Portland State University

PDXScholar

University Honors Theses University Honors College

5-15-2020

Laurent Series Expansion and its Applications

Anna Sobczyk
Portland State University

Follow this and additional works at: https://pdxscholar.library.pdx.edu/honorstheses

Part of the Mathematics Commons

Let us know how access to this document benefits you.

Recommended Citation
Sobczyk, Anna, "Laurent Series Expansion and its Applications" (2020). University Honors Theses. Paper
839.
https://doi.org/10.15760/honors.858

This Thesis is brought to you for free and open access. It has been accepted for inclusion in University Honors
Theses by an authorized administrator of PDXScholar. Please contact us if we can make this document more
accessible: pdxscholar@pdx.edu.
 Laurent Series Expansion and its Applications

By

Anna Sobczyk

An undergraduate honors thesis submitted in partial fulfillment of the

requirements for the degree of

Bachelor of Science

In

University Honors

And

Mathematics

Thesis Advisor

Dr. Bin Jiang

Portland State University

2020
 Abstract

The Laurent expansion is a well-known topic in complex analysis for its application in
obtaining residues of complex functions around their singularities. Computing the
Laurent series of a function around its singularities turns out to be an efficient way to
determine the residue of the function as well as to compute the integral of the function
along any closed curves around its singularities. Based on the theory of the Laurent
series, this paper provides several working examples where the Laurent series of a
function is determined and then used to calculate the integral of the function along any
closed curve around the singularities of the function. A brief description of the Frobenius
method in solving ordinary differential equations is also provided.

Section I. Introduction
The method of Laurent series expansions is an important tool in complex analysis. Where a
Taylor series can only be used to describe the analytic part of a function, Laurent series allows us
to work around the singularities of a complex function. To do this, we need to determine the
singularities of the function and can then construct several concentric rings with the same center
𝑧0 based on those singularities and obtain a unique Laurent series of 𝑧 − 𝑧0inside each ring
where the function is analytic.

1
The construction of Laurent series is important because the coefficient corresponding to the 𝑧−𝑧
0
term gives the residue of the function. The calculation of the integral of the function along any
closed curve can be done efficiently by using such residue based on the Residue Theorem. Not
only does the Laurent series create an efficient method for the integration, it also has many other
applications in physics and engineering.

While the residue of the function has been used extensively in calculating both complex and real
1
integration, we seldom investigate the coefficient of the 𝑧−𝑧 term that occur in the outer rings of
0
a Laurent series expansion. This paper serves to speak on the significance of this coefficient in
the outer rings by providing several working examples of the Laurent series outside of the center
annulus and using them to compute the integral of the function along any closed curve outside of
the center annulus. This paper also describes the Frobenius method, a method very similar to
Laurent series, in solving second-order ordinary differential equations around their singularities.

2
Section II. Background Theory

First, let us define a Laurent series.1

Theorem 1.1 Let 0 ≤ 𝑟1 < 𝑟2 and 𝑧0 ∈ ℂ. Suppose 𝑓(𝑧) is analytic on the region 𝐴 =
{𝑧 ∈ ℂ| 𝑟1 < |𝑧 − 𝑧0 | < 𝑟2 } where 𝑟1 ≥ 0 and 𝑟2 ≤ ∞. Then we have:
∞ ∞
𝑏𝑛
𝑓(𝑧) = ∑ 𝑎𝑛 (𝑧 − 𝑧0 )𝑛 + ∑
(𝑧 − 𝑧0 )𝑛
𝑛=0 𝑛=1

which converge absolutely on 𝐴. This series for 𝑓(𝑧) is called the Laurent series or
Laurent expansion around 𝑧0 in the annulus 𝐴.

The coefficients can be determined by:

1 𝑓(𝜁)
𝑎𝑛 = ∫ 𝑑𝜁 , 𝑛 = 0, 1, 2, …
2𝜋𝑖 𝛾 (𝜁 − 𝑧0 )𝑛+1
1
𝑏𝑛 = ∫𝑓(𝜁)(𝜁 − 𝑧0 )𝑛−1 𝑑𝜁 , 𝑛 = 1, 2, …
2𝜋𝑖 𝛾
Where 𝛾 can be any circle with center 𝑧0 and radius 𝑟, 𝑟1 < 𝑟 < 𝑟2 . Furthermore, 𝑎𝑛 and
𝑏𝑛 are unique.

However, the equations for finding the coefficients 𝑎𝑛 and 𝑏𝑛 in the Laurent Series are impractical
for a given function. Although the formulas exist, they are seldom used as there are always easier
tricks to obtain the Laurent expansion. For instance, we can use well-known Taylor expansions
of some fundamental function to obtain the Laurent series. Another common trick is to creatively
manipulate a function 𝑓 to fit into the form of a common geometric series. See the working
examples for demonstrations of these tricks.

Next, let us introduce the classification of singularities, which are central to constructing a
Laurent series. Simply put, a singularity is a point 𝑧0 in which a function is differentiable at
points arbitrarily close to but not including 𝑧0 . Singularities are not always easily for a
complicated function. There are different types and classifications of singularities that we will
define now.

Definition 1.1 If a function 𝑓 is analytic on a region 𝐴 = {𝑧| 0 < |𝑧 − 𝑧0 | <

𝑟2 } 𝑤𝑖𝑡ℎ 𝑟1 = 0, which is a deleted neighborhood of 𝑧0 , then 𝑧0 is called an isolated
singularity2 and is expressed as:
𝑏 𝑏
𝑓(𝑧) = ⋯ + (𝑧−𝑧𝑛 )𝑛 + ⋯ + 1 + 𝑎𝑛 + 𝑎1 (𝑧 − 𝑧0 ) + ⋯ 𝑖𝑛 {𝑧| 0 < |𝑧 − 𝑧0 | < 𝑟2 }
0 𝑧−𝑧0

1
Theorem adapted from Marsden & Hoffman textbook.
2
Definition taken from Marsden & Hoffman and Apelian & Surace text books.

3
The convenience of Laurent series is that we can always find a Laurent expansion centered at an
isolated singularity in an annulus that omits that point.3 The Laurent expansion allows for a
series representation in both negative and positive powers of (𝑧 − 𝑧0 ) in a region excluding
points where 𝑓 is not differentiable. If 𝑓 is differentiable in the entire region, then it is analytic
and the Laurent series centered at 𝑧0 will reduce to the Taylor series of the function below:
∞
𝑓 𝑘 (𝑐)
𝑓(𝑧) = ∑ (𝑧 − 𝑐)𝑘 𝑓𝑜𝑟 |𝑧 − 𝑐| < 𝑟
𝑘!
𝑛=0

There are different classifications of isolated singularities as below:

Definition 1.2 Let 𝑧0 be an isolated singularity of 𝑓. If all but a finite number of the 𝑏𝑛
terms are zero, then 𝑧0 is called a pole of 𝑓. The order of the pole is determined by the
highest integer k such that 𝑏𝑘 ≠ 0 and is called a pole of order k. A pole of order one is
commonly referred to as a simple pole or single pole.

Definition 1.3 If an infinite number of 𝑏𝑘 are nonzero, 𝑧0 is called an essential

singularity. This 𝑧0 is also sometimes called a pole of infinite order.

Definition 1.4 We call 𝑧0 a removable singularity if all 𝑏𝑘 ’s are zero. A Taylor series
expansion always exists for removable singularities.

We focus on the main application of Laurent series: finding the residue of a function. While
some complex functions have handy formulas for calculating the residue, it mainly depends on
the type of singularity you are dealing with. For instance, there is no efficient way to find the
residue of a function with an essential singularity. To find the residue in this case, you must find
𝒃
the Laurent expansion of the function and locate the 𝑧−𝑧𝟏 term:
0

𝑏 𝒃
𝑓(𝑧) = ⋯ + (𝑧−𝑧𝑛 )𝑛 + ⋯ + 𝑧−𝑧𝟏 + 𝑎𝑛 + 𝑎1 (𝑧 − 𝑧0 ) + ⋯ 𝑖𝑛 {𝑧| 0 < |𝑧 − 𝑧0 | < 𝑟2 }
0 0
1
In the Laurent expansion, 𝒃𝟏 is the coefficient of the 𝑧−𝑧 term in the series. Even if the residue
0
can be found easily through a formula, one might want to look at the Laurent series obtained in
the outer annuli associated with the singularities. Once the Laurent expansion is known, we can
𝒃
find the corresponding 𝑧−𝑧𝟏 term. It can help us with calculating the integration of 𝑓(𝑧) along
0
any closed curve located within those annuli.

3
See Apelian & Surace.

4
 Theorem 1.2 Residue Theorem4: Let 𝐴 be a region and let 𝑧1, 𝑧2 , … , 𝑧𝑛 be 𝑛 distinct
points in A. Let 𝑓 be analytic in 𝐴 except for at the isolated singularities at 𝑧1, 𝑧2 , … , 𝑧𝑛 .
Let  be a closed curve in 𝐴 homotopic to a point in 𝐴. Assume no 𝑧𝑖 lies on 𝛾. Then:
𝑛

∫ 𝑓(𝑧)𝑑𝑧 = 2𝜋𝑖 ∑[𝑅𝑒𝑠(𝑓; 𝑧𝑖 )]𝐼(𝛾; 𝑧𝑖 )

𝛾 𝑖=1

Here 𝑅𝑒𝑠(𝑓; 𝑧𝑖 ) is the residue of 𝑓 at 𝑧𝑖 and 𝐼(𝛾; 𝑧𝑖 ) is the index of 𝛾 with respect to 𝑧𝑖 .
For simplicity, we assume 𝐼(𝛾; 𝑧𝑖 ) = 1 for this paper.

Simply put, for a clockwise 𝛾, the integral equals 2𝜋𝑖 times the sum of the residues of 𝑓 inside 𝛾.
In complex analysis, residues play a critical role in allowing us to easily compute integrals.
Using the Residue Theorem to solve a line integral is significantly more efficient than other
methods because it greatly reduces computational time. The Residue Theorem in complex
analysis also makes the integration of some real functions feasible without need of numerical
approximation.

The examples in this paper focus on obtaining the residue from a Laurent series. The residues
obtained from the Laurent series would speed up the complex integration on closed curves.

4
Theorem adapted from Marsden & Hoffman textbook.

5
Section III. Working Examples
In the following examples we will refer to a helpful geometric series and some common known
Taylor expansions.

Geometric Series
∞

∑ 𝑧𝑛 , |𝑧| < 1
1 𝑛=0
= ∞
1−𝑧 1
−∑ 𝑛 , |𝑧| > 1
{ 𝑛=1 𝑧

Table 1.1
Known Function Expansions
Function Taylor Series around 0 Valid for
∞
𝑧3 𝑧5 𝑧 2𝑛−1 all 𝑧
sin (𝑧) 𝑧− + − ⋯ = ∑(−1)𝑛+1
3! 5! (2𝑛 − 1)!
𝑛=1
∞
cos (𝑧) 𝑧 2
𝑧 𝑧 4 6
𝑧 2𝑛 all 𝑧
𝑛
1 − + − + ⋯ = ∑(−1)
2! 4! 6! (2𝑛)!
𝑛=1
𝑧 ∞
e 𝑧2 𝑧3 𝑧𝑛 all 𝑧
1+𝑧+ + +⋯ = ∑
2! 3! 𝑛!
𝑛=0

6
Example 1
1
Let 𝑓(𝑧) = 2+𝑧. Determine the Laurent series around 𝑧 = 1.

Solution:

Obviously, we have a simple pole at 𝑧 = −2. Hence, we are dealing with a radius of 3 and want
to find the Laurent series for both |𝑧 − 1| < 3 𝑎𝑛𝑑 |𝑧 − 1| > 3. The Laurent series will reduce
to a Taylor series inside |𝑧 − 1| < 3 where 𝑓(𝑧) is analytic.

For |𝑧 − 1| < 3, we refer to the well-known geometric series. We begin by trying to create a
(𝑧 − 1) term in the denominator.

1 1 1 1 1
𝑓(𝑧) = = = =( )
2 + 𝑧 2 + 𝑧 − 1 + 1 3 + (𝑧 − 1) 3 −(𝑧 − 1)
1−( )
3
−(𝑧−1)
Since | 3 | < 1, we can now represent the function as a series:

∞ ∞
1 −(𝑧 − 1)𝑛 (−1)𝑛 (𝑧 − 1)𝑛
⇒ 𝑓(𝑧) = ∑ = ∑ for |𝑧 − 1| < 3
3 3𝑛 3𝑛+1
𝑛=0 𝑛=0

Which is just a Taylor series as the function is analytic inside the region.
3
For |𝑧 − 1| > 3, we can use |𝑧−1| < 1 and follow our previous work to obtain:

∞ ∞
1 1 1 1 (−1)𝑛 3𝑛 (−1)𝑛 3𝑛
𝑓(𝑧) = = = ∑ = ∑
3 + (𝑧 − 1) 𝑧 − 1 1 − ( −3 ) 𝑧 − 1 (𝑧 − 1)𝑛 (𝑧 − 1)𝑛+1
𝑛=0 𝑛=0
𝑧−1
∞ ∞
(−1)𝑛 3𝑛 (−1)𝑛−1 3𝑛−1
⟹ 𝑓(𝑧) = ∑ = ∑
(𝑧 − 1)𝑛+1 (𝑧 − 1)𝑛
𝑛=0 𝑛=1

In this case, we have obtained the Laurent expansion. The generalized residue for the outer ring
1
|𝑧 − 1| > 3 is the coefficient of , that is 𝑏1 = 1.
𝑧−1

Remarks:
a) Integrating along a closed path inside the inner ring |𝑧 − 1| < 3 means we should use the
residue obtained from the Laurent series in |𝑧 − 1| < 3. By Theorem 1.1, we can
compute
∫ 𝑓(𝑧)𝑑𝑧 = 2𝜋𝑖(𝑏1 ) = 2𝜋𝑖(0) = 0.
|𝑧−1|=1

7
b) Integrating along a non-simple closed path inside the outer ring |𝑧 − 1| > 3 means we
must use the generalized residue obtained from the Laurent series in |𝑧 − 1| > 3. By
Theorem 1.1, we can compute
∫ 𝑓(𝑧)𝑑𝑧 = 2𝜋𝑖(𝑏1 ) = 2𝜋𝑖(1) = 2𝜋𝑖 .
|𝑧−1|=4
c) By combining a) and b), we conclude that the Laurent series computed in each ring,
either inner ring or outer ring can be used directly with coefficient 𝑏1 to compute the
integral along any closed curve inside that ring.

8
Example 2

This example will use the same method as in Example 1 based on the known geometric series to
find the Laurent Expansion. However, this is a more complicated example and will require extra
techniques—a substitution and partial fraction decomposition. Suppose:

1
𝑓(𝑧) =
𝑧(𝑧 − 2)(𝑧 − 5)

By observation, we can easily see that 𝑓(𝑧) has simple poles at 𝑧 = 0, 𝑧 = 2 𝑎𝑛𝑑 𝑧 = 5. These
poles determine the radii of the annuli produced with the Laurent Series. We have 0 < |𝑧| < 2 as
the center annulus, 2 < |𝑧| < 5 as the middle annulus, and 5 < |𝑧| < ∞ as the outermost infinite
“ring.”

Figure 1.1

In this example, we will find the Laurent expansion valid on 𝑅 = {𝑧 | |𝑧 − 3| < 1}. This is the
circle of radius 1 centered at 𝑧 = 3 where 𝑓(𝑧) is analytic.

Step 1 In order to construct the Laurent series of this function, we first use a substitution such
that 𝑤 = 𝑧 − 3. Then we can center 𝑤 around 𝑤 = 0 within |𝑤| < 1.

1
𝑓(𝑧) =
(𝑤 + 3)(𝑤 + 1)(𝑤 − 2)

9
Step 2 We now use partial fraction decomposition:

1 𝐴 𝐵 𝐶
𝑓(𝑤) = = + +
(𝑤 + 3)(𝑤 + 1)(𝑤 − 2) 𝑤 + 3 𝑤 + 1 𝑤 − 2

⟹ 𝐴(𝑤 2 − 𝑤 − 2) + 𝐵(𝑤 2 + 𝑤 − 6) + 𝐶(𝑤 2 + 4𝑤 + 3) = 1

𝐴+𝐵+𝐶 =0
−𝐴 + 𝐵 + 4𝐶 = 0
−2𝐴 − 6𝐵 + 3𝐶 = 1

From the above equations, we obtain the following coefficients:

1 1 1
𝐴 = 10 ; 𝐵 = − 6 ; 𝐶 = 15.

We will denote the partial fraction components of 𝑓(𝑧) as:

1 −1 1
𝑓1 (𝑤) = 10(𝑤+3) , 𝑓2 (𝑤) = 6(𝑤+1) , 𝑓3 (𝑤) = 15(𝑤−2).

Step 3 Determine the Laurent expansion valid for |𝑤| < 1 based on the above partial fractions.
1
Let us start with 𝑓1 (𝑤) = 10(𝑤+3). For |𝑤| < 1:

1 1 1 1 1 1 1 (−1)𝑛 𝑤 𝑛
= 10 3−(−𝑤) = 10 3 −𝑤 = 10 ∑∞
𝑛=0
10(𝑤+3) 1−( ) 3𝑛+1
3

We obtain the following geometric series and then substitute 𝑤 = 𝑧 − 3.

(−1)𝑛 𝑤 𝑛 (−1)𝑛 (𝑧−3)𝑛

𝑓1 (𝑤) = ∑∞ ∞
𝑛=0 10(3𝑛+1 ) , |𝑤| < 1 ⟹ 𝑓1 (𝑧) = ∑𝑛=0 , |𝑧 − 3| < 1.
10(3𝑛+1 )

−1
For 𝑓2 (𝑤) = 6(𝑤+1) on |𝑤| < 1:
−1 1 1 1
= − 6 1−(−𝑤) = − 6 ∑∞ 𝑛 𝑛
𝑛=0(−1) 𝑤
6(𝑤+1)

We obtain the following geometric series and substitute back in for 𝑤 = 𝑧 − 3.

(−1)𝑛 𝑤𝑛 (−1)𝑛 (𝑧−3)𝑛

𝑓2 (𝑤) = − ∑∞
𝑛=0 , |𝑤| < 1 ⟹ 𝑓2 (𝑧) = − ∑∞
𝑛=0 , |𝑧 − 3| < 1
6 6

10
 1
For 𝑓3 (𝑤) = 15(𝑤−2) on |𝑤| < 1:
1 1 1 −1 1 1 𝑤𝑛
= 15(−2−(−𝑤)) = 15 𝑤 = − 15 ∑∞
𝑛=0 2𝑛+1
15(𝑤−2) 2 (1− )
2

We obtain the following geometric series and substitute back in for 𝑤 = 𝑧 − 3.

1 𝑤𝑛 (𝑧−3)𝑛
𝑓3 (𝑤) = − 15 ∑∞ ∞
𝑛=0 2𝑛+1 |𝑤| < 1 ⟹ 𝑓3 (𝑧) = − ∑𝑛=0 15(2𝑛+1 ) , |𝑧 − 3| < 1.

Step 4 Determine which parts are relevant for the Laurent series. In this case, we only
want |𝑤| < 1, or |𝑧 − 3| < 1. We combine the above geometric series and obtain:
∞ ∞ ∞
(−1)𝑛 (𝑧 − 3)𝑛 (−1)𝑛 (𝑧 − 3)𝑛 (𝑧 − 3)𝑛
𝑓(𝑧) = ∑ −∑ −∑
10(3𝑛+1 ) 6 15(2𝑛+1 )
𝑛=0 𝑛=0 𝑛=0

∞
(−1)𝑛 (𝑧 − 3)𝑛 (−1)𝑛 (𝑧 − 3)𝑛 (𝑧 − 3)𝑛
⟹ 𝑓(𝑧) = ∑ ( − − )
10(3𝑛+1 ) 6 15(2𝑛+1 )
𝑛=0

∞
(−1)𝑛 (−1)𝑛 1
⟹ 𝑓(𝑧) = ∑ ( 𝑛+1
− − ) (𝑧 − 3)𝑛
10(3 ) 6 15(2𝑛+1 )
𝑛=0

∞
{(−1)𝑛 (2𝑛+1 )[3 − 5(3𝑛+1 )] − 2(3𝑛+1 )}
⟹ 𝑓(𝑧) = ∑ 𝑛+1
(𝑧 − 3)𝑛
30(6 )
𝑛=0

Note that this results in the analytic part of the Laurent expansion without principal part. This
means our Laurent series has reduced to a Taylor series as in Example 1.

11
Example 3
1
Let us find the Laurent series of 𝑓(𝑧) = 𝑧(𝑧−2)(𝑧−5) valid in the region 2 < |𝑧| < 5 around 𝑧 = 0.

Step 1 we start with partial fraction decomposition.

1 𝐴 𝐵 𝐶
𝑓(𝑧) = = + +
𝑧(𝑧 − 2)(𝑧 − 5) 𝑧 𝑧 − 2 𝑧 − 5

⟹ 𝐴(𝑧 2 − 7𝑧 + 10) + 𝐵(𝑧 2 − 5𝑧) + 𝐶(𝑧 2 − 2𝑧) = 1

𝐴+𝐵+𝐶 =0
−7𝐴 − 5𝐵 − 2𝐶 = 0
10𝐴 = 1

From the above equations, we obtain the following coefficients:

1 1 1
𝐴 = 10 ; 𝐵 = − 6 ; 𝐶 = 15.

We will denote the partial fraction components of 𝑓(𝑧) as:

1 −1 1
𝑓1 (𝑧) = 10𝑧 , 𝑓2 (𝑧) = 6(𝑧−2) , 𝑓3 (𝑧) = 15(𝑧−5).

Step 2 Determining the geometric series to each corresponding partial fraction.

We consider the region 2 < |𝑧| < 5. In order to use our geometric series, we must represent the
1
function in the region in terms of 𝑧 𝑜𝑟 𝑧 based on the fact that

2 𝑧
2 < |𝑧| ⟹ | | < 1 𝑎𝑛𝑑 |𝑧| < 5 ⟹ | | < 1.
𝑧 5
2 𝑧
The Laurent series representation is found by looking for 𝑧 for 𝑓2 (𝑧) and 5 for 𝑓3 (𝑧):

−1 −1 1 1 2𝑛 1 2𝑛
a. 𝑓2 (𝑧) = 6(𝑧−2) = 2 = − 6𝑧 ∑∞ ∞
𝑛=0 𝑧 𝑛 = − 6 ∑𝑛=0 𝑧 𝑛+1
6𝑧 (1− )
𝑧

1 1 1 1 1 1 𝑧𝑛 1 𝑧𝑛
b. 𝑓3 (𝑧) = 15(𝑧−5) = 15 5 ( 𝑧 )=− ∑∞
𝑛=0 = − 15 ∑∞
𝑛=0 5𝑛+1
−1+ 15 5 5𝑛
5

1 1 1
Note that 𝑓1 (𝑧) = 10𝑧 = 10 𝑧 is the form of series representation.

12
Step 3 Build the Laurent expansion from the above geometric series.
∞ ∞
1 1 1 2𝑛 1 𝑧𝑛
𝑓(𝑧) = − ∑ 𝑛+1 − ∑ 𝑛+1
10 𝑧 6 𝑧 15 5
𝑛=0 𝑛=0

∞ ∞
𝑧𝑛 1 1 2𝑛
⟹ 𝑓(𝑧) = − ∑ + −∑
15(5𝑛+1 ) 10 𝑧 6(𝑧 𝑛+1 )
𝑛=0 𝑛=0

∞ ∞
𝑧𝑛 1 1 2𝑛−1 1
⟹ 𝑓(𝑧) = − ∑ − −∑
15(5𝑛+1 ) 15 𝑧 6 𝑧𝑛
𝑛=0 𝑛=2

𝑏1
Step 4 Find the generalized residue. We only need to collect the term.
𝑧
1
Therefore, the generalized residue of 𝑓(𝑧) is − 15 in the region 2 < |𝑧| < 5.

13
Example 4

To demonstrate application of how the generalized residue can be used to solve complex
function integration, we will use 𝑓(𝑧) from Example 3. Suppose we want to integrate 𝑓(𝑧)
along the path |𝑧| = 4.

We can solve this problem using the Residue Theorem with two methods:
a. Finding the residues using known formulas and apply the Residue Theorem.
b. Use our previously constructed Laurent series and apply Theorem 1.1 directly.

For (a), we can find the residue of a pole using the formula 𝑅𝑒𝑠(𝑓, 𝑧0 ) = lim (𝑧 − 𝑧0 )𝑓(𝑧).5
𝑧→𝑧0
We will only need to find the residue of the poles 𝑧 = 0 and 𝑧 = 2 because they are the only two
simple poles that lie within the closed path. The other simple pole 𝑧0 = 5 lies outside the given
path, and will not be considered for this integral.

1
𝑅𝑒𝑠(𝑓, 0) = lim(𝑧 − 0)𝑓(𝑧) =
𝑧→0 10
1
𝑅𝑒𝑠(𝑓, 2) = lim(𝑧 − 2)𝑓(𝑧) = −
𝑧→2 6
1 1 2𝜋𝑖
Applying the Residue Theorem, we have ∫|𝑧|=4 𝑓(𝑧)𝑑𝑧 = 2𝜋𝑖 (10 − 6) = − .
15

For (b), we can use Theorem 1.1 directly. This means we can use the generalized residue found
in our Laurent expansion obtained in Example 3 because the Laurent series was constructed in
the ring containing |𝑧| = 4. We simply take the generalized residue, and multiply it by 2𝜋𝑖:

1 2𝜋𝑖
∫ 𝑓(𝑧)𝑑𝑧 = 2𝜋𝑖 (− )=−
|𝑧|=4 15 15

Remarks

If we wanted to integrate along any path |𝑧| > 5, then we need the Laurent expansion for the
outmost ring |𝑧| > 5. The Laurent series is the only way to obtain the generalized residue
associated with this region. We skip the details as its approach is similar to Example 3.

5
See Marsden & Hoffman textbook for this formula and others to calculate residues.

14
Example 5

The following two functions are further examples on how to use a well-known series expansion
to find the Laurent series. Each of these functions contains essential singularities. Notice that
they have infinitely many 𝑏𝑛 terms.
1
1 1 1 1 1
1. 𝑓(𝑧) = 𝑧 2 𝑒 𝑧 = 𝑧 2 (1 + 1!𝑧 + 2!𝑧 2 + ⋯ ) = 𝑧 2 + 𝑧 + 2 + 3!𝑧 + 4!𝑧 2 … 𝑓𝑜𝑟 |𝑧| > 0
𝑠𝑖𝑛𝑧 1 𝑧3 𝑧5 1 𝑧3 𝑧5
2. 𝑔(𝑧) = = 𝑧 2 (𝑧 − + − +⋯) = 𝑧 − + − +⋯
𝑧2 3! 5! 3! 5!
1
For 𝑓(𝑧), you simply multiply a 𝑧 2 term through the Taylor Series of 𝑒 𝑧 . In order to find the
1
residue, you need to find the constant associated with the 𝑧 term. Clearly, that term for 𝑓(𝑧)
1 1
is 3!𝑧. Therefore 𝑏1 = 6 is the residue of 𝑓(𝑧) around 𝑧 = 0.

For 𝑔(𝑧), we divide by 𝑧 2 the Taylor Series of sin (𝑧). We immediately get our residue from the
1
first term 𝑧. Then 𝑏1 = 1 is the residue of 𝑔(𝑧) around 𝑧 = 0.

15
Section IV. The Frobenius Method

The Frobenius method is used to solve linear differential equations with variable coefficients.

Theorem 1.36 Any differential equation that has the form:

𝑝(𝑥) 𝑞(𝑥)
(𝟏) 𝑦 ′′ + 𝑦′ + 𝑦=0
𝑥 𝑥2

Here the functions 𝑝(𝑥) and 𝑞(𝑥) are analytic at 𝑥 = 0 will have at least one solution that
can be represented by:

( 𝟐) 𝑦 ( 𝑥 ) = 𝑥 𝑟 ∑∞ 𝑚 𝑟 2
𝑛=0 𝑎𝑚 𝑥 = 𝑥 (𝑎0 + 𝑎1 𝑥 + 𝑎2 𝑥 + ⋯ )

The exponent r can be any real or complex number and is chosen so that 𝑎0 ≠ 0.

Solving (1) involves expanding 𝑝(𝑥) and 𝑞(𝑥) in power series and differentiating (2) term by
term. The process results in a very important quadratic equation called the indicial equation.

We begin by expanding y(x) as defined in (2) along with its first and second derivatives:

i. 𝑦(𝑥) = 𝑎0 𝑥 𝑟 + 𝑎1 𝑥 𝑟+1 + 𝑎2 𝑥 𝑟+2 + ⋯

ii. 𝑦′(𝑥) = 𝑟𝑎0 𝑥 𝑟−1 + (𝑟 + 1)𝑎1 𝑥 𝑟 + ⋯
iii. 𝑦 ′′ (𝑥) = 𝑎0 𝑟(𝑟 − 1)𝑥 𝑟−2 + 𝑎1 𝑟(𝑟 + 1)𝑥 𝑟−1 + ⋯

If we multiply (1) by 𝑥 2 , we get 𝑥 2 𝑦 ′′ + 𝑥𝑝(𝑥)𝑦′ + 𝑞(𝑥)𝑦 = 0.

Now plug the expansions of y(x) from (i), (ii), and (iii) into our simplified (1) and let 𝑥 → 0
then 𝑝(𝑥) → 𝑝0 𝑎𝑛𝑑 𝑞(𝑥) → 𝑞0 . From this, we get the equation:

𝑎0 𝑟(𝑟 − 1)𝑥 𝑟 + 𝑝0 𝑎0 𝑟𝑥 𝑟 + 𝑞0 𝑎0 𝑥 𝑟 = 0

Here 𝑎0 𝑟(𝑟 − 1)𝑥 𝑟−2 of 𝑦′′(𝑥), 𝑎0 𝑟𝑥 𝑟−1 of 𝑦′(𝑥), and 𝑎0 𝑥 𝑟 of y(x) are the major terms.

We factor out and divide by the common term to arrive at the indicial equation:

(𝑟(𝑟 − 1) + 𝑝0 𝑟 + 𝑞0 )(𝑎0 𝑥 𝑟 ) = 0

⟹ 𝑟(𝑟 − 1) + 𝑝0 𝑟 + 𝑞0 = 0

6
Theorem adapted from Kreyszig’s Theorem 1 (p. 216).

16
 Definition 1.57 The indicial equation is a quadratic equation defined that takes on the
form:

𝑟(𝑟 − 1) + 𝑝0 𝑟 + 𝑞0 = 0, where

𝑝(𝑥) = ∑∞ 𝑘 ∞ 𝑘
𝑘=0 𝑝𝑘 𝑥 , 𝑞(𝑥) = ∑𝑘=0 𝑞𝑘 𝑥 .

The roots r1 and r2 of the indicial equation can present in three different cases:

1. Distinct roots not differing by an integer.

2. Double root 𝑟1 = 𝑟2 = 𝑟.
3. Roots differing by an integer, which may or may not have a logarithmic term.

Let us show the framework of the method of Frobenius.8 Suppose we have the equation:

𝑥 2 𝑦 ′′ + 𝑥𝑝(𝑥)𝑦 ′ + 𝑞(𝑥)𝑦 = 0

Here 𝑥 = 0 is a regular singular point. Then p(x) and q(x) are analytic at the origin and have
power series expansions
∞ ∞
𝑘
𝑝(𝑥) = ∑ 𝑝𝑘 𝑥 , 𝑞(𝑥) = ∑ 𝑞𝑘 𝑥 𝑘 , |𝑥| < 𝜌
𝑘=0 𝑘=0

Those are convergent for some 𝜌 > 0. Let r1, r2 be the roots of the indicial equation

𝐹(𝑟) = 𝑟(𝑟 − 1) + 𝑝0 𝑟 + 𝑞𝑜 = 0.

The power series given in each of the following solution forms are convergent at least in the
interval |𝑥| < 𝜌.

Case 1: Distinct roots not differing by an integer (i.e. 𝑟1 − 𝑟2 ≠ 𝑛, 𝑛 ∈ ℤ).

The solutions will have the form:

∞

𝑦1 (𝑥) = 𝑥 ∑ 𝑎𝑘 (𝑟1 )𝑥 𝑘
𝑟1

𝑘=0
∞

𝑦2 (𝑥) = 𝑥 𝑟2 ∑ 𝑏𝑘 (𝑟2 )𝑥 𝑘
𝑘=0

Where 𝑎𝑘 (𝑟1 ) and 𝑏𝑘 (𝑟2 ) are determined by substituting 𝑦1 (𝑥) or 𝑦2 (𝑥) into the original ODE.

7
Indicial equation and form of solutions taken from Kreyszig.
8
Framework adapted from Phinney’s lecture notes

17
 Case 2: Repeated root (i.e. 𝑟1 = 𝑟2).

The solutions will have the form:

∞

𝑦1 (𝑥) = 𝑥 ∑ 𝑎𝑘 (𝑟1 )𝑥 𝑘
𝑟1

𝑘=0

𝑦2 (𝑥) = 𝑦1 (𝑥)𝑙𝑜𝑔𝑥 + 𝑥 𝑟1 ∑ 𝑏𝑘 (𝑟1 )𝑥 𝑘

𝑘=0

Let us show that the formula for 𝑦2 (𝑥) is fesible by demonstrating why the term 𝑦1 (𝑥)𝑙𝑜𝑔𝑥 is
needed, based on the fact that 𝑦1 (𝑥) and 𝑦2 (𝑥) need to be linearly independent.

Let 𝐸2 (𝑥) = 𝑥 𝑟1 ∑∞ 𝑘
𝑘=0 𝑏𝑘 (𝑟1 )𝑥 denote the second term of 𝑦2 (𝑥).

Begin by differentiating:
1
𝑦2′ (𝑥) = 𝑦1′ (𝑥)𝑙𝑜𝑔𝑥 + 𝑦1 (𝑥) + 𝐸2 ′(𝑥)
𝑥
′′ (𝑥) ′′ (𝑥)𝑙𝑜𝑔𝑥 ′ (𝑥)
1 1
𝑦2 = 𝑦1 + 2𝑦1 − 𝑦1 (𝑥) 2 + 𝐸2′′ (𝑥)
𝑥 𝑥

Recall:

𝑥 2 𝑦2 ′′ + 𝑥𝑝(𝑥)𝑦2 ′ + 𝑞(𝑥)𝑦2 = 0

Plugging in for 𝑦2 (𝑥), 𝑦2′ (𝑥), and 𝑦2′′ (𝑥) we get:

2𝑦1 (𝑥) 𝑦1 (𝑥) 𝑦1 (𝑥)

𝑥 2 [𝑦1′′ (𝑥)𝑙𝑜𝑔𝑥 + − 2 + 𝐸2 ′′(𝑥)] + 𝑥𝑝(𝑥) [𝑦1′ (𝑥)𝑙𝑜𝑔𝑥 + + 𝐸2′ (𝑥)]
𝑥 𝑥 𝑥
+ 𝑞(𝑥)[𝑦1 (𝑥)𝑙𝑜𝑔𝑥 + 𝐸2 (𝑥)] = 0

⟹ 𝑙𝑜𝑔𝑥[𝑥 2 𝑦1′′ (𝑥) + 𝑥𝑝(𝑥)𝑦1′ (𝑥) + 𝑞(𝑥)𝑦1 (𝑥)] + 2𝑥𝑦1′ (𝑥) − 𝑦1 (𝑥) + 𝑝(𝑥)𝑦1 (𝑥)
+ (𝑥 2 𝐸2′′ (𝑥) + 𝑥𝑝(𝑥)𝐸2′ (𝑥) + 𝑞(𝑥)) = 0

Since 𝑦1 (𝑥) is already a solution, the first term above is zero. The second term 𝐸2 (𝑥) of 𝑦2 (𝑥)
satisfies:

𝑥 2 𝐸2′′ (𝑥) + 𝑥𝑝(𝑥)𝐸2′ (𝑥) + 𝑞(𝑥) = 1 − 𝑝(𝑥)𝑦1 (𝑥) − 2𝑥𝑦1′ (𝑥)

This is nonhomogeneous Frobenius equation for 𝐸2 (𝑥). Since 𝑦1 (𝑥) is already expressed in a
power series, a similar method can be used to solve 𝐸2 (𝑥). Adding 𝐸2 (𝑥) back into 𝑦1 (𝑥)𝑙𝑜𝑔𝑥
will ensure 𝑦2 (𝑥) is linearly independent with 𝑦1 (𝑥).

18
 Case 3: Roots differing by an integer (i.e. 𝑟1 − 𝑟2 = 𝑛, 𝑛 ∈ ℤ+ ).

The solutions will have the form:

∞

𝑦1 (𝑥) = 𝑥 ∑ 𝑎𝑘 (𝑟1 )𝑥 𝑘
𝑟1

𝑘=0

𝑦2 (𝑥) = 𝑐𝑦1 (𝑥)𝑙𝑜𝑔𝑥 + 𝑥 𝑟2 ∑ 𝑏𝑘 (𝑟2 )𝑥 𝑘

𝑘=0

And c may be the value zero. The constant 𝑏𝑛 (𝑟2 ) is arbitrary and may be set to zero.

Let us justify the feasibility of second solution 𝑦2 (𝑥):

Let 𝐸2 (𝑥) = 𝑥 𝑟2 ∑∞ 𝑘
𝑘=0 𝑏𝑘 (𝑟2 )𝑥 denote the second term of 𝑦2 (𝑥).

Begin by differentiating:
1
𝑦2′ (𝑥) = 𝑐 (𝑦1′ (𝑥)𝑙𝑜𝑔𝑥 + 𝑦1 (𝑥) ) + 𝐸2 ′(𝑥)
𝑥
1 1
𝑦2′′ (𝑥) = 𝑐 (𝑦1′′ (𝑥)𝑙𝑜𝑔𝑥 + 2𝑦1′ (𝑥) − 𝑦1 (𝑥) 2 ) + 𝐸2′′ (𝑥)
𝑥 𝑥

Recall:

𝑥 2 𝑦2 ′′ + 𝑥𝑝(𝑥)𝑦2 ′ + 𝑞(𝑥)𝑦2 = 0

Plugging in for 𝑦2 (𝑥), 𝑦2′ (𝑥), and 𝑦2′′ (𝑥) we get:

2𝑦1 ′(𝑥) 𝑦1 (𝑥) 𝑦1 (𝑥)

𝑥 2 [𝑐 (𝑦1′′ (𝑥)𝑙𝑜𝑔𝑥 + − 2 ) + 𝐸2 ′′(𝑥)] + 𝑥𝑝(𝑥) [𝑐 (𝑦1′ (𝑥)𝑙𝑜𝑔𝑥 + ) + 𝐸2′ (𝑥)]
𝑥 𝑥 𝑥
+ 𝑞(𝑥)[𝑐𝑦1 (𝑥)𝑙𝑜𝑔𝑥 + 𝐸2 (𝑥)] = 0

⟹ 𝑙𝑜𝑔𝑥[𝑐𝑥 2 𝑦1′′ (𝑥) + 𝑐𝑥𝑝(𝑥)𝑦1′ (𝑥) + 𝑐𝑞(𝑥)𝑦1 (𝑥)] + 2𝑐𝑥𝑦1′ (𝑥) − 𝑐𝑦1 (𝑥) + 𝑐𝑝(𝑥)𝑦1 (𝑥)
+ (𝑥 2 𝐸2′′ (𝑥) + 𝑥𝑝(𝑥)𝐸2′ (𝑥) + 𝐸2 (𝑥)𝑞(𝑥)) = 0

19
Since 𝑦1 (𝑥) is already a solution, the first term above is zero. The second term 𝐸2 (𝑥) of 𝑦2 (𝑥)
satisfies:

𝑥 2 𝐸2′′ (𝑥) + 𝑥𝑝(𝑥)𝐸2′ (𝑥) + 𝑞(𝑥)𝐸2 (𝑥) = 𝑐(𝑦1 (𝑥) − 𝑝(𝑥)𝑦1 (𝑥) − 2𝑥𝑦1′ (𝑥))

From here, we must explore two scenarios with the value of c. First, assume 𝑐 = 0, and then
solve the above homogeneous ODE to see if there exist two linearly independent solutions 𝑦1 (𝑥)
and 𝑦2 (𝑥) = 𝐸2 (𝑥) . Linear dependence occurs if 𝑦2 (𝑥) turns out to be a multiple of 𝑦1 (𝑥). Then
we must choose 𝑐 = 1 to solve for 𝑦2 (𝑥).

For example, assume 𝑟1 = 3.5 and 𝑟2 = 1.5. If 𝑦1 (𝑥) is already solved, then:

𝑦1 (𝑥) = 𝑥 3.5 (𝑎0 + 𝑎1 𝑥 + 𝑎2 𝑥 2 + ⋯ )

Assume for 𝑐 = 0, we get

𝑦2 (𝑥) = 𝑥1.5 (𝑏0 + 𝑏1 𝑥 + 𝑏2 𝑥 2 + ⋯ )

If 𝑏0 ≠ 0 𝑜𝑟 𝑏1 ≠ 0, then 𝑦1 (𝑥) and 𝑦2 (𝑥) are surely linearly independent and we are done.
Otherwise we have 𝑦2 (𝑥) = 𝑥1.5 (𝑏2 𝑥 2 + 𝑏3 𝑥 3 … ) = 𝑥 3.5 (𝑏2 + 𝑏3 𝑥 … ) = 𝑦1 (𝑥). Then we need
to choose 𝑐 = 1 to solve for 𝑦2 (𝑥) as in Case 2 with the repeated root.

20
Section V. Conclusion

The Laurent series are central to complex analysis and its applications. Although the residue of
some functions can be found easily with well-known formulas, this is not true in general. For
example, for a function with an essential singularity—where no such easy formula exists—the
Laurent expansion is the only way to determine the function residue at such a singularity. The
residues found through the Laurent series will greatly simplify complex integration, which can
be applied in many fields outside of pure mathematics. A similar series expansion called the
Frobenius method is also useful in solving second-order ordinary different equations with
singularities.

21
 References

Apelian, C. & Surace, S. (2010). Real and complex analysis. Chapman & Hall/CRC.

Campbell, A. & Daners, D. (2013). Linear Algebra via Complex Analysis. The American

Mathematical Monthly. 120:10, 877-892. DOI: 10.4169/amer.math.monthly.120.10.877

Kreyszig, E. (1993). Advanced engineering mathematics (Seventh ed.). New York: Wiley.

Loughborough University. Complex variables: Laurent series handout. SYMBOL.

http://sym.lboro.ac.uk/resources/Handout-Laurent.pdf.

Marsden, J. E., & Hoffman, M. J. (1999). Basic Complex Analysis (3rd ed.). New York, NY:

W.H. Freeman and Company.

Phinney, Sterl. “The Method of Frobenius (Lecture Notes).” Introductory Methods of Applied

Mathematics, California Institute of Technology,

www.its.caltech.edu/~esp/acm95b/frobenius.pdf.

Weisstein, Eric W. "Frobenius Method." From MathWorld--A Wolfram Web Resource

https://mathworld.wolfram.com/FrobeniusMethod.html

22