Conformal Mapping

UNIT 9 CONFORMAL MAPPING

Structure
9.1 Introduction
Objectives

9.2 Geometrical Representation of Function of z

9.3 The Conformal Mapping
9.4 The Bilinear Transformation
9.5 The Schwarz-Christoffel Transformation
9.6 Summary
9.7 Answers to SAQs

9.1 INTRODUCTION
Earlier we define that the quantity x + i y, in which x and y are independent real variables,
is known as a complex variable and is, generally denoted by z. Also we defined that if
there is a second complex variable w = u + i v, which has one or more values for each
value of z, then w is called a function of z and is written as w = f (z). In this unit, we shall
study functions mappings or transformations which preserve certain properties. We first
of all give geometrical representation of a function of z. We then discuss some Simple
Transformations, followed by Bilinear Transformations. The transformations or
mappings which preserve angles between any two oriented curves interesting at a point,
known as Conformal Mappings, form an important part of this unit. The analytic property
of f (z), at z = z0 with f′ (z0) ≠ 0 maps the mapping f (z) as conformal at z0. In addition, a
harmonic function remains harmonic under as one-to-one conformal mapping. This
property of conformal mapping has applications in solving
two-dimensional Laplace’s equation with given boundary values. We shall discuss all
these in detail in this unit. Finally, we shall study some special mappings which map real
line in the z-plane into a polygon into the w-plane.
Objectives
After studying this unit, you should be able to
• determine the images of a given set under the mapping by a function of a
complex variable,
• find some special mappings using bilinear transformations,
• define conformal mapping and discuss its properties and applications, and
• map certain given regions to certain other given regions using some special
transformations.

9.2 GEOMETRICAL REPRESENTATION OF

FUNCTION OF z
In the conventional method of Cartesian plotting, a complex number z = x + i y is
geometrically represented in an Argand diagram by a point (x, y) in the xy-plane; a curve
is also associated with a real function of one variable y = g (x) in xy-plane and we
associate a surface in xyz-rectangular system with a real function of two variables
z = h (x, y). In the complex domain, a relation w = f (z), defined for all z, that is
w = u + i v = f ( x + i y)

113
Complex Variables-II involves four real variables, namely, two independent variables x and y and the two
dependent variables u and v. Hence, a space of four dimensions is required if we are to
pole w = f (z) in the cartesian fashion. To avoid difficulties inhert in such a device, we
make a decided departure from the conventional method of cartesian plotting and choose
to proceed as follows :
We plot the independent variables x, y in one-plane, called the z-plane. The dependent
variables u, v is plotted in another plane and call this second plane as w-plane
(Figure 9.1).
y y

z (x, y) w (u, v)

O z-Plane x O′ w-Plane u

Figure 9.1 : Argand Diagrams Representing z-plane and w-plane

A function w = f (z) is now represented not by a focus of point in a space of four

dimensions, but by a correspondence between the points of these two cartesian planes.
Wherever a point z = x + i y is given in the z-plane, the function w = u + i v = f (z)
determines one or more values of u + i v and, hence, one ore more point in w-plane. As z
ranges over any configuration in the z-plane (called the domain of function w = f (z)), the
corresponding point u + i v describes some configuration in the w-plane. The function
w = f (z) this defines a Mapping or a transformation of the z-plane into the w-plane and,
in turn, is represented geometrically by this mapping.
If the point z0 maps into the point w0 = f (z0) in w-plane, then z0 is known as the image
of z0.
Consider for example the relation w = z2, i.e.

w = u + i v = z 2 = ( x + i y ) 2 = x 2 − y 2 + i 2 xy

This given u = x 2 − y 2 , v = 2 xy . . . (9.1)

These are the equation of transformation from the z-plane to the w-plane. From them,
many features of the mapping can easily be infessed.
Here the origin in the z-plane maps into the origin in the w-plane; the points A (2, 0),
B (2, 2) and C (0, 2) in the z-plane respectively map into A′ (4, 0), (0, 8) and (− 4, 0). The
curves

x 2 − y 2 = c, 2 xy = c′

in the z-plane map into straight lines u = c, v = c′ parallel to the coordinate axes in the
w-plane. Figure 9.2 shows the lines u = 1, u = 4, u = 2, u = 4 and the corresponding
curves in the z-plane corresponding to there. The shaded region in the two figures of
Figure 9.2 map into each other.
Again, lines parallel to x-axis, i.e. lines with equation x = c, map into curves in w-plane
whose parametric equation, from (9.1), are

u = c12 − y 2 , v = 2c1 y

114 y
v
B
u=1 u=4
C
v=4
 Conformal Mapping

Figure 9.2

Eliminating the parameter y, we obtain the equation

v2
u = c12 −
4c12
This defines a family of parabolas having the origin of w-plane as focus and the line
v = 0 as axis, all opening to the left (Figure 9.3).
v

y x=½ x=1 x=1 y=1

x = 1/2 y = 1/2
y=1

y = 1/2
u

O x

Figure 9.3 : The Mapping of the Cartesian Lines by the Function w = z2

1
Similarly, lines parallel to the x-axis, i.e. lines with equations y = map into curves in
2
the w-plane whose parametric equation are

u = x 2 − c12 , v = 2c1 x
Eliminating x, we obtain
u2
u= − c22 ,
4c22
which is the equation of a family of parabolas having the origin as focus and the line.
v = 0 as axis, but this time all opening are to the right. The image of other curves, or
regions, can be found in the same fashion with varying degrees or difficulty. For
instance, if we wish to find the curves into which the line
y = 2x + 1
is transformed, we must eliminate x and y between the equation and the equation of
transformation. To do this, we first substitute for y in Eq. (9.1) getting

u = x 2 − (2 x + 1) 2 = − 3 x 2 − 4 x − 1
115
Complex Variables-II
v = 2 x (2 x + 1) = 4 x 2 + 2 x
If we regard these as two simultaneous equations in quantities x and x2, we can solve
getting
4u + 3v + 4 u + 2v + 1
x= , x2 +
− 10 3
2
u + 2v + 1 ⎛ 4u + 3v + 4⎞
Hence, =⎜ ⎟
3 ⎝ − 10 ⎠
or 16u 2 + 24u v + 9v 2 + 12u − 16v = 4
which is the equation of a parabola.
For the converse relation w = z2, although w is a single-valued function of z, the converse
is not true. In fact, when w is given, z may be either of two square roots of w. Because of
this, the mapping from the z-plane to the w-plane covers the latter twice, as Figure 9.4
suggests. This, of course, is nothing but the graphic representation of the familiar fact
Z-plane
y W-plane
v

x u

y
v

x u

y v

x u

y v

x u

Figure 9.4 : The Two-valued Character of the Mapping z = w1/2

that the angle of complex numbers are doubled when the number are squared. We now
take up some example of simple transformations.
116
Example 9.1 Conformal Mapping

Discuss the mapping w = z + z0, when z0 is a complex constant.

Solution
Let z0 = x0 + i y0 , z = x + i y

Then w = u + i v = z + z0 = ( x + i x ) + ( x0 + i y0 ) = ( x + x0 ) + i ( y + y0 )

Thus the point z = (x, y) is translated to the point w = (u , v ) = ( x + x0 , y + y0 ) .

The image of any figure in the z-plane is exactly the same figure in the w-plane,
except that the location is changed with respect to the origin.
The transformation w = z + z0 preserves both distances and angles and is mere
translation of axes. If z0 = 0, the mapping becomes w = z and is called the Identity
mapping. It maps every point to itself.
y v

z = x + iy w = z + zo

y + yo

o z o′ x + xo u

Figure 9.5 : Transition of Axes

Example 9.2
Discuss the mapping w = b z, where
(a) b is a real constant.
(b) b is a complex constant.
Solution
(a) Let w = b z, where b is a real constant.
Then |w|=|bz|=b|z|
and arg w = arg (b z) = arg (b z) = arg b + arg z = arg z since arg b = 0
because b is real.
Hence the image in the w-plane of a figure in z-plane is similar and similarly
located about the origins, but the figure in w-plane is magnified b times
(Figure 9.6). This mapping is called a magnification.
y v w
z

θ br
θ
r

x u
o o′
Figure 9.6 : Magnification (w = b z, b is a Real Constant)

(b) Let w = b z, where b is a complex constant. Taking the polar form, let
x = r ei θ, b = r, ei α, and w = p ei φ. 117
Complex Variables-II
Then w = ρ ei φ = b z = r1 ei α . r ei . θ = rr1 ei ( α + θ) so that ρ = rr1 and
φ = α + θ. Thus the transformation consists of a rotation of the z-plane
through an angle α and a uniform magnification of length by a factor r1 (see
Figure 9.7).
y v w

r1 r = p

z
r

θ+α=φ
θ
x u
o o′
Figure 9.7 : Rotation and Magnification (w = b z, b is a Complex Constant)
Example 9.3
1
Discuss the mapping w = .
z
Solution
In polar coordinates, let w = ρ ei φ , z = r ei θ . Then the transformation gives
1 −i θ
ρ ei φ = e
r
i
⇒ ρ= , φ=θ
r
1
Hence | w|= and arg w = − arg z.
|z|
Therefore, points outside the unit circle map into points inside it and vice-versa.
This is one-to-one mapping : one point of z-plane correspond to one point of
w-plane. The exception in the origin, which is a singular point. A small region near
1
origin inside the circle r = maps into the exterior of the circle by taking R
R
sufficiently large, we can map any small neighbourhood of origin into the exterior
of the circle f = R. It is convenient to express this by saving ‘z = 0 maps into the
“point at infinity”.’
Thus the image of a point in the z-plane is obtained first by inverting the point with
respect to the circle | z | = 1 and then reflecting it in the real axis (see
Figure 9.8). This mapping is called an inversion.
y v

x u
O′

Figure 9.8 : Inversion

In cartesian coordinates, we have

118
 1 u−iv Conformal Mapping
x+i y= = 2
u + i v u + v2

u −v
so that x= , y=
u +v
2 2
u + v2
2

Hence the equation

a ( x 2 + y 2 ) + bx + cy + d = 0 . . . (9.2)
transforms of
⎛ u 2 + v2 ⎞ bu cv
a=⎜ 2 + − +d =0
⎜ (u + v 2 ) 2 ⎟⎟ u 2 + v 2 u 2 + v 2
⎝ ⎠

i.e. (u 2 + v 2 ) + bu − cv + a = 0
This shows that a circle in the z-plane, in general, transforms to another circle in
the w-plane. The exceptional case occurs when d = 0. A circle through the origin
transforms to a straight line. Again, when a = 0, Eq. (9.2) represents a straight line
and we see that straight lines in the z-plane, not passing through the origin,
transform into circles in w-plane, while straight lines through the origin transform
into straight lines.
Example 9.4

Discuss the mapping w = e z .

Solution
Let z = x + i y, w = u + i v so that u + i v = ex + i y = ex cos y + i ex sin y

i.e. u = e x cos y and v = e x sin y

Also if w = ρ ei φ , then f = ex and φ = y

Therefore lines x = c transform to circles f = ec and the lines y = c′ transform to

radial lines φ = c′.
In particular, the y-axis (x = 0) transforms into the unit circle f = 1.
The x-axis (y = 0) transforms into positive u-axis (φ = 0) and the line y = π
transforms into the negative u-axis (φ = π). Thus the region between the lines
y = 0 and y = π maps into the upper half of w-plane. Similarly, the region between
the lines y = 0 and y = − π maps into the lower half of w-plane.
In fact ez is a periodic function with period 2π i. Therefore any horizontal strip
between the lines y = c and y = c + 2π will transform into the whole w-plane.
Let us use the knowledge gained to attempt the following SAQs.
SAQ 1
(a) What is the image of y = x – 2 under the mapping w = z2?
(b) What is the image of circle x2 + y2 = a2 under the mapping w = z2?

SAQ 2
1
What is the image of | z | < | and | z | > | under the mapping w = ?
z

119
Complex Variables-II SAQ 3
Discuss the mapping w = z3. Plot the image of the line u – 1. What is the equation
of the image of the line x = 1?

In the section, we saw that various functions of a complex variable map the xy-plane onto
the uv-plane. In the next section we purpose to investigate in more general terms the
character of transmission for which the mapping function w = u ( x, y ) + i v ( x, y ) is
analytic. This is done through the study of conformal mapping.

9.3 THE CONFORMAL MAPPING

It is important to know when the transformation equations can be solved (at least
theoretically) for x and y as single-valued functions of u and v, that is, when the
transformation has a single-valued inverse. This is achieved through the following
theorem.
Theorem 1
Statement
If f (z) is analytic, the function w = f (z) will have a single valued inverse in the
neighbourhood of any point where the derivative of the mapping function is
different from zero.
Proof
If z = x + i y and w = u + i v = f (z) = f (x + i y), then the condition for function f (z)
to have a single-valued inverse, as established in most text on advanced calculus
(Refer to pp. 363-366, Advanced Calculus, by R. C. Buck, 3rd Edition,
McGraw-Hill, New Your, 1978) is simply that the Jacobian determinant of the
transformation is different from zero, that is,
∂u ∂v
⎛ u, v ⎞ ∂x ∂y
J =⎜ ⎟= ≠0
⎝ x, y ⎠ ∂v ∂v
∂x ∂y

Since w = u + i v = f (z) is assumed to be analytic, therefore, u and v must satisfy

the Cauchy-Reimann equations, i.e.
∂u ∂v ∂u ∂v
= and =−
∂x ∂y ∂y ∂x
Hence substituting into the Jacobian determinant, we have
∂u − ∂v
⎛ u, v ⎞ ∂x ∂y ⎛ ∂u ⎞
2
⎛ ∂v ⎞
2
J =⎜ ⎟= =⎜ ⎟ +⎜ ⎟
⎝ x, y ⎠ ∂v ∂u ⎝ ∂x ⎠ ⎝ ∂x ⎠
∂x ∂y
2
∂u ∂v
= +i
∂x ∂x

= | f ′ ( z ) |2
120
 and hence the condition for existence of single-valued inverse for the Conformal Mapping
transformation w = f (z), when f (z) is analytic, becomes | f ′ ( z ) | ≠ 0 , that is, in the
neighbourhood of any point, the derivative of the mapping function is different
from zero.
This establishes Theorem 1.
Exceptional points where f ′ (z) = 0 are known as Critical Points of the
transformation.
We next take some other property of transformation, which is given by Theorem 2.
Theorem 2
Statement
In the mapping defined by an analytic function ut = f (z), the lengths of the
infinitesimal segments, regardless of their direction, are altered by a factor | f ′ ( z ) |
which depends only on the point from which the segments are drawn.
Proof
Consider a value of z and its image w = f (z), where f (z) is analytic.

Let Δ z = | Δ z | ei θ and Δ w = | Δ w | ei φ be the corresponding increments of z and

w (Refer Figure 9.9).
y

v
Δz

θ φ
Δw

w
u
o′

o x

Figure 9.9 : Δ z and its Image Δ w under a Mapping w = f (z)

Δw | Δ w | ei φ Δ w i ( φ − θ)
Then f ′ ( z ) = Lt = Lt iθ
= Lt e
Δ z →0 Δz Δ z →0 | Δz | e Δ z →0 Δz

From this, it is apparent that

Δw
Lt = | f ′ ( z ) | and Lt (φ − θ) = arg f ′ ( z )
Δ→0 Δz Δ z →0

or, to an arbitrary degree of approximation,

| Δ w | = | f ′ ( z) | . | Δ z | . . . (9.3)

and φ = θ + arg f ′ ( z )

or arg Δ w = Δ z + arg f ′ . . . (9.4)

The fact that f ′ ( z ) exist [which, of course, is a fact, since f (z) is assumed to be
analytic] means that both | f ′ ( z ) | and arg | f ′ ( z ) | are independent of the manner
in which Δ z → 0. In other words, they depend solely on z and not on the limiting
orientation of the increment Δ z.
Hence from (9.3), we draw the conclusions of Theorem 2.
121
Complex Variables-II Since infinitesimal lengths are magnified by the factor | f ′ ( z ) | , it follows that
⎛ u, v ⎞
infinitesimal areas are magnified by the factor | f ′ ( z ) |2 , that is, by J = ⎜ ⎟.
⎝ x, y ⎠
Similarly, we conclude from (9.4), that, in general, the difference between the
angles of an infinitesimal segment and its image is independent of the direction of
the segment and depends only on the point from which the segment is drawn. In
particular, two infinitesimal segments forming an angle will both be rotated in the
same direction by the same amount. Hence, the measure of the angle between them
will, in general, be left invariant by the transformation.
However, when f ′ ( z ) = 0 , then arg f ′ ( z ) is undefined and we cannot assert that
angles are preserved. To investigate this case, suppose that f ′ ( z ) has n-fold zero
at z = z0. Then f ′ ( z ) must contain the factor (z – z0)n, and hence we cannot write

f ′ ( z ) = ( n + 1) a ( z − z0 ) n + ( n + 2) b ( z − z0 ) n + 1 + . . .

where a, b are complex coefficients, of no common to us, and the factors

(n +1) (n + 2) . . . have been inserted for convenience in integrating f ′ ( z ) to
obtain f (z).
Integrating, we get

f ( z ) = f ( z0 ) + a ( z − z0 ) n + 1 + b ( z − z0 ) n + 2 + . . .

Let ( z − z0 = Δ z , f ( z ) − f ( z0 ) = Δ w and dividing by a ( z Δ ) n + 1 , we get

Δw b
n +1
=1+ Δz + ...
a (Δ w) a

As Δ z → 0, the right hand side approaches 1. Therefore,

Lt (arg Δ w) − Lt [arg a (Δ z )n + 1 ] = arg 1 = 0

Δ z →0 Δ z →0

or, to an arbitrary degree of approximation,

arg (Δ w) = arg + (n + 1) arg (Δ z ) . . . (9.5)

Now let Δ z1 and Δ z2 be two infinitesimal segments which make an angle θ with
each other, and let Δ w1 and Δ w2 be their images. From Eq. (9.5), we have
arg Δ w1 = arg a + ( n + 1) arg Δ z1

arg Δ w2 = arg a + ( n + 1) arg Δ z2

Hence subtracting, we get

arg Δ w2 − arg Δ w1 = (n + 1) (arg Δ z2 − arg Δ z1 ) = (n + 1)θ

Thus we have established that.

Theorem 3
In the mapping defined by an analytic function w = f (z), angle are, in general,
preserved in magnitude and in sence. The only exception to this occurs when the
vertex of the angle is an n-fold zero of f ′ ( z ) , in which case the measure of the
angle is altered by the factor n + 1.
122
Proof Conformal Mapping

Example considered in Section 9.2, namely, the transformation w = z 2 is an

excellent illustration of the behaviour described in Theorem 3 above. The mapping
function w = f ( z ) = z 2 is everywhere analytic, and as Figures 9.2 and 9.3
suggest, angle measures are in general preserved. However, the derivative
f ′ ( z ) = 2 z has a simple zero at z = 0 and as Figure 9.4 indicates, the angle with
vertex at origin are not preserved but are doubled. We now give some definitions.
Definition 1
A transformation which preserves the magnitude of the angles is said to be
ISOGONAL.
Definition 2
A transformation which preserves the sense as well as the magnitude of the
angles is said to be CONFORMAL.
Alternatively, A transformation w = f (z) is said to be CONFORMAL to z0 if a small
triangle in the neighbourhood of z0 transforms into a similar triangle in the w-plane.
This is not true, however, for large configurations, which may bear little or no
resemblance to their images.
We now prove some theorems on such transformations.
Theorem 4
A transformation w = f (z) is conformal if f (z) is analytic and f ′ ( z ) is not zero
at z = z0.

y v

(z2) (w2)
P2 P1 P′ 2 P′ 1
(z1) (w1)

Po P′ o
(zo) (wo)

o x o′ w-plane u
z-plane

Figure 9.10

Proof
Let the triangle p0 p1 p2 in the z-plane have its vertices at z0 z1 z2 where z1 and z2 are
points in the neighbourhood of z0.
Let the corresponding points P0′ P1′ P2′ in w-plane be w0, w1, w2. Then upto a first
approximation.
Δ w w1 − w0 w2 − w0
f ′ ( z0 ) = = = . . . (9.6)
Δz z1 − z0 z 2 − z0

provided f ′ ( z ) is analytic at z0.

It f ′ ( z ) ≠ 0 , this gives

w2 − w0 z − z0
= 2 . . . (9.7)
w1 − w0 z1 − z0
123
Complex Variables-II w2 − w0 z − z0
and arg = = arg 2 . . . (9.8)
w1 − w0 z1 − z0
Now | w2 – w0 | gives the length P0′ P2′ and (w2 – w0) the angle φ2 which P0′ P2′
makes with the horizontal axis. Similar interpretations hold for other quantities.
Therefore relation (9.7) means that the sides P0′ P2′ and P0′ P1′ are in the same ratio
as P0 P2 and P0 P1.
w2 − w0
Again, arg = = arg ( w2 − w0 ) − arg ( w1 − w0 )
w1 − w0
= φ2 − φ1 = ∠ P2′ P0′ P1′
Therefore, by (9.8), ∠ P2′ P0′ P1′ = ∠ P2 P0 P1
Hence the transformation is conformal.
Note : We see that while a conformal transformation may rotate or change the
dimensions of any small figures, it preserves the angles and the shape.
This, however, is not true at a point where f ′ ( z ) = 0 . At such a points,
w2 – w0 and w1 – w0 are of an order smaller than z2 – z0 and z1 – z0 and
relations (9.7) and (9.8) cannot be derived from relation (9.6). We next
take up another theorem on conformal mapping.
Theorem 5
A harmonic function h (x, y) remains harmonic under a one-to-one conformal
mapping w = f (z).
Proof
Since the mapping w = f (z) is one-to-one, the inverse mapping z = f −1 ( w) exist,
giving x = x (u, v) and y = (u, v). Since the mapping w = f (z) is conformal, f ′ ( z )
exist and is non-zero. Hence
dz 1 1
= =
dw dw f ′ ( z )
dz

exists at every point. Therefore f − 1 ( m) is analytic.

Let g (x, y) be the conjugate harmonic function of h (x, y).
Then h + i g = H (z), where H is an angle analytic function. Under the
transformation w = f (z) this becomes
h + i g = H [ f −1 ( w)]

which is analytic since H and f − 1 are both analytic functions. Hence the real part
h [ x (u , v), y (u , v)] is a harmonic function. This proves the theorem.
One important reason for studying conformal transformations is that the solutions
of Laplace’s equation in two-dimensions remain solutions of Laplace’s equation
when subjected to a conformal transformation. More precisely, we have the
following theorem.
Theorem 6
If φ (x, y) is a solution of the equation
∂ 2φ ∂ 2φ
+ =0
∂ x2 ∂ y2
124
 then when is φ (x, y) transformed into a function of u and v by a conformal Conformal Mapping
transformation, it will satisfy the equation

∂ 2φ ∂ 2φ
+ =0
∂ u2 ∂ v2

everywhere except possibly at the images of the points where the derivatives of the
mapping function is equal to zero.
Proof
Let w = u ( x, y ) + i v ( x, y ) define a conformal transformation by means of which
φ (x, y) is transformed into a function of u and v. Then
∂φ ∂φ ∂u ∂φ ∂v ∂φ ∂φ ∂u ∂φ ∂v
= . + . and = +
∂x ∂u ∂x ∂v ∂x ∂y ∂u ∂y ∂v ∂y

∂φ ∂φ
A second differentiation of each of and yields
∂x ∂y

∂ 2φ ∂φ ∂ 2u ⎛ ∂ 2 φ ∂u ∂ 2 φ ∂v ⎞ ∂u ∂x ∂ 2 v
= . 2 +⎜ 2 . + . ⎟ +
∂u ∂x ⎜ ⎟
∂x 2 ⎝ ∂u ∂x ∂v ∂u ∂x ⎠ ∂x ∂v ∂x
2

⎛ ∂ 2 φ ∂u ∂ 2 φ ∂v ⎞ ∂v
+⎜ +
⎜ ∂u ∂v ∂x ∂v 2 ∂x ⎟⎟ ∂x
⎝ ⎠

∂ 2φ ∂φ ∂ 2u ⎛ ∂ 2 φ ∂u ∂ 2 φ ∂v ⎞ ∂u ∂y ∂ 2 v
= + ⎜ + ⎟ +
∂y 2 ∂u ∂y 2 ⎜⎝ ∂u 2 ∂y ∂v ∂u ∂y ⎟⎠ ∂y ∂v ∂y 2

⎛ ∂ 2 φ ∂u ∂ 2 φ ∂v ⎞ ∂v
+⎜ +
⎜ ∂u ∂v ∂y ∂v 2 ∂y ⎟⎟ ∂y
⎝ ⎠
Adding, we get

∂ 2φ ∂ 2φ
∂φ ⎛ ∂ 2u ∂ 2u ⎞ ∂ 2φ ⎡⎛ ∂u ⎞ 2 ⎛ ∂u ⎞2 ⎤
+ 2 = + + ⎢⎜ ⎟ + ⎜ ⎟ ⎥
∂x 2
∂y ∂u ⎜⎜ ∂x 2 ∂y 2 ⎟⎟ ∂u 2 ⎢⎝ ∂x ⎠ ⎝ ∂y ⎠ ⎥⎦
⎝ ⎠ ⎣

∂ 2φ ⎛ ∂u ∂v ∂u ∂v ⎞ ∂φ ⎡ ∂ 2 v ∂ 2 v ⎤ ∂ 2 φ ⎡⎛ ∂v ⎞2 ⎛ ∂v ⎞ 2 ⎤
+2 ⎜ + ⎟+ ⎢ 2 + 2⎥ + 2 ⎢⎜ ⎟ + ⎜ ⎟ ⎥
∂u ∂v ⎝ ∂x ∂x ∂y ∂y ⎠ ∂v ⎢⎣ ∂x ∂y ⎥⎦ ∂v ⎢⎝ ∂x ⎠ ⎝ ∂y ⎠ ⎥⎦
⎣
Since w = u + i v is analytic, by hypothesis, therefore u and v themselves satisfy
Laplace’s equation. Hence, the first and fourth group of terms on the right vanish
identically. Moreover, u and v satisfy the Cauchy-Riemann equations, hence the 3rd
group of term also vanishes identically. By using the Cauchy-Riemann equations
again in the remaining terms, we get

∂ 2 φ ⎡⎛ ∂u ⎞ ⎛ ∂v ⎞ ⎤ ∂ φ ⎡⎛ ∂v ⎞ 2 ⎛ ∂u ⎞ 2 ⎤
2 2
∂ 2φ ∂ 2φ 2
+ = ⎢ ⎜ ⎟ + ⎜ − ⎟ ⎥ + ⎢⎜ ⎟ + ⎜ ⎟ ⎥
∂x 2 ∂y 2 ∂u 2 ⎢⎣⎝ ∂x ⎠ ⎝ ∂x ⎠ ⎥⎦ ∂v 2 ⎢⎣⎝ ∂x ⎠ ⎝ ∂x ⎠ ⎥⎦

⎡ ∂ 2φ ∂ 2φ ⎤ ⎡⎛ ∂u ⎞ 2 ⎛ ∂v ⎞ 2 ⎤ ⎡ ∂ 2φ ∂ 2φ ⎤
=⎢ 2 + 2⎥ ⎢⎜ ⎟ + ⎜ ⎟ ⎥ = | f ′ ( z ) |2 ⎢ 2 + 2 ⎥
⎢⎣ ∂u ∂v ⎥⎦ ⎢⎣⎝ ∂x ⎠ ⎝ ∂x ⎠ ⎥⎦ ⎣⎢ ∂u ∂v ⎥⎦

Thus, at any point where the transformation is conformal, that is, where f ′ ( z ) ≠ 0 ,
125
Complex Variables-II ∂ 2φ ∂ 2φ ∂ 2φ ∂ 2φ
+ = 0 implies + =0
∂x 2 ∂y 2 ∂u 2 ∂v 2

as asserted.
Suppose now that it is required to solve Laplace’s equation subject to certain
boundary conditions, within a region R. Unless R is of very simple shape, the
solution of the problem, directly, will usually be exceedingly difficult. However, it
may be possible to find a conformal transformation which will convert region R
into some simple region R1, such as a circle or a half-plane, in which Laplace’s
equation can be solved, subject, of course, to transformed boundary conditions. If
this is the case, the resulting solution, when carried back to R by the inverse
transformation, will be the required solution of the original problem.
Let us take up some examples on conformal mapping.
Example 9.5
For the conformal transformation w = z2, show that

(a) coefficient of magnification at z = 2 + i is 2 5 .

(b) the angle of relation at z = 2 + i as tan– 1 (0.5).

(c) the circle | z – 1 | = 1 transforms into the cardiod ρ = 2 (1 + cos φ) ,
where w = ρ ei φ in the w-plane.

Solution
(a) For conformal transformation w = f (z), the coefficient of a magnification at
z = z0 is given by | f ′ ( z0 ) | .

Here f ( z ) = z 2 , ∴ f ′ ( z ) = 2 z

∴ Coefficient of magnification at z = 2 + i
= | 2 (2 + i )| = | 4 = 2i |

= 42 + 22 = 2 5 . Ans. I

(b) For conformal transformation w = f (z), the angle of solution at z = z0 is

arg { f ′ ( z0 )} .

Here f ( z ) = z 2 , ∴ f ′ ( z ) = 2 z

∴ Angle of relation at z = 2 + i

⎛2⎞
arg f ′ (2 + i ) = arg (4 + 2i ) = tan −1 ⎜ ⎟ = tan −1 (0.5) Ans. II
⎝4⎠

(c) Let z = r ei θ , w = ρ ei φ

Then for given transformation w = z2, we get

ρ ei φ = r 2 e 2i φ = ρ = r 2 and φ = 2θ

Now the circle | z – 1 | = 1 is transformed to

| z − 1 | = 1 ⇒ | r ei θ − 1 | ⇒ | r cos θ − 1 | + i ( r sin θ) | = 1

⇒ r 2 cos 2 θ − 2r cos θ + r 2 sin 2 θ + 1 = 1

126
 Conformal Mapping
⇒ r 2 = 2r cos θ

⇒ r = 2 cos θ (substituting of r and θ in terms of ρ and φ)

φ φ
= ρ = 2 cos ⇒ ρ2 ⇒ 4 cos 2 = ρ = 2 (1 + cos φ) Ans. III
2 2
Example 9.6
For the conformal mapping w = z2,
(a) Find the images of lines x = c and y = d and find the angle of
intersection of the image curves.
(b) Find the images of lines y = x and x = z and find the angle of
intersection of the image curves.
Solution
The mapping function is

w = z 2 = ( x + i y ) 2 = x 2 − y 2 + 2i xy
If, w = u + i v, then

u = x 2 − y 2 , v = 2 xy
(a) For the line x = c, the mapping becomes

u = c 2 − y 2 , v = 2cy
Eliminating y from these two relations, the image of lines x = c in w-plane
becomes
v2
u = c2 −
4c 2
Similarly, the lines y = d are mapped in w-plane to
v2
u= 2
− d2
4d
2
Since w = z is a conformal mapping, and for (c, d), the point of intersection
of x = c and y = d, f ′ ( z ) = 2 z is not zero, hence the angle of intersection of
lines x = c and y = d in z-plane will be same as the angle of intersection of
their mapping in w-plane.
Since the lines x = c and y = d intersect at eight angles, we can immediately
conclude that the two families of parabolas, which are images of x = c and
y = d in w-plane, intersect at right angles.
(b) Now for line y = x, the mapping under w = z2 becomes,

u = x2 − x2 = 0
The line x = 2 is mapped into

u = u − y2 , v = 2 y
Eliminating y, the mapping of x = 2 under mapping function w = z2 becomes,
y2
u=4− or v 2 = − 16 (u − 4)
16
which is a parabola.

127
Complex Variables-II π
The line y = x and x = 2 intersect at and their point of intersection
4
is 2 + 2i.
Also f ′ ( z ) |2 + 2i = 2 z |2 + 2i = 4 + 4i ≠ 0

∴ By conformality of w = z2, we conclude that the images u = 0 and

π
v 2 = − 16 (u − 4) of the lines y = x and x = 2 also intersect at an angle in
4
w-plane.
You may now try the following SAQs.
SAQ 4
Find the images of lines x = 0 and y = 0 under the transformation w = z2 and find
the angle of their intersection.

SAQ 5
If w = f (z) is analytic at z = z0 and f ′ ( z0 ) ≠ 0 , then f has an inverse in some
neighbourhood of z0. Show that this inverse is an analytic function. Further show
that the inverse mapping is also conformal.

SAQ 6
Determiner the critical points of the transformation w = f (z), where

(a) w = z2 + 2z
(b) w = cos z

SAQ 7
Find the magnification coefficient of the conformal transformation w = z3 at the
point 1 + i.

The simplest class of conformal transformations, yet one of the most important, is the
class of bilinear transformations, which we take up in the next section.

9.4 THE BILINEAR TRANSFORMATION

Let us first give the definition of bilinear transformation.
128
Definition Conformal Mapping

The transformation
az + b
w= , ad − bc ≠ 0 . . . (9.9)
cz + d

where a, b, c, d are real or complex constants, is called a “Bilinar or Linear

Fractional or Mobins Transformations.”
Note 1
This transformation was first studied by German geometer A. F. Mobins
(1790-1868) and is also named for him.
Note 2
a c
The condition ad – bc ≠ 0 is essential, because if ad – bc = 0, then = and the
b d
transformation becomes
⎛a⎞
cz + b b ⎜ ⎟ z +1 b
.⎝ ⎠
b
w= = =
cz + d d ⎛c⎞ d
⎜ ⎟ z +1
⎝d⎠
which is a constant.
Note 3
cz + b
If a = b = c = d = 0, the expression w = becomes meaningless.
cz + d

The bilinear transformation is frequently denoted by w = T (z). The inverse

az + b
transformation T – 1 of T is obtained by solving w = for z. it is easily seen that
cz + d

− wd + b
z= = T −1 ( w)
cw − a

Thus T– 1 (w) is also a bilinear mapping.

We may obverse that transformations defined in Examples 9.1, 9.2 and 9.3 of this unit
represent bilinear transformations. For example, if we take a = 1, b = z0, c = 0, d = 0, we
get the bilinear transformation given by (9.9) becomes w = z + z0 of Example 9.1.
−d
Except of z = to every value of z, these corresponds first one finite value of w, i.e.
c
the transformation (9.9) gives a one-to-one mapping of the z-plane into the w-plane. The
−d
point z = maps into the ‘point at infinity.’
c
The fixed points of bilinear transformation (i.e. points which do not change their
position) are given by
(az + b)
z=
(cz + d )

i.e. cz 2 − ( a − d ) z − b = 0

This is quadratic equation and has at the most two roots. Hence a bilinear
transformation has at most two fixed points.
129
Complex Variables-II Exceptions are
(a) c = 0, or
(b) c = b = 0, a = d.
In case (a), transformation (9.9) becomes a linear transformation and there is one fixed
point.
In case (b), the transformation (9.9) reduces to identity transformation and every point is
a fixed point.
Geometrically, the image of a fixed point of a transformation under the same
transformation is the point itself.
Before we prove a theorem on circles under bilinear transformation, we shall discuss the
equation of circles in terms of its inverse points.
We know that the equation of a circle with centre a and radius r is
|z–a|=r
We also know that two points p and q are converse points with respect to a circle of
radius r if p and q are colinear with the centre of the circle and lie on the same side of it
and if the product of their distances from the centre of the circle is equal to the square of
the radius of the circle (Figure 9.11).

r
p 2
C Cp . Cq = r

Figure 9.11 : p and q are Inverse Points of the Circle

r 2 i θ0
This if p = a + l ei θ0 , then q = a + e .
l

If z = a + r ei θ is any point in the circle, then

z− p z − a − l ei θ0
=
z−q r 2 i θ0
z−a− e
l

r ei θ − l ei θ0
= 2
, since | z − a | = r ⇒ z − a = r ei θ for any point.
r i θ0
r ei θ − e
l

l r ei θ − l ei θ0 l
= =
r l ei θ − r ei θ0 r

d l
If we assume that p is inside the circle, then so that ≠ 1 .
r r
130
 Conformal Mapping
z− p l
Thus = , where p and q are inverse points, is an alternatively form of the
z−q r
equation of the circle | z – a | = r.
This circle is called the Circle of Appllonius.

z− p
Conversely any equation = k (k ≠ 1) represent a circle with p and q as inverse
z−q
points.
To prove this, the equation gives
2
z− p
= k 2 or | z − p |2 = k 2 | z − p |2
z−q

or ( z − p) ( z − q) = k 2 ( z − q) ( z − q)

or ( z − p) ( z − p ) = k 2 ( z − q) ( z − q )

or | z |2 − ( z p + z p ) + | p |2 = k 2 [| z |2 − ( z q + z q ) + | q |2 ]

or | z |2 − 2 Re ( p z ) + | p |2 = k 2 [| z |2 − 2 Re ( z q ) + | q |2 ]

2 Re [( p − k 2 q ) z ] [| p |2 − k 2 | q |2
or | z |2 − + =0
1 − k2 1 − k2
2
p − k2 q | p − k 2 q |2 | p |2 − k 2 | p |2
or z− = −
1 − k2 (1 − k 2 ) 2 1 − k2

| p − k 2 q |2 − (1 − k 2 ) [| p |2 − k 2 | q |2
=
(1 − k 2 ) 2

( p − k 2 q ) ( p − k 2 q ) − (1 − k 2 ) ( pp − k 2 qq )
=
(1 − k 2 ) 2

since zz = | z |2

k 2 | p − q |2
=
(1 − k 2 )2

p − k2 q k| p−q|
Hence z − =
1− k 2
1 − k2

This equation is of the form | z – a | = r and hence it represents a circle with centre
p − k2 q k|q− p|
a= and radius .
1− k 2
1 − k2

k 2 (q − p) q− p
Also p − a = and q − a = .
1− k 2
1 − k2

p−a
Since = k 2 , which is real and positive; it follows that p and q are on the same side
q−a
of centre a. 131
Complex Variables-II k 2 (q − p)2
Also | p − a || q − a | = = r2
(1 − k )
2 2

Hence p and q are inverse points of the circle.

If k = 1, then | z – p | = | z – q |. This means that z is equidistant from p and q, and
therefore lies on the perpendicular bisector giving p and q.
Remark
The inverse point of the centre of the circle is the point at infinity.
We now prove an interesting theorem.
Theorem 9.7
Under a general bilinear transformation circles and straight lines are mapped into
circles and straight lines.
Remark
By this theorem, a circle is transformed only to a circle or a straight line by a
bilinear transformation. Similarly a straight line is transformed only to a circle or a
straight line.
We given two proofs of this theorem.
Proof 1
It is convenient to investigate the general bilinear transformation by considering
first the three special cases :
(a) w = z + z0
(b) w=bz
1
(c) w=
z
In case (a), w is found by adding a constant z0 to each z. This transformation is just
a translation in the direction defined by arg z0 through a distance equal to | z0 | and
this rigid motion necessarily transforms circles and straight lines into straight lines.
In case (b), it is found by rotating each z through a fixed angle equal to arg b and
then multiplying its length by the factor | b |. The equation straight circle in z-plane
is

a1 ( x 2 + y 2 ) + b1 x + c1 y + d1 = 0, a1 b1 c1 d1 real and b12 + c12 ≥ 4a1 d1 . . . (9.10)

z+z z+z
We know that x = , y= , x2 + y 2 = z z .
2 2i
Then Eq. (9.10), becomes
b1 + i c1 b + i c1
a1 z z + z+ 1 z + d1 = 0
2 2i
or, renaming the coefficients,
( A + A) z z + BZ + BZ + ( D + D) = 0 . . . (9.11)

where now A, B and D are arbitrary complex number and the condition
b12 + c12 ≥ 4a1 d1 becomes BB ≥ ( A + A) ( D + D) − this condition ensure that
radius of the circle is real.
132
 w Conformal Mapping
Now from the transformation, w = bz, if we substitute w = in (9.11), we obtain
b
the transformed equation of the circle given by (9.10) as
w w w W
( A + A) . +B +B + ( D + D) = 0
b b b b
or ( A + A) w w + ( B b ) w + ( B b ) w + ( D + D ) b b = 0 . . . (9.12)

Since the coefficients of the first and last term in Eq. (9.12) are real and since the
coefficients of linear terms w and w are conjugate of each other, thus Eq. (9.12)
has the same structure as of Eq. (9.11) and hence will also represent a circle
provided its coefficients satisfy the condition for radius to be real. For the focus
described by Eq. (9.12), this condition is
( B b ) ( B b ) ≥ ( A A) ( D + D ) b b

or, dividing throughout by b b , which is necessarily positive,

B B ≥ ( A A) ( D + D ) ,
which is true by hypothesis.
If a1 = 0, so that A + A = 0 , then both the given circle (9.10) are and its image
(9.12) are straight lines.
In case (c) we can write
1 z
w= =
z zz

1
which shows that w is of length and has the direction of z .
|z|
To show that circles are also transformed into circles in case (c), let the
1
substitution z = be made in the self conjugate form of equation of circle (9.10).
w
This gives
1 1 B B
( A + A) . + + + ( D + D) = 0
w w W W
or ( D + D) ww + B w + B w + ( A + A) = 0
which is also the equation of a circle with real radius.
If A + A = 0 ⇒ a1 = 0 , then the original circle (9.9) reduces to a straight line
whose image becomes
( D + D) w w + B w + B w = 0
which is a circle passing through the origin. Conversely, any circle passing
through the origin is transformed into straight line.
The above three special transformation (a), (b) and (c) can be used to construct, or
synthesise, the general bilinear transformation. To verify this, two cases can be
considered
(a) c≠0
(b) c = 0.
When c ≠ 0, then the general bilinear transformation is equivalent to the following
chain of special transformations :
133
Complex Variables-II d d
w1 = z + − a shift of origin to .
c c
w2 = cw1 = cz + d − a rotation and magnification by the factor c.

1 1
w3 = = − a reciprocal transformation.
w2 cz + d

bc − ad bc − ad
w4 = w3 = − a rotation and change of scale due to factor
c c (cz + d )
bc − ad
.
c
a bc − ad a az + b a
w = w4 + = + = − finally a shift of origin to .
c c (cz + d ) c cz + d c

On the other hand, when c = 0, it is clear from the restriction ad – bc ≠ 0 that

neither a nor d can zero. Hence we can take
b b
w1 = z + − a shift of origin to .
a a
a a ⎛ b ⎞ az + b
w= w1 = ⎜z + ⎟ = − a rotation and change of scale due to
d d ⎝ a⎠ d
a
factor .
d
Thus we have shown that in all cases, the general bilinear transformation can be
obtained as the composition of a succession of simple transformation of types (a),
(b) and (c). Since each of these is known to transform circles, including straight
lines as special cases, we have thus established the theorem.
Alternative Proof II

z− p
Let = k be a circle (or a straight line), with p and q as inverse points.
z−q

(when a circle becomes a straight line, inverse points become symmetrical about
the line).
az + b dw − b
Let T ( z ) = w = , z= be the general bilinear transformations.
cz + d − cw + a

The image of the inverse points p and q under T (z) are

ap + b aq + b
T ( p) = and T (q ) = . . . (9.13)
cp + d cq + d

The circle is, then, transformed to

dw − b
− p
− cw + a
=k
dw − b
−q
− cw + a

dw − b − p (− cw) − ap
or =k
dw − b − q (− cw) − aq
134
 ap + b Conformal Mapping
w−
cp + d cq + b
or =k
aq + b cp + d
w−
cq + a

w − T ( p) cq + d
or =k
w − T (q) cp + d

using Eq. (9.13). This is a circle with T (p) and T (q) as inverse points. This proves
the theorem.
Remark
A general bilinear transformation maps inverse point of a circle (or a straight line)
into inverse points.
We now take up an example.
Example 9.7
Find all bilinear mappings of the upper half-plane (when Im z ≥ 0) which maps
into the unit disk | w | ≤ |.
Solution
az + b
Let the required transformation be w = .
cz + d
Since the boundaries of corresponding regions must correspond under the
transformation, the unit circle in the w-plane must be the image of the real axis in
the z-plane. Therefore for all real values of z, we must have
b
|a| z +
az + b az + b a
| w|= = = =1
cz + d cz + d d
|c| z +
c
y v

o x o′ u
z-plane w-plane

Figure 9.12 : The Upper Half of the z-plane to be Mapped onto the
Interior of Unit Circle in the w-plane

In particular, from the limiting case | z | → ∞, we find

|a|
=1
|c|

a
In other words, is a complex number of absolute value 1, say ei θ. From this, we
c
conclude that for all real values of z.
135
Complex Variables-II b d ⎛ b⎞ ⎛ d⎞
z+ = z+ or z − ⎜− ⎟ = z − ⎜− ⎟
a c ⎝ a ⎠ ⎝ c⎠

The above expression shows the fact that the complex numbers
⎛ b⎞ ⎛ d⎞
⎜ − ⎟ and ⎜ − ⎟ are equally far from an arbitrary point z on real axis – which is
⎝ a⎠ ⎝ c⎠
possible if and only if the real axis in the z-plane is perpendicular bisector of the
⎛ b⎞ ⎛ d⎞ ⎛ b⎞ ⎛ d⎞
segment joining the points ⎜ − ⎟ and ⎜ − ⎟ . Thus ⎜ − ⎟ and ⎜ − ⎟ must be
⎝ a⎠ ⎝ c⎠ ⎝ a⎠ ⎝ c⎠
conjugate complex numbers, say λ and λ . Hence we can write,

b
z+
az + b a c = a z − λ = ei θ z − λ
w= = . . . . (9.14)
cz + d c z + d c z − λ z−λ
c
So far we have enforced only the condition that the boundaries of the two regions
correspond. It is necessary to make sure that the regions themselves corresponds as
required and ensure that the upper half of z-plane has not been mapped onto the
outside of the unit circle in w-plane. This is not most easily verified by
determining the image of some convenient points, say z = λ. Early the point z = λ
maps into the point w = 0, which is certainly inside the circle | w | = 1. Thus if λ is
restricted to lie in the upper half of z-plane, the solution is complete.
As a special case, let ei θ = − 1 and let λ be a pure imaginary, say i. Then
z−i
w=− . . . (9.15)
z+i

( w − w) 1 ⎛z −i z + i⎞ ⎛ z+z ⎞
Now Im w = =− ⎜ − ⎟=⎜ ⎟
2i 2i ⎝ z + i z − i ⎠ ⎝ ( z + i) ( z − i) ⎠

The denominator of the last fraction is the product of (z + i) and its conjugate
( z − i ) and hence is a positive quantity. Thus the imaginary part of w will be
positive if and only if z + z is positive. Since z + z is twice the real part of z, this
shows that the transformation (9.7) not only maps the upper half of z-plane into the
interior of the unit circle | w | = 1 but does it in such a way that the first quadrant of
z-plane [where Re (z) > 0] corresponds to the upper half of the circle [where In (w)
> 0] and the second quadrant of the z-plane corresponds to the lower half of the
circle.
On the opposite direction, the inverse transformatin
w −1
z=−i
w +1

maps the interior of the circle | w | = 1 into the upper half of the z-plane in such a
way that the upper half of the circle maps into the first quadrant of z-plane.
We take another example.
Example 9.8
π
Find a transformation which will map an infinite sector of angle into the
4
interior of the unit circle.
136
Solution Conformal Mapping

Since the boundary of the sector consists of two straight lines while the image is to
be a single circle, it is apparent that the mapping cannot be accomplished by a
bilinear transformation above. However, a simple combination of a power function
and bilinear transformation will define a suitable transformation. Specifically, the
transformation t = z4 will open out the given sector in the z-plane into the upper
half of the auxiliary t-plane (Figure 9.13).
Now we map the upper half of t-plane into the unit circle by any transformation of
t−i
the family (9.14) which we obtained in Example 9.7 above, say w = .
t+i
y

o u
x O1 O′′

z-plane t-plane w-plane

Figure 9.13 : The Two Transformation Needed

Taking the composition of these two mappings, we obtain the required

z4 − i
transformation as w = 4 .
z +i
You may now try the following SAQs.
SAQ 8
Find all the transformations of the unit disc | z | ≤ 1 into the unit disk | w | ≤ 1.

SAQ 9
Find a transformation which will map a 60o sector of the unit circle in z-plane into
the upper half of the w-plane.

az + b
The general bilinear transformation w = depends on three unknown ratios
cz + d
a : b : c : d, namely the ratios of any three of the coefficients a, b, c, d to the fourth.
Hence it is evident that three conditions are necessary to determine a bilinear
transformation. In particular, the requirement that three distinct of z, say z1, z2, z3 have
specified distinct images w1, w2, w3 leads to a unique transformation. The transformation
which sends three given points into these specified image points can be found by
137
Complex Variables-II az + b
imposing these condition on the general transformation equation w = and then
cz + d
solving the resulting three equations for the ratios of the coefficient. However it is
generally simpler to make use of the cross-ratios of four points and its invariant property
under a bilinear transformation, which we take up next. We first define the cross-ratio.
Definition
( z1 − z2 ) ( z3 − z4 )
The fraction is called the cross-ratio or a harmonic ratio of
( z1 − z4 ) ( z3 − z2 )
the four members z1, z2, z3, z4.
We now take up the following theorem on cross-ratio.
Theorem 9.7
The cross-ratio of four points is variant under bilinear transformation.
Proof
If w1, w2, w3, w4 are, respectively, the images of z1, z2, z3, z4, then we have to
az + b
establish that (for a bilinear transformation w = .
cz + d

( w1 − w2 ) ( w3 − w4 ) ( z1 − z2 ) ( z3 − z4 )
=
( w1 − w4 ) ( w3 − w2 ) ( z1 − z4 ) ( z3 − z2 )

azi + b azi + b (ad − bc) ( zi − z j )

Here wi − w j = − =
czi + d cz j + d (czi + d ) (cz j + d )

(ad − bc) ( z1 − z2 ) (ad − bc) ( z3 − z4 )

.
( w − w2 ) ( w3 − w4 ) (cz1 + d ) (cz2 + d ) (cz3 + d ) (cz4 + d )
Hence 1 =
( w1 − w4 ) ( w3 − w2 ) (ad − bc) ( z1 − z4 ) (ad − bc) ( z3 − z2 )
.
(cz1 + d ) (cz4 + d ) (cz3 + d ) (cz2 + d )

( z1 − z2 ) ( z3 − z4 )
=
( z1 − z4 ) ( z3 − z2 )
This proves the theorem.
Suppose now that it is required to find the transformation which sends z1, z2, z3 into
w1, w2, w3, respectively. If w is the image of a general point z under this
transformation, then according to Theorem 9.7, the cross-ratio of w, w1, w2, w3
must equal the cross-ratio of z, z1, z2, z3, i.e.
( w1 − w2 ) ( w3 − w) ( z1 − z2 ) ( z3 − z )
= . . . (9.16)
( w1 − w) ( w3 − w2 ) ( z1 − z ) ( z3 − z2 )
This equation is clearly bilinear in w and z and is satisfied by the three pairs of
corresponding values (z1, w1), (z2, w2), (w3, z3). Moreover, in it everything is known
except the variables w and z themselves. Hence it is necessary only to solve for w
in terms of z to obtain the required transformation in standard form.
We now prove that this (9.16) is the only bilinear map which maps z1, z2, z3
respectively to w1, w2, w3. If possible, let w = g (w) be another bilinear
transformation which maps z1, z2, z3 respectively to w1, w2, w3. Then g– 1 (w) maps
w1 to z1, w2 to z2 and w3 to z3. The composite map

H = g − 1 [ f ( z )]
will then map z1 to z1, z2 to z2 and z3 to z3.
138
 Now g −1 [ f ( z )] is also a bilinear map. The bilinear map H will then have three
Conformal Mapping

fixed points z1, z2 and z3. But by the property that ‘the bilinear transformation has
utmost two fixed points unless the transformation is the identity mapping.’ If a
bilinear map has more than two fixed points, it must be the identity map.
Hence H ( z ) = g −1 [ f ( z )] = z
or g ( z) = f ( z)
Thus there is exactly one bilinear mapping which maps three given distinct points
z1, z2, z3 to three given distinct points w1, w2, w3.
Let us take up some examples.
Example 9.9
Find the bilinear transformation that maps the points – i, 0, 1 into the points
– I, I2, I.
Solution
The bilinear transformation that maps z1, z2, z3 into the points w1, w2, w3 is given by
( w1 − w2 ) ( w3 − w) ( z1 − z2 ) ( z3 − z )
=
( w1 − w) ( w3 − w2 ) ( z1 − z ) ( z3 − z2 )
Substituting the values of z1, z2, z3 as – i, 0, i and w1, w2, w3 as – 1, i, 1 in the
above, we get
( − 1 − i ) (1 − w) (− i − 0) (i − z )
=
( − 1 − w) (1 − i ) (− i − z ) (i − 0)
Simplifying, we get
i (1 − z )
w=
1+ z
as the required bilinear mapping.
Example 9.10
Find the bilinear mapping that maps the points z1 = ∞, z2 = i, z3 = 0 into the points
w1 = 0, w2 = i and w3 = ∞.
Solution
We cannot apply bilinear transformation of z1, z2, z3 into w1, w2, w3 given by
w − w1 w2 − w3 z − z1 z2 − z3
. = .
w − w3 w2 − w1 z − z3 z2 − z1
directly in this case. However using it as
w2 z
−1 −1
w − w1 w3 z1 z − z3
. = . 2
− 1 w2 − w1 z − z3
w z2
−1
w3 z1
and substituting for the points, we get
w − 0 0 −1 0 −1 i − 0
. = .
0 − 1 i − 0 z − 0 z − z3
− w −1
or =
−t −z
1
or w=−
z
as the required mapping.
You may try the following SAQs. 139
Complex Variables-II SAQ 10
Find the bilinear transformation that maps
(a) z = 0, − i, − 1 into w = i, 1, 0 respectively.
(b) i, 1, − 1 into 1, 0, ∞.
(c) − i, 0, I into ∞, − 1, 0.

SAQ 11
Find the bilinear transform whose fixed points are 1 and 2.

We give below a few additional examples on mapping and its applications to physical
problems.
Example 9.11
1
Find the image of | z – 2i | = 1 under the mapping w = .
z
Solution
1
The circle | z – 2i | = 1 transforms in w-plane, under mapping w = , into
z
1
− 2i = 1
w
or | 1 − 2i w | = | w |, or | − i | . | 2 w + i | = | w |
or | 2w + i | = | w |

or 4u 2 + (2v + 1) 2 = u 2 + v 2

or 3 (u 2 + v 2 ) + 4v + 1 = 0

or 9u 2 + (3v + 2) 2 = 1

or | 3w + 2i |2 = 1

Hence the required image is the circle | 3w + 2i |2 = 1 .

Example 9.12
1
Discuss the conformal mapping w = z2 .
Solution
We are given
1
w= z2

or u + iv = x +iy

or (u + i v ) 2 = x + i y

or u 2 − v 2 + 2i u v = x + i y

so that, u 2 − v 2 = x, 2uv = y . . . (9.17)

140
 On eliminating v from above, we have Conformal Mapping

v2
u2 − = x or y 2 = − 4u 2 ( x − u 2 ) . . . (9.18)
4u 2
Similarly, on eliminating, we get
y 2 = 4v 2 ( x + v 2 ) . . . (9.19)
1
On taking the polar form, w = z 2 gives

ρ ei θ = r e i θ
1
1
Therefore, p = r2, φ = θ . . . (9.20)
2
We see from relation (9.20) that the whole of z-plane maps into the half of the
1
w-plane. A radial line θ = α maps into a radial line φ = α , inclined at half the
2
angle to the initial line. The circle r = a maps into the circle ρ = a , the unit circle
mapping into the unit circle.
We see from (9.17) that the lines x = a and y = c′ map respectively into the
rectangular hyperbolas
u 2 − v 2 = c and z u v = c′
which cut each other at right-angles.
We see from (9.18) that the line u = b maps into the parabola
y 2 = − 4b 2 ( x − b 2 ) . . . (9.21)
having its vertex at (b2, 0), latus rectum 4b2 and axis towards ox′.
Similarly, by (9.19), the line v = d maps into the parabola
y 2 = 4d 2 ( x + d 2 ) . . . (9.22)
with vertex (– d2, 0), latus rectum 4d2, and axis towards ox. The parabolas (9.21)
and (9.22) cut each other orthogonally (see Figure 9.14).
y
v

u= 2 v = 2 x=−4
x=−1
u=1 v=1 x=1
x=4
y=4
y=2

u
o x o′
y = −2
y=−4

z-plane
w-plane

Figure 9.14

Example 9.13
141
Complex Variables-II Discuss the transformation z = c cosh w, where c is a real number. Why physical
problems can be studied with the help of this transformation.
Solution
Here z = c cosh w
⇒ x + i y = c cosh (u + i v)
= c cosh u cosh i v + c sinh u sinh i v
= c cosh u cos v + i c sinh u sin v
Therefore x = c cosh u cos v
y = c sinh u sin v
Eliminating v, we get
x2 y2
+ = cos 2 v + sin 2 v = 1 . . . (9.23)
c 2 cosh 2 u c 2 sinh 2 u
Hence lines u = a transform into
x2 y2
+ . . . (9.24)
c 2 cosh 2 a c 2 sinh 2 a
which is an ellipse with centre at origin and axis along ox, oy of length 2c cosh a,
2c sinh c. the distance of the focus from the centre is

e 2 cosh 2 a − c 2 sinh 2 a = c
Thus the various ellipses for different a have the same foci.
Eliminating u from (9.23), we get,
x2 y2
2 2
− 2 2
= cos2 u − sinh 2 u = 1
c cos v c sin v
so the liens u = b transform into the hyperbolas
x2 y2
− =1
c 2 cos 2 b c 2 sin 2 b
with centre at origin and same foci (± c, 0).
Some of these conics are shown in Figure 9.15 below.
y

s O s x

Figure 9.15

142
 For small values of a, Eq. (9.24) gives a flat ellipse. As a → 0, the ellipse tends to Conformal Mapping
become circle with centre origin and radius as c. Similarly, when b → 0 (or π), the
hyperbolas (9.25) approach the line sx (or s′ x′).
Since any of the curves corresponding to u = a (or v = b) can be taken as an
equipotential, therefore, the transformation can be used to study the problem of
determining the potential
(a) between two confocal elliptic cylinders;
(b) between two confocal hyperbola cylinders;
(c) between two plates yy′ and sx;
(d) between two plates s′ x′ and sx; etc.
when the two surfaces are kept at different uniform potentials.
In general, the conformal transformation of one given region onto another is exceedingly
difficult. However, the existence of such a transformation is assured by the following
theorem due to Riemann.
Theorem 9.8
Either of two bounded simply-connected regions can be mapped conformally onto
the other.
The determination of the specific function which accomplishes a required mapping,
however, is usually out of question. In fact, the only class of regions for which conformal
transformations of practical interest exist are those bounded by polygons having a finite
number of vertices (one or more of which may lie even at infinity). These can always be
mapped onto a half-plane (and hence onto any region into which a
half-plane can be transformed) by means of a transformation, known as the
Schwarz-Christoffel Transformation, which we shall now discuss.

9.5 THE SCHWARZ-CHRISTOFFEL

TRANSFORMATION
We recall the mapping properties of the power function
w = zm
This transformation has the property that it alerts by the factor m any angle with vertex at
the origin. Thus it follows that the transformation
⎛ α1 ⎞ α
⎜ ⎟ dw α1 1 −1
w − w1 = ( z − x1 )⎝ π ⎠ , = ( z − x1 ) π . . . (9.25)
dz π
will take a segment of the x-axis containing x in its interior, i.e. a straight angle with
vertex at x, and fold it into an angle of
α1
. π = α1
π
with vertex at w1. Clearly, if we do it simultaneously for a number of points x1, x2, . . . xn
on the x-axis, x-axis would be mapped into a polygon whose angles were, respectively,
α1, α2, . . . , αn and conversely.
This is actually possible, and the transformation which accomplishes it as suggested by
the form if derivative of the function in (9.25) is defined by
⎛ α1 ⎞ ⎛ α2 ⎞ ⎛ αn ⎞
dw ⎜ ⎟ −1 ⎜ ⎟ −1 ⎜ ⎟ −1
= k ( z − x1 )⎝ π ⎠
( z − x2 )⎝ π ⎠ . . . ( z − xn ) π⎠
⎝ . . . (9.26)
dz
In order to verify this, we begin with a point z on the x-axis to the left of the first of the
given points x1, x2, . . . , xn and investigate the locus of its image as it moves to the right
of x-axis (see Figure 9.16). 143
Complex Variables-II

π − α3
w
w w3
α3 π − α2

α1 w2 α2
w1
z π − α1

x1 X2 X3

z-plane w-plane

Figure 9.16 : The Mapping of the Real Axis in z-plane into a Polygon
with Prescribed Angles in the w-plane

In our seasoning in this fashion, we are actually considering the inverse problem of
mapping the upper half of the z-plane onto a polygon region in w-plane. As a
consequence, in our applications, we will be mapping a polygon given in the w-plane
onto the upper half of the z-plane. This interchange of the usual roles of z and w planes
should, however, cause no confusion.
From relation (9.26), we immediately obtain
⎛α ⎞ ⎛α ⎞
arg dw = arg k + ⎜ 1 − 1⎟ arg ( z − x1 ) + ⎜ 2 − 1⎟ arg ( z − x2 )
⎝ π ⎠ ⎝ π ⎠
⎛α ⎞
+ ⎜ n − 1⎟ arg ( z − xn ) + arg dz . . . (9.27)
⎝ π ⎠
From above, it is apparent that until z reaches x1, every term on the right remains constant
(since z – x1, z – x2, . . . , z – xn) are all negative real numbers and hence have π for their
respective arguments and since dz is positive and therefore has O as its argument). Thus
the image point w traces a straight line (since the argument of the increment dw remains
constant). However, as z passes through x1, the difference z – x1 changes abruptly from
negative to positive and thus arg (z – x1) decreases abruptly from π to 0. Hence, arg (dw)
changes by the amount.
⎛ α1 ⎞
⎜ π − 1⎟ ( − π) = π − α1
⎝ ⎠
But, from Figure 9.16(b), it is evident that this is the precise amount through which it is
necessary to turn if w is to begin to move in the direction of the next side of the polygon.
As z moves from x1 to x2, the similar situation exists. The arg dw remains constant, and
thus w moves in a straight line until z reaches x2. Here (z – x2) changes abruptly from
negative to positive, is argument jumps immediately from π to 0, and as a consequence,
arg (dw) decreases by the amount (π − α2), which is just the amount of rotation required
to give the direction of the next side of the polygon.
Thus as z traverses the x-axis, it is clear that w moves along the boundary of a polygon
whose interior angles are precisely the given angles α1, α2, . . . , αn. Moreover, it is
evident that the origin which is mapped into the half-plane is the region which contains
these angles.
The required transformation will be obtained if we can ensure that the lengths of the sides
of the polygon, as well as its angles, have the correct values.
Now, the mapping function w, obtained by integrating (9.26)
⎛ α1 ⎞ ⎛ α2 ⎞ ⎛ αn ⎞
⎜ ⎟ −1 ⎜ ⎟ −1 ⎜ π ⎟ −1
w=k ∫ ( z − x1 ) π⎠
⎝ ( z − x2 ) π ⎠ ,...,
⎝ ( z − xn ) ⎠
⎝ dz + c . . . (9.28)

144
Transformation (9.28) can be thought of as the composition of two transformations, Conformal Mapping
namely,
⎛ α1 ⎞ ⎛ α2 ⎞ ⎛ αn ⎞
⎜ ⎟ −1 ⎜ ⎟ −1 ⎜ π ⎟ −1
t= ∫ ( z − x1 ) π⎠
⎝ ( z − x2 ) π ⎠ ,...,
⎝ ( z − xn ) ⎠
⎝ dz . . . (9.29)

and w = kt + c . . . (9.30)
The first of these transform the x-axis into same polygon which then the second
translates, rotates, and either stretches or shrinks, as the case may be. If the polygon
determined by transformation (9.29) is similar to the given polygon, the constants in
(9.30) can always be determined so as to make the two polygons coincide.
For two polygons to be similar, not only must corresponds angles be congruent by
corresponding sides must also be proportional. For triangles, this is automatically the
case. For quadrilaterals, one further condition is required, namely, the two pairs of
corresponding sides have the same ratio. For pentagons, two such conditions are
required.
And, in general, for polygons of n sides (n – 3) conditions over and above the congruence
of corresponding angles, are necessary to ensure similarity.
Hence, in mapping a polygon of n sides into a half-plane, three of the points
x1, x2, . . . , xn, which are the images of the vertices of the polygon, can be assigned
arbitrarily, following which the remaining (n – 3) image points are determined by the
conditions of similarity.
In many important problems, one vertex of the polygon, often an infinite one, will
correspond to z = ∞. This, of course, accounts for one of the three assignments of the
image points which can be made arbitrarily. As a consequence, there will be one less
dw
finite image point in the expression (9.26) for , and only two of the finite image
dz
points x1, x2, . . . , xn – 1, can be specified arbitrarily. In either case, the resulting
transformation is known as the Schwarz-Christoffel Transformation, named for the
German mathematician H. A. Schwarz (1843-1921) and the Swiss mathematician
E. B. Christoffel (1829-1900) who discovered it independently in about 1865.
Since w is analytic everywhere, except possibly at the points z = x1, x2, . . . , xn, the
transformation is conformal over the interior of the two regions.
Due to the complexity of the integral which defines the mapping function w, the
usefulness of the Schwarz-Christoffel transformation is often limited.
We now take up some examples.
Example 9.14
Given a semi-infinite strip in the z-plane between the lines y = 0, y = a for which
x ≥ 0. Find the Schwarz-Christoffel transformation which transforms the boundary
of the strip into the real axis of w-plane such that z = 0, i a are mapped at
w = 1, − 1 respectively.
Solution
Here the planes of w and z have been interchanged. The polygon in the z-plane is
shown by the shaded area in the Figure 9.17(a).
π
In this case there are two interior angles, α1 = = α2
2
⎛π⎞
⎜ ⎟
⎛ α1 ⎞ ⎝2⎠ 1 ⎛α ⎞
So ⎜ − 1 ⎟ = − 1 = − ⎜ 2 − 1⎟
⎝ π ⎠ π 2⎝ π ⎠
Hence the transformation is given by
1 1
dw − − A
= A ( w − 1) 2 ( w + 1) 2 =
dz w −1
2

145
Complex Variables-II A dw
∴ z= ∫ w −1
2
= A cosh −1 w + c

A
π/2

π/2
A′ O′
o x u
−1 O1 1
z-plane
y-plane

Figure 9.17 : A Semi-infinite Strip to be Mapped onto a Half-plane

Since z = 0 corresponds to w = 1 (i.e. A → 0′), we have

0 = A cosh −1 1 + c ⇒ c = 0

Also z = i a is to correspond to w = − 1 (i.e. A → A′), we have

a
i a = A cosh −1 (− 1) = A (i π) ⇒ A =
π
Therefore, the required transformation is

⎛a⎞ ⎛π z⎞
z = ⎜ ⎟ cosh −1 ( w) or w = cosh ⎜ ⎟
⎝2⎠ ⎝ a ⎠
Example 9.15
Figure 9.18 gives the cross-section of a dam consisting of a flat bed AB resting on
porous soil and a pile CD of depth b below the soil. The dam is so long in the
z-direction, that the problem may be considered two-dimensional.

h1

h2

x′ x
A C E B

a1 a2

D
2π
y
Figure 9.18
146
 Obtain the transformation which will convert the contour ACDEB into a straight Conformal Mapping
lines. Hence obtain the stream function and velocity potential for the seepage of
water from the upstream bed Ax′ to downstream bed Bx.
Solution
π
We can consider ACDEB as an incomplete polygon with internal angles at C = ,
2
π
at D = π and at E = . Let this transform into the real line in ξ-plane with the
2
three vertices – 1, 0, 1 respectively. Then the required transformation is
1 1
dz −
= A (ξ + 1) 2 (ξ − 0) (ξ − 1) 2
dξ
Aξ
=
ξ2 − 1

This gives on integration

z = A ξ 2 − 1) + B

C1 −1 o
1 C2
ζ′ ζ
A′ C′ D′ E′ B′

Figure 9.19

Since z = 0 corresponds to z = ± 1, therefore B = 0. Also if CD + b, ξ = 0

corresponds to z = i b. Therefore
ib = A − 1 or A=b

Hence z = b ξ 2 − 1

⎛ z2 ⎞
or ξ = 1+ ⎜ 2 ⎟
⎜b ⎟
⎝ ⎠
We see that the points A and B, i.e. z = − a1 and a2 transform in ξ-plane to the
points ξ = − c1 and c2, where

⎛ a2 ⎞
c1 = 1 + ⎜ 12 ⎟
⎜b ⎟
⎝ ⎠

⎛ a2 ⎞
and c2 = 1 + ⎜ 22 ⎟
⎜b ⎟
⎝ ⎠
Thus in ξ-plane, the seepage is from the surface A′ ξ′ to B′ ξ and the velocity
potential u and stream function v both satisfy Laplace’s equation. Here A′ ξ′ and B′
ξ are both equipotential (i.e. equipressure) lines with pressures due to heads h1 and
h2 of water. The situation is therefore similar to the figure of Example 9.13 with
dotted lines (v = constant) as equipotentials and ellipses (u = constant) as stream
147
Complex Variables-II lines and s′ s (u = 0) as the flat bed. Adjusting the constants to suit the current
problem, we can write
1 1
ξ= (c1 + c2 ) cosh w − (c1 − c2 )
2 2
= c cosh w − d , say

⎛ z2 ⎞
Then c cosh (u + i v) = d + 1 + ⎜ 2 ⎟ gives the velocity potential v and stream
⎜b ⎟
⎝ ⎠
function u in the z-plane. Also u = 0 will give the stream line ACDEB, v = 0 will
give the equipotential BX and v = π the equipotential AX′. The pressure p at any
point is, therefore, given (in term of water column) by
⎛v⎞
p = h1 + (h2 − h1 ) ⎜ ⎟
⎝u⎠
You may now try the following SAQs.
SAQ 12
Find the transformation which maps the interior of the triangle shown in
Figure 9.20 onto the upper half of the z-plane, as indicated.
v y

P
α θ Q P′ Q′
O u
O x

Figure 9.20 : A Triangle Mapped onto a Half-plane

SAQ 13
Find the transformation which maps the infinite region shown in Figure 9.21(a)
onto the upper half of the z-plane, as indicated in Figure 9.21(b).
v y

B A α1 = 2π

α2 = 0
A′ B′
C x
O u O
−1

w-plane z-plane

Figure 9.21 : A Semi-infinite Channel to be Mapped onto a Half-plane

148
 Conformal Mapping

SAQ 14
Find a suitable transformation for obtaining the flow in the channel shown in
Figure 9.22. The velocity at u = − ∞ is given to be U.
v

A′ C′

a
U

B C

A u
O

Figure 9.22 : Channel Flow

[Hint : Map w = − ∞ at z = 0, w = 0 at z = c2 and B at z = 1. Conditions of flow

will be given c = h.]

9.6 SUMMARY
We now summarise the main results of this unit.
• A function w = f (z) is represented graphically by a correspondence between
the points of two cartesian planes, z-plane and w-plane. Whenever a point
z = x + i y is given in the z-plane, the function w = u + i v = f (z) determine
one or more values of u + i v and hence one or more points in w-plane. As z
ranges over any configuration in the z-plane, the corresponding point u + i v
describes some configuration in w-plane. The function w = f (z) thus defines
a mapping of the z-plane into the w-plane and, in turn is represented
graphically by this mapping.
• The mapping w = z + z0 preserves both distances and angles and is more
translation of axes. If z0 = 0, the mapping w = z is called the identity
mapping.
• The mapping w = b z is called magnification, when b is real constant, but
when b is a complex constant, it represents rotation and magnification.
1
• The mapping w = is called inversion. Under inversion,
z
(a) a point at z = 0 is mapped into the point at infinity
(b) a circle in the z-plane is transformed to another circle in w-plane
(c) a circle through the origin in z-plane is transformed to a straight line
(d) straight lines, not passing through the origin, are transformed into
circles into w-plane 149
Complex Variables-II (e) straight lines through the origin in z-plane transform into straight lines
in w-plane.
• Under the mapping w = ez,
(a) the y-axis transforms into the unit circle
(b) the region between the lines y = 0 and y = π maps into the upper half
of the w-plane
(c) the region between the lines y = 0 and y = − π maps into the lower half
of the w-plane
(d) any horizontal strip between y = c and y = c + 2π will transform into
the whole of w-plane.
• If f (z) is analytic, the function w = f (z) will have a single-valued inverse in
the neighbourhood of any point where the derivative of the mapping
function is different from zero.
Exceptional points, where f′ (z) = 0 are called critical points of the
transformation.
• If f (z) is analytic, in the mapping w = f (z), the lengths of infinitesimal
segments, regardless of their direction, are already by a factor | f′ (z) |, which
depends only on the point from which the segments are drawn.

⎛ u, v ⎞
Infinitesimal areas are magnified by the factor | f ′ ( z ) |2 = J ⎜ ⎟.
⎝ x, y ⎠

• In the mapping w = f (z), defined by an analytic functions, angles are, in

general, preserved in magnitude and in sense. The only exception is when
the vertex of the angle is an n-fold zero of f′ (z), in which case the measure
of the angle is altered by the factor (n + 1).
• A transformation which preserves the magnitude of the angle is called
Isogonal.
• A transformation which preserves the sense as well as the magnitude of the
angles is called Conformal.
A transformation w = f (z) is said to be conform at z0 if a small triangle in the
neighbourhood of z0 transforms into a similar triangle in w-plane. This is not
true, however, for large configurations.
• A transformation w = f (z) is conformal if f (z) is analytic and f′ (z) is not
zero at z = z0.
• A harmonic function h (x, y) remains harmonic under a one-to-one
conformal mapping w = f (z).
• If φ (x, y) is a solution of 2-D Laplace equation, then when φ (x, y) is
transformed into a function of u and v by a conformal transformation, it will
∂ 2φ ∂ 2φ
satisfy 2-D Laplace equation in w-plane, i.e. + = 0 everywhere
∂u 2 ∂v 2
except possibly at the images of the points where derivatives of the mapping
function is equal to zero.

150
 (az + b) Conformal Mapping
• The transformation w = , ad − bc ≠ 0 , where a, b, c, d are real or
(cz + d )
complex constants, is called a Bilinear or Linear Fractional or Mobius
Transformation.
b
If ab – bc = 0, the transformation becomes w = = constant. A bilinear
d
(az + b)
transformation w = has at the most two fixed points given by
(cz + d )

[cz 2 − ( a − d ) z − b] = 0

In case c = 0, transformation is linear and there is one fixed point.

In case c = 0 = b, transformation reduces to identity transformation, and
every point is a fixed point.

z−i
• The equation = k (k ≠ 1) represents a circle with p and q as inverse
z−q
points – it is called Circle of Apollonius.
• Under a general bilinear transformation, circles and straight lines are
mapped into circles and straight lines.
{( z1 − z2 ) ( z3 − z3 )}
• The fraction is called the cross-ratio or an harmonic
{( z1 − z4 ) ( z3 − z2 )}
ratio of the four numbers z1, z2, z3, z4.
• The cross-ratio of four points in invariant under bilinear transformation.
• The bilinear transformation which maps z1, z2, z3 into w1, w2, w3 is
( w − w1 ) ( w2 − w3 ) ( z − z1 ) ( z − z3 )
=
( w − w3 ) ( w2 − w1 ) ( z − z3 ) ( z2 − z1 )

• The bilinear transformation

az + b a ⎛ bc − ad ⎞ 1
w= = +⎜ ⎟
cz + d c ⎝ c 2 ⎠ z + d
c
can be considered as a result of the following successive transformations :
d
(a) a shift of origin to
c
(b) a reciprocal transformation
(bc − ad )
(c) a rotation and change of scale due to the factor and
c2
a
(d) a shift of origin to .
c
• The transformation given by
⎛ α1 ⎞ ⎛ α2 ⎞ ⎛ αn ⎞
dw ⎜ ⎟ −1 ⎜ ⎟ −1 ⎜ ⎟ −1
= A ( z − x1 )⎝ π ⎠
(z − x2 )⎝ π ⎠ . . . ( z − xn ) π⎠
⎝
dz
where A is a constant (real or complex), α1, α2, α3, . . . , αn are real numbers
and x1, x2, . . . , xn are points on the real line in the z-plane (in increasing 151
Complex Variables-II order of magnitude) maps the real line in the z-plane into a polygon in
w-plane with interior angles α1, α2, . . . , αn is called Schwarz-Christoffel
Transformation.
Since w is analytic everywhere, except possibly at the points x1, x2, . . . , xn,
this transformation is conformal over the interior of the two regions.

9.7 ANSWERS TO SAQs

SAQ 1

(a) The given mapping is w = z2, or u + i v = ( x + i y ) 2 = x 2 − y 2 + 2ixy .

This implies u = x 2 − y 2 , v = 2 xy . . . (9.31)

We have to find the image of y = x – 2
Eq. (9.31) with the help of Eq. (9.32) become

u = x 2 − ( x 2 − 2) 2 , v = 2 x ( x − 2)

or u = 4x − 4 , v = 2 x2 − 4 x
Eliminating the parameter x, we obtain
2
⎛u + 4⎞
v=2⎜ ⎟ − (u + 4)
⎝ 4 ⎠

or 8v = u 2 + 8u + 16 − 8u − 32

or u 2 = 8 (v + 2)
This defines a parabola with v + 2 = 0 as tangent at the vertex, the line u = 0
as axis and origin as focus and latus – rectum = 2. Thus the image of
y = x – 2 under the mapping w = z2 is u2 = 8 (v + 2).
v

y
2
u = 8 (v+2)

y = x−2

x u
o o

(0, −2) v+2 = 0

z-plane w-plane

Figure 9.23 : Image of y = x – z under the Mapping w = z2

(b) The given mapping is w = z2, i.e. u + i v = ( x + i y ) 2 = x 2 − y 2 + 2ixy

⇒ u = x 2 − y 2 , v = 2 xy .
v
y

a=2
1 a=1
4
o 1
152 x u
o

2
 Conformal Mapping

Figure 9.24

We have to find the image of x 2 + y 2 = a 2 .

Eliminating x from above relation, we get

u = a 2 − 2 y 2 , v 2 = 4 y 2 (a 2 − y 2 )

Eliminating the parameter y, we get

⎛ a2 − u ⎞
v2 + 2 (a 2 − u ) ⎜ a 2 − ⎟⎟ = ( a − u ) (2a − a − u ) = ( a − u ) ( a + u )
2 2 2 2 2
⎜ 2
⎝ ⎠

⇒ u 2 + v 2 = a 4 = ( a 2 ) 2 , which is the required image.

If the image of circle x 2 + y 2 = a 2 under the mapping w = z2 is the circle

u 2 + v 2 = a 4 , where centre is at the origin, but radius a2, i.e. square of
radius of original circle.
SAQ 2
1
Image of | z | > 1 under the mapping w = is the interior of circle | w | = 1, i.e.
z
| w | < 1.
1
Image of | z | < 1 under the mapping w = is the exterior of unit circle | w | = 1,
z
i.e. | w | > 1.
SAQ 3
The given mapping is w = z3
Taking the polar forms, we get

ρ ei φ = r 3 e3i φ

so that ρ = r 3 , φ = 3θ

We see that the unit circle transforms into the unit circle, and the initial line
(θ = 0) transforms into the initial line (φ = 0).

The circle r = r0 transforms into the circle ρ = r03 . The line θ = θ0 transforms to the
line φ = θ n0. This shows that the angles are not preserved at the origin. In fact that
π
angular region between the lines θ = 0 and θ = maps into the upper half of the
3
w-plane (see Figure 9.25). (Ans. I).

y v

π/3 π 153
x u
o O′
Complex Variables-II

Figure 9.25

The mapping is w = z 3 ⇒ u + i y = ( x + i y )3 = x3 − i y 3 + 3ix 2 y − 3 xy 2

⇒ u = x3 − 3 xy 2 , v = 3 x 2 y − y 3

Now image of u = 1 under the mapping w = z 3 is x3 − 3 xy 2 = 1 (Ans. II).

To find the image of line x = 1, relations are u = 1 − 3 y 2 , v = 3 y − y 3 .

Eliminating y, we get the required image as

1−u ⎛ u − 1⎞
v= ⎜3 + ⎟
3 ⎝ 3 ⎠

or 27v 2 = 64 − 48u − 15u 2 − u 3 (Ans. III).

SAQ 4
The given transformation is

w = u + i v = z 2 = ( x + i y ) 2 = x 2 − y 2 + 2ixy

and thus u = x 2 − y 2 and v = 2 xy

Now image of the line x = 0 is

u = − y 2 and v = 0 . . . (9.33)
y v

v=4
x=0

v=1
x u
O y=0 O′

u=−4 u=−1
z-plane w-plane

Figure 9.26

This defines a family of a parallel lines, parallel to v-axis for different values of y.
Similarly, the image of the line y = 0 is

u = x2 , v = 0 . . . (9.34)

This defines a family of parallel lines, parallel to u-axis for different values of x.
π
These two images (9.33) and (9.34) intersect at an angle of .
2
SAQ 5
Let z = g ( w) = x (u , v) + i y (u , v) be the inverse transformation of
154
 w = u + i v = f ( z) = f ( x + i y) Conformal Mapping

Differentiating the relations

u = u ( x, y ) and v = v ( x, y )

with respect to u and then with respect to v, show that

1
xu (u , v) = V y ( x, y )
⎛ u, v ⎞
J⎜ ⎟
⎝ x, y ⎠
−1
xv (u , v) = U y ( x, y )
⎛ u, v ⎞
J⎜ ⎟
⎝ x, y ⎠
−1
yu (u , v) = Vx ( x, y ) . . . (9.35)
⎛ u, v ⎞
J⎜ ⎟
⎝ x, y ⎠
1
yv (u, v) = U x ( x, y )
⎛ u, v ⎞
J⎜ ⎟
⎝ x, y ⎠
Now, since w = f (z) is analytic, therefore, u, v satisfy C. R. Equations,
∂u ∂v ∂u ∂v
i.e. = and =− . . . (9.36)
∂x ∂y ∂y ∂x

substituting from relations (9.35) in (9.36), we get

∂x 1 1 ∂y
xu = = V y ( x, y ) = U x = yv =
∂x ⎛ u, v ⎞ ⎛ y, v ⎞ ∂v
J⎜ ⎟ J⎜ ⎟
⎝ x, y ⎠ ⎝ x, y ⎠
∂x −1 1 ∂v
and xv = = U y ( x, y ) = Vx ( x, y ) = − yu = −
∂v ⎛ u, v ⎞ ⎛ u, v ⎞ ∂u
J⎜ ⎟ J⎜ ⎟
⎝ x, y ⎠ ⎝ x, y ⎠
Hence z = g ( w) = x (u , v) + i y (u , v) satisfy C. R. Equations, showing that this
inverse is an analytic function.
SAQ 6
Critical points of the transformation w = f (z) are given by f (z) = 0

(a) ∴ Critical points of w = z2 + 2z are z 2 + 2 z = 0 ⇒ z = 0, − 2

π
(b) Critical points of w = cos z are cos z = 0 ⇒ z = for
2
n = ± 1, ± 3, ± 5, . . . .

SAQ 7
Here w = Z3

∴ f ( z ) = z 3 , f ′ ( z ) = 3z 2

∴ f ′ ( z ) at (1 + i ) = 3 (1 + i ) 2 = 3 (1 − 1 + 2i ) = 6i ≠ 0 155
Complex Variables-II ∴ Magnificent coefficient of conformal transformation w = z3 at the point (1 + i) is
| f ′ (1 + 6i ) | = | 6i | = 6 (Ans.)

SAQ 8
Consider the transformation
b
z+
az + b a a
w= = . . . . (9.37)
cz + d c z + d
c
1
Here w = 0 and w = ∞ must correspond to the inverse points, say z = α and z = ,
α
where α is interior point of circle | z | = 1, i.e. where | α | < 1.
b 1 d
α+ +
a a c α c
Hence 0 = . and 0 = .
c α+d a 1 +d
c α c
b 1 d
⇒α=− and =− . . . (9.38)
a α c
From (9.37) and (9.38), we get
a z − α aα z − α
w= . = . . . . (9.39)
c z− 1 c αz −1
α
The point z = 1 must correspond to
|w | = 1
Hence (9.39) yields
aα 1− α aα
|w|= =1
c α −1 c

aα
Thus if we take = ei λ , where λ is real, then the above condition is satisfied.
c
aα
Hence, Eq. (9.39) with = ei λ gives the required transformation. To verify it, if
c
z = ei θ, α = ei β, then

ei θ − b ei β
|w|= =1
b e(i θ − β) − 1

Thus | z | = 1 is mapped to | w | = 1.
Again if z = r ei θ, where r < 1, is a point of | z | < 1, then

| z − α |2 − | α z − 1 |2 = [ r 2 − 2rb cos (θ − β) + b 2 ]

− [b 2 r 2 − 2br cos (θ − β) + 1]

= ( r 2 − 1) (1 − b 2 ) < 1
Hence | z − α | < | α z − 1 |

156
 Conformal Mapping
z−α
and so | w|= <1
αz −1

Thus the unit circle disc | z | ≤ 1 corresponds to | w | ≤ 1.

SAQ 9
We first map the sector into a semi-circle by the transformation t1 = z3 and then
map the semi-circle from the t1-plane onto the first quadrant of the t2-plane by
t −1
means of the transformations t2 = − i 1 . Finally, let us open out the first
t1 + 1
quadrant of the t2-plane into the upper half of the w-plane by the transformation
w = + t22 .

O2
B

o
B′at ∞
60

O A B1 O1 A1 A2 O′ A′

z-plane t1-plane t2-plane w-plane

Figure 9.27 : The Sequence of Transformation Necessary to Map a Circular Sector to Half-plane

Taking the composition of these three transformations, we find

2
⎛ z3 − 1 ⎞
w=−⎜ 3
⎜ z + 1 ⎟⎟
⎝ ⎠
as the required mapping.
Note : Had we solved the problem simply by opening the given sector into a full
circle by the transformation t = z6 and then mapping the circle from the t-plane to
the upper half of the w-plane by means of inverse of one of the transformation of
z−λ w −1
the family w = ei θ , i.e. by means of the transformation say z = − i ,
z+λ w +1
then we would have faced the difficulty of mapping the circular region consisting
of a simple boundary + interior obtained in t-plane in this case (instead of interior
of a circle cut along a radius, as the radius OA1 ≡ OB1 is actually image of the two
boundaries radius OA and OB of the given sector (see Figure 9.28 below).

A′
x
O A B′

z-plane

t-plane

Figure 9.28 : A Circular Sector Opened Out into a Circular Region Cut along a Radius

SAQ 10
The bilinear transformation mapping z1, z2, z3, w1, w2, w3 is given by 157
Complex Variables-II ( w1 − w) ( w3 − w2 ) ( z1 − z ) ( z3 − z2 )
=
( w1 − w2 ) ( w3 − w) ( z1 − z2 ) ( z3 − z )

(a) Here z1 = 0, z2 = − i, z3 = − i, w1 = i, w2 = 1, w3 = 0 . Then, in this case

(i − w) (0 − 1) (0 − z ) (− 1 + i ) w−i z (1 − i )
= ⇒ =
(i − 1) (0 − w) (0 + i ) (− 1 − z ) w (1 − i ) − i (1 + z )

i z (1 − i )2 1 z (1 − 1 − 2i )
⇒ 1− = ⇒ =−i−
w − i (1 + z ) w (1 + z )

1 i + i z − 2iz i ( z − 1) z +1
⇒ =− = ⇒w=i
w 1+ z z +1 1− z

(b) Here z1 = i, z2 = 1, z3 = − 1, w1 = 1, w2 = 0, w3 = ∞ writing the

transformation as
⎛ w ⎞
( w1 − w) ⎜1 − 2 ⎟
⎝ w3 ⎠ ( z1 − z ) ( z3 − z2 )
=
⎛ w ⎞ ( z1 − z2 ) ( z3 − z )
( w1 − w2 ) ⎜1 − ⎟
⎝ w3 ⎠

and substituting for z1, z2, z3, w1, w2, w3, we get
(i − w) (1 − 0) (i − z ) (− 1 − 1) 2 ( z − i)
= ⇒i−w=
(i − 0) (1 − 0) (i − 1) (− 1 − z ) (1 − i ) ( z + 1)

( z + i ) (1 + i )
⇒w=i−
( z + 1)

(c) Here z1 = − i, z2 = 0, z3 = i, w1 = ∞, w2 = − 1 + w3 = 0 .
The required transformation is given by
(1 − 0) (0 + 1) (− i − z ) (i − 0) −1
= ⇒
(1 − 0) (0 − w) (− i − 0) (i − z ) w

z+i z −1
⇒ ⇒ w=
i−z z +1
SAQ 11
az + b
Let the transformation be w = .
cz + d
The fixed points of the transformation are given by
(az + b)
z= ⇒ cz 2 − (a − d ) z − b = 0
(cz + d )

a−d b
= z1 + z2 = , z1 z2 = −
c c
∴ Fixed points are 1 and 2,
a−d
∴ 1+ 2 =
c
b d a
and 2=− ⇒− =3−
158 c c c
 b Conformal Mapping
and − =2
c
a
Let = λ , then required transformation is the family of curves given by
c
a b
+
λz − 2
w = cz c =
z+
d z+λ−3
c
SAQ 12
With the given assignments of images, remembering that the vertex R is mapped
into the point at infinity in the z-plane, we have
⎛α⎞ ⎛β⎞ ⎛β⎞
dw ⎜ ⎟ −1 ⎜ ⎟ −1 ⎜ ⎟ −1
= k ( z − 0)⎝ π ⎠ ( z − 1)⎝ π ⎠ (1 − z )⎝ π ⎠
dz
⎛β⎞
⎜ ⎟ −1
where k1 = k (− 1)⎝ π ⎠ . It is convenient in this case to introduce the dummy
variable t and integrate between the limit 0 to z. This gives
z ⎛α⎞
⎜ ⎟ −1
w = k1 ∫ t⎝ π ⎠ t+c
0

Since w = 0 corresponds to z = 0, it follows that C = 0. Likewise since w = 1

corresponds to z = 1, we have
1 ⎛α⎞ ⎛β⎞
⎜ ⎟ −1 ⎜ ⎟ −1
1 = k1 ∫ t⎝ π ⎠ (1 − t )⎝ π ⎠ dt
0

⎛α β⎞
The last integral is the Beta-Function B ⎜ , ⎟ and its value is
⎝ π π⎠
⎛α⎞ ⎛β⎞
Γ⎜ ⎟Γ⎜ ⎟
⎝ π⎠ ⎝π⎠
⎛α + β⎞
B⎜ ⎟
⎝ π ⎠
1 ⎛α β⎞
Thus k1 = ⎜ , ⎟ and the transformation becomes
B ⎝ π π⎠
z ⎛α⎞ ⎛β⎞
1 ⎜ ⎟ −1 ⎜ ⎟ −1
w=
⎛ α β⎞ ∫ t⎝ π ⎠ (1 − t )⎝ π ⎠ dt
B⎜ , ⎟ 0
⎝ π π⎠
In the last expression, z is complex and the integral is very difficult to evaluate.
When z is real number between 0 to 1, the integral is the so-called incomplete Beta
Function, which is of considerable interest in statistics.
SAQ 13
With images assigned as shown and with angle at the finite vertex A identified as
α1 = 2 π and the angle at infinite vertex b identified as α2 = 0, we have
⎛ 2π ⎞ ⎛o⎞
dw ⎜ ⎟ −1 ⎜ ⎟ −1 ⎛ 1⎞
π ⎠
= k ( z + 1)⎝ z⎝ π ⎠ = k ⎜1 + ⎟
dz ⎝ z⎠ 159
Complex Variables-II and integrating, we get
w = k ( z + ln z ) + c . . . (9.40)
To determine the constants k and c, we write (9.40) in the form
(u + i v) = ( k1 + i k2 ) ( z + i y + ln | z | + i arg z ) + c1 + i c2 . . . (9.41)
Equating imaginary parts, we obtain
v = k1 y + k2 x + k2 ln | z | + k1 arg z + c2 . . . (9.42)
When w becomes infinite along AB, in which v = π, the image point z approaches
zero along the negative real axis, on which y = 0 and arg z = π. Hence from (9.42),
we get
π = Lt (k1.0 + k2 x + k2 ln | z | + k1 π + c2 )
z →0

Obviously k2 must be zero to keep ln | z | from making the right hand side infinite.
Hence π = k1 π + c2 . . . (9.43)
Also, as w becomes infinite along OB, on which y = 0, the image point z
approaches zero along positive real axis, in which y = 0 and arg z = 0. Hence, using
(9.42) again, we have
0 = Lt (k1.0 + c2 ) = c2
z →0+

Therefore, c2 = 0 and from (9.43) we find that k1 = 1. Thus relation (9.40) reduces
to w = z + ln z + c1. Finally, the point A, that is, w = i π must map into the point A′,
that is, z = − 1. Hence from the last relation
i π = − 1 + ln ( − 1) + c1 = − 1 + i π + c1 ⇒ c1 = 1
The required mapping function is, therefore,
w = z + ln z + 1
Note : The curve in w-plane which correspond to the semi-circle r = 1, 2, 3 and the
π π π
rays θ = , , 3 in the z-plane as shown in Figure 9.29.
4 2 4

θ = 3π/4
θ = π/2
θ = π/4

r=2 r=3
r=1

Figure 9.29 : Typical Streamlines for Fluid Flow from a Straight Channel
This figure represents either the lines of equal velocity potential and the
streamlines for the flow of an ideal incompressible fluid from an infinite channel
or the lines of flux and equipotential lines for a parallel-plate capacitor.
SAQ 14
Using hint, the suitable transformation is
1+ s h+s
w = h log − log
1− s h−s
160
 Conformal Mapping
z − h2
where s2 = .
z −1

FURTHER READING
Kreyszig Erwirn, Advanced Engineering Mathematics, John Wiley and Sons.
Jeffrey Alan, Mathematics for Engineers and Scientists, Van Nostrand Reinhold (UK)
Co. Ltd.
Grewal B. S., Higher Engineering Mathematics, Khanna Publishers, Delhi.
Deshpande J. V., Complex Analysis, Tata McGraw-Hill Publishing Co. Ltd., New Delhi.
Gonzalez Mario O., Complex Analysis, Marcel DEKKER Inc. New York.
Moore Theral O. and Hadlock Edwin H., Complex Analysis, World Scientific Publishing
Company.
Flanigan Francis J., Complex Variables, Dover Publishing Inc., New York.
Churchill Ruel V. and Brown James Ward, Complex Variables and Applications,
International Student Edition, McGraw-Hill Book Company.

161
Complex Variables-II

162
 Conformal Mapping

COMPLEX VARIABLES-II
In Block 1, we dealt with sequences and series and studied the convergence of infinite
series and properties of Fourier Series. Block 2 acquainted us with the complex numbers
and there we studied some important features of theory of functions of complex
variables. In continuation to Block 2, some more features of theory of complex variables
are developed and explained in Block 3.
Block 3 has been divided into three units.
Unit 7 deals with the expansion of functions in power series of complex terms.
Application of Taylor’s series and Laurent’s series as convergent series of complex
functions are discussed in detail. Further, zeros and singularities of analytical function
and application of residues for evaluation of real integrals are explained. Jordan’s
inequality and Jordan’s Lemma are discussed with appropriate example.
Unit 8 discusses evaluation of definite and improper integrals. It also describes the use of
contour integration, a powerful and elegant tool to evaluate real integrals of
single-valued and many-valued functions.
Unit 9 deals with the conformal mapping, the bilinear transformations and their
applications. Some special mappings which map certain given regions into certain other
given regions are also discussed.
You will find that each unit contains a number of Self Assessment Questions (SAQs) to
help you to self-monitor the progress made in the process of knowledge acquisition. If
you go through the text attentively and attempt the SAQs independently, you will
definitely develop a confidence to solve any problem related to this course.

163