See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/323551388

A Physicist’s Look at the Multi-Valued Complex Functions: Theory and

Examples

Presentation · March 2018

DOI: 10.13140/RG.2.2.35836.23682

CITATIONS READS

0 46

1 author:

Spiros Konstantogiannis
Ronin Institute
21 PUBLICATIONS 9 CITATIONS

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Quantum Harmonic Oscillator View project

Quasi-Exactly Solvable Potentials View project

All content following this page was uploaded by Spiros Konstantogiannis on 05 March 2018.

The user has requested enhancement of the downloaded file.

A Physicist’s Look at the Multi-Valued Complex Functions:
Theory and Examples

Spiros Konstantogiannis
spiroskonstantogiannis@gmail.com
5 March 2018

1 5 March 2018
Copyright © 2018 by Spiros Konstantogiannis.
All rights reserved.
Only non-commercial use of content from the present document is permitted, provided
that a link to the document is included.

2 5 March 2018
 Contents
A Physicist’s Look at the Multi-Valued Complex Functions: Theory and
Examples ........................................................................................................... 1
Contents ............................................................................................................ 3
The infinite Riemann sheets of the complex argument resulting from the
degeneracy of the complex numbers .................................................................. 4
Definitions......................................................................................................... 7
The polar form of the complex numbers as a means of doing rotations on the
complex plane.................................................................................................... 8
Examples of multi-valued functions................................................................... 8
Branch point at infinity .....................................................................................12
General examples .............................................................................................16
References ........................................................................................................33

3 5 March 2018
 The infinite Riemann sheets of the complex argument resulting
from the degeneracy of the complex numbers
In polar – or geometric – form, a complex number z is written as

z = z exp ( i arg z ) (1)

where arg z is the argument of z , and it holds that

arg z = Argz + 2np (2)

where n = 0, ±1, ±2,..., ±¥ .

By Argz we denote the principal argument of z , which is, intuitively speaking,

the “first” argument of z , i.e. the angle belonging to the first positive rotation
around zero, with the condition that the initial point of the rotation circle is
included once only, i.e. the range of Argz is a closed-open or open-closed interval
of width 2p .

Usually, we choose

-p < Argz £ p or 0 £ Argz < 2p .

In the first case, the cut is taken along the negative real axis, i.e. we cannot cross the
negative real axis, while in the second case, the cut is taken along the positive real
axis, i.e. we cannot cross the positive real axis.
However, the cut can be taken along an arbitrary direction, provided that it contains
zero.
For instance, choosing as cut line the positive or the negative imaginary axis, the
range of Argz is, respectively,

3p p p 3p
- < Argz £ or - < Argz £ .
2 2 2 2

Since the absolute value of zero is zero, we cannot define an argument for the
complex number z = 0 , and thus 0 is excluded from the domain of arg z . In other
words, z ¹ 0 .
Also, considering only counterclockwise – or positive – rotations, the number n is
non-negative, i.e. n = 0,1,...

4 5 March 2018
Besides, we have

exp ( i arg z ) = cos ( arg z ) + i sin ( arg z ) = cos ( Argz + 2np ) + i sin ( Argz + 2np ) =
= cos ( Argz ) + i sin ( Argz ) = exp ( iArgz )

That is

exp ( i arg z ) = exp ( iArgz ) (3)

By means of (3), (1) is written as

z = z exp ( iArgz ) (4)

We see that z does not depend on n , i.e. changing n leaves z unchanged, which is
geometrically obvious.

This property can be thought of as degeneracy, inherent or built-in degeneracy,

of all non-zero complex numbers.

On the other hand, as seen from (2), a change of n by – let’s say – Dn , changes the
argument of z by 2Dnp , and thus the same complex number z has more than one
arguments argz .
Actually, every non-zero complex number z has infinite – but countable – arguments,
as each value of n corresponds to a different argument.
The function arg z is an example of a multi-valued complex function.
However, the function Argz , i.e. the principal argument, is a single-valued complex
function, because every non-zero complex number has only one principal argument.
Then, for every fixed value of n , the function Argz + 2np is also single-valued, and
thus, by (2), we derive that, for every fixed value of n , the function arg z is single-
valued too.

Every fixed value of n corresponds to a branch, or a Riemann sheet, of the

function arg z .
The function arg z has thus infinite – but countable – branches, or Riemann
sheets.

5 5 March 2018
All Riemann sheets of arg z constitute the Riemann surface of the function arg z ,
which is thus infinite-sheeted.
The function arg z maps the domain of z , which is the whole complex plane
except zero, to infinite Riemann sheets, which are
Argz, Argz + 2p , Argz + 4p ,...,
with Argz being the principal argument of z .

As mentioned, in defining the principal argument of z , we choose a cut line, which

necessarily contains zero. Thus, there is no way of doing a full rotation around zero,
and thus returning to the same point on the complex plane, i.e. z ® z , without
crossing a cut line.
After a full rotation, the value of arg z changes, and thus the function arg z moves to
another branch or Riemann sheet.

The point z = 0 is called a branch point of the function arg z , and the cut lines
from zero to infinity are called branch cuts of arg z .

On the other hand, if z0 is a non-zero complex number, there exists a radius e > 0

such that the point z = 0 , i.e. the origin, lies outside the circle z - z0 = e , and then we

can do a full rotation around z0 without crossing a cut line, i.e. there exists at least

one cut line that does not intersect the circle z - z0 = e .

Then, doing full rotations around z0 , the function arg z does not change branch, i.e. it

remains on the same Riemann sheet.

Thus, z0 ¹ 0 is not a branch point of the function arg z .

Note

If we consider the function arg ( z - z1 ) , then setting z ¢ = z - z1 , the previous

function is written as arg z ¢ , and thus it has a branch point at z ¢ = 0 , i.e. at z = z1 ,

and its branch cuts extend from z ¢ = 0 , i.e. from z = z1 , to infinity.

6 5 March 2018
 Definitions

Generally, a point z0 is a branch point of a complex function f ( z ) if in every

neighborhood of z0 , f ( z ) is multi -valued.

A neighborhood of z0 is any set of complex numbers containing the open disc

z - z0 < e , for some e > 0 .

If in the neighborhood of a branch point z0 , the function f ( z ) is n -valued, z0 is a

branch point of order n - 1 . Thus, for example, the point z = 0 is an infinite-order

branch point of the function arg z .
The following excerpt is from http://mathworld.wolfram.com/BranchPoint.html
A branch point of an analytic function is a point on the complex plane whose complex
argument can be mapped from a single point in the domain to multiple points in the
range.
The following excerpt is from http://mathworld.wolfram.com/BranchCut.html
A branch cut is a curve (with end possibly open, closed, or half-open) on the complex
plane across which an analytic multi-valued function is discontinuous.
For convenience, branch cuts are often taken as lines or segments.
Usually, a branch cut connects two branch points, but this is not necessary, as the
endpoints of branch cuts may not be branch points.
The following excerpt is from https://en.wikipedia.org/wiki/Branch_point
Roughly speaking, branch points are the points where the various sheets of a multi-
valued function come together.
The branches of the function are the various sheets of the function,
A branch cut is a curve on the complex plane such that it is possible to define a single
analytic branch of a multi-valued function on the complex plane minus that curve.
Branch cuts are usually, but not always, taken between pairs of branch points.
Branch cuts allow one to work with a collection of single-valued functions, “glued”
together along the branch cut instead of a multi-valued function.

7 5 March 2018
 The polar form of the complex numbers as a means of doing
rotations on the complex plane

Another way of seeing the polar form (1) is as representing a circle of radius z with

center at the origin.

The argument of z is given by (2), and by changing the number n , we do rotations
around the origin.
Thus, using the polar form (1), we can examine if the origin, i.e. the point z = 0 , is a
branch point of a given complex function f ( z ) .

Similarly, using the polar form of z - z0 , which is

z - z0 = z - z0 exp ( i arg ( z - z0 ) ) , with arg ( z - z0 ) = Arg ( z - z0 ) + 2np ,

we do rotations around the point z = z0 , and thus we can examine if the point z = z0

is a branch point of a given complex function f ( z ) .

Again, considering only counterclockwise – or positive – rotations, the number n is

non-negative, i.e. n = 0,1,...

Examples of multi-valued functions

1. The function z
We consider the function

f (z) = z

By means of the polar form (1), f ( z ) is written as

1 1
æ arg z ö
f ( z ) = ( z exp ( i arg z ) ) 2 = z 2 exp ç i ÷
è 2 ø

Substituting (2) into the last equality, we obtain

1
æ Argz + 2np ö 1
æ iArgz ö
f ( z ) = z 2 exp ç i ÷ = z 2 exp
ç + inp ÷ =
è 2 ø è 2 ø
1
1
æ Argz ö
( inp3) = ( -1) ( z exp ( iArgz ) ) 2
n
= z 2 exp ç i ÷ exp
è 2 ø 14 24
( -1)
n

Then, using (4), we end up to

8 5 March 2018
f ( z ) = ( -1)
n
z (5)

with n = 0,1, 2,... , as we consider only positive rotations.

From (5), we see that
a) If n is even, i.e. if n = 0, 2, 4,... , then f ( z ) = z . This is the principal branch of

the complex square root function.

b) If n is odd, i.e. if n = 1, 3,... , then f ( z ) = - z . This is the second branch of the

complex square root function.

Thus, the function z has two branches – two Riemann sheets – its first, or principal,
branch is z , and its second branch is - z .
The Riemann surface of z is thus two-sheeted. The even-numbered Riemann sheets
of arg z correspond to the first Riemann sheet of z , i.e. to z , while the odd-

numbered Riemann sheets of arg z correspond to the second Riemann sheet of z,

i.e. to - z .

The function z maps the infinite sheets of arg z to two sheets, which are ± z .

The point z = 0 is a first-order branch point of the function z.

From the previous, it is obvious that the square roots of a positive real number are
± x.

We thus see that the plus and minus square roots of a positive real number arise
naturally from the two branches – the two Riemann sheets – of the complex
square root function.

k
2. The function z
We consider the function

f ( z ) = k z , with k = 2,3,...

Using again the polar form (1), f ( z ) is written as

1 1 1
æ arg z ö
f ( z ) = z k = ( z exp ( i arg z ) ) k = z k exp ç i ÷
è k ø

Substituting (2) into the last equality, we obtain

9 5 March 2018
 1
æ Argz + 2np ö 1
æ iArgz 2np i ö
f ( z ) = z k exp ç i ÷ = z k exp ç + ÷=
è k ø è k k ø
æ 2np i ö æ 2np i ö æ 2np i ö k
1
1
æ iArgz ö 1
= z k exp ç ÷ exp ç ÷ (
= z exp ( i Argz ) ) ç k ÷
k exp = exp ç ÷ z
è k ø è k ø è ø è k ø

That is

æ 2np i ö k
f ( z ) = exp ç ÷ z (6)
è k ø

From (6), we obtain the branches

n = 0 , f ( z ) = k z (principal branch)

æ 2p i ö k
n = 1 , f ( z ) = exp ç ÷ z (second branch)
è k ø

æ 2 ( k - 1) p i ö k
n = k - 1 , f ( z ) = exp ç ÷ z ( k -th branch)
è k ø

For n = k , we obtain again the principal branch, as (6) gives

æ 2kp i ö k
f ( z ) = exp ç ÷ z = exp ( 2p3i ) k z = k z
è k ø 14 24
1

In general, for n = k + n¢ , with n¢ = 0,1,..., k - 1 , from (6) we obtain

æ 2 ( k + n¢ ) p i ö k æ 2kp i + 2n¢p i ö k æ 2n¢p i ö k

f ( z ) = exp ç ÷ z = exp ç ÷ z = exp ç 2p i + ÷ z=
è k ø è k ø è k ø
æ 2n¢p i ö k æ 2n¢p i ö k
= exp ( 2p i ) exp ç ÷ z = exp ç ÷ z
1424 3 è k ø è k ø
1

That is

æ 2n¢p i ö k
f ( z ) = exp ç ÷ z
è k ø

Then, if n¢ = 0 , we obtain the principal branch, if n¢ = 1 , we obtain the second branch,

and so on.
k
Thus, the function z has k branches, or k Riemann sheets, and its Riemann
surface is k -sheeted.

10 5 March 2018
 k
The function z maps the infinite sheets of arg z to k sheets, which are

k æ 2p i ö k æ 2 ( k - 1) p i ö k
z , exp ç ÷ z ,..., exp ç ÷ z.
è k ø è k ø

The point z = 0 is a ( k - 1) -th order branch point of the function k

z.

k
Observe that all branches of z result from the multiplication of the principal
k
branch z with the k k -th roots of unity.

The k -th roots of unity

Let us remember how the k -th roots of unity are derived.
If k ³ 1 , a k -th root of unity is a complex number z satisfying the equation

zk = 1
Using the polar form (1), the previous equation becomes

( z exp ( i arg z ) ) = 1 Þ z exp ( ik arg z ) = 1 Þ z exp ( ik arg z ) = exp ( 2np i )

k k k

with n = 0, ±1, ±2,...

From the last equation, we obtain
2np
z = 1 and k arg z = 2np Þ arg z = ,
k
and then

æ 2np i ö
z = exp ç ÷ , with n = 0, ±1, ±2,...
è k ø
To number the different roots z – let us denote them by zn – we consider only
positive rotations, and we choose the principal argument of z to be such that
0 £ Argz < 2p .
Then, the k -th roots of unity are

æ 2np i ö
zn = exp ç ÷ (7)
è k ø
with n = 0,1,..., k - 1
We thus see that there are k k -th roots of unity.
If k = 1 , n = 0 , and we have only one first root of unity, which is z0 = 1 .

If k = 2 , n = 0,1 , and we have two second – or square – roots of unity, which are

11 5 March 2018
 æ 2p i ö
z0 = 1 and z1 = exp ç ÷ = exp (p i ) = -1 , i.e. the square roots of unity are
è 2 ø
±1 .
If k = 3 , n = 0,1, 2 , and we have three third roots of unity, which are

æ 2p i ö æ 4p i ö
z0 = 1 , z1 = exp ç ÷ , and z2 = exp ç ÷.
è 3 ø è 3 ø

3. The logarithmic function

The logarithm of a non-zero complex number is given by

log z = ln z + i ( Argz + 2np ) ,

or, using (2),

log z = ln z + i arg z (8)

From (8), we see that there is a one-to-one correspondence between the Riemann
sheets of arg z and those of log z , and thus the Riemann surface of log z has infinite
sheets, it is infinite-sheeted.
The function log z maps the infinite sheets of arg z to infinite sheets.
The principal branch of arg z , i.e. the principal argument, corresponds to the principal
branch of log z , which is denoted by Logz , and thus

Logz = ln z + iArgz (9)

The point z = 0 is an infinite-order branch point of the function log z .

Branch point at infinity

A complex function f ( z ) has a k -th order branch point at infinity if the function

æ1ö
f ç ÷ has a k -th order branch point at zero.
èzø
Let us see some examples.
1. The function arg z has an infinite-order branch point at infinity.
Indeed, using the property

æz ö
arg ç 1 ÷ = arg z1 - arg z2 ,
è z2 ø

12 5 March 2018
we obtain for z1 = 1 and z2 = z ,

æ1ö
arg ç ÷ = arg1 - arg z ,
è zø

and since arg1 = 0 , we end up to

æ1ö
arg ç ÷ = - arg z
è zø

Then, since the function arg z has an infinite-order branch point at z = 0 , the function

æ1ö
arg ç ÷ also has an infinite-order branch point at z = 0 , and thus the function arg z
èzø
has an infinite-order branch point at infinity.
2. Since log z = ln z + i arg z , the logarithmic function also has an infinite-order

branch point at infinity.

3. f ( z ) = k z , with k = 2,3,...

We have

æ1ö 1 1
f ç ÷= k = k
è zø z z

Since the function k

z has a ( k - 1) -th order branch point at z = 0 , the function

æ1ö
f ç ÷ also has a ( k - 1) -th order branch point at z = 0 , and thus the function
èzø

f ( z ) = k z has a ( k - 1) -th order branch point at infinity.

4. f ( z ) = z ( z - z1 )

We have

1 - zz1 } 1 - zz1
*
æ1ö 1æ1 ö 1 - zz1
f ç ÷= ç - z1 ÷ = 2
= =
è zø zè z ø z z2 z

z1 z1
* In £ , it holds that = , for any complex numbers z1 , z2 , with z2 being
z2 z2
non-zero. In ¡ , this property holds only for positive numbers.

13 5 March 2018
That is

æ1ö 1 - zz1
f ç ÷=
è zø z

æ1ö
We see that the function f ç ÷ has not a branch point at z = 0 , and thus the function
èzø

f ( z ) = z ( z - z1 ) has not a branch point at infinity.

æ1ö
Actually, the function f ç ÷ has a simple pole at z = 0 , and thus the function
èzø

f ( z ) = z ( z - z1 ) has a simple pole at infinity.

5. f ( z ) = z ( z - z1 )( z - z2 )

We have

æ1ö
f ç ÷=
1æ1 öæ 1 ö (1 - zz1 )(1 - zz2 ) = (1 - zz1 )(1 - zz2 )
ç - z1 ÷ç - z2 ÷ =
è zø zè z øè z ø z3 z z

That is

æ1ö (1 - zz1 )(1 - zz2 )

f ç ÷=
è zø z z

æ1ö
Since the function z has a first-order branch point at z = 0 , the function f ç ÷ also
èzø
has a first-order branch point at z = 0 , and then the function

f ( z ) = z ( z - z1 )( z - z2 ) has a first-order branch point at infinity.

6. f ( z ) = ( z - z1 )( z - z2 ) ... ( z - zm ) , with m = 1, 2,...

We have

æ1ö æ1 öæ 1 ö æ1 ö
f ç ÷ = ç - z1 ÷ ç - z2 ÷ ... ç - zm ÷ =
(1 - zz1 )(1 - zz2 ) ... (1 - zzm ) =
è zø èz øè z ø èz ø zm
(1 - zz1 )(1 - zz2 ) ... (1 - zzm )
= m
z2

14 5 March 2018
That is

æ1ö (1 - zz1 )(1 - zz2 ) ... (1 - zzm )

f ç ÷= m
èzø
z 2

m
If m is a non-zero even number, i.e. then = 1, 2,... , and thus z m 2 is single-valued,
2
æ1ö
and then f ç ÷ has not a branch point at z = 0 , and thus
èzø

f (z) = ( z - z1 )( z - z2 ) ... ( z - zm ) has not a branch point at infinity.

m -1
= 0,1,... , and thus z m 2 = z ( ) z , with the function z ( )
m -1 2 m -1 2
If m is odd, then
2
being single-valued and the function z having a first-order branch point at z = 0 .
Thus, the function z m 2 has a first-order branch point at z = 0 , and then the function
æ1ö
f ç ÷ also has a first-order branch point at z = 0 , which means that the function
èzø

f (z) = ( z - z1 )( z - z2 ) ... ( z - zm ) has a first-order branch point at infinity.

Therefore, if m is a non-zero even number, the previous function has not a branch
point at infinity, while if m is an odd number, the previous function has a first-order
branch point at infinity.
7. On the contrary, the function

f ( z ) = z - z1 + z - z2 + ... + z - zm , with m = 1, 2,... ,

has always, i.e. for every value of m , a first-order branch point at infinity.
Indeed, we have

æ1ö 1 1 1 1 - zz1 1 - zz2 1 - zzm

f ç ÷= - z1 + - z 2 + ... + - zm = + + ... + =
è zø z z z z z z
1 - zz1 1 - zz2 1 - zz m 1 - zz1 + 1 - zz2 + ... + 1 - zz m
= + + ... + =
z z z z

That is

æ1ö 1 - zz1 + 1 - zz2 + ... + 1 - zz m

f ç ÷=
èzø z

15 5 March 2018
We see that for every value of m , there is a square root of z (in the denominator),
which has a first-order branch point at z = 0 .
æ1ö
Thus, the function f ç ÷ has a first-order branch point at z = 0 , and then the function
èzø

f ( z ) = z - z1 + z - z2 + ... + z - zm has a first-order branch point at infinity, for

every non-zero positive integer m .

Thus, for instance, the function ( z - z1 )( z - z2 ) has two first-order branch points, at

z = z1 and at z = z2 , while the function z - z1 + z - z2 has three first-order branch

points, at z = z1 , at z = z2 , and at infinity.

General examples

1. Find the branches of the function f ( z ) = z 2 3 .

Using (1) and (2), f ( z ) is written as

æ 2iArgz 2np i ö
2
æ 2 ö
( )
2 2
f ( z ) = z exp ( i ( Argz + 2np ) ) = z 3 exp ç i ( Argz + 2np ) ÷ = z 3 exp ç
3
+ ÷=
è 3 ø è 3 3 ø
2
æ 2iArgz ö æ 2np i ö æ 2np i ö 2
= z 3 exp ç ÷ exp ç ÷ = exp ç ÷( z exp ( i Argz ) )3
è 3 ø è 3 ø è 3 ø

Then, making use of (4), we end up to

æ 2np i ö 3
2
f ( z ) = exp ç ÷z (10)
è 3 ø

Thus, we have the branches

2
n = 0 , f ( z ) = z (principal branch)
3

æ 2p i ö 3
2

n = 1 , f ( z ) = exp ç ÷ z (second branch)

è 3 ø

æ 4p i ö 3
2

n = 2 , f ( z ) = exp ç ÷ z (third branch)

è 3 ø

For n = 3 ,

æ 6p i ö 3
2 2 2
f ( z ) = exp ç ÷ z = exp ( 2p i ) z = z ,
3 3

è 3 ø

16 5 March 2018
i.e. we obtain again the principal branch.
Thus, the function z 2 3 has three branches, which correspond to n = 0,1, 2 .

The point z = 0 is a second-order branch point of the function z 2 3 .

The function z 2 3 maps the infinite sheets of arg z to three sheets, which are

z 2 3 , exp ( 2p i 3) z 2 3 , exp ( 4p i 3) z 2 3 .

2. Find the branches of the function f ( z ) = z m1 m2 , where m1 , m2 are positive

integers.
m1 m
If = 1, 2,... , i.e. if 1 is a positive integer, then the function f ( z ) is single-
m2 m2
valued, i.e. it has only one branch.
m1 m
If ¹ 1, 2,... , i.e. if 1 is a rational number (not an integer), then using (1) and (2),
m2 m2

f ( z ) is written as

m1 m1
æm ö
( ) ( exp ( i ( Argz + 2np ) ) )
m1 m1
f ( z ) = z exp ( i ( Argz + 2np ) ) m2
= z m2 = z m2 exp ç 1 i ( Argz + 2np ) ÷ =
m2

è m2 ø
m1 æm 2m np i ö m1 æm ö æ 2m np i ö
= z m2 exp ç 1 iArgz + 1 ÷ = z m2 exp ç 1 iArgz ÷ exp ç 1 ÷=
è m2 m2 ø è m2 ø è m2 ø
m1
æ ö m2 æ 2m1np i ö
m
æ 2m1np i ö m12
= z exp ( iArgz )
ç ÷ exp ç ÷ = exp ç ÷z
ç 144244 3÷ è m2 ø è m2 ø
è z ø

That is
m
æ 2m np i ö m12
f ( z ) = exp ç 1 ÷z (11)
è m2 ø

Thus, we have the branches

m1

n = 0 , f (z) = z m2
(principal branch)

m
æ 2m1p i ö m12
n = 1 , f ( z ) = exp ç ÷ z (second branch)
è m2 ø

17 5 March 2018
 æ 2m ( m - 1) p i ö m12
m

n = m2 - 1 f ( z ) = exp ç 1 2 ÷ z ( ( m2 - 1) -th branch)

è m2 ø

For n = m2 ,

m m
æ 2m m p i ö 1 1

f ( z ) = exp ç 1 2 ÷ z m2 = exp ( 2m1p i ) z m2

è m2 ø

Since m1 is an integer, then exp ( 2m1p i ) = 1 , and thus

m1

f (z) = z , m2

i.e. we obtain again the principal branch.

Thus, if m1 m2 is a rational number (not an integer), the function z m1 m2
has m2

branches, which correspond to n = 0,1,..., m2 - 1 .

The point z = 0 is then a ( m2 - 1) -th order branch point of the function z m1 m2

.

The function z m1 m2
maps the infinite sheets of arg z to m2 sheets, which are

z m1 m2 , exp ( 2m1p i m2 ) z m1 m2
,..., exp ( 2m1 ( m2 - 1) p i m2 ) z m1 m2
.

Observe that the number of branches of the function z m1 m2

does not depend on m1 , it

depends only on m2 , i.e. on the denominator of the exponent m1 m2 , provided that

the exponent is a rational number (not an integer).

As a consequence, each of the functions z1 m , z 2 m ,..., z ( m -1) m , with m = 2, 3,.... , has m

branches, and the point z = 0 is an ( m - 1) -th order branch point of each of these

functions, and each of these functions maps the infinite sheets of arg z to m sheets.

3. How many branches does the function f ( z ) = z 2

have?

Using (1) and (2), f ( z ) is written as

(
f ( z ) = z exp ( i ( Argz + 2np ) ) ) ( )
2
2i ( Argz + 2np ) =
2
= z exp

= z
2
exp ( 2iArgz + 2 2np i = z ) 2
exp ( ) (
2iArgz exp 2 2np i = )
2
æ ö
(
= ç z exp ( iArgz ) ÷ exp 2 2np i = exp 2 2np i z
ç 14 4244 3÷ ) ( ) 2

è z ø

18 5 March 2018
That is

(
f ( z ) = exp 2 2np i z ) 2
(12)

For n = 0 , from (12) we obtain the principal branch, which is z 2 .

We observe that there is no other n ¹ 0 , such that f ( z ) becomes again z 2 , i.e. we

cannot return to the principal branch.

Indeed, to return to the principal branch, there must be an n ¹ 0 such that the phase

2 2np becomes an integer multiple of 2p , i.e. 2 2np = 2mp , with m being a

positive integer.
But then we have

} m
n ¹0

2 2np = 2mp Þ 2n = m Þ 2 =
n

Since both m and n are integers, m n is rational, while 2 is irrational. Thus, the
previous equation is impossible, and then we cannot return to the principal branch.
Similarly, we can show that if m, n are different positive integers, the respective

branches of f ( z ) – let us denote them by f m ( z ) and f n ( z ) , respectively – are also

different.
Indeed, if f m ( z ) = f n ( z ) , then

}
*

(
exp 2 2mp i z ) 2
(
= exp 2 2np i z ) 2
( ) (
Þ exp 2 2mp i = exp 2 2np i Þ )
Þ 2 2mp = 2 2np + 2kp , with k being an integer.

* We want the equation to hold for z ¹ 0 .

Then, we have

} m¹ n
k
2 2 ( m - n ) p = 2kp Þ 2 ( m - n ) = k Þ 2 =
m-n

Since m, n, k are integers, k ( m - n ) is rational, while 2 is irrational, and thus the

previous equation is impossible, and then f m ( z ) ¹ f n ( z ) .

19 5 March 2018
Therefore, any two different values of n correspond to different branches of the

function z 2 .
Thus, there is a one-to-one correspondence between the positive integers and the

branches z 2 , which means that z 2

has infinite – but countable – branches.

The point z = 0 is then an infinite-order branch point of z 2 .

2
The function z maps the infinite sheets of arg z to infinite sheets.

In the same way we can show that the function z q , with q being an irrational
number, has infinite branches.

4. i) Find the branches of f ( z ) = ( z - 1) .

34

ii) Choosing a branch cut parallel to the positive imaginary axis, find the values
the previous function takes at z = 0 on its different branches, and at z = 2 on its
last branch.
i) Setting z ¢ = z - 1 , the function becomes f ( z ¢ ) = z ¢3 4 . Then, we have the case

z¢( m-1) m ( m = 4 ) , we discussed in the example 2 of this section. Thus, the function
z¢3 4 has a third-order branch point at z = 0 , which means that the function ( z - 1)
34

has a third-order branch point at z = 1 .

Then, the function f ( z ) has four branches.

In polar form, z - 1 is written as

z - 1 = z - 1 exp ( i arg ( z - 1) )

with

arg ( z - 1) = Arg ( z - 1) + 2np

where n = 0,1,... , as we consider only positive rotations.

Substituting the last equation into the next-to-last, we obtain

(
z - 1 = z - 1 exp i ( Arg ( z - 1) + 2np ) ) (13)

For n = 0 , (13) gives

20 5 March 2018
z - 1 = z - 1 exp ( iArg ( z - 1) ) (14)

Using (13), f ( z ) is written as

( ( )) æ3 ö
3
f ( z ) = z - 1 exp i ( Arg ( z - 1) + 2np ) = z - 1 4 exp ç i ( Arg ( z - 1) + 2np ) ÷ =
4

è4 ø
3
æ3 6np i ö 3
æ3 ö æ 3np i ö
= z - 1 4 exp ç iArg ( z - 1) + ÷ = z - 1 4 exp ç iArg ( z - 1) ÷ exp ç ÷=
è4 4 ø è4 ø è 2 ø
æ 3np i ö
3

(
= z - 1 exp ( iArg ( z - 1) ) 4 exp ç )
è 2 ø
÷

Then, using (14), we end up to

æ 3np i ö 3
f ( z ) = exp ç ÷ ( z - 1) 4 (15)
è 2 ø

Thus, we have the branches

3
n = 0 , f 0 ( z ) = ( z - 1) 4 (principal branch)

æ 3p i ö 3 3
n = 1 , f1 ( z ) = exp ç ÷ ( z - 1) 4 = -i ( z - 1) 4 = -if ( z )
0
1424è 23ø
-i

3 3
n = 2 , f 2 ( z ) = exp ( 3p i )( z - 1) 4 = - ( z - 1) 4 = - f 0 ( z )

n = 3,
æ 9p i ö 3
æ 8p i p i ö 3
æ pi ö 3
f3 ( z ) = exp ç ÷ ( z - 1) 4 = exp
ç + ÷ ( z - 1) 4 = exp 4p i +
ç ÷ ( z - 1) 4 =
è 2 ø è 2 2ø è 2ø
æ pi ö 3 3
= exp ( 4p i ) exp ç ÷ ( z - 1) 4 = i ( z - 1) 4 = if 0 ( z )
1424 3 è 23ø
1 1424
i

For n = 4 ,

æ 12p i ö 3 3 3
f 4 ( z ) = exp ç ÷ ( z - 1) 4 = exp ( 6p i )( z - 1) 4 = ( z - 1) 4 = f 0 ( z ) ,
è 2 ø 1424 3
1

i.e. we obtain again the principal branch.

The branches of f ( z ) = ( z - 1)
34
are then

3
f 0 ( z ) = ( z - 1) 4 (principal branch)

21 5 March 2018
f1 ( z ) = -if 0 ( z ) (second branch)

f 2 ( z ) = - f 0 ( z ) (third branch)

f3 ( z ) = if 0 ( z ) (fourth branch)

The function ( z - 1)
34
has four branches – four Riemann sheets – which correspond

to n = 0,1, 2,3 .

The function ( z - 1)
34
maps the infinite sheets of arg z to four sheets, which are

( z - 1) , -i ( z - 1) , - ( z - 1) , i ( z - 1) .
34 34 34 34

The Riemann surface of ( z - 1)

34
is thus four-sheeted.

ii) Choosing the branch cut means choosing the range of the principal argument,
i.e. the values the principal argument takes.

Thus, choosing the branch cut parallel to the positive imaginary axis means that

3p p 3p p
- < Arg ( z - 1) £ or - £ Arg ( z - 1) < .
2 2 2 2

Using (14), the principal branch of ( z - 1)

34
is written as

3
æ3 ö
( )
3
f 0 ( z ) = z - 1 exp ( iArg ( z - 1) ) 4
= z - 1 4 exp ç iArg ( z - 1) ÷ =
è4 ø
3 æ 3iArg ( z - 1) ö
= z - 1 4 exp ç ÷
è 4 ø

That is

3 æ 3iArg ( z - 1) ö
f 0 ( z ) = z - 1 4 exp ç ÷ (16)
è 4 ø
3 3
For z = 0 , Arg ( z - 1) = Arg ( -1) = -p and z - 1 4 = -1 4 = 1 .

Substituting into (16), we obtain that the principal value of ( z - 1)

34
at z = 0 is

æ 3p i ö
f 0 ( 0 ) = exp ç - ÷
è 4 ø

22 5 March 2018
But

æ 3p i ö æ 3p ö æ 3p
ö ö æ 3p
æ 3p ö
exp ç - ÷ = cos ç - ÷ + i sin ç - ÷ = cos ç
÷ - i sin ç ÷=
è 4 ø è 4 ø è 4
ø ø è 4 øè 4
æp p ö æp p ö æp ö æp ö 2 2 2
= cos ç + ÷ - i sin ç + ÷ = - sin ç ÷ - i cos ç ÷ = - -i =- (1 + i )
è2 4ø è2 4ø è4ø è4ø 2 2 2

That is

æ 3p i ö 2
exp ç - ÷=- (1 + i )
è 4 ø 2

Thus, the principal value of ( z - 1)

34
at z = 0 is

2
f0 (0 ) = - (1 + i ) (17)
2

Then, the values of ( z - 1)

34
on its other branches, at z = 0 , are

æ 2 ö 2 2 2
f1 ( 0 ) = -if 0 ( 0 ) = -i çç - (1 + i ) ÷÷ = ( i + i 2 ) = ( i - 1) = - (1 - i )
è 2 ø 2 2 2

That is

2
f1 ( 0 ) = - (1 - i )
2

2
f 2 ( 0 ) = - f0 ( 0 ) = (1 + i )
2

That is

2
f2 ( 0 ) = (1 + i )
2

æ 2 ö 2 2 2
f3 ( 0 ) = if 0 ( 0 ) = i çç - (1 + i ) ÷÷ = - ( i + i 2 ) = - ( i - 1) = (1 - i )
è 2 ø 2 2 2

That is

2
f3 ( 0 ) = (1 - i )
2

For z = 2 , Arg ( z - 1) = Arg (1) = 0 and z - 1

34 34
=1 =1.

23 5 March 2018
Substituting into (16), we obtain that the principal value of ( z - 1)
34
at z = 2 is

f 0 ( 2 ) = exp ( 0 ) = 1

Then, using that f3 ( z ) = if 0 ( z ) , we obtain

f3 ( 2 ) = if 0 ( 2 ) = i

Thus, the value of ( z - 1)

34
at z = 2 , on its last branch, is i .

5. We’ll examine the functions f ( z ) = ( z - 1)( z - 2) and g ( z ) = z - 1 + z - 2 .

As we saw in the section “Branch point at infinity”, the function f ( z ) has not a

branch point at infinity, while the function g ( z ) has.

Thus, f ( z ) has two first-order branch points, at z = 1 and at z = 2 , while g ( z ) has

three first-order branch points, at z = 1 , at z = 2 , and at infinity.

Since* ( z - 1)( z - 2) = z - 1 z - 2 , then f ( z ) = z - 1 z - 2 .

* z1 z2 = z1 z2 , for any complex numbers. We note that, in ¡ , this property

holds only for positive numbers.

The function z - 1 has two branches, ± z - 1 , and similarly, the function z-2
has two branches, ± z - 2 .
Thus, the function f ( z ) has two branches, which are

± z -1 z - 2 ,

while the function g ( z ) has four branches, which are

z -1 + z - 2 , - z - 1 + z - 2 , z -1 - z - 2 , - z - 1 - z - 2 .

Rotating around the point z = 1 , with the point z = 2 being outside the circle of
rotation, changes the sign of z - 1 , while z - 2 does not change.
Rotating around the point z = 2 , with the point z = 1 being outside the circle of
rotation, changes the sign of z - 2 , while z - 1 does not change.

24 5 March 2018
Rotating around the point z = 1 or z = 2 , with the other branch point being inside the
circle of rotation, changes the sign of both z - 1 and z-2.
Thus
i. Rotating around the point z = 1 , with the point z = 2 being outside the circle of
rotation, gives both branches of f ( z ) , and the branches z - 1 + z - 2 (principal

branch) and - z - 1 + z - 2 of g ( z ) .

ii. Rotating around the point z = 2 , with the point z = 1 being outside the circle of
rotation, gives again both branches of f ( z ) , and the branches z -1 + z - 2

(principal branch) and z - 1 - z - 2 of g ( z ) .

iii. Rotating around the point z = 1 or z = 2 , with the other branch point being inside
the circle of rotation, gives only the branch z - 1 z - 2 (principal branch) of f ( z ) ,

and the branches z - 1 + z - 2 (principal branch) and - z - 1 - z - 2 of g ( z ) .

From the previous, it is clear that a branch cut from the branch point z = 1 to the
branch point z = 2 is a valid branch cut of the function z - 1 z - 2 , as this
function is single-valued around any circle containing both branch points, but it
is NOT a valid branch cut of the function z - 1 + z - 2 , as this function is not
single-valued on the circle iii, and generally, it is not single-valued on every circle
with the same center and bigger radius.

Compared to the function z - 1 z - 2 , the function z - 1 + z - 2 has an extra

branch point, at infinity, and because of that its valid branch cuts must extend to
infinity.
With respect to the function z -1 z - 2 :
One rotation around the circle iii is equivalent to two rotations around one of the
circles i or ii.
One rotation around the circle iii is equivalent to one rotation around both of the
circles i and ii.
With respect to the function z -1 + z - 2 :
One rotation around the circle iii is equivalent to one rotation around both of the
circles i and ii.

25 5 March 2018
6. We’ll examine the function f ( z ) = 3 z ( z - 1) .

The function f ( z ) is written as

f ( z ) = 3 z 3 z -1

k
As we saw in the section “Examples of multi-valued functions”, the function z,
with k = 2,3,... , has a ( k - 1) -th order branch point at z = 0 .

Thus, the function 3

z has a second-order branch point at z = 0 , and similarly, the
function 3
z - 1 has a second-order branch point at z = 1 .
Also, we have
1
æ1ö 1æ1 ö 1 æ 1 - z ö 3 1 - z 3 1 - z (1 - z ) 3
f ç ÷ = 3 ç - 1÷ = 3 ç ÷= = = 2
è zø zè z ø zè z ø z2 3 2
z z3

That is
1
æ 1 ö (1 - z ) 3
f ç ÷= 2
èzø
z3

As we saw in the example 1 of this section, the function z 2 3 has a second-order

æ1ö
branch point at z = 0 , and thus f ç ÷ has a second-order branch point at z = 0 ,
èzø
which means that the function f ( z ) has a second-order branch point at infinity.

Therefore, the function f ( z ) = 3 z ( z - 1) has three second-order branch points, at

z = 0 , at z = 1 , and at infinity.
Let us now find the branches of f ( z ) .
k
As we saw in the section “Examples of multi-valued functions”, the branches of z,
with k = 2,3,... , are
k
z , exp ( 2p i k ) k z ,..., exp ( 2 ( k - 1) p i k ) k z ,
k
and they are given by the product of the principal branch z with the k k -th roots
of unity.
For k = 3 , we obtain the branches of 3
z , which are

26 5 March 2018
3
z , exp ( 2p i 3) 3 z , exp ( 4p i 3) 3 z .

Similarly, the branches of 3

z - 1 are
3
z - 1, exp ( 2p i 3) 3 z - 1, exp ( 4p i 3) 3 z - 1 .

This is easily shown if we set z ¢ = z - 1 . Then, the function 3

z - 1 becomes 3
z¢ ,
which is the function 3
z with different variable. Thus, doing rotations around z ¢ = 0 ,
i.e. around z = 1 , we obtain the branches of 3
z ¢ , which are the same as the branches
of 3
z if we replace z with z ¢ .
Since the k -th roots of unity form an abelian group under multiplication, the
multiplication of two k -th roots of unity gives a k -th root of unity, and the result
does not change if we change the order of the two roots.
Then, the function f ( z ) = 3 z 3 z - 1 has three branches, which are

3
z 3 z - 1 (principal branch)

æ 2p i ö 3 3
exp ç ÷ z z - 1 (second branch)
è 3 ø

æ 4p i ö 3 3
exp ç ÷ z z - 1 (third branch)
è 3 ø

Rotating around the origin, with z = 1 being outside the circle of rotation, we obtain
all branches of f ( z ) .

Similarly, rotating around z = 1 , with the origin being outside the circle of rotation,
we obtain again all branches of f ( z ) .

Now, let us rotate around the origin or around z = 1 , with the other of the two branch
points being inside the circle of rotation, and let us assume that the initial value of
f ( z ) is 3
z 3 z - 1 , i.e. we are on the principal branch.

After a full positive rotation, the value 3

z changes to exp ( 2p i 3) 3 z , while the

value of 3
z - 1 changes to exp ( 2p i 3) 3 z - 1 .

Thus, the value of 3

z 3 z - 1 changes to

æ 2p i ö 3 æ 2p i ö 3 æ 4p i ö 3 3
exp ç ÷ z exp ç ÷ z - 1 = exp ç ÷ z z -1 ,
è 3 ø è 3 ø è 3 ø

27 5 March 2018
i.e. we obtain the third branch of f ( z ) .

Similarly, after one more rotation, the value exp ( 4p i 3) 3 z 3 z - 1 changes to

æ 4p i ö æ æ 2p i ö 3 ö æ æ 2p i ö 3 ö æ 8p i ö 3 3 æ 6p i + 2p i ö 3 3
exp ç ÷ ç exp ç ÷ z ÷ ç exp ç ÷ z - 1 ÷ = exp ç ÷ z z - 1 = exp ç ÷ z z -1 =
è 3 øè è 3 ø øè è 3 ø ø è 3 ø è 3 ø
æ 2p i ö 3 3 æ 2p i ö 3 3 æ 2p i ö 3 3
= exp ç 2p i + ÷ z z - 1 = exp1424
( 2p i ) exp ç
3 ÷ z z - 1 = exp ç ÷ z z -1 ,
è 3 ø è 3 ø è 3 ø
1

i.e. we obtain the second branch of f ( z ) .

Doing one more rotation, the value exp ( 2p i 3) 3 z 3 z - 1 changes to

æ 2p i ö æ æ 2p i ö 3 öæ æ 2p i ö 3 ö æ 6p i ö 3 3
exp ç ÷ ç exp ç ÷ z ÷ ç exp ç ÷ z - 1 ÷ = exp ç ÷ z z -1 =
è 3 øè è 3 ø øè è 3 ø ø è 3 ø
= exp ( 2p i ) 3 z 3 z -1 = 3 z 3 z -1 ,
1424 3
1

i.e. we return to the principal branch.

Thus, after three full rotations, the function 3
z 3 z - 1 returns to its principal branch.
Schematically, rotating around the “big” circle, i.e. around the circle containing both
points z = 0 and z = 1 , we obtain

}
1 full rotation
æ 4p i ö 3 3
3
z z -1
3
® exp ç ÷ z z -1
è 3 ø

}
2 full rotations
æ 2p i ö 3 3
3
z z -1
3
® exp ç ÷ z z -1
è 3 ø

}
3 full rotations
3
z z -1
3
® 3
z 3 z -1

Obviously, if we increase the radius of the circle to infinity, the above scheme does
not change, and this explains why the function 3
z 3 z - 1 has a second-order branch
point at infinity. We need three full rotations on the “big” circle to return to the
principal branch of 3
z 3 z -1 .

From the above, it is clear that a branch cut from the branch point z = 0 to the

branch point z = 1 is NOT a valid branch cut of the function 3

z 3 z - 1 , because
rotating around a circle centered at one of these points with radius big enough to

28 5 March 2018
encircle the other point changes the value of the function 3
z 3 z - 1 , i.e. the
function 3
z 3 z - 1 is not single-valued on that circle.

Because of the presence of a branch point at infinity, the valid branch cuts of
3
z 3 z - 1 must extend to infinity.
7. We’ll examine the function f ( z ) = z 3 z - 1 .

The function z has a first-order branch point at z = 0 , and its branches are ± z .
As we saw in the previous exercise, the function 3
z - 1 has a second-order branch
point at z = 1 and its branches are 3
z - 1, exp ( 2p i 3) 3 z - 1, exp ( 4p i 3) 3 z - 1 .

To examine if f ( z ) has a branch point at infinity, we do the transformation z ® 1 z ,

and then

æ1ö 13 1 1 3 1- z 1 3 1 - z 3 1- z
f (z) ® f ç ÷ = -1 = = = =
èzø z z z z z 3z z3 z
1 1 1

=
(1 - z ) 3 = (1 - z ) 3 = (1 - z ) 3
1 1 1 1 5
+
z z
2 3
z 2 3
z 6

That is
1
æ 1 ö (1 - z ) 3
f ç ÷= 5
èzø
z6

As we saw in the example 2 of this section, the function z ( m-1) m has a ( m - 1) -th

order branch point at z = 0 . Thus, the function z 5 6 has a fifth-order branch point at
æ1ö
z = 0 , and then the function f ç ÷ also has a fifth-order branch point at z = 0 . Then,
èzø
the function f ( z ) has a fifth-order branch point at infinity.

Thus, the function f ( z ) = z 3 z - 1 has a first-order branch point at z = 0 , a second-

order branch point at z = 1 , and a fifth-order branch point at infinity.

The branches of f ( z ) result from the multiplication of the branches of z and
3
z -1 .
Thus, the branches of f ( z ) are

29 5 March 2018
 z 3 z - 1 (principal branch)

æ 2p i ö 3
exp ç ÷ z z -1
è 3 ø

æ 4p i ö 3
exp ç ÷ z z -1
è 3 ø

- z 3 z -1

æ 2p i ö 3
- exp ç ÷ z z -1
è 3 ø

æ 4p i ö 3
- exp ç ÷ z z -1
è 3 ø

We’ll calculate ± exp ( 2p i 3) and ± exp ( 4p i 3) explicitly to show that the six

branches are indeed different. We have

æ 2p i ö æ 3p i - p i ö æ pi ö æ pi ö æ æ pö æ p öö
exp ç ÷ = exp ç ÷ = exp ç p i - ÷ = exp ( p i ) exp ç - ÷ = - ç cos ç - ÷ + i sin ç - ÷ ÷ =
è 3 ø è 3 ø è 3ø 1 424 3 è 3ø è è 3ø è 3 øø
-1

æ æp ö æ p öö æ1 3ö 1 3
= - ç cos ç ÷ - i sin ç ÷ ÷ = - çç - i ÷÷ = - + i
è è3ø è 3 øø è2 2 ø 2 2

That is

æ 2p i ö 1 3
exp ç ÷ = - +i
è 3 ø 2 2

And

æ 4p i ö æ 3p i + p i ö æ pi ö æ pi ö
exp ç ÷ = exp ç ÷ = exp ç p i + ÷ = exp
1424
(p3
i ) exp ç ÷ =
è 3 ø è 3 ø è 3ø
-1
è 3ø
æ æp ö æ p öö æ1 3ö
= - ç cos ç ÷ + i sin ç ÷ ÷ = - çç + i ÷
è è3ø è 3 øø è2 2 ÷ø

That is

æ 4p i ö 1 3
exp ç ÷ = - -i
è 3 ø 2 2

Thus, the branches of f ( z ) = z 3 z - 1 are also written as

30 5 March 2018
 z 3 z - 1 (principal branch)

æ 1 3ö 3
çç - + i ÷ z z -1
è 2 2 ÷ø

æ 1 3ö 3
çç - - i ÷ z z -1
è 2 2 ÷ø

- z 3 z -1

æ1 3ö 3
çç - i ÷ z z -1
è2 2 ÷ø

æ1 3ö 3
çç + i ÷ z z -1
è2 2 ÷ø

Now, let us imagine a circle centered at the origin or at z = 1 , with radius big enough
so that it encircles the other point.
Assuming that we start from the principal branch, i.e. from the value z 3 z - 1 , after
a full positive rotation, the value of z changes to - z , while the value of 3
z -1
changes to exp ( 2p i 3) 3 z - 1 .

Thus, the value z 3 z - 1 changes to

æ 2p i ö 3 æ 2p i ö 3
- z exp ç ÷ z - 1 = - exp ç ÷ z z -1 ,
è 3 ø è 3 ø

or, schematically,

}
1 full rotation
æ 2p i ö 3
z z -1
3
® - exp ç ÷ z z -1
è 3 ø

In the same way, doing one more rotation, the value - exp ( 2p i 3) z 3 z - 1 changes

to

æ 2p i ö æ æ 2p i ö 3 ö æ 4p i ö 3
- exp ç ÷ ç - z exp ç ÷ z - 1 ÷ = exp ç ÷ z z -1 .
è 3 øè è 3 ø ø è 3 ø

Thus

31 5 March 2018
 }
2 full rotations
æ 4p i ö 3
z z -1
3
® exp ç ÷ z z -1
è 3 ø

Then, doing successive rotations, we have

}
1 full rotation
æ 2p i ö 3
z 3 z -1 ® - exp ç ÷ z z -1
è 3 ø

}
2 full rotations
æ 4p i ö 3
z z -1
3
® exp ç ÷ z z -1
è 3 ø

}
3 full rotations
æ 4p i ö æ æ 2p i ö 3 ö æ 6p i ö 3
z 3 z -1 ® exp ç ÷ ç - z exp ç ÷ z - 1 ÷ - exp ç ÷ z z -1 = - z z -1
3

è 3 øè è 3 ø ø 1424 è 33ø
exp( 2p i ) =1

}
4 full rotations
æ 2p i ö æ 2p i ö 3
z 3 z -1 ® - exp ç
è 3 ø
( )
÷ - z z - 1 = exp ç
3

è 3 ø
÷ z z -1

}
5 full rotations
æ 2p i ö æ æ 2p i ö 3 ö æ 4p i ö 3
z z -1
3
® - exp ç ÷ ç exp ç ÷ z z - 1 ÷ = - exp ç ÷ z z -1
è 3 øè è 3 ø ø è 3 ø

}
6 full rotations
æ 2p i ö æ æ 4p i ö 3 ö
z z -1
3
® - exp ç ÷ ç - exp ç ÷ z z -1 ÷ = z z - 1
3

è 3 øè è 3 ø ø

After six full rotations, we return to the initial value, i.e. we return to the principal
branch z 3 z -1 .
Thus, rotating around the circle containing the origin and the point z = 1 , we
successively obtain the branches

z 3 z - 1, - exp ( 2p i 3) z 3 z - 1, exp ( 4p i 3) z 3 z - 1, - z 3 z - 1, exp ( 2p i 3) z 3 z - 1

and - exp ( 4p i 3) z 3 z - 1 , which are all the branches of z 3 z - 1 , arranged in

different order.
Obviously, we can increase the radius of the circle to infinity without changing the
previous scheme, and this explains why the function z 3 z - 1 has a fifth-order
branch point at infinity – we need six full rotations on the big circle to return to the
principal branch of z 3 z -1 .

From the above, it is clear that, as in the previous example, a branch cut from
the branch point z = 0 to the branch point z = 1 is NOT a valid branch cut of the

32 5 March 2018
 function z 3 z - 1 , because rotating around a circle centered at one of these two
points, with radius big enough to encircle the other point, changes the value of
the function z 3 z - 1 , and thus the function z 3 z - 1 is not single-valued on
that circle.

Because of the presence of a branch point at infinity, the valid branch cuts of
z 3 z - 1 must extend to infinity.

References
[1] James Ward Brown, Ruel V. Churchill, Complex Variables and Applications
(McGraw-Hill, Eighth Edition, 2009).
[2] http://mathworld.wolfram.com/BranchPoint.html.
[3] http://mathworld.wolfram.com/BranchCut.html.
[4] https://en.wikipedia.org/wiki/Branch_point.
[5] Francesco Hautmann, Functions of a complex variable,
http://www-thphys.physics.ox.ac.uk/people/FrancescoHautmann/ComplexVariable/.
[6] Anna Maltsev,
https://people.maths.bris.ac.uk/~maavm/mathmethods_files/branch_cuts.pdf.

33 5 March 2018

View publication stats