Você está na página 1de 28

In the last chapter we introduced the notion of complex integration.

An important part of
our development was the statement of Cauchy’s integral formula. In this chapter we’re
going to extend this technique using reside theory. This is an elegant formulation that not
only allows you to calculate many complex integrals, but also gives you a trick you can
use to calculate many real integrals. We begin by stating some theorems related to
Cauchy’s integral formula.

Theorems Related To Cauchy’s Integral Formula


We begin the chapter by writing down another form of Cauchy’s integral formula. First
let’s write (6.16) in the following way:

1 f  z
f  a  
Ñ dz
2 i  z  a

Now let’s take the derivative of this expression, with respect to a. This gives:

d  1 f  z  1 d  f  z  1 f  z
f  a   
Ñ dz   
Ñ  dz  
Ñ dz
da  2 i  z  a  2 i da  z  a  2 i   z  a  2
 

We can repeat this process multiple times. That is, take the derivative again. Each time
the exponent, which is negative, cancels out the minus sign we pick up by computing the
derivative with respect to a of z  a . For example the second derivative is:

d  1 f  z  1 f  z
f   a     z  a  2   i Ñ   z  a  3 dz
da  2 i Ñ
 dz  

This process can be continued. For an arbitrary n, we obtain a second Cauchy Integral
formula for the nth derivative of f  a  :

n! f  z
f  n  a   
Ñ dz
2 i   z  a  n 1

for n  1, 2,3,... . There are two facts you should come away with from the Cauchy
Integral formulas:

 If a function f  z  is known on a simple closed curve  , then that function is


known at all points inside  . Moreover, all of the functions derivatives can be
found inside  .
• If a function is analytic in a simply connected region of the complex plane, and
hence has a first derivative, all of its higher derivatives exist in that simply
connected region.

Now we turn to a statement known as Cauchy’s inequality. This statement is related to ,


which gives us an expression we can use to calculate the derivative of an analytic
function in a simply connected region. Consider a circle of radius r which has the point
z  a at its center, and suppose that f  z  is analytic on the circle and inside the circle.
Let M be a positive constant such that f  z   M in the region z  a  r . Then:

Mn !
f  n  a  
rn

The next theorem, which is due to Liouville, tells us that an entire function cannot be
bounded unless it is a constant. This statement is called Liouville’s theorem but it was first
proved by Cauchy. So maybe we should call it the Cauchy-Liouville theorem. In any
case, it simply says that if f  z  is analytic and bounded in the entire complex plane, i.e. t
f  z   M for some constant M, then f  z  is a constant.

Liouville’s theorem implies the Fundamental Theorem of Algebra. Consider a polynomial


with degree n  1 and coefficient an  0 :

P  z   a0  a1 z  a2 z 2  L  an z n

The fundamental theorem of algebra tells us that every polynomial P  z  has at least one
root. The proof follows from Liouville’s theorem and a proof by contradiction. Suppose
that instead P  z   0 for all z. Then

1
f  z 
P z

is analytic throughout the complex plane and is bounded outside some circle z  r .
Moreover, the assumption that P  z   0 implies that f  1/ P is also bounded for z  r .
Hence 1/ P  z  is bounded in the entire complex plane. Using Liouville’s theorem,
1/ P  z  must be a constant. This is a contradiction, since
P  z   a0  a1 z  a2 z 2  L  an z n is clearly not constant. Therefore P  z  must have at
least one root such that P  z   a0  a1 z  a2 z  L  an z  0 is satisfied.
2 n
Next we state the Maximum modulus theorem and the Minimum modulus theorem. The
maximum modulus theorem tells us the following. Let f  z  be a complex valued
function which is analytic inside and on a simple closed curve  . If f  z  is not a
constant, then the maximum value of f  z  is found on the curve  .

Now we state the minimum modulus theorem. Assume once again that f  z  is a complex
valued function which is analytic inside and on a simple closed curve  . If f  z   0
inside  , then f  z  assumes its minimum value on the curve  .

The next theorem is the Deformation of path theorem. Consider a domain D in the
complex plane, and two curves in D we call  1 and  2 . We suppose that
 1 is larger than or lies outside of  2 , and that  1 can be deformed into  2 without leaving
the domain D (that is, we can shrink the first curve down to the second one without
crossing any holes or discontinuities in the domain-see Figure 7-1). If f  z  is analytic in
D then

i f  z  dz  i f  z  dz
1 2

1
Can have hold inside
second curve.

2 If first curve has to


cross hole to deform
into second curve,
theorem does not
work.

Figure 7-1: A graphic illustration of the deformation of path theorem.

Next, we state Gauss’ mean value theorem. Consider a circle  of radius r centered at the
point a. Let f  z  be a function which is analytic on and inside  . The mean value of
f  z  on  is given by f  a  :
1
f  a  rei  d
2
f  a  
2 0

Once again, let f  z  be a function which is analytic on and inside a simple, closed curve
 . Now assume that f  z  has a finite number of poles inside  . If M is the number of
zeros of f  z  inside  and N is the number of poles inside  , the argument theorem
states that:

1 f  z

Ñ
2 i  f  z 
dz  M  N

Next is a statement of Rouche’s theorem. Let f  z  and g  z  be two functions which are
analytic inside and on a simple closed curve  . If g  z   f  z  on  , then
f  z   g  z  and f  z  have the same number of zeros inside  .

Finally, we end our whirlwind tour of theorems and results related to the Cauchy integral
formula with a statement of Poisson’s integral formula for a circle. This expresses the
value of a harmonic function inside of a circle in terms of its values on the boundary. Let
f  z  be analytic inside and on the circle  , centered at the origin with radius R. Suppose
that z  rei is any point inside  . Then

f  z 
1 2 R 2
 r 2  f  Rei 
2 0 R 2  2 Rr cos       r 2
d

This example illustrates the solution of Laplace’s equation on a disk. First, show that


u  r ,    a0   an r n cos n  bn r n sin n
n 1

is the solution of Laplace’s equation on the disc 0  r  1 with Dirichlet boundary


conditions:

1   u  1  2u
 r    0, 0  r  1, 0    2
r r  r  r 2 r 2

u  1,   f    , u  r ,   bounded as r  0
Show the coefficients in the series expansion are given by:

1 2 1 2 1 2
a0   f    d , an   f    cos n d , bn   f    sin n d
2 0  0  0

Use the result to deduce Poisson’s integral formula for a circle of radius 1:

1 2 1 r2
u  r ,    f    d
2 0 1  2r cos       r 2

We try separation of variables. Let u  r ,    R  r      . Then it follows that:

u R  2u  2 R  2u  2
   ,    ,  R r
r r r 2 r 2  2  2

The statement of the problem tells us that:

 2u 1 u 1  2u
  0
r 2 r r r 2  2
Hence:

2 R 1 R 1  2
0   
     2   2
  R r
r 2 r r r 

We divide every term in this expression by u  r ,    R  r      . This allows us to write:

r 2  2 R r R 1  2
 
R r 2 R r   2

The left hand side and the right hand side are functions of r only and  only respectively.
Therefore they can be equal only if they are both equal to a constant. We call this constant
n 2 . Then we have the equation in  :

1 d 2 d 2
  n ,
2
 n2  0
 d 2
d 2

Note that partial derivatives can be replaced by ordinary derivatives at this point, since
each equation involves one variable only. This familiar differential equation has solution
given by:

     an cos n  bn sin n
Now, turning to the equation in r, we have:

r 2 d 2 R r dR 2
2 d R dR
2
  n 2
,  r 2
r  n2 R  0
R dr R dr dr dr

You should also be familiar with this equation from the study of ordinary differential
equations. It has solution:

R  r   cn r n  c n r  n

The total solution, by assumption is the product of both solutions, i.e.


u  r ,    R  r      . So we have:

u  r ,     cn r n  c n r  n   an cos n  bn sin n 

The condition that u  r ,   is bounded as r  0 imposes a requirement that the constant


c n  0 since:

c n
  as r  0
rn

Therefore, we take u  r ,     cn r   an cos n  bn sin n  . We can just absorb the


n

constant cn into the other constants, and still designate them by the same letters. Then:

u  r ,    r n  an cos n  bn sin n 

The most general solution is a superposition of such solutions which ranges over all
possible values of n. Therefore we write:

 
u  r ,     r n  an cos n  bn sin n   a0   r n  an cos n  bn sin n 
n 0 n 1
To proceed, the following orthogonality integrals are useful:

2   mn for n  0

0
sin m sin n d  
 0 for n  0

2   mn for n  0
0
cos m cos n d  
 2 mn for n  0
2
 0
sin m cos n d  0

Here,  mn  1 for m  n, 0 otherwise is the Kronecker delta function. Now we apply the
boundary condition u  1,   f    for 0    2 :


f     a0   r n  an cos n  bn sin n 
n 1

Multiply through this expression by sin m and integrate. We obtain:

 
2 2  2 2
 f    sin m d  a0  sin m d   an cos n sin m d   bn sin n sin m d
0 0 0 0
n 1

 
 2 
 bn sin n sin m d  bn mn   bm
0
n 1 n 1
where - were used. We conclude that:

1 2
bn   f    sin n d
 0

Now we return to , and multiply by cos m and integrate. This time:

 
2 2  2 2
 f    cos m d  a0  cos m d   an cos n cos m d   bn sin n cos m d
0 0 0 0
n 1

 
 2 
 an sin n cos m d  an mn   am
0
n 1 n 1
Hence:

1 2
am   f    cos m d
 0

To obtain the constant a0 , we integrate without first multiplying by any trig functions, i.e.
 
2 2  2 2
 f    d  a0  d   an cos n d   bn sin n d
0 0 0 0
n 1

 a0 2

1 2
 a0   f    d
2 0

This should be obvious since:

2 1 2
 0
cos n d 
n
sin n
0
0

2 1 2
 0
sin n d   cos n
n 0
0

Now we are in a position to derive Poisson’s formula. We have:

   

1 2  1 2 1 2 
u  r ,    f    d   r n  f    cos n d cos n  f    sin n d sin n 
2 0
n 1   0  0

We can move the summation inside the integrals:

1 2 2 1 
 2 1 

u  r ,  
2 0
f    d  
0
f 
 

n 1
r n cos n cos n d   f    
 0 
r
n 1
n
sin n sin n d

1 2   

  d f     1  2 r n cos n cos n  2 r n sin n sin n 
2 0
 n 1 n 1 

1 2  

  d f     1  2 r n  cos n cos n  sin n sin n  
2 0
 n 1 

Now recall that:

cos n cos n  sin n sin n  cos  n      

Its also true that:



1 r2
1  2 r n cos  n       
n 1 1  2r cos       r 2

So we arrive at the Poisson formula for a disc of radius 1:

1 2 1 r2
u  r ,    f    d
2 0 1  2r cos       r 2

This tells us that the value of a harmonic function at a point inside the circle is the
average of the boundary values of the circle.

The Cauchy Integral Formula as a Sampling Function


The Dirac delta function has two important properties. First if we integrate over the
entire real line then the result is unity:


   x  dx  1


Second, it acts as a sampling function-that is, it picks out the value of a real function
f  x  at a point:


 f  x    x  a  dx  f  a 


In complex analysis, the function 1/ z plays an analogous role. It has a singularity at


z  0 , and:

1 dz  0 if 0 is not in the interior of 



Ñ 
2 i  z  1 if 0 is in the interior of 

It also acts as a sampling function for analytic functions f  z  in that:

1 f  z  dz
f  a  
Ñ
2 i  z  a

Some Properties of Analytic Functions


Now we are going to lay some more groundwork before we state the residue theorem. In
this section, we consider some properties of analytic functions.
Suppose that a function f  z  is analytic inside a disc centered at a point a of radius r:
z  a  r . Then f  z  has a power series expansion given by:


f  z    an  z  a 
n

n0

The coefficients of the expansion can be calculated using the Cauchy integral formula :

f  n  a 
an 
n!

Note the following result:

 0 if m  1
  z  a
m
dz  
  ln  z  a  if m  1

Hence:

 f  z  dz   a   z  a 
n
n dz
 n0 

since n is never equal to -1.

f(z)
Consider the punctured disc of radius r centered at the point a. We denote this by writing
0  z  a  r . If f  z  is analytic in this region, it is analytic inside the disc but not at the
point a. In this case, the function has a Laurent expansion:


f  z   a  z  a
n
n
n 

As stated in chapter 5, we can classify the points at which the function blows up or goes
to zero. A removable singularity is a point a at which the function appears to be
undefined, but it can be shown by writing down the Laurent expansion that in fact the
function is analytic at a. In this case the Laurent expansion assumes the form


f  z    an  z  a 
n

nk

where k  0 . Then it turns out the point z  a is a zero of order k.


On the other hand, suppose that the series expansion retains terms with n  0 :


f  z   a  z  a
n
n
n  k

Then we say that the point z  a is a pole of order k. Simply put, a pole is a point that
1
behaves like the point z  0 for g  z   . That is, as z  a , then f  z    . A function
z
might have multiple poles. For example, in Figure 7-2 we illustrate the poles of the
modulus of the gamma function,   z  , which are points where the function blows up.

Figure 7-2: When the real part of z is negative, the modulus of the gamma function blows
to infinity at several points. These points are the poles of the function.

A Laurent series expansion of this type can be split into two parts:

 1 
f  z   an  z  a    an  z  a    an  z  a   F  G
n n n

n  k n  k n 0

The second series, which we have denoted by G, looks like a plain old Taylor expansion.
The other series, which we have denoted by F, is called the principal part and it includes
the singularities (the real ones-the poles) of the function.

sin z
Is the point z  0 a removable singularity of f  z   ?
z
At first glance, the behavior of the function at z  0 can’t really be determined. To see
what’s going on we expand the sin function in Taylor:

sin z 1  1 1 
f  z    z  z3  z5  L 
z z 3! 5! 

1 2 1 4
1 z  z L
3! 5!

From this expression, its easy to see that:

sin z 1 1
lim f  z   lim  lim1  z 2  z 4  L  1
z 0 z 0 z z 0 3! 5!

Therefore, the point z  0 is a zero of order 1.

ez
Describe the nature of the singularities of f  z   .
z

We follow the same procedure used in Example 7-2. First expand in Taylor:

ez 1  z 2 z3  1 z z2
f  z    1 z   L  1  L
z z 2! 3!  z 2 6

The principal part of this series expansion is given by:

1
z

It follows that the point z  0 is a pole of order 1.

sin z
Is the point z  0 a removable singularity of f  z   ?
z4

Contrast this solution with that found in Example 7-2. Expanding in Taylor we find:
sin z 1  1 1 1 
f  z  4
 4  z  z3  z5  z 7 L 
z z  3! 5! 7! 

1 1 1 1
 3
  z  z3 L
z 6 z 5! 7!

This time, the singularity cannot be removed. So the point z  0 is a pole. The principal
part in this series expansion is:

1 1

z3 6z

The leading power (most negative power) in the expansion gives the order of the pole.
Hence z  0 is a pole of order 3.

Next we consider the essential singularity. In this case the Laurent series expansion of the
function includes a principal part that is non-terminating. That is, all terms out to minus
infinity are included in the Laurent expansion with negative n, i.e. there are no non-zero
terms in the expansion for n  0 :


f  z   a  z  a
n
n
n 

1
Describe the nature of the singularity at z  0 for f  z   e z .

This function is the classic example used to illustrate an essential singularity. We just
write down the series expansion:
1
f  z  e z

2 3 4 5
1 1 1 1 1 1  1 1  1
 1             L
z 2 z 6 z 4!  z  5!  z 

1 2 1 3 1 4 1 5
 1  z 1  z  z  z  z L
2 6 4! 5!

This series expansion has a non-terminating principal part. Therefore z  0 is an essential


singularity.
The Residue Theorem
Now we’re in a position where we can describe one of the central results of complex
analysis, the residue theorem. We consider a function f  z  in a region enclosed by a
curve  that includes isolated singularities at the points z1 , z2 ,..., zk . The function is
analytic everywhere on the curve and inside it except at the singularities. This is
illustrated in Figure 7-3.

 zk

 z1
 z0

Figure 7-3: A function f  z  is analytic in a certain region enclosed by a curve, except at


a set of isolated singularities.

We can use the deformation of path theorem to shrink the curve down. In fact, we can
shrink it down into isolated curves enclosing each singularity. This is shown in figure 7-4.

 zk
k
 z1  2
 z0
1
Figure 7-4: If the region is simply connected, we can apply the deformation of path
theorem to shrink the curve down, until we have circles around each isolated singular
point.

After application of the deformation of path theorem, the integral is broken up into a sum
of integrals about each singular point:

 f  z  dz    f  z  dz
 j 1  j

This expression can be written in terms of the Laurent expansion. Note that there will be
a series expansion (which is local) about each singular point:

 
f  z  dz   anj  z  z j  dz   anj   z  z j  dz  aj1 2 i
n n

j  j n  n  j

j
We call the coefficient in the expansion a1 the residue. Summing over all of the integrals
for each singular point, we get the residue theorem. This states that the integral is
proportional to the sum of the residues:

 f  z  dz  2 i residues
Ñ  j 1

Residues are computed by finding the limit of the function f  z  as z approaches each
singularity. This is done for a singularity at z  a as follows:

1 d k 1 
 z  a  f  z  
k
residue  lim
z  a  k  1 ! dz k 1 

where k is the order of the singularity.

Compute the integral

5z  2
 z  z  2  dz
Ñ

where  is a circle of radius r  3 centered at the origin.

The singularities of this function are readily identified to be located at z  0, 2 . Both


singularities are enclosed by the curve, since z  3 in both cases. To find each residue,
we compute the limit of the function for each singularity. The residue corresponding to
is:

5z  2 5 z  2 2
lim z  lim  1
z 0 z  z  2 z  0  z  2  2
The residue corresponding to the singularity at z  2 is:

5z  2 5z  2 8
lim  z  2   lim  4
z 2 z  z  2 z  2 z 2

Therefore using the integral evaluates to:

5z  2
 z  z  2  dz  2 i  residues  2 i  1  4   10 i
Ñ

cosh z
Compute the integral of 
 z3
dz where  is the unit circle centered about the origin.

The function has a singularity at z  0 of 3d order. Using:

n! f  z
f  n  a   
Ñ dz
2 i   z  a  n 1

We have:

cosh z 2 i d 2
 z3
Ñ dz   cosh z  z 0   i  cosh z  z 0  i
2! dz 2

Evaluation of Real, Definite Integrals


One of the most powerful applications of the residue theorem is in the evaluation of
definite integrals of functions of a real variable. We start by considering:

2
 f  cos  ,sin   d
0

Now write the complex variable z in polar form on the unit circle, that is let z  ei .
Notice that:
1 i
dz  iei d ,  d  dz   dz
iz z

As  increases from 0 to 2 , one sees that the complex variable z moves around the unit
circle in a counter clockwise direction. Using Euler’s formula, we can also rewrite cos 
and sin  in terms of complex variables. In the first case:

ei  e  i  i  e  1  ei 2  1 z 2  1
i 2
cos   e   
2  2  2ei 2z

Similarly, we find that:

z2 1
sin  
2iz

2
Taking these facts together, we see that  f  cos  ,sin   d can be rewritten as a
0
contour integral in the complex plane. We only need to include residues that are inside the
unit circle.

2 d
Compute 0 24  8cos 
.

i z2 1
Using d   dz together with cos   we have:
z 2z

2 d dz
0 24  8cos 
 i i
  z2 1  
z  24  8  
  2z  

dz
 i i
24 z  4 z 2  4

dz
 i i
4 z  24 z  4
2

i dz
 
i
4 z  6z 1
2
We will choose the unit circle for our contour. To find the singularities, we find the roots
of the denominator. Some algebra shows that they are located at:

z  3 2 2

The first residue is given by:

lim  z 3 2 2 1

1
z  3 2 2
 z 3 2 2  z 3 2 2 4 2

The residue corresponding to z  3  2 2 is given by:

lim  z 3 2 2 1

1
z  3 2 2
 z 3 2 2  z 3 2 2 4 2

You should always check that your singularities lie inside the curve you are using to
integrate. If they do not, they do not contribute to the integral. In this case both residues
do not contribute. This is because

z  3 2 2 1

lies outside the unit circle. So we will only include the second residue, because the
singularity it corresponds to, z  3  2 2  1 and so is inside the unit circle.

Using we have:

dz  1  i
i  2 i  residues  2 i    
z  6z  1
2
 4 2 2 2

Hence:

2 d i dz i i  
 0
 i 2
24  8cos  4 z  6 z  1
  
4 2 2  8 2 

The next type of definite integral we consider is one of the form:

  cos mx 
 f  x   dx

 sin mx 

This type of integral can be converted into a contour integral of the form:
 f  z e
Ñ
imz
dz

To obtain the desired result, we take the real or imaginary part of depending on whether
or not a cos or sin function is found in the original integral. A useful tool when evaluating
integrals of the form is called Jordan’s Lemma. Imagine that we choose  to be a semi-
circle located at the origin and in the upper half plane, as illustrated in Figure 7-3.

Figure 7-3: A semi-circle in the upper half plane, of radius R.

Jordan’s lemma states that:

lim  f  z  e mz dz  0
R 
C1

Jordan’s lemma does not hold in all cases. To use , if m  0 then it must be the case that
f  z   0 as R   . We can also apply it in the following case:

lim  f  z  dz  0
R 
C1

provided that f  z   0 faster than 1/z as R   .

cos kx

Compute  x2
dx .

We can compute this integral by computing:


cos kx

 x2
dx  Re I z ,

Where:

 eikz
Iz  P dz
 z 2

The P stands for principal part. To do the integral, we will take a circular contour in the
upper half plane which omits the origin. This is illustrated in Figure 7-4.

Figure 7-4: We use a semicircular contour in the upper half plane, omitting the origin
using a small semicircle or radius r that gives us a curve that omits the origin.

Now we can write out the integral piecewise, taking little chunks along the curves
C1 and C2 . Note that when directly on the real axis, we set z  x . This gives:

eikz r e
ikx
eikz Re
ikx
eikz
 z 2 dz   R x 2 dx  C z 2 dz  r x 2 dx C z 2 dz  0
Ñ
2 1

The entire sum of these integrals equals zero because the contour encloses no singular
points. However, individual integrals in this expression are not all zero. By Jordan’s
lemma:

eikz
C z 2 dz  0
1

So, we only need to calculate the residues for the curve C2 . This curve is in the clockwise
direction, so we need to add a minus sign when we do our calculation. Also, up to this
point, we have been using full circles in our calculations. The curve in this case is a semi-
circle, so is written as:

 f  z  dz   i  residues
Ñ
 j 1

The singularity at z  0 is inside the curve C2 , of order 2. The residue corresponding to


this singularity is:

d  2 eikz 
 z 2   ike  ik
ikz
z 0
dz  z  z 0

Therefore:

eikz
 z 2 dz  i  ik    k
C2

Now,

eikz r e
ikx
Re
ikx
eikz
 z2
Ñ dz   R x2 dx  r x2 C z 2 dz  0,
dx 
1

eikx r e
ikx
Re
ikx
P dx   R x 2 dx  r x 2 dx, as r  0, R   
 x 2

eikx eikz
P dx   z 2 dz  0
 x 2
C1

Therefore we find that:

ikx
 cos kx  e eikz
 x 2 dx  Re I z  P  x 2 dx   C z 2 dz   k
1

Integral of a Rational Function


The integral of a rational function f  x  :


 f  x  dx

can be calculated by computing:

 f  z  dz
Ñ
using the contour shown in Figure 7-3, which consists of a line along the x axis from –R
to R and a semicircle above the x axis the same radius. Then we take the limit R   .

Consider the Poisson kernel:

1 y
py  x  
 x  y2
2

Treating y as a constant, use the residue theorem to show that it’s Fourier transform is
1 k y
given by e .
2

The Fourier transform of a function f  x  is given by the integral:

1 
F  k   f  x  e ikx dx
2 

So, we are being asked to evaluate the integral:

1  1 y
I
2    x y
2 2
e ikx dx

We do this by considering the contour integral:

1 y
 2
Ñ 2
z y
2 2
e  ikz dz

First, note that:

1 y 1 y
 2
2 x  y
2 2 2
2  z  iy   z  iy 

Therefore, there are two simple poles located at z  iy . These lie directly on the y axis,
one in the upper half plane and one in the lower half plane. To get the right answer for the
integral we seek, we need to compute using both cases. First we consider the pole in the
upper half plane. The residue corresponding to z  iy is:
ye ikz ye ky 1
a1   z  iy    2 e ky
2 2  z  iy   z  iy  2  2iy  4 i
z  iy 2

Applying the residue theorem, we find that:

ky
1  1 y  1 ky  e
  x 2  y 2 
 ikx
I e dx  2 i  e  
2  4 i  2
2

However, now let’s consider enclosing the other singularity, which would be an equally
valid approach. The singularity is located at z  iy , which is below the x axis, so we
would need to use a semicircle in the lower half plane to enclose it. This time the residue
is:

ye  ikz 1  ky
a1   z  iy   e
2 2  z  iy   z  iy 
z  iy
4 2i

Using this result, we obtain:

 ky
1 1 y  1  ky  e
  x 2  y 2 e dx  2 i  4 2i e   2
 ikx
I
2

Combining both results gives the correct answer, which is:

e k y
I
2

Compute the integral given by:

 x2
I  dx
 1  x 4

This integral is given by:

I  2 i  residues in upper half plane

We find the residues by considering the complex function:

z2
f  z  4
z 1

The singularities are found by solving the equation:


z4 1  0

This equation is solved by z   1 . But, remember that 1  ei .


1/4

That is, there are four roots given by:

 ei /4
 i 3 /4
 e
z  e1/4 i   2 n 1   i 5 /4
 e
 ei 7 /4

These are shown in Figure 7-5. Notice that two of the roots are in the upper half plane,
while two of the roots are in the lower half plane. We reject the roots in the lower half
plane because we are choosing a closed semicircle in the upper half plane (as in Figure 7-
3) as our contour. We only consider the singularities that are inside the contour, the others
do not contribute to the integral.

ei 3 /4 ei / 4

ei 5 /4 ei 7 / 4

Figure 7-5: An illustration of the roots in Example 7-11. For our contour, we will enclose
the upper half plane-so we ignore the roots that lie in the lower half plane.

We proceed to compute the two residues. They are all simple so in the first case we have:
lim
 z e  z i /4 2

z  ei /4
 z e i /4
  z e   z e
i 3 /4 i 5 /4
  z e i 7 /4

z2
 limi /4
z e
 z  ei3 /4   z  ei5 /4   z  ei 7 /4 
i

e i /4
e i 3 /4
e i /4
 ei 5 /4   ei /4  ei 7 /4 

i 1
   ei 3 /4
2e i /4
e i /2
e i 3 /2
 4

And so:

A similar calculation shows that the residue corresponding to the pole at


1
z  ei 3 /4  ei /4 . Hence:
4

1 1
 residues   4 e i /4
 ei 3 /4
4

1
  cos  / 4  i sin  / 4  cos 3 / 4  i sin 3 / 4 
4

1 1 1 i
  1  i 1  i    2i  
4 2 4 2 2 2

Therefore the integral evaluates to:

 i  
I  2 i  residues in upper half plane  2 i   
 2 2 2

 cos x
Compute  x  2x  2
2
dx

We can compute this integral by considering:


eiz eiz dz
 z2  2z  2 Ñ
Ñ dz    z   1 i   z   1 i 

The root z  1  i lies in the upper half plane, while the root z  1  i lies in the lower half
plane. We choose a contour which is a semicircle in the upper half plane, enclosing the
first root. This is illustrated in Figure 7-6.

Figure 7-6: The contour used in Example 7-12.

The residue is given by:

 eiz 
lim   z   1  i  
z i 1
  z   1  i    z   1  i   

eiz
 lim
z i 1 z   1  i 

e 1ei

2i

Therefore we have:

eiz dz  e 1ei   i
 z2  2z  2
Ñ  2 i 
 2i

 e
e

But, using Euler’s identity, we have:


ei  cos1  i sin1

And so:

eiz dz  
 z 2  2 z  2  e cos  1  i e sin  1
Ñ
Now, we have:

R eix dx eiz dz
 R x 2  2 x  2 C z 2  2 z  2
1

R cos xdx R sin xdx eiz dz


  i  R x 2  2 x  2 C z 2  2 z  2

 R x2  2x  2
1

 
 cos  1  i sin  1
e e

Now we let R   . By Jordan’s lemma:

eiz dz
 z 2  2z  2  0
C1

So we have:

cos xdx  sin xdx  


  i 2  cos  1  i sin  1
 x  2 x  2
2  x  2 x  2 e e

Equating real and imaginary parts gives the result we are looking for:

cos xdx 
  cos  1
 x  2 x  2
2
e

Chapter Summary
By computing the Laurent expansion of an analytic function in a region containing one or
more singularities, we were able to arrive at the residue theorem which can be used to
calculate a wide variety of integrals. This includes integrals of complex functions, but the
residue theorem can also be used to calculate certain classes of integrals involving
functions of a real variable.
Chapter Quiz
sinh z
1. Compute 
 z3
dz

sinh z
2. Compute 
 z4
dz

1
3. Find the principal part of f  z  
1 z 3 2

sin z
4. What are the singular points and residues of  5  ?
z z 
 2 

sin z
5. What are the singularities and residues of ?
z    z
2

2 d
6. Evaluate 
0 24  6sin 

 sin 2 x
7. Using the technique outlined in Example 7-9, compute 0 x2
dx

R dx
8. Use the residue theorem to compute lim  R
R  1  x2

x2  3
9. Compute  2 dx
 x  1  x 2  4 
cos x

10. Compute  1  x 2
dx

Você também pode gostar