x x
consider the value of the integral
1
; x fŸt dt.
Given any x  ¡0, 1¢, then there exists a positive N such that 2 "N"1  x t 2 "N . So,
1 1 2 "N
; x fŸt dt  ; 2 "N
fŸt dt  ;
x
fŸt dt
N
 2 1" 1  2 "N Ÿ2 "N " x  by Remark in (a)
3 4
N N
 2  1 1 " 1 x.
3 3 4 2
So,
N N
x" 1 1
FŸx   1
3 4 2
 2 "N x " 1 2 "2N
3
 xAŸx  " 1 AŸx  2
3
where AŸx   2 "N . Note that 2 "N"1  x t 2 "N , we have
N  log 1/2 x , where ¡y¢ is the Gauss symbol.
Hence,
AŸx   2 "¡"log x/ log 2¢ .
Remark: (1) The reader should give it a try to show it directly by considering ¡0, x¢,
where 0 t x t 1.
(2) Here is a good exercise. We write it as a reference. Suppose that f is defined on
¡0, 1¢ by the following
1 j
2n
if x  2n
where j is an odd integer and 0  j  2 n , n  1, 2, . . . ,
fŸx  
0 otherwise.
Show that f  R on ¡0, 1¢ and has the value of the integral 0.
Proof: In order to show this, we consider the Riemann’’s condition with respect to
)Ÿx   x as follows. Given a partition
j n
P  x 0  0  20n , x 1  21n , x 2  22n , . . . , x j  2 n , . . . , x 2 n  22 n  1 , then the upper
sum
2n
UŸP, f   ! Mk xk
k1
2n
 1n
2 ! Mk
k1

 1n 2n  n " 1 v 0 as n v ..
2 2
So, f satisfies the Riemann’’s condition on ¡0, 1¢.
Note: (1) The reader should give it a try to show that the set of discontinuities of f has
measure zero. Thus by Theorem 7.48 (Lebesgue’’s Criterion for Riemann Integral), we
know that f  R on ¡0, 1¢. In addition, by the fact, the lower Riemann integral equals the
Riemann integral, we know that its integral is zero.
(2) For the existence of Riemann integral, we summarize to be the theorem: Let f be a
bounded function on ¡a, b¢. Then the following statements are equivalent:
(i) f  R on ¡a, b¢.
(ii) f satisfies Riemann’’s condition on ¡a, b¢.
b b
(iii) ; fŸx dx  ; fŸx dx
a a"
(iv) the set of discontinuities of f on I has measure zero.
P.S.: The reader should see the textbook, pp 391; we have the general discussion.
7.27 Assume f has a derivative which is monotonic decreasing and satisfies
f U Ÿx  u m  0 for all x in ¡a, b¢. Prove that
b
; a cos fŸx dx 2.
t m

Hint: Multiply and divide the integrand by f U Ÿx  and use Theorem 7.37(ii).
1
Proof: Since f U Ÿx  u m  0, and fU
is monotonic increasing on ¡a, b¢, we consider
b b cos fŸx  U
; a cos fŸx dx  ; a f U Ÿx 
f Ÿx dx
b
 1
f U Ÿb 
; c ¡cos fŸx ¢f U Ÿx dx, by Theorem 7.37(ii)
fŸb 
 1 ; cos udu, by Change of Variable
f U Ÿb  fŸc 
sin fŸb  " sin fŸc 

f U Ÿb 
which implies that
b
; a cos fŸx dx 2.
t m

7.28 Given a decreasing seq uence of real numbers £GŸn ¤ such that GŸn  v 0 as
n v .. Define a function f on ¡0, 1¢ in terms of £GŸn ¤ as follows: fŸ0   1; if x is
irrational, then fŸx   0; if x is rational m/n (in lowest terms), then fŸm/n   GŸn .
Compute the oscillation F f Ÿx  at each x in ¡0, 1¢ and show that f  R on ¡0, 1¢.
Proof: Let x 0  Q c 9 ¡0, 1¢. Since lim nv. GŸn   0, given /  0, there exists a
positive integer K such that as n u K, we have |GŸn |  /. So, there exists a finite number
of positive integers n such that GŸn  u /. Denote S  £x : |fŸx | u /¤, then #ŸS   ..
Choose a -  0 such that Ÿx 0 " -, x 0  - Ÿ– ¡0, 1¢  does NOT contain all points of S. Note
that fŸx 0    0. Hence, we know that f is continuous at x 0 . That is, F f Ÿx   0 for all
x  Q c 9 ¡0, 1¢.
Let x 0  0, then it is clear that F f Ÿ0   1Ÿ fŸ0    0. So, f is not continuous at
0.  Q 9 ¡0, 1¢.
Let x 0  Q 9 Ÿ0, 1¢, say x 0  M N
(in lowest terms). Since £GŸn ¤ is monotonic
decreasing, there exists a finite number of positive integers n such that GŸn  u GŸN .
Denote T  £x : |fŸx | u GŸN ¤, then #ŸT   .. Choose a -  0 such that
Ÿx 0 " -, x 0  -  9 ¡0, 1¢ does NOT contain all points of T. Let h  Ÿ0, - , then
sup£fŸx  " fŸy  : x, y  Ÿx 0 " h, x 0  h  9 ¡0, 1¢¤  fŸx 0    GŸN .
So, F f Ÿx 0    GŸN . That is, F f Ÿx   fŸx  for all x  Q 9 Ÿ0, 1¢.
Remark: (1) If we have proved f is continuous on Q c 9 ¡0, 1¢, then f is automatically
Riemann integrable on ¡0, 1¢ since DŸ– Q 9 ¡0, 1¢ – Q , the set of discontinuities of f has
measure zero.
(2) Here is a good exercise. We write it as a reference. Given a function f defined on
Ÿa, b , then the set of continuities of f on Ÿa, b  is G - set.
Proof: Let C denote the set of continuities of f on Ÿa, b , then
C  £x : F fŸx   0¤
 9 .k1 £x : F f Ÿx   1/k¤
and £x : F fŸx   1/k¤ is open. We know that C is a G - set.
Note: (i) We call S a G - set if S  9 .n1 O n , where O n is open for each n.
(ii) Given y  £x  Ÿa, b  : F f Ÿx   1/k¤ : I, then F f Ÿy   1/k. Hence, there exists a
d  0, such that
( f ŸBŸy, d    1/k, where BŸy, d  – Ÿa, b 
For z  BŸy, d , consider a smaller - so that BŸz, -  – BŸy, d . Hence,
( f ŸBŸz, -    1/k
which implies that
F f Ÿz   1/k.
Hence, BŸy, d  – I. That is, y is an interior point of I. That is, I is open since every point
of I is interior.
7.29 Let f be defined as in Exercise 7.28 with GŸn   1/n. Let gŸx   1 if
0  x t 1, gŸ0   0. Show that the composite function h defined by hŸx   g¡fŸx ¢ is not
Riemann-integrable on ¡0, 1¢, although both f  R and g  R on ¡0, 1¢.
Proof: By Exercise 7.28, we know that
0 if x  Q c 9 ¡0, 1¢
hŸx  
1 if x  Q 9 ¡0, 1¢
which is discontinuous everywhere on ¡0, 1¢. Hence, the function h (Dirichlet Function) is
not Riemann-integrable on ¡0, 1¢.
7.30 Use Lebesgue’’s theorem to prove Theorem 7.49.
(a) If f is of bounded variation on ¡a, b¢, then f  R on ¡a, b¢.
Proof: Since f is of bounded variation on ¡a, b¢, by Theorem 6.13 (Jordan Theorem),
f  f 1 " f 2 , where f 1 and f 2 are increasing on ¡a, b¢. Let D i denote the set of discontinuities
of f i on ¡a, b¢, i  1, 2. Hence, D, the set of discontinuities of f on ¡a, b¢ is
D – D1 : D2.
Since |D 1 |  |D 2 |  0, we know that |D|  0. In addition, f is of bounded on ¡a, b¢ since f
is bounded variation on ¡a, b¢. So, by Theorem 7.48, f  R on ¡a, b¢.
(b) If f  R on ¡a, b¢, then f  R on ¡c, d¢ for every subinterval ¡c, d¢ – ¡a, b¢, |f|  R
and f 2  R on ¡a, b¢. Also, f  g  R on ¡a, b¢ whenever g  R on ¡a, b¢.
Proof: (i) Let D ¡a,b¢ and D ¡c,d¢ denote the set of discontinuities of f on ¡a, b¢ and ¡c, d¢,
respectively. Then
D ¡c,d¢ – D ¡a,b¢ .
Since f  R on ¡a, b¢, and use Theorem 7.48, |D ¡a,b¢ |  0 which implies that |D ¡c,d¢ |  0.
In addition, since f is bounded on ¡a, b¢, f is automatically is bounded on ¡c, d¢ for every
compact subinterval ¡c, d¢. So, by Theorem 7.48, f  R on ¡c, d¢.
(ii) Let D f and D |f| denote the set of discontinuities of f and |f| on ¡a, b¢, respectively,
then
D |f| – D f .
Since f  R on ¡a, b¢, and use Theorem 7.48, |D f |  0 which implies that |D |f| |  0. In
addition, since f is bounded on ¡a, b¢, it is clear that |f| is bounded on ¡a, b¢. So, by
Theorem 7.48, |f|  R on ¡a, b¢.
(iii) Let D f and D f 2 denote the set of discontinuities of f and f 2 on ¡a, b¢, respectively,
then
D f2 – D f .
Since f  R on ¡a, b¢, and use Theorem 7.48, |D f |  0 which implies that D f 2  0. In
addition, since f is bounded on ¡a, b¢, it is clear that f 2 is bounded on ¡a, b¢. So, by
Theorem 7.48, f 2  R on ¡a, b¢.
(iv) Let D f and D g denote the set of discontinuities of f and g on ¡a, b¢, respectively.
Let D fg denote the set of discontinuities of fg on ¡a, b¢, then
D fg – D f : D g
Since f, g  R on ¡a, b¢, and use Theorem 7.48, |D f |  |D g |  0 which implies that
|D fg |  0. In addition, since f and g are bounded on ¡a, b¢, it is clear that fg is bounded on
¡a, b¢. So, by Theorem 7.48, fg  R on ¡a, b¢.
(c) If f  R and g  R on ¡a, b¢, then f/g  R on ¡a, b¢ whenever g is bounded away
from 0.
Proof: Let D f and D g denote the set of discontinuities of f and g on ¡a, b¢, respectively.
Since g is bounded away from 0, we know that f/g is well-defined and f/g is also bounded
on ¡a, b¢. Consider D f/g , the set of discontinuities of f/g on ¡a, b¢ is
D f/g – D f : D g .
Since f  R and g  R on ¡a, b¢, by Theorem 7.48, |D f |  |D g |  0 which implies that
|D f/g |  0. Since f/g is bounded on ¡a, b¢ with |D f/g |  0, f/g  R on ¡a, b¢ by Theorem
7.48.
Remark: The condition that the function g is bounded away from 0 CANNOT omit.
For example, say gŸx   x on Ÿ0, 1¢ and gŸ0   1. Then it is clear that g  R on ¡0, 1¢, but
1/g  R on ¡0, 1¢. In addition, the reader should note that when we ask whether a function
is Riemann-integrable or not, we always assume that f is BOUNDED on a COMPACT
INTERVAL ¡a, b¢.
(d) If f and g are bounded functions having the same discontinuities on ¡a, b¢, then
f  R on ¡a, b¢ if, and only if, g  R on ¡a, b¢.
Proof: (´) Suppose that f  R on ¡a, b¢, then D f , the set of discontinuities of f on ¡a, b¢
has measure zero by Theorem 7.48. From hypothesis, D f  D g , the set of discontinuities
of g on ¡a, b¢, we know that g  R on ¡a, b¢ by Theorem 7.48.
(²) If we change the roles of f and g, we have proved it.

(e) Let g  R on ¡a, b¢ and assume that m t gŸx  t M for all x  ¡a, b¢. If f is
continuous on ¡m, M¢, the composite function h defined by hŸx   f¡gŸx ¢ is
Riemann-integrable on ¡a, b¢.
Proof: Note that h is bounded on ¡a, b¢. Let D h and D g denote the set of dsicontinuities
of h and g on ¡a, b¢, respectively. Then
Dh – Dg.
Since g  R on ¡a, b¢, then |D g |  0 by Theorem 7.48. Hence, |D h |  0 which implies that
h  R on ¡a, b¢ by Theorem 7.48.
Remark: (1) There has a more general theorem related with Riemann-Stieltjes
Integral. We write it as a reference.
(Theorem)Suppose g  RŸ)  on ¡a, b¢, m t gŸx  t M for all x  ¡a, b¢. If f is
continuous on ¡m, M¢, the composite function h defined by hŸx   f¡gŸx ¢  RŸ)  on
¡a, b¢.
Proof: It suffices to consider the case that ) is increasing on ¡a, b¢. If )Ÿa   )Ÿb ,
there is nothing to prove it. So, we assume that )Ÿa   )Ÿb . In addition, let
K  sup x¡a,b¢ |hŸx |. We claim that h satisfies Riemann condition with respect to ) on
¡a, b¢. That is, given /  0, we want to find a partition P such that
n
UŸP, h, )  " LŸP, h, )   !¡M k Ÿh  " m k Ÿh ¢ ) k  /.
k1
/
Since f is uniformly continuous on ¡m, M¢, for this /  0, there is a 2ŸK1   -  0 such
that as |x " y|  - where x, y  ¡m, M¢, we have
|fŸx  " fŸy |  / . 1
ʳʳʳʳ
2¡)Ÿb  " )Ÿa ¢
Since g  RŸ)  on ¡a, b¢, for this -  0, there exists a partition P such that
n
UŸP, g, )  " LŸP, g, )   !¡M k Ÿg  " m k Ÿg ¢ )k  -2. 2
ʳʳʳʳ
k1
Let P  A : B, where A  £x j : M k Ÿg  " m k Ÿg   -¤ and
B  £x j : M k Ÿg  " m k Ÿg  u -¤, then
- ! )k t !¡M k Ÿh  " m k Ÿh ¢ ) k t - 2 by (2)
B B
which implies that
! ) k t -.
B
So, we have
UŸA, h, )  " LŸA, h, )  t !¡M k Ÿh  " m k Ÿh ¢ )k
A

t //2 by (1)
and
UŸB, h, )  " LŸB, h, )  t !¡M k Ÿh  " m k Ÿh ¢ )k
B

t 2K ! ) k
B
t 2K-
 //2.
It implies that
UŸP, h, )  " LŸP, h, )   UŸA, h, )  " LŸA, h, )   UŸB, h, )  " LŸB, h, ) 
 /.
That is, we have proved that h satisfies the Riemann condition with respect to ) on ¡a, b¢.
So, h  RŸ)  on ¡a, b¢.
Note: We mention that if we change the roles of f and g, then the conclusion does
NOT hold. Since the counterexample is constructed by some conclusions that we will learn
in Real Analysis, we do NOT give it a proof. Let C be the standard Cantor set in ¡0, 1¢ and
C U the Cantor set with positive measure in ¡0, 1¢. Use similar method on defining Cantor
Lebesgue Function, then there is a continuous function f : ¡0, 1¢ v ¡0, 1¢ such that
fŸC   C U . And Choose g  X C on ¡0, 1¢. Then
h  g ( f  X CU
which is NOT Riemann integrable on ¡0, 1¢.
(2) The reader should note the followings. Since these proofs use the exercise 7.30 and
Theorem 7.49, we omit it.
(i) If f  R on ¡a, b¢, then |f|  R on ¡a, b¢, and f r  R on ¡a, b¢, where r  ¡0, . .
(ii) If |f|  R on ¡a, b¢, it does NOT implies f  R on ¡a, b¢. And if f 2  R on ¡a, b¢, it
does NOT implies f  R on ¡a, b¢. For example,
1 if x  Q 9 ¡a, b¢
f
"1  Q c 9 ¡a, b¢
(iii) If f 3  R on ¡a, b¢, then f  R on ¡a, b¢.
7.31 Use Lebesgue’’s theorem to prove that if f  R and g  R on ¡a, b¢ and if
fŸx  u m  0 for all x in ¡a, b¢, then the function h defined by
hŸx   fŸx  gŸx 
is Riemann-integrable on ¡a, b¢.
Proof: Consider
hŸx   expŸh log f ,
then by Theorem 7.49,
f  R ´ log f  R ´ h log f  R ´ expŸh log f   h  R.

7.32 Let I  ¡0, 1¢ and let A 1  I " Ÿ 13 , 23   be the subset of I obtained by removing
those points which lie in the open middle third of I; that is, A 1  ¡0, 13 ¢ : ¡ 23 , 1¢. Let A 2
be the subset of A 1 obtained by removing he open middle third of ¡0, 13 ¢ and of ¡ 23 , 1¢.
Continue this process and define A 3 , A 4 , . . . . The set C  9 .n1 A n is called the Cantor set.
Prove that
(a) C is compact set having measure zero.
Proof: Write C  9 .n1 A n . Note that every A n is closed, so C is closed. Since A 1 Ÿp C 
is closed and bounded, A 1 is compact and C – A 1 , we know that C is compact by
Theorem 3.39.
n
In addition, it is clear that |A n |  Ÿ 23   for each n. Hence, |C| t lim nv. |A n |  0, which
implies that C has a measure zero.
.
(b) x  C if, and only if, x  ! n1 a n 3 "n , where each a n is either 0 or 2.
Proof: (´)Let x  C  9 .n1 A n , then x  A n for all n. Consider the followings.
(i) Since x  A 1  ¡0, 13 ¢ : ¡ 23 , 1¢, then it implies that a 1  0 or 2.
(ii) Since x  A 2  Ÿ¡0, 19 ¢ : ¡ 29 , 39 ¢  : Ÿ¡ 69 , 79 ¢ : ¡ 89 , 99 ¢ , then it implies that a 2  0
or 2.
.
Inductively, we have a n  0 or 2. So, x  ! n1 a n 3 "n , where each a n is either 0 or 2.
.
(²) If x  ! n1 a n 3 "n , where each a n is either 0 or 2, then it is clear that x  A n for
each n. Hence, x  C.
(c) C is uncountable.
Proof: Suppose that C is countable, write C  £x 1 , x 2 , . . . ¤. We consider unique ternary
.
expansion: if x  ! n1 a n 3 "n , then x : Ÿa 1 , . . . , a n , . . .  . From this definition, by (b), we
have
x k  Ÿx k1 , x k2 , . . . x kk, . . .   where each component is 0 or 2.
Choose y  Ÿy 1 , y 2 , . . .   where
2 if x jj  0
yj 
0 if x jj  2.
By (b), y  C. It implies that y  x k for some k which contradicts to the choice of y.
Hence, C is uncountable.
Remark: (1) In fact, C  C U means that C is a perfect set. Hence, C is uncountable.
The reader can see the book, Principles of Mathematical Analysis by Walter Rudin, pp
41-42.
.
(2) Let C  x : x  ! n1 a n 3 "n , where each a n is either 0 or 2 . Define a new
function C : C v ¡0, 1¢ by
.
CŸx   ! Ÿa2n n /2 ,
n1
then it is clear that C is 1-1 and onto. So, C is equivalent to ¡0, 1¢. That is, C is
uncountable.
(d) Let fŸx   1 if x  C, fŸx   0 if x  C. Prove that f  R on ¡0, 1¢.
Proof: In order to show that f  R on ¡0, 1¢, it suffices to show that, by Theorem 7.48,
f is continuous on ¡0, 1¢ " C since it implies that D – C, where D is the set of
discontinuities of f on ¡0, 1¢.
Let x 0  ¡0, 1¢ " C, and note that C  C U , so there is a -  0 such that
Ÿx 0 " -, x 0  -  9 C  C, where Ÿx 0 " -, x 0  -  – ¡0, 1¢. Then given /  0, there is a
-  0 such that as x  Ÿx 0 " -, x 0  - , we have
|fŸx  " fŸx 0  |  0  /.
Remark: (1) C  C U :Given x  C  9 .n1 A n , and note that every endpoints of A n
belong to C. So, x is an accumulation point of the set y : y is the endpoints of A n . So,
C – C U . In addition, C U – C since C is closed. Hence, C  C U .
(2) In fact, we have
f is continuous on ¡0, 1¢ " C and f is not continuous on C.
Proof: In (d), we have proved that f is continuous on ¡0, 1¢ " C, so it remains to show
that f is not continuous on C. Let x 0  C, if f is continuous at x 0 , then given /  1/2, there
is a -  0 such that as x  Ÿx 0 " -, x 0  -  9 ¡0, 1¢, we have
|fŸx  " fŸx 0  |  1/2
which is absurb since we can choose y  Ÿx 0 " -, x 0  -  9 ¡0, 1¢ and y  C by the fact C
does NOT contain an open interval since C has measure zero. So, we have proved that f is
not continuous on C.
Note: In a metric space M, a set SŸ– M  is called nonwhere dense if intŸclŸS    C.
Hence, we know that C is a nonwhere dense set.
Supplement on Cantor set.
From the exercise 7.32, we have learned what the Cantor set is. We write some
conclusions as a reference.
(1) The Cantor set C is compact and perfect.
(2) The Cantor set C is uncountable. In fact, #ŸC   #ŸR .
(3) The Cantor set C has measure zero.
(4) The Cantor set C is nonwhere dense.
.
(5) Every point x in C can be expressed as x  ! n1 a n 3 "n , where each a n is either 0 or
2.
(6) X C : ¡0, 1¢ v £0, 1¤ the characteristic function of C on ¡0, 1¢ is Riemann integrable.
The reader should be noted that Cantor set C in the exercise is 1 "dimensional case. We
can use the same method to construct a n "dimensional Cantor set in the set
£Ÿx 1 , . . . , x n   : 0 t x j t 1, j  1, 2, . . n¤. In addition, there are many researches on Cantor
set. For example, we will learn so called Space-Filling Curve on the textbook, Ch9, pp
224-225.
In addition, there is an important function called Cantor-Lebesgue Function related
with Cantor set. The reader can see the book, Measure and Integral (An Introduction to
Real Analysis) written by Richard L. Wheeden and Antoni Zygmund, pp 35.
7.33 The exercise
n
outlines a proof (due to Ivan Niven) that = 2 is irrational. Let
fŸx   x n Ÿ1 " x  /n!. Prove that:
(a) 0  fŸx   1/n! if 0  x  1.
Proof: It is clear.
(b) Each kth derivative f Ÿk  Ÿ0  and f Ÿk  Ÿ1  is an integer.
Proof: By Leibnitz Rule,
k
f Ÿk  Ÿx   1
n! !Ÿ kj  £¡n   Ÿn " j  1 ¢x n"j ¤ Ÿ"1  k"j ¡n   Ÿn " k  j  1 ¢Ÿ1 " x  n"kj
j0

which implies that


0 if k  n
f Ÿk  Ÿ0   1 if k  n .
Ÿ kn  Ÿ"1  k"n ¡n   Ÿ2n " k  1 ¢ if k  n
So, f Ÿk  Ÿ0   Z for each k  N. Similarly, f Ÿk  Ÿ1   Z for each k  Z.
Now assume that = 2  a/b, where a and b are positive integers, and let
n
FŸx   bn !Ÿ"1  k f Ÿ2k  Ÿx = 2n"2k .
k0
Prove that:
(c) FŸ0  and FŸ1  are integers.
Proof: By (b), it is clear.
d
(d) = 2 a n fŸx  sin =x  dx
£F U Ÿx  sin =x " =FŸx  cos =x¤
Proof: Note that
n n
F UU Ÿx   = 2 FŸx   bn ! Ÿ"1  k f Ÿ2k2  Ÿx = 2n"2k  =2bn !Ÿ"1  k f Ÿ2k  Ÿx = 2n"2k
k0 k0
n"1
 b n !Ÿ"1  k f Ÿ2k2  Ÿx = 2n"2k  b n Ÿ"1  n f Ÿ2n2  Ÿx 
k0
n
 = 2 b n !Ÿ"1  k f Ÿ2k  Ÿx = 2n"2k  = 2 b n fŸx = 2n
k1
n"1
 b n ! Ÿ"1  k f Ÿ2k2  Ÿx = 2n"2k  Ÿ"1  k1 f Ÿ2k2  Ÿx = 2n"2k
k0
 b n Ÿ"1  n f Ÿ2n2  Ÿx   = 2 b n fŸx = 2n
 = 2 a n fŸx  since f is a polynomial of degree 2n.
So,
d £F U Ÿx  sin =x " =FŸx  cos =x¤
dx
 Ÿsin =x ¡F UU Ÿx   = 2 FŸx ¢
 = 2 a n fŸx  sin =x.
1
(e) FŸ1   FŸ0   =a n ; fŸx  sin =xdx.
0

Proof: By (d), we have


1
= 2 a n ; fŸx  sin =xdx  F U Ÿx  sin =x " =FŸx  cos =x| 10
0
 ¡F U Ÿ1  sin = " =FŸ1  cos =¢ " ¡F U Ÿ0  sin 0 " =FŸ0  cos 0¢
 =¡FŸ1   FŸ0 ¢
which implies that
1
FŸ1   FŸ0   =a n ; fŸx  sin =xdx.
0

(f) Use (a) in (e) to deduce that 0  FŸ1   FŸ0   1 if n is sufficiently large. This
contradicts (c) and show that = 2 (and hence =) is irrational.
Proof: By (a), and sin x  ¡0, 1¢ on ¡0, =¢, we have
1 1
0  =a n ; fŸx  sin =xdx t =a ; sin =xdx  2a v 0 as n v ..
n n

0 n! 0 n!
So, as n is sufficiently large, we have, by (d),
0  FŸ1   FŸ0   1
which contradicts (c). So, we have proved that = 2 (and hence =) is irrational.
Remark: The reader should know that = is a transcendental number. (Also, so is e). It
is well-known that a transcendental number must be an irrational number.
In 1900, David Hilbert asked 23 problems, the 7th problem is that, if )Ÿp 0, 1  is an
algebraic number and * is an algebraic number but not rational, then is it true that ) * is a
transcendental number. The problem is completely solved by Israil Moiseevic Gelfand in
1934. There are many open problem now on algebraic and transcendental numbers. For
example, It is an open problem: Is the Euler Constant
+  lim 1  1 . . .  1
n " log n
2
a transcendental number.
7.34 Given a real-valued function ), continuous on the interval ¡a, b¢ and having a
finite bounded derivative ) U on Ÿa, b . Let f be defined and bounded on ¡a, b¢ and assume
that both integrals
b b
; a fŸx d)Ÿx  and ; a fŸx ) U Ÿx dx
exists. Prove that these integrals are equal. (It is not assumed that ) U is continuous.)
Proof: Since both integrals exist, given /  0, there exists a partition
P  £x 0  a, . . . , x n  b¤ such that
b
SŸP, f, )  " ; fŸx d)Ÿx   //2
a
where
n
SŸP, f, )   ! fŸt j   ) j for t j  ¡x j"1 , x j ¢
j1
n
 ! fŸt j  ) U Ÿs j   x j by Mean Value Theorem, where s j  Ÿx j"1 , x j   *
ʳʳʳʳ
j1

and
b
SŸP, f) U   " ; fŸx ) U Ÿx dx  //2
a
where
n
SŸP, f) U    ! fŸt j  ) U Ÿt j   x j for t j  ¡x j"1 , x j ¢ **
ʳʳʳʳ
j1

So, let t j  s j , then we have


SŸP, f, )   SŸP, f) U  .
Hence,
b b
; a fŸx d)Ÿx  " ; a fŸx ) U Ÿx dx
b b
t SŸP, f, )  " ; fŸx d)Ÿx   SŸP, f) U   " ; fŸx ) U Ÿx dx
a a

 /.
So, we have proved that both integrals are equal.
7.35 Prove the following theorem, which implies that a function with a positive
integral must itself be positive on some interval. Assume that f  R on ¡a, b¢ and that
b
0 t fŸx  t M on ¡a, b¢, where M  0. Let I  ; fŸx dx, let h  12 I/ŸM  b " a , and
a
assume that I  0. Then the set T  £x : fŸx  u h¤ contains a finite number of intervals,
the sum of whose lengths is at least h.
n
Hint. Let P be a partition of ¡a, b¢ such that every Riemann sum SŸP, f   ! k1 fŸt k   x k
satisfies SŸP, f   I/2. Split SŸP, f  into two parts, SŸP, f   ! kA  ! kB , where
A  £k : ¡x k"1 , x k ¢ – T¤, and B  £k : k  A¤.
If k  A, use the inequality fŸt k   t M; if k  B, choose t k so that fŸt k    h. Deduce that
! kA x k  h.
Proof: It is clear by Hint, so we omit the proof.
Remark: There is another proof about that a function with a positive integral must
itself be positive on some interval.
Proof: Suppose NOT, it means that in every subinterval, there is a point p such that
fŸp  t 0. So,
n
LŸP, f   ! mj x j t 0 since m j t 0
j1

for any partition P. Then it implies that


b
sup LŸP, f  
P
; a fŸx dx t 0
which contradicts to a function with a positive integral. Hence, we have proved that a
function with a positive integral must itself be positive on some interval.
Supplement on integration of vector-valued functions.
(Definition) Given f 1 , . . . , f n real valued functions defined on ¡a, b¢, and let
f  Ÿf 1 , . . . , f n   : ¡a, b¢ v R n . If ) { on ¡a, b¢. We say that f  RŸ)  on ¡a, b¢ means that
f j  RŸ)  on ¡a, b¢ for j  1, 2, . . . , n. If this is the case, we define
b b b
; a fd)  ; a f 1 d), . . . , ; a f n d) .

From the definition, the reader should find that the definition is NOT stranger for us. When
we talk f  Ÿf 1 , . . . , f n    RŸ)  on ¡a, b¢, it suffices to consider each f j  RŸ)  on ¡a, b¢ for
j  1, 2, . . . , n.
For example, if f  RŸ)  on ¡a, b¢ where ) { on ¡a, b¢, then PfP  RŸ)  on ¡a, b¢.
Proof: Since f  RŸ)  on ¡a, b¢, we know that f j  RŸ)  on ¡a, b¢ for j  1, 2, . . . , n.
Hence,
n
! f 2j  RŸ)  on ¡a, b¢
k1
which implies that, by Remark (1) in Exercise 7.30,
n 1/2

PfP  ! f 2j  RŸ)  on ¡a, b¢  RŸ)  on ¡a, b¢.


k1
Remark: In the case above, we have
b b
; a fd) t ; a PfPd).
Proof: Consider
b b b b
PyP 2   ; a f 1 d), . . . , ; a f n d), ; a f 1 d), . . . , ; a f n d) 
n b b
 ! ; a f j d) ; a f j d) ,
j1
b
which implies that, (let y j  ; f j d), y Ÿy 1 , . . . , y n  ),
a
n b
PyP  2
! y j ; a f j d)
j1
n b
 ! ; a f j y j d)
j1

b n
 ;a ! fjyj d)
j1
b
t ; a PfPPyPd)
b
 PyP ; PfPd)
a
which implies that
b
PyP t ; a PfPd).
Note: The equality holds if, and only if, fŸt   kŸt y.
Existence theorems for integral and differential equations
The following exercises illustrate how the fixed-point theorem for contractions.
(Theorem 4.48) is used to prove existence theorems for solutions of certain integral and
differential equations. We denote by C¡a, b¢ the metric space of all real continuous
functions on ¡a, b¢ with the metric
dŸf, g   Pf " gP  max |fŸx  " gŸx |,
x¡a,b¢

and recall the C¡a, b¢ is a complete metric space.


7.36 Given a function g in C¡a, b¢, and a function K is continuous on the rectangle
Q  ¡a, b¢ • ¡a, b¢, consider the function T defined on C¡a, b¢ by the equation
b
TŸI Ÿx   gŸx   5 ; KŸx, t IŸt dt,
a
where 5 is a given constant.
(a) Prove that T maps C¡a, b¢ into itself.
Proof: Since K is continuous on the rectangle Q  ¡a, b¢ • ¡a, b¢, and IŸx   C¡a, b¢,
we know that
b
; a KŸx, t IŸt dt  C¡a, b¢.
Hence, we prove that TŸI Ÿx   C¡a, b¢. That is, T maps C¡a, b¢ into itself.
(b) If |KŸx, y | t M on Q, where M  0, and if |5|  M "1 Ÿb " a  "1 , prove that T is a
contraction of C¡a, b¢ and hence has a fixed point I which is a solution of the integral
b
equation IŸx   gŸx   5 ; KŸx, t IŸt dt.
a

Proof: Consider
b
PTŸI 1  Ÿx  " TŸI 2  Ÿx P  5 ; KŸx, t ¡I 1 Ÿt  " I2Ÿt ¢dt
a
b
t |5| ; |KŸx, t ¡I 1 Ÿt  " I2Ÿt ¢|dt
a
b
t |5|M ; |I 1 Ÿt  " I2Ÿt |dt
a
t |5|MŸb " a PI 1 Ÿt  " I2Ÿt P. *
ʳʳʳʳ
"1 "1
Since |5|  M "1 Ÿb " a  , then there exists c such that |5|  c  M Ÿb " a  . Hence, by
"1

(*), we know that


PTŸI 1  Ÿx  " TŸI 2  Ÿx P  +PI 1 Ÿt  " I2Ÿt P
where 0  cMŸb " a  : +  1. So, T is a contraction of C¡a, b¢ and hence has a fixed
b
point I which is a solution of the integral equation IŸx   gŸx   5 ; KŸx, t IŸt dt.
a

7.37 Assume f is continuous on a rectangle Q  ¡a " h, a  h¢ • ¡b " k, b  k¢, where


h  0, k  0.
(a) Let I be a function, continuous on ¡a " h, a  h¢, such that Ÿx, IŸx    Q for all x in
¡a " h, a  h¢. If 0  c t h, prove that I satisfies the differential equation y U  fŸx, y  on
Ÿa " c, a  c  and the initial condition IŸa   b if, and only if, I satisfies the integral
equation
x
IŸx   b  ; fŸt, IŸt  dt on Ÿa " c, a  c .
a
Proof: (´)Since I U Ÿt   fŸt, IŸt   on Ÿa " c, a  c  and IŸa   b, we have,
x  Ÿa " c, a  c 
x
IŸx   IŸa   ; I U Ÿt dt
a
x
 IŸa   ; fŸt, IŸt  dt on Ÿa " c, a  c .
a
(²)Assume
x
IŸx   b  ; fŸt, IŸt  dt on Ÿa " c, a  c ,
a
then
I U Ÿx   fŸt, IŸx   on Ÿa " c, a  c .
(b) Assume that |fŸx, y | t M on Q, where M  0, and let c  min£h, k/M¤. Let S
denote the metric subspace of C¡a " c, a  c¢ consisting of all I such that |IŸx  " b| t Mc
on ¡a " c, a  c¢. Prove that S is closed subspace of C¡a " c, a  c¢ and hence that S is itself
a complete metric space.
Proof: Since C¡a " c, a  c¢ is complete, if we can show that S is closed, then S is
complete. Hence, it remains to show that S is closed.
Given f  S U , then there exists a sequence of functions £f n ¤ such that f n v f under the
sup norm P. P. So, given /  0, there exists a positive integer N such that as n u N, we
have
max |f n Ÿx  " fŸx |  /.
x¡a"c,ac¢

Consider
|fŸx  " b| t |fŸx  " f N Ÿx |  |f N Ÿx  " b|
t PfŸx  " f N Ÿx P  Pf N Ÿx  " bP
 /  Mc
which implies that
|fŸx  " b| t Mc for all x
since / is arbitrary. So, f  S. It means that S is closed.
(c) Prove that the function T defined on S by the equation
x
TŸI Ÿx   b  ; fŸt, IŸt  dt
a
maps S into itself.
Proof: Since
x
|TŸI Ÿx  " b|  ; a fŸt, IŸt  dt
x
t ; a |fŸt, IŸt  |dt
t Ÿx " a M
t Mc
we know that TŸI Ÿx   S. That is, T maps S into itself.
(d) Now assume that f satisfies a Lipschitz condition of the form
|fŸx, y  " fŸx, z | t A|y " z|
for every pair of points Ÿx, y  and Ÿx, z  in Q, where A  0. Prove that T is a contraction of
S if h  1/A. Deduce that for h  1/A the differential equation y U  fŸx, y  has exactly one
solution y  IŸx  on Ÿa " c, a  c  such that IŸa   b.
Proof: Note that h  1/A, there exists 5 such that h  5  1/A. Since
PTŸI 1  Ÿx  " TŸI 2  Ÿx P
x
t ; a |fŸt, I 1 Ÿt   " fŸt, I 2 Ÿt  |dt
x
t A ; |I 1 Ÿt  " I 2 Ÿt |dt by |fŸx, y  " fŸx, z | t A|y " z|
a
t AŸx " a PI 1 Ÿt  " I 2 Ÿt P
t AhPI 1 Ÿt  " I 2 Ÿt P
 +PI 1 Ÿt  " I 2 Ÿt P
where 0  5A : +  1. Hence, T is a contraction of S. It implies that there exists one and
only one I  S such that
x
IŸx   b  ; fŸt, IŸt  dt
a
which implies that
I U Ÿx   fŸx, IŸx  .
That is, the differential equation y U  fŸx, y  has exactly one solution y  IŸx  on
Ÿa " c, a  c  such that IŸa   b.
Supplement on Riemann Integrals
1. The reader should be noted that the metric space ŸRŸ¡a, b¢ , d  is NOT complete,
where
b
dŸf, g   ; a |fŸx  " gŸx |dx.
We do NOT give it a proof. The reader can see the book, Measure and Integral (An
Introduction to Real Analysis) by Richard L. Wheeden and Antoni Zygmund, Ch5.
2. The reader may recall the Mean Value Theorem: Let f be a continuous function on
¡a, b¢. Then
b
; a fŸx dx  fŸx 0  Ÿb " a 
where x 0  ¡a, b¢. In fact, the point x 0 can be chosen to be interior of ¡a, b¢. That is,
x 0  Ÿa, b .
Proof: Let M  sup x¡a,b¢ fŸx , and m  inf x¡a,b¢ fŸx . If M  m, then it is clear. So, we
may assume that M p m as follows. Suppose NOT, it means that x 0  a or b. Note that,
fŸx 1    m t fŸx 0   : r t M  fŸx 2  
by continuity of f on ¡a, b¢. Then we claim that
fŸx 0    m or M.
If NOT, i.e.,
fŸx 1    r  fŸx 2  
it means that there exists a point p  Ÿx 1 , x 2   such that fŸp   r by Intermediate Value
Theorem. It contradicts to p  a or b. So, we have proved the claim. If fŸa   m, then
b b
; a fŸx dx  mŸb " a  ´ 0  ; a ¡fŸx  " m¢dx
which implies that, by fŸx  " m u 0 on ¡a, b¢,
fŸx   m forall x  ¡a, b¢.
So, it is impopssible. Similarly for other cases.
Remark: (1) The reader can give it a try to consider the Riemann-Stieltjes Integral as
follows. Let ) be a continuous and increasing function on ¡a, b¢. If f is continuous on
¡a, b¢, then
b
; a fŸx d)Ÿx   fŸc ¡)Ÿb  " )Ÿa ¢
where c  Ÿa, b .
Note: We do NOT omit the continuity of ) on ¡a, b¢ since
0 if x  0
fŸx   x on ¡0, 1¢; )Ÿx   .
1 if x  Ÿ0, 1¢

(2) The reader can see the textbook, exercise 14.13 pp 404.
Exercise: Show that
=/2
= 
2
;0 dx
1 2
 = .
2
1" 2
sin x

Proof: It is clear by the choice of x 0  Ÿ0, =/2 .


3. Application on Integration by parts for Riemann-integrable function. It is
well-known that
; fŸx dx  xfŸx  " ; xdfŸx . *
ʳʳʳʳ
If fŸx  has the inverse function gŸy   x, then (*) implies that
; fŸx dx  xfŸx  " ; gŸy dy.
For example,
; arcsin xdx  x arcsin x " ; sin ydy.
4. Here is an observation on Series, Differentiation and Integration. We write it as a
table to make the reader think it twice.
Series : Summation by parts Cesaro Sum ?

Differentiation : Ÿfg  U  f U g  fg U Mean Value Theorem Chain Rule

Integration : Integration by parts Mean Value Theorem Change of Variable .

You might also like