Chapter 5: Improper Integrals

1 Generalities
In the last chapter, we have considered the case of integrals of bounded functions on bounded
intervals. The main goal of this chapter is to study integrals of unbounded functions or
integrals of functions on unbounded intervals. As example consider the case of the function
1
f (x) = p on the interval (0; 1] : Clearly the function f is unbounded on (0; 1] but we have
2 x
Z 1
1 p
lim+ p dx = lim 1 " = 1:
"!0 " 2 x "!0+

R1 1
This suggests to set 0
p dx = 1:
2 x
1
Similarly, consedering the function g(x) = on [0; +1), we have
1 + x2
Z A
1
lim dx = lim arctan (A) = :
A!+1 0 1 + x2 A!+1 2
R +1 1
Hence, we set 0
dx = :
1 + x2 2
The above examples show that we can extend the concept of integrability to unbounded
functions or to unbounded intervals.

De…nition 1 Let f : [a; b) ! R be a function. We say that f is locally integrable on [a; b)

if f is integrable on [a; c] for any c 2 (a; b) :

Remark 1 In the above de…nition b may be equals to +1:

Remark 2 Similarly, if f : (a; b] ! R is a function, then we say that f is locally integrable

on (a; b] if f is integrable on [c; b] for any c 2 (a; b) :

In all this chapter we denote by Rloc ([a; b)) the set of all function that are locally inte-
grable on [a; b) :
Rb
De…nition 2 Let f 2 Rloc ([a; b)) : We say that the integral f (x)dx is convergent if
Rt a
limt!b a f (x)dx exists and is …nite. In this case we set
Z b Z t
f (x)dx = lim f (x)dx
a t!b a
Rb
and we say that the integral a f (x)dx is improper (or singular) at b:
Rb
If the integral a f (x)dx is not convergent, we say that it is divergent.

1
 Rb
Remark 3 Similarly, if f 2 Rloc ((a; b]) : We say that the integral f (x)dx is convergent
Rb a
if limt!a+ t f (x)dx exists and is …nite. In this case we set
Z b Z b
f (x)dx = lim+ f (x)dx:
a t!a t

Example 1 Let be a real number and set

Z 1 Z +1
dx dx
I( ) = and J( ) = :
0 x 1 x
We have
Z 1
dx ln (") , if = 1; +1, if 1;
lim+ = lim+ 1 = 1
"!0 " x "!0 1
1 " 1 1 ; if 6= 1 1
; if <1
and Z A
dx ln (A) , if = 1; +1, if 1;
lim = lim+ 1 1 = 1
A!+1 1 x "!0 1 A 1 1 ; if 6= 1 1
; if > 1:
Therefore
I( ) converges , < 1;
J( ) converges , > 1:
The integrals I( ) and J( ) are commonly called Riemann integrals.
Z 1
dt
Example 2 We have p = , Indeed
0 1 t2 2
Z a
dt
lim p = lim arcsin(t) = :
a!1 0 1 t2 a!1 2
Example 3 We have
Z +1 Z A
t
e dt = lim e t dt = lim 1 e A
= 1:
0 A!+1 0 A!+1

Exercise 1 Calculate the following improper integrals.

Z +1 Z +1 Z +1 Z +1
px x dx earctgx
I1 = e dx (p > 0); I2 = xe dx; I3 = ; I4 = dx:
0 0 2 x2 1 0 1 + x2
A
RA px 1 1 e px
1) I1 = limA!+1 0
e dx = limA!+1
= limA!+1 1 e pA = :
p 0 p p
RA x px A
RA x A
2) I2 = limA!+1 0 xe dx = limA!+1 [ xe ]0 + 0 e dx = limA!+1 Ae +
1 e A = 1:
R A dx A 1 1
3) I3 = limA!+1 2 2 = limA!+1 ln ln = ln(3):
x 1 A+1 3
R A earctgx
4) I4 = limA!+1 0 dx = limA!+1 earctg(A) 1 = e =2 1:
1 + x2

2
 Z b
Proposition 1 Let f 2 Rloc ((a; b]) and let c 2 (a; b) : The integral f (x)dx is convergent
Rb a
if and only if c f (x)dx is convergent and we have
Z b Z c Z b
f (x)dx = f (x)dx + f (x)dx:
a a c
Proof. Indeed, for all " 2 (c; b) we have
Z " Z c Z "
f (x)dx = f (x)dx + f (x)dx:
a a c

Remark 4 The above proposition says that Chasles relation holds when an improper integral
converges. Moreover, we understand that the nature of an improper depends on the behavior
of f near b:
n Rb o
Proposition 2 Let E = f 2 Rloc ((a; b]) : a f (x)dx is convergent . Then E is a real
Rb
vector space and the mapping f ! a f (x)dx is a linear form on E: In other words, for all
f; g 2 E and all ; 2 R we have
Z b Z b Z b
( f (x) + g(x)) dx = f (x)dx + g(x)dx:
a a a
Proof. Indeed, for all t 2 (a; b) we have
Z t Z t Z t
( f (x) + g(x)) dx = f (x)dx + g(x)dx:
a a a
Passing to the limits as t ! b , we obtain the desired.
Proposition 3 (Integration by parts) Let f; g : [a; b) ! R be two functions of class C 1 :
Rb Rb
If limt!b f (t)g(t) exists and is …nite, then the integrals a f 0 (t)g(t)dt and a f (t)g 0 (t)dt have
the same nature and we have
Z b Z b
0
f (t)g(t)dt = lim f (t)g(t) f (a)g(a) f (t)g 0 (t)dt:
a t!b a
Proof. Indeed, for all s 2 (a; b) we have
Z s Z s
0
f (t)g(t)dt = f (s)g(s) f (a)g(a) f (t)g 0 (t)dt:
a a
Rs 0 Rs
This shows that lims!b a f (t)g(t)dt exists and is …nite if and only if lims!b a f (t)g 0 (t)dt
is: Moreover, if one of this integrals converges then
Z b Z s
0
f (t)g(t)dt = lim f 0 (t)g(t)dt
a s!b a
Z s
= lim f (s)g(s) f (a)g(a) f (t)g 0 (t)dt
s!b a
Z b
= lim f (s)g(s) f (a)g(a) f (t)g 0 (t)dt:
s!b a

3
Example 4 Because for all A > 1 we have
Z A Z A
sin(x) cos(A) cos(x)
dx = + cos (1) dx
1 x A 1 x2
cos(x) R +1 sin(x) R +1 cos(x)
and limx!+1 = 0, the integrals 1 dx and 1 dx have the same
x x x2
nature.
t
Example 5 Since for any > 0, limt!+1 te = 0; we have
Z +1 Z
t te t
1 +1 1
te dt = lim e t dt = 2 :
0 t!+1 0

Proposition 4 (Change of variables) Let f 2 Rloc ([a; b)) and let ' : [ ; ) ! [a; b) be a
Rb R
bijective function of class C 1 : The integrals a f (x)dx and f ('(t))'0 (t)dt have then the
Rb R
same nature and in the case where one converges then a f (x)dx = f ('(t))'0 (t)dt:
Rc R ' 1 (c)
Proof. Indeed, for all c 2 (a; b) we have a f (x)dx = f ('(t))'0 (t)dt:
Example 6 Setting x = t2 , we get
Z A Z A2
2 sin(x)
sin(t )dt = p dx for all A > 0:
1 1 2 x
R +1 R +1
Hence, the integrals 1
sin(t2 )dt and 1 sin(x)
p dx have the same nature.
2 x
Rb
Example 7 Let I ( ) = a (b dtt) . By the change of variables x = b t, we obtain I( ) =
R b a dx
0
: Hence, we conclude that I( ) is convergent if and only if < 1:
x
Theorem 5 (Cauchy criterion) Let f 2 Rloc ([a; b)) : The following assertions are equiv-
alents.
Rb
1. a f (t)dt is convergent.
2. For all > 0 there exists c 2 (a; b) such that for all t1 ; t2 2 (a; b), t1 ; t2 > c implies
R t2
t1
f (t)dt < :
Rb Rx
Proof. Since a f (t)dt is convergent if and only if limx!+1 a f (t)dt = l 2 R; applying
Theorem 28 of Chapter Real functions: Limits, we get the desired.
Remark 5 An integral can be improper at many points. In this we divide the original inte-
gral into many integrals and then we investigate the nature of each integral. ZThe original inte-
+1
dx
gral converges if all the integrals are convergents. As example, consider I = p :
1 jxj (x2 + 1)
Since I is improper at 1; 0 and +1, we write
Z1 Z0 Z1 Z
+1
dx dx dx dx
I= p + p + p + p :
jxj (x2 + 1) jxj (x2 + 1) jxj (x2 + 1) jxj (x2 + 1)
1 1 0 1

Since all the integrals are convergents we conclude that I is convergent.

4
2 Convergence of integrals of positive functions
Theorem 6 Let f 2 Rloc ([a; b)) such that f 0 on (c; b) for some c 2 (a; b) : If the function
Rx Rb
x ! a f (t)dt is upper bounded, then a f (t)dt is convergent.
Rb
Proof. By Proposition
Rx 1 we have to show that c
f (t)dt is convergent. Since f 0 on (c; b)
the function x ! c f (t)dt is nondecreasing. Moreover, for all x 2 (c; b) we have
Z x Z x Z c Z x Z c
f (t)dt = f (t)dt f (t)dt sup f (t)dt f (t)dt:
c a a x2(a;b) a a

Rx Rb
This shows that limx!b c
f (t)dt exists and is …nite and so c
f (t)dt is convergent.

Theorem 7 (Comparaison) Let f; g 2 Rloc ([a; b)) such that 0 f g on (c; b) for some
c 2 (a; b) :
Rb Rb
1. Ifg(t)dt is convergent, then a f (t)dt is convergent.
a
Rb Rb
2. If a f (t)dt is divergent, then a g(t)dt is divergent.
Rx
Proof. We have x ! c f (t)dt is increasing on (c; b) and for all x 2 (c; b)
Z x Z x Z b
f (t)dt sup g(t)dt = g(t)dt:
c x2(c;b) c c

Rx Rb
This shows that limx!b c f (t)dt exists and is …nite and so c f (t)dt is convergent. By
Rb
Proposition 1, a f (t)dt is convergent.

R +1 2 + cos(t) R +1 1 2 + cos(t) 1 1
Example 8 Let I = 1 2
dt and J = e 2
dt: Since ; 2
1+t t ln(t) 1 + t2 2
t t ln t
1 R +1 1
2
and 1 dt is convergent, we conclude that I and J are convergent.
t t2
Corollary 8 Let f; g 2 Rloc ([a; b)) such that f; g 0 on [c; b) for some c 2 (a; b) and
f (x)
suppose that f = (g) at b (i.e. limx!b = 0). Then
g(x)
Rb Rb
a) f (t)dt is convergent if a g(t)dt is,
a
Rb Rb
b) a g(t)dt is divergent if a f (t)dt is.

f (x)
Proof. Let > 0 small enough. From limx!b = 0, we deduce existence of > 0 such
g(x)
f (x)
that for all x 2 (b ; b). Therefore, for all x 2 (c ; b) where c = max(b ; c)
g(x)
we have f (x) g(x): Hence, the corollary is obtained by applying the above theorem.

5
Example 9 Set for ; 2R
Z +1
dt
B( ; )= Bertrand integral.
2 t ln (t)

We distinguish three cases.

1 1
a) > 1: Let > 0 be such that > 1. Set f (t) = and g(t) = : Hence, we
t ln (t) t
have
f (t) 1
lim = lim = 0:
t!+1 g(t) t!+1 t ln (t)

Applying Corollary 8, we obtain B ( ; ) is convergent.

1 1
b) < 1: Let > 0 be such that + < 1. Set f (t) = +
and g(t) = : Hence, we
t t ln (t)
have
f (t) ln (t)
lim = lim = 0:
t!+1 g(t) t!+1 t
Applying Corollary 8, we obtain B ( ; ) is divergent.

c) = 1: In this case for all A > 2, we have

Z A
dt ln(ln (A)) ln (ln(2)) ; if = 1;
=
2 t ln (t) 1
1
ln1 (A) ln1 (2) ; if 6= 1:

Hence, we obtain (
Z A
dt +1; if 1;
lim = ln1 (2)
A!+1 2 t ln (t) 1
; if > 1;
showing that B (1; ) converges if and only if > 1:

As a conclusion, B ( ; ) converges if and only if > 1 or = 1 and > 1:

Theorem 9 Let f; g 2 Rloc ([a; b)) such that 0 f g on (c; b) for some c 2 (a; b) : If
f (t) Rb Rb
limt!b = 1 (i.e. f g near b), then a f (t)dt and a g(t)dt have the same nature.
g(t)

Proof. Let 2 (0; 1) : There exists > 0 such that jf (t) g(t)j jg(t)j for all t 2
(b ; b) : Therefore, for all t 2 (c ; b) ; (1 ) g(t) f (t) (1 + ) g(t), where c =
max(c; b ): Applying Theorem 6, we get the desired.
p
1 1 sin x 1
Example 10 Since p 3=2
near +1 and p near 0, we conclude
3
1+ x +1 x x x
Z +1 Z 1p
dx sin x
that the integrals p and dx are convergent.
0 1+ x +13
0 x

6
 Z +1
x
Example 11 Let ; be positive real numbers and consider the integral I( ; ) = x e dx:
0
1
Since limx!+1 x +2 e x = 0, there exists A > 0 such that x e x for all x > A: Be-
Z +1 x2
dx
cause is convergent, I( ; ) is convergent for all ; > 0:
A x2

Corollary
R +1 10 Let f 2 Rloc ([a; +1)) and suppose that limt!+1 t f (t) = l > 0. Then
a
f (t)dt converges if and only if > 1:

l
Proof. The limit in the corollary means that f (t) near +1: Hence, the corollary is
t
obtained by a simple application of Theorem 9.

Exercise 2 Discuss the nature of the following integrals.

Z +1 Z 0 Z +1
dx dx ln(x + 1)
I1 = p ; I 2 = p ; I3 = dx;
1 x2 1 + x 1 x
2 1+x
0 x
Z +1 p Z 1 Z +1 p Z +1 p
te t 1 ln t
I4 = 2
dt; I5 = cos2 ( )dt; I6 = e ln t dt; I7 = p dt:
0 1+t 0 t 1 1 (1 t2 ) t

1 1 R +1 dx
1) I1 is improper at +1 and we have p at +1; with 1
converges.
x2 1+x x3=2 x3=2
Hence, I1 is convergent.
Z 1 2) I2 is improper at 1 and by the chnage Z 1of variables y = x + 1, we get I2 =
dy 1 1 dy
2 p with 2p p at 0 and p converges. Hence, I2 is convergent.
0 (y 1) y (y 1) y y 0 y
ln(x + 1)
3) I3 is improper at +1 and it is not improper at 0: Indeed, we have limx!0 =
x
ln(x + 1) 1
1 (the integrated function is bounded at 0). Since at +1, and
x x(ln x) 1
R +1 1
2
dx is divergent (Bertrand integral with = 1 and = 1), I3 is divergent.
x(ln x) 1 p p Z +1 pt Z +1
te t e t e
4) I4 is improper at +1 and we have p and p dt = 2 e x dx =
1 + t2 t 0 t 0
2: Therefore, I4 is convergent.
1
5) I5 is not improper since cos2 1:
t
p 1 1 1
6) I6 is improper at +1 and we have e ln t = p ln t
= near +1: Hence, I6 is
Z +1 e ln t e t
dt
divergent since is.
1 t p
Z 2
ln t
7) I7 is improper at 1 and at +1: Let I7 = I70 + I7+1 where I71 = p dt and
p p 1 (1
p t2 ) t
Z +1 Z 1
+1 ln t 1 ln(1 + x) ln(1 + x)
I7 = p : We have I7 = p dx with p
2 (1 t2 ) t 0 x (x + 2) 1+x x (x + 2) 1 + x

7
 p
1 1 ln t 1
p at 0: So I7 is convergent. We have also p at +1: Since the
2 x (1 t2 ) t t5=2 (ln t)1=2
Z +1
dt
Bertrand integral 1=2
converges, I7+1 converges. All the above shows that I7
5=2
t (ln t)
is convergent.

3 Improper integrals of changing sign functions

The main goal of this section is the study of the convergence of improperZ +1integrals of functions
sin(x)
changing in…nitely many times near the singularity. As an example dx:
1 x
Z b Z b
De…nition 3 Let f 2 Rloc ([a; b)) : We say that f (t)dt converges absolutely if jf (t)j dt
a a
is convergent.
Rb Rb
Proposition 11 Let f 2 Rloc ([a; b)) : If a
f (t)dt converges absolutely, then a
f (t)dt is
convergent.
Proof. Let > 0: By R y Cauchy criterion, there is c 2 (a; b) such that for all x; y 2 (a; b) ;
x; y 2 (c; b) implies x jf (t)j dt : Hence, we have for all x; y 2 (a; b) ; x; y 2 (c; b) implies
Z y Z y
f (t)dt jf (t)j dt :
x x
Rb
This shows that a
f (t)dt is convergent.
Z +1 Z +1
sin(x) sin(x) 1 1
Example 12 Let I = dx. We have and dx is convergent.
1 x2 x2 x2 1 x2
Hence, I is absolutely convergent and so, I is convergent.
Z +1
Example 13 Let I ( ; p) = sin( t)e pt dt with ; p > 0. We have jsin( t)e pt j e pt
Z +1 0
1
and e pt dt = : Hence, I ( ; p) is absolutely convergent and so, I ( ; p) is convergent.
0 p
Lemma 12 (The second mean theorem) Let f; g : [a; b] ! R be two continuous func-
tions and suppose that g is positive and nonincreasing. Then there exists c 2 [a; b] such
that Z b Z c
f (t)g(t)dt = g(a) f (t)dt:
a a

Proof. Let (xi )i=n

i=0 be the uniform subdivision of the interval [a; b] : We have then
Z b i=n Z xi
X
f (t)g(t)dt = f (t)g(t)dt
a i=1 xi 1

X
i=n Z xi i=n Z
X xi
= g(xi 1 ) f (t)dt + f (t) (g(t) g(xi 1 )) dt:
i=1 xi 1 i=1 xi 1

8
Set
X
i=n Z xi i=n Z
X xi
Un = g(xi 1 ) f (t)dt and vn = f (t) (g(t) g(xi 1 )) dt:
i=1 xi 1 i=1 xi 1

We have
i=n Z
X xi i=n Z
X xi
jVn j jf (t)j (g(xi 1 ) g(t)) dt sup jf (t)j (g(xi 1 ) g(xi )) dt
i=1 xi 1 t2[a;b] i=1 xi 1

b a
sup jf (t)j (g(a) g(b)) :
n t2[a;b]

This shows thatR limn!+1 Vn = 0:

x
Let F (x) = a f (t)dt: Also we have

X
i=n Z xi X
i=n
Un = g(xi 1 ) f (t)dt = g(xi 1 ) (F (xi ) F (xi 1 ))
i=1 xi 1 i=1
= g(a)F (x1 ) + g(x1 )F (x2 ) + g(x2 )F (x3 ) + : : : + g(xn 1 )F (b)
g(a)F (a) g(x1 )F (x1 ) g(x2 )F (x2 ) : : : g(xn 1 )F (xn 1 )
= F (x1 ) (g(x0 ) g(x1 )) + F (x2 ) (g(x1 ) g(x2 )) + : : : + F (xn 1 ) (g(xn 2 ) g(xn 1 ))
+g(xn 1 )F (b)
X1
i=n
= g(xn 1 )F (b) + F (xi ) (F (xi 1 ) F (xi )) :
i=1

Set M = maxt2[a;b] F (t) = F (t+ ) and m = mint2[a;b] F (t) = F (t ) where t+ ; t 2 [a; b] : We

obtain then from the above calculations

g(a)F (t ) = mg(a) Un M g(a) = g(a)F (t+ ):

Since the function F is continuous the intermediate

Rc value theorem guarantees the existence
of cn 2 [a; b] such that Un = g(a)F (cn ) = g(a) a n f (t)dt:
Because (cn ) admits a convergent subsequence (cnk ) ; by letting k ! +1 in
Z b Z cnk
f (t)g(t)dt = g(a) f (t)dt + Vnk
a a

we obtain the desired.

Theorem 13 (Abel’s rule) Let f; g : [a; b) ! R be two continuous functions and suppose

i) f is nonincreasing with limx!b f (x) = 0 and

Rx
ii) there exists M > 0 such that a g(t)dt M for all x 2 (a; b) :
Rb
Then a
f (t)g(t)dt converges.

9
Proof. Let > 0: Since limx!b f (x) = 0, there is c 2 (a; b) such that for all t 2 (c; b) ;
jf (t)j : Let t1 ; t2 2 (c; b) with t1 < t2 : By the second mean theorem, there is 2 (t1 ; t2 )
2M
such that
Z t2 Z Z Z t1
f (t)g(t)dt = f (t1 ) g(t)dt f (t1 ) g(t)dt + f (t1 ) g(t)dt
t1 t1 a a
Z Z t1
jf (t1 )j g(t)dt + jf (t1 )j g(t)dt
a a

M +M = :
2M 2M
Rb
By Cauchy criterion a
f (t)g(t)dt converges.

Example 14 Let ; be positive real numbers and let the integrals

Z +1 Z +1
cos( t) sin( t)
A1 ( ; ) = dt and A2 ( ; ) = dt
1 t 1 t
called Abel’s integrals. Applying Abel’s rule with f (t) = t1 and g(t) = cos( t) (resp. g(t) =
sin( t)), we obtain that A1 ( ; ) (resp. A2 ( ; )) is convergent for all ; > 0:

We cannot use comparison for improper integrals of functions changing sign. There is
no
Z +1result in this sense. But we can use LDs as follows. Consider the improper integral
sin x
p dx: We have
1 x + sin x
sin x sin x 1
p = p :
x + sin x px
x 1 + sinx

sin x sin x X
Set X = p : Since limx!+1 p = 0, we use the LD at 0 of the function h(X) = :
x x 1+X
Since
1
h(X) = = X X2 + X3 + X3 ;
1+X
we obtain
sin x sin x sin2 x sin3 x sin3 x
p = p + p + p
x + sin x x x x x x x
sin x 1 1 1 cos (2x) sin3 x sin3 x
= p + + p + p :
x 2x 2 x x x x x
Z +1 Z +1 Z +1
sin x cos (2x) cos (2x)
By Abel’s rule, the integrals p dx; dx and dx are con-
1 x 1 x 1 Z x
+1
sin3 x 1 sin3 x 1 dx
vergent. Moreover, since p p ; p p and p is conver-
Z +1 x x Z x x x x x x 1 x x
+1
sin3 x sin3 x
gent, we have p dx and p dx are convergent. Hence, we conclude
1 x x 1 x x

10
 Z +1 Z +1 Z +1
sin x dx sin x
that p dx and have the same nature, that is p dx is
1 x + sin x 1 x 1 x + sin x
divergent. Z +1 Z +1
sin x sin x sin x sin x
Notice that p p near 0 but p dx and p dx have
x + sin x x 1 x + sin x 1 x
opposite nature.
Exercise 3 Discuss following the real parameters the nature of the following integrals.
Z 1 Z +1
dx ax
I( ; ) = ; I(a) = dx; a > 0;
0 x (1 x) 1 1 + a2x
Z +1 t Z +1 Z +1
e 1 t sin t t ln t
J1 ( ) = dt; J2 ( ) = dt; J3 ( ) = dt;
0 t 0 t 0 (1 + t2 )
Z +1
sin x
J4 ( ) = ln 1 + dx with > 0::
2 x

1) We have
Z 1=2 Z 1
dx dx
I( ; )= +
0 x (1 x) 1=2 x (1 x)
1 1 1 1
with near 0 and near 1: We conclude that I ( ; )
x (1 x) x x (1 x) (1 x)
converges if and only if < 1 and < 1:
ax x x ln(1=a) ax 1 ax
2) We have a = e ; if a 2 (0; 1) ; = 2
if a = 1 and
1 + a2x 1 + a2x 1 + a2x
x x ln a
a =e ; if a 2 (1; +1) : Hence, I(a) converges if and only if a 6= 1:
e t 1 1 e t 1 1
3) We have 1
at 0 and at +1. Hence, J1 ( ) converges if
t t t t
and only if 2 (1; 2) :
t sin t 1 t sin t 1
4) We have 3
at 0 and at +1. Therefore, J2 ( )
t 6t t t 1
converges if and only if 2 (2; 4) :
t ln t 1
5) J3 ( ) is improper at +1 and we have at +1: We conclude
2
(1 + t ) t2 1 (ln t) 1
that J3 ( ) converges if and only if > 1:
6) We have for > 1=2;
sin x sin x sin2 x sin2 x
ln 1 + = +
x x 2x2 x2
Z +1
sin2 x 1 sin2 x 1 R +1 dx sin x
with ; 0 ; 1 x 2 is convergent. Since dx is
x2 x2 x2 x2 1 x
convergent, J4 ( ) converges for > 1=2:
1 1
Suppose now that < n 2; In this case we have
n+1 2
sin x X 1 sin2k+1 (x) X 1 sin2k x sinn+1 x
ln 1 + = + :
x 2k+1 n+1
2k + 1 x(2k+1) 2k n+1
2k x2k x(n+1)

11
 Z +1
sin2k+1 (x)
Since by Abel’s criterion dx converges for all k;
1 x(2k+1)
X 1 sin2k x 1 sin2 x 1 1 1 cos (2x)
=
2k n+1
2k x2k 2 x 2 4 x2 4 x2
Z +1 Z +1
dx cos(2x)
with diverges and dx converges, we conclude that J4 ( ) diverges for
1 x2 1 x2
1=2:

12