Mohamad A Mehdi Conformal Mappings and Applications
Mohamad A Mehdi Conformal Mappings and Applications
Applications
Mohamad A. Mehdi
THESIS
1
.
2
Conformal Mapping and its
Applications
Mohamad A. Mehdi
Approved by:
-
Dr. Bassem Ghalayini
3
.
4
This thesis is dedicated to my two lovely daughters Tala and Sara. I love
you both a lot.
5
.
6
Acknowledgments
I would also like to thank all the members of the math department at Notre Dame
University-Louaize. Dr. Melhab Keirouz, I would like to thank you for the continuous
inspiration and eagerness you provided me with. I am also thankful to Dr. Georges Eid,
Dr. Bassem Ghalayini, Dr. Holem Saliba, and Dr. Joesph Malkoun.
I must express my gratitude to Zeinab Dia, my wife, for her continuous support
and encouragement. Also I am forever in debt to my mother, father, brothers, and
sister who were there for me with all the ups and downs of my research. Special thanks
are extended to my student Zahraa for her willingness to proof read countless pages of
meaningless mathematics. I would like to thank Dr. Yasser Fadlallah for standing by
my side throughout the research time.
Completing this work would not have happened without the support and help of my
close friends Sara, Hussein, Rida and Mohanad. I shall always be in their dept.
7
.
8
Introduction
Conformal mappings have long played an important role in hydrodynamics and aero-
dynamics; this can be projected through Halseys work [Hal]. Siegel and Snyder suc-
ceeded in applying conformal mappings to the analysis of heat transfers in porous regions
with curved boundaries [Sie-Sny] and Theodorsen stressed on the importance of numeri-
cal conformal mapping when working on wing sections [Thd]. Contributions in applying
conformal mapping techniques to determine the exact frequencies of vibrating plates of
complicated boundary shapes carrying concentrated masses were made by Schinzinger
and Laura [Sch-Lau]. The application of conformal mapping can also conquer the pat-
9
tern recognition field. In fact, Wechsler and Zimmerman mapped the arrangement of
pixels on a machine-vision image from a log-spiral configuration into a rectangular plane
[Wec-Zim]. Conformal mapping methods has been also successfully applied in a wide
variety of fields varying from acoustic wave guide studies to plate buckling theory, from
the analysis of vibrating printed circuits to the determination of the temperature field
in the nuclear field arrangement of fast breeder reactors, from studies on airflow past
multielement airfoils to the solution of geophysical and oceanographic problems, and
from microstrip design to analysis of the earth-ionosphere wave guide.
The goal of this thesis is to introduce conformal mappings, prove their prop-
erties and discuss some of their applications.
10
Contents
2 Conformal Mapping 25
2.1 Mappings by elementary functions . . . . . . . . . . . . . . . . . . . . . . 25
2.1.1 Linear transformations . . . . . . . . . . . . . . . . . . . . . . . . 25
2.1.2 The transformation w = z1 . . . . . . . . . . . . . . . . . . . . . . 26
2.1.3 The mapping w = sin z . . . . . . . . . . . . . . . . . . . . . . . . 27
2.1.4 Linear fractional transformations . . . . . . . . . . . . . . . . . . 28
2.1.5 Linear fractional transformations of the upper half plane . . . . . 29
2.2 Conformal mappings . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
2.2.1 Definition of conformal mappings . . . . . . . . . . . . . . . . . . 31
2.2.2 Examples of conformal mappings . . . . . . . . . . . . . . . . . . 33
2.3 Properties of conformal mappings . . . . . . . . . . . . . . . . . . . . . . 36
11
12
Chapter 1
It is known that a complex number z = x + iy can be written in polar form (or expo-
nential form) as
z = rei = r cos + ir sin ,
p
where r is the length of z, r = |z| = x2 + y 2 and is the angle that z makes with the
positive real axis when z is interpreted as a radius vector. has an infinite number of
possible values that differ by integral multiples of 2. Each value is called an argument
of z and the set of all such values is denoted by arg z. The principal value of arg z
denoted by Arg z is that unique value such that < . Evidently, then,
arg z = Arg z + 2n (n = 0, 1, 2, ).
Using the polar form of a complex number, f (z) can be also written as
f (z) = f (rei ) = U (r, ) + iV (r, ),
where U (r, ) and V (r, ) are real functions of r and .
13
A neighborhood of a complex number z0 is the set of complex numbers z such that
|z z0 | < for some > 0.
Definition 1.1.1. Let f be a complex function defined for all points z in a certain neigh-
borhood of z0 . We denote the limit of f as z approaches z0 is w0 in C by limz z0 f (z) =
w0 . It means that
If > 0, > 0 such that |f (z) L| < whenever |z| > , we say that the limit of f
when z tends to infinity is L. It is denoted by
1
lim f (z) = lim f = L. (1.1)
z + z 0 z
A function f is continuous at z0 if limzz0 f (z) exists, f (z0 ) exists and limzz0 f (z) =
f (z0 ).
f (z) f (z0 )
lim (1.2)
z z0 z z0
exists. This limit (when it exists) is denoted by f 0 (z0 ).
Proposition 1.1.1. Let the function f (z) = U (x, y) + iV (x, y) be defined through-
out some neighborhood of a point z0 = x0 + iy0 . Suppose that the first-order partial
derivatives of the functions U and V with respect to x and y exist everywhere in that
neighborhood and those partial derivatives are continuous at (x0 , y0 ) and satisfy the
Cauchy-Riemann equations, i.e.,
Ux = Vy ,
(1.3)
Uy = Vx ,
at (x0 , y0 ). Then f 0 (z0 ) exists and we have f 0 (z0 ) = Ux (x0 , y0 ) + iVx (x0 , y0 ). Here Ux
(res. Uy ) denotes the partial derivative of U w.r.t the variable x (resp. w.r.t the variable
y) and Vx (res. Vy ) denotes the partial derivative of V w.r.t the variable x (resp. w.r.t
the variable y).
Proposition 1.1.2. Let the function f (z) = U (r, ) + iV (r, ) be defined throughout
some neighborhood of a point z0 = r0 ei0 . Suppose that the first-order partial derivatives
of the functions U and V with respect to r and exist everywhere in that neighborhood
14
and those partial derivatives are continuous at r0 ei0 and satisfy the Cauchy-Riemann
equations, i.e.,
rUr = V ,
(1.4)
U = rVr ,
at (r0 , 0 ). Then f 0 (z0 ) exists and we have f 0 (z0 ) = ei0 (Ur (r0 , 0 ) + iVr (r0 , 0 )). Here
Ur (resp. U ) denotes the partial derivative of U w.r.t the variable r (resp. w.r.t the
variable ) and Vr (res. V ) denotes the partial derivative of V w.r.t the variable r (resp.
w.r.t the variable ).
A subset D of the complex plane is a domain if D is open and connected. It means
that D is open and any two points in D can be joined by a broken line segment in D.
A domain D is said to be star-shaped with respect to z0 if for all points z1 in D, the
straight line segment joining z0 and z1 is contained in D. A domain D is said to be
star-shaped if it is star-shaped with respect to at least one of its points.
For instance, the function f (z) = z1 is analytic at each nonzero point in the finite
plane but the function f (z) = |z|2 is not analytic at any point since its derivative exists
only at z = 0 and not throughout any neighborhood. An entire function is a function
that is analytic at each point in C. Since the derivative of a polynomial exists every-
where, it follows that every polynomial is an entire function.
15
where k Z and is the principal argument of z. For example, for z = 1 i 3,
we have log z = ln 2 + i(
3
+ 2k), where k Z.
The principal value of log z is the value obtained from Equation (2) when k = 0
and it is denoted by Log z, i.e., Log z = ln r + i. Note that
The function f (z) = Logz is analytic on D with f 0 (z) = z1 , where D denotes the
set of the complex plane deprived of the points which are real and non-positive.
It is called the principal branch of Logz. More general, the function f (z) = log z
is analytic on D with f 0 (z) = z1 and where D denotes the entire complex plane
deprived of the points on the semi line starting from the origin and making an
angle with the positive x-axis, i.e., D = C {z, |z| =
6 0, < arg z < + 2}.
d n
3. It is clear that f (z) = z n is entire and that dz
z = nz n1 .
4. We may use the exponential function to define the complex sine and cosine of
complex number. In fact, we have
known as Laplaces equation. Here Hxx (resp. Hyy ) denotes the second partial derivative
of H with respect to x (resp. y).
Example 1.2.1. [Bro-Chu] The function T (x, y) = ey sinx is harmonic in any domain
of the xy-plane and, in particular, in the semi-infinite vertical strip 0 < x < , y > 0 .
16
It also assumes the values on the edges of the strip. More precisely, it satisfies all of the
conditions
Txx + Tyy = 0
Thus, T is harmonic in any domain of the xy-plane particularly in the semi - infinite
vertical strip (0 < x < , y > 0). More precisely, it satisfies all the conditions:
Txx (x, y) + Tyy (x, y) = 0,
T (0, y) = 0,
T (, y) = 0, (1.7)
T (x, 0) = sin x,
limy T (x, y) = 0,
which describe steady temperatures T (x, y) in a thin homogeneous plate in the xy-plane
that has no heat sources or sinks and is insulated except for the stated conditions along
the edges.
The next theorem provides a source of harmonic functions. In fact, using the Cauchy-
Riemann equations, we can state the following:
Theorem 1.2.1. If a function f = U + iV is analytic in a domain D, then U and V
are harmonic.
If two given functions U and V are harmonic in a domain D and their first order
partial derivatives satisfy the Cauchy-Reimann equations throughout D, then V is said
to be a harmonic conjugate of U .
Theorem 1.2.2. A function f (z) = U + iV is analytic in a domain D if and only if V
is the harmonic conjugate of U .
Remark 1.2.1. Let U and V be 2 functions in a domain D.
1. If V is a harmonic conjugate of U , then it is not always true that U is a har-
monic conjugate of V . Consider U (x, y) = x2 y 2 and V (x, y) = 2xy. A simple
calculation shows that Uxx + Vyy = 0, Ux = Vy and Uy = Vx , and as U and V are
continuous it follows that V is a harmonic conjugate of U and f (z) = U + iV = z 2
is analytic on any domain D. On the other hand g(z) = 2xy + i(x2 + y 2 ) is not
analytic on any domain D, which means that in this case U is not a harmonic
conjugate of V .
2. The harmonic conjugate of a function is unique up to adding a constant. Consider
the continuous function U (x, y) = y 3 3x2 y. We want to find a function V which
is a harmonic conjugate of U. Knowing that U and V satisfy the Cauchy-Reimann
equations, we get Ux = Vy which means that Vy = 6xy and thus V = 3xy 2 + (x).
Similarly, Uy = Vx so 3y 2 3x2 = 3y 2 0 (x) and 0 (x) = 3x2 and hence (x) =
x3 . Now we can write the harmonic conjugate of U which is V = 3xy 2 + x3 + C
where C is any real constant. It means that all the functions:
fC (z) = y 3 3x2 y + i(3xy 2 + x3 + k) = i(z 3 + C)
are analytic for all values of C R.
17
1.3 Complex integration
We are now able to define the integral of a complex function f . Complex integration
affords a lot of freedom due to how different the complex plane is from R.
Let f (t) be a complex-valued function of a real variable t, i.e., f (t)= U (t) + iV (t),
then the definite integral of f (t) over the interval a t b is given by
Z b Z b Z b
f (t)dt = U (t)dt + i V (t)dt.
a a a
An arc (C) is smooth if it has non vanishing, continuous derivative. Finally a con-
tour, or piecewise smooth arc (C) is an arc consisting of a finite number of smooth arcs
joined end to end. When only the initial and final values of z(t) are the same, a contour
C is called a simple closed contour. The length of a contour or a simple closed contour
is the sum of the lengths of the smooth arcs that make up the contour.
18
Theorem 1.3.1. [Greens Theorem] Let (C) be a positively oriented simple closed
curve in a plane, and let D be the region bounded by (C). If P and Q are two functions of
x and y defined on an open region containing D and have continuous partial derivatives
then, Z ZZ
P Q
P dx + Qdy = dxdy. (1.8)
(C) D x y
Definition 1.3.2. P (x, y) and Q(x, y) are continuously differentiable functions.
P Q
1. The differential P dx + Qdy is said to be closed on a domain D if y
= x
on D.
2. The differential of P dx + Qdy is said to be exact if there exists a differentiable
complex function h where dh = P dx + Qdy.
From the previous definition, we can say that exact differentials are closed (which
follows immediately from Cauchy-Riemann equations applied to the function h). More-
R (x ,y )
over, the integral (x01,y01) P dx + Qdy is independent of the path from (x0 , y0 ) to (x1 , y1 )
if the differential P dx + Qdy is exact, and in this case
Z x1 ,y1 )
P dx + Qdy = h(x1 , y1 ) h(x0 , y0 )
(x0 ,y0 )
19
2. Cauchy integral formula[Bro-Chu]. Let f be an analytic function everywhere
inside and on a simple closed contour (C) taken in the positive sense. If z0 is any
point interior to (C) then:
Z
1 f (z)dz
f (z0 ) =
2i (C) z z0
and more general, we have
Z
(n) n! f (z)dz
f (z0 ) =
2i (C) (z z0 )n+1
2. Let (C) is a positively oriented circle with equation |z| = 2, then by Cauchy integral
z
formula applied to the function f (z) = 9z 2 we get
Z Z z i
zdz 9z 2
2
= dz = 2if (i) = 2i =
(C) (9 z )(z + i) (C) z (i) 10 5
20
Suppose that f is analytic throughout an annular domain D : R1 < |z z0 | < R2 ,
and let (C) denote any positively oriented simple closed contour around z0 lying in D,
then the Laurent series representation of f on D is
X
n
X bn
f (z) = an (z z0 ) +
n=0 n=1
(z z0 )n
1
R f (z)dz 1
R f (z)dz
where an = 2i (C) (zz0 )n+1
and bn = 2i (C) (zz0 )n+1
.
1 1 1
2. [Bro-Chu] Consider the function f (z) = (z1)(z2) = z1 z2 which clearly has
two singular points z = 1 and z = 2. The function f is analytic in D1 P : |z| < 1,
D2 : 1 < |z| < 2 and D3 : 2 < |z| < . In the domain D1 , we have 1z = 1 n
n=0 z ;
now as | z2 | < 1, then 1(1 z ) = n=0 ( z2 )n . Noting that f (z) = 1z
1
+ 12 11 z , we get
P
2 2
that the Maclaurin expansion of f is:
X 1 X z n X n1
n
f (z) = z + ( ) = (2 1)z n
n=0
2 n=0 2 n=0
1 1 1 1
Similarly, in D2 we can write f (z) = z 1 z1
+ 2 1 z2
to get for 1 < |z| < 2 the series
X zn X 1
f (z) = +
n+0
2n+1 n=1 z n
1 1 1 1
In D3 , f (z) = z 1 z1
z 1 z2
and so
X 1 2n1
f (z) =
n+0
zn
3. Let be a closed positively oriented contour inside D1 : |z|R< 1 around 0 and let
1 1
f be the function defined by f (z) = (z1)(z2) . The integral z2 (z1)(z2) dz can be
computed directly using the Maclaurin series of f in D1 as
1
Z Z
f (z)dz
2
dz = = 2ia1
z (z 1)(z 2) z2
1
Z 3 1
2
dz = 2i = i
z (z 1)(z 2) 4 2
21
1
R
We can also compute z2 (z1)(z2) dz using the Cauchy integral formula. Take
1
f (z) = (z1)(z2) which is analytic everywhere inside and on (positively oriented).
Now as 0 is an interior point of , then by Cauchy integral formula:
1
Z Z
f (z)dz 2i 00
2
dz = 2
= f (0)
z (z 1)(z 2) z 2!
00 2 2 00
Knowing that f (z) = (z1)3
(z2)3
, we get f (0) = 12 , so:
1
Z
2i 00 1
dz = f (0) = i
z 2 (z 1)(z 2) 2! 2
Proposition 1.4.1. [Bro-Chu] Let (C)Pdenotes any contour interior to the circle of
convergence of the power series S(z) = n
n=0 an (z z0 ) .
1. The series formed by multiplying each term of the power series by g(z) can be
integrated term by term over (C) where g(z) is any continuous complex function
on (C). That is:
Z
X Z
g(z)S(z)dz = an g(z)(z z0 )n dz
(C) n=0 (C)
z = i.
z0 and m N. Moreover,
22
S (m1) (z0 )
(b) Resz=z0 f (z) = (m1)!
when m 2.
Theorem 1.4.1 (Cauchys residue theorem). [Bro-Chu]. Let (C) be a simple closed
positively oriented contour. If a function f is analytic inside (C) and on (C) except on
a finite number of singular points zk (k = 1, 2, . . . , n) inside (C) then:
Z X n
f (z)dz = 2i Res f (z). (1.9)
(C) z=zk
k=1
5z2
R
Example 1.4.2. Consider (C) z(z1) dz, where (C) is the circle |z| = 2. The function
5z2
f (z) = z(z1) has two isolated singularities inside (C) : z = 0 and z = 1, now we need to
find r1 and r2 which are the respective residues of f at z = 0 and z = 1 . As 0 < |z| < 1,
1
it is clear that f (z) = 5z2
z 1z
= (2 z2 )(1 z z 2 . . .), and so r1 = 2 as it is b1 in
the Laurent series. Similarly in 0 < |z 1| < 1, we can write the Laurent series to get
r2 = b2 = 3 and thus
Z X2
f (z)dz = 2i Res f (z) = 2i(r1 + r2 ) = 2i(5) = 10i
(C) z=zk
k=1
Another approach for finding r1 and r2 is by noting that z = 0 and z = 1 are both poles
5z2
z1 5z2
of order 1 with respect to f , as f (z) = z
, so r1 = g(0) = 2, where g(z) = z1
. To
5z2
5z2
find r2 , take f (z) = z
z1
and so r2 = h(1) = 3, where h(z) = z
.
Definition 1.4.2. A number z0 is a zero of f with order m if the function f is analytic
at z0 , all derivatives f (n) (z) at z0 exist, f (z0 ) = 0, f (k) (z0 ) = 0 (k = 1, 2, . . . m 1) and
f (m) (z0 ) 6= 0 .
Next, we provide some theorems relating poles and zeroes of analytic functions.
Theorem 1.4.2. [Bro-Chu] A function f analytic at z0 ) has z0 a zero of order m if and
only if there exists a function g, which is analytic and nonzero at z0 such that
f (z) = (z z0 )m g(z)
From the previous theorem, if two functions P (z) and Q(z) are both analytic at z0
and P (z0 ) 6= 0 and z0 is a zero with order m of P (z), then z0 is a pole of order m for
P (z)
f (z) = Q(z) . Now, if P and Q are two analytic functions such that P (z0 ) 6= 0 and z0 is
P (z)
a zero with order 1 for Q(z), then z0 is a pole with order 1 for f (z) = Q(z)
and
P (z0 )
Res f (z) =
z=z0 Q0 (z0 )
Example 1.4.3. To find the residue of the function f (z) = z4z+4 at the isolated singu-
larity z = 1+i, we take P (z) = z and Q(z) = z 4 +4 as P (1+i) = 1+i 6= 0, Q(1+i) = 0
and Q0 (1 + i) = 8 + 8i 6= 0, then 1 + i is a zero of order 1 for Q and so a pole of order
1 for f . Hence,
P (1 + i) 1+i i
Res f (z) = 0 = =
z=1+i Q (1 + i) 8 + 8i 8
23
Definition 1.4.3. A function f is meromorphic on D if f (z) has only isolated singu-
larities in D, which are all poles.
f 0 (z)
Z
1
dz = N0 (f ) Np (f ) (1.10)
2i (C) f (z)
Proof. Since the zeroes and poles of f are isolated, we can deform (C) into a set
of small disjoint external circles (C ) centered at each zero or pole. It follows that it
is enough to prove the theorem for each circle separately and then add the results. In
such a circle (C ) centered at z0 we can write f (z) = (z z0 )m h(z), where m > 0 if z0
is a zero with order m and m < 0 if z0 is a pole with order m and where h(z0 ) 6= 0.
Thus, we have
f 0 (z)
Z Z
1 1 m
dz = +0=m
2i (C ) f (z) 2i (C ) z z0
24
Chapter 2
Conformal Mapping
And so if f (z) = z 2 the image of A(1,1) is the point A(0,2) and so A has coordinates
(1,1) in the z-plane and A has coordinates (0,2) in the w-plane.
A general linear transformation transforming a point with affix z into an image point
with affix w is given by w = Az + B, where A and B are nonzero complex constants.
We begin with the study of the transformation with complex form w = z + B.
25
(xV , yV ). The translation maps all figures in the z-plane into congruent figures in
the w-plane.
2. Consider the transformation defined by its complex form w = Az, with the con-
dition that A = rei 6= 0, so we can write w = aei z and it follows directly that
this transformation expands or contracts radius vectors with a scale |A| = r and
rotates them with angle = arg(A) about the origin. So this transformation maps
all figures in the z-plane onto similar figures extended/contracted and rotated in
the w-plane.
Proof. Let (~a, ~b) be an oriented angle in the z-plane and (a~0 , b~0 ) be its image in the
w-plane by a translation thus ~a = a~0 and ~b = b~0 and so (~a, ~b)= (a~0 , b~0 ) up to adding a
multiple of 2 and thus a translation preserves oriented angles, a similar argument but
now the mapping w = Az is considered where A = rei then:
z Az z
b~0 ~b ~b
(a~0 , b~0 ) = arg = arg = arg = (~a, ~b) + 2ki where k Z.
za~0 Az~a z~a
1
2.1.2 The transformation w = z
Consider for this section that all complex numbers involved are different than zero.
The mapping with complex form w = z1 will be described as the composite of two
transformations with respective complex forms w1 = z1 and w2 = z. Clearly, the first
transformation is an inversion with pole O and power 1, or simply an inversion with
respect to the unit circle, and the second transformation is the symmetry with respect
to the axis of abscissas. Noting that an inversion transforms lines into lines and cir-
cles and circles into lines and circles depending whether or not O belongs to the curve,
while the symmetry conserves geometric figures but not oriented angles, it will lead
to the result that the mapping w = z1 will transform circles into lines and circles and
circles into lines and circles. The mapping w = z1 is an involution, which means that
it is its own inverse mapping. Now, we study how this mapping maps circles and
26
straight lines. As an initiative, we need to find the analytic form of this mapping. Let
z = x + iy and w = u + iv. By comparing real parts and imaginary part we get that
x y
u = x2 +y 2 and v = x2 +y 2 . As this mapping is an involution we can directly conclude
u v
that x = u2 +v 2 and y = u2 +v 2 . The general equation of a line or a circle is given by the
2. The image of a circle (A 6= 0) which does not pass through the origin (D 6= 0) is a
circle with equation D(u2 + v 2 ) + Bu Cv + A = 0 which does not pass through
the origin.
4. The image of straight line (A = 0) which does not pass through the origin (D 6= 0)
is a circle with equation D(u2 + v 2 ) + Bu Cv = 0 which passes through the ori-
gin. We could have foreseen this result with the mapping w = z1 being its own
inverse, and from part (1) we know that circles passing through the origin will
be mapped onto straight lines not passing through origin and so straight lines
not passing though the origin should be mapped onto circles passing through the
origin. All the above curves are such that they are deprived of the origin in case
they pass through the origin.
27
interior of this strip; each point in the region 2 < x < 2 and y 0 lies on one of
the vertical semi line x = c1 with y 0 where c1 ] 2 , 2 [ and those semi lines will
be mapped onto u = sin c1 cosh y and v = cos c1 sinh y where y [0, +[, noting that
cosh2 y sinh2 y = 1 we get the the image of the lines are branches of the hyperbolas
on the upper side of the v-axis in the w-plane (as v = cos c1 sinh y 0) with equation:
u2 v2
=1
sin2 c1 cos2 c1
p
whose foci have at the points sin2 c1 + cos2 c1 = 1 in the w-plane; see Figure 2.1.
So we notice that the images of those semi lines are distinct and constitutes the entire
half plane v 0. More precisely if we are moving c1 from 0 to 2 say a point L then
L will be on a right branch of a hyperbolas which will be opening wider and its vertex
(sin c1 , 0) is moving towards the origin w = 0, a similar argument is made for a point M
c1 is moving from 2 to 0, and this achieves our aim.
3. T 1 () = when c = 0.
4. T 1 ( ac ) = if c 6= 0.
5. T 1 () = dc if c 6= 0.
28
Example 2.1.2. The linear fractional transformation with complex form:
z
(z) = with || < 1
z 1
maps the unit disk into itself, moving the origin z=0 into = . To illustrate this we
note the following relations:
which is negative whenever || < 1 and |z| < 1 which means that || < 1.
(1 z0 )(1 z0 ) = (1 z1 )(1 z1 ).
29
It is very clear to see the result geometrically. If z is above the x-axis, then the distance
joining z to z0 is smaller than the distance from z to z0 as z0 is situated above the
x-axis, and so for any point z above the x-axis, its image w will have |w| < 1 and thus
the region y > 0 will be mapped onto |w| < 1; while if a point z is below the x-axis, its
image w will have |w| > 1, and if a point z = k where k R is on the x-axis, we have
|k zo | = |k z0 | = |k z0 | and thus its image w will be on |w| = 1.
Example 2.1.3. [Bro-Chu, page 327-328]
1. The transformation w = iz
i+z
zi
can be written as w = ei zi
and thus maps the upper
half plane Im z > 0 onto the open disk |w| < 1 and the boundary Im z = 0 into the
boundary |w| = 1.
z1
2. Consider the transformation w = z+1 . Note that this transformation doesnt satisfy
the conditions for a linear fractional transformation that maps the upper half plane
Im z > 0 onto the open disk |w| < 1 and the boundary Im z = 0 into the boundary
|w| = 1 as, here, z0 = 1 with Im z0 = 0. Writing z = x + iy and w = u + iv gives
that this mapping maps the upper half plane y > 0 onto the upper half plane v > 0
2y
and the x-axis onto the u-axis. To make this clear, v = Im w= |z+1| 2 which means
w = Log Z, (2.5)
where z = x + iy, Z = X + iY and w = u + iv. We have seen (part 2 above) that
z1
the mapping Z = z+1 maps the upper region y > 0 into the upper Y > 0; now, we
check the image of the upper region by Log Z. We can write Z = Rei and
Log Z = ln R + i,
where pi < < and R > 0.
We see that as a point Z = Rei0 (0 < 0 < ) moves outward from the origin
along the ray = 0 , its image is the point whose rectangular coordinates in the
w-plane are (ln R, 0 ). That image evidently moves to the right along the entire
length of the horizontal line v = 0 . Since these lines fill the strip 0 < v <
as the choice of 0 varies between 0 and , the mapping of the half plane Y > 0
onto the strip is, in fact, one to one. This shows that the composition (2.3) of
the mappings (2.4-2.5) transforms the plane y > 0 onto the strip 0 < v < (see
Figure 2.2).
30
Figure 2.2: w = Log Z
Example 2.2.1. 1. The angle between the line (m) with equation x = 0 and the line
(n) with equation y = x at the origin is clearly 4 , now we check the angle of their
images under f (z) = z 2 in the w-plane. The image (m0 ) of (m) has an equation
v = 0 with u < 0 while the image (n0 ) of (n) has an equation u = 0 with v > 0,
meeting at the origin and thus making a right angle at the origin.
2. Consider the two curves (C1 ) and (C2 ) with respective equations y = x and y = x2
meeting at the point A(1, 1). An equation of the tangent to (C1 ) at A is y = x and
s1 s2
to (C2 ) is y = 2x , now the angle between the curves is such that = | 1+s 1 s2
|,
where s1 and s2 are the respective slopes of the tangents to (C1 ) and (C2 ). We
have tan = 13 and = tan1 ( 13 ). In general we do not need to find the equation
dy
of the tangent as all we need is to find dx at the point A which represents the slope
of the tangent.
3. Consider the mapping f (z) = z 2 from the z-plane onto the w-plane and lets see
the angle between the image of the vertical line (d) with equation x = 1 and the
image of the horizontal line (L), y = 2 in the w-plane at the image B of the point
A(1, 2). First, writing z = x + iy, we get that w = u + iv = x2 y 2 + 2xyi and
so u = x2 y 2 and v = 2xy. The image of (d) is found by letting x = 1 and so
u = 1 y 2 and v = 2y and so the image of (d) is the parabola (P ) in the w-plane
with equation v 2 = 4u + 4. Similarly for (L), we get its image in the w-plane
which is the parabola (Q) with equation v 2 = 16u + 64 meeting at B(3, 4). Clearly
dv
(d) and (L) meet. The slope of tangent to (P ) at B in the w-plane is du and as we
have the relation between u and v given implicitly, we use implicit differentiation
dv
to get 2v du dv
= 4, meaning that du = 2
v
= 21 at the point B. Similarly for (Q),
dv 8
we get du = v = 2 at the point B, and it is clear that they meet at a right angle
as 12 2 = 1.
31
If two curves (C1 ) and (C2 ) meet at a point A making an angle then it is not
necessary that the images of those curve (meeting at the image of A ) under a mapping
f meet making the same angle as we saw in the previous examples. This motivates us
to define the mappings which preserve angles.
Definition 2.2.2. A mapping f : U V is conformal (or angle preserving) at a point
z0 U if it preserves the oriented angles between all curves through z0 . A mapping
f : U V is said to be conformal over U if it is conformal at every point z in U.
A more useful definition for conformal mapping is as follows:
Definition 2.2.3. [Olv, page 2.] Let f : U R2 defined by f (x, y) = u(x, y)+iv(x, y),
be a a mapping with continuous partial derivatives and let (x0 , y0 ) be any point in U ,
then f is said to be conformal at (x0 , y0 ) if r R+ and [0, 2[ such that for any
differentiable curve : [a, b] U with (0) = (x0 , y0 ) we have:
This means that when any differentiable curve through (x0 , y0 ) is mapped by f, its
tangent vector at (x0 , y0 ) is stretched by a factor r and rotated by an angle where r
and are independent of the curve. This means that oriented angles will be preserved.
In complex terms the word conformality is nothing but differentiability as we will
see in the next proposition.
Proposition 2.2.1. [Bro-Chu] Let f : U V where U, V C be a complex function
with continuous partial derivatives and let z0 U , then f is conformal at z0 if and only
if f is analytic at z0 and f 0 (z0 ) 6= 0.
Proof. Suppose that a given function f is analytic at z0 with f 0 (z0 ) 6= 0. Let
r = |f 0 (z0 )| , and let = arg(f 0 (z0 )) where [0, 2[. If is any differentiable curve
with (0) = z0 and using the chain rule we get
32
Similarly, taking the curve (t) = z0 + it then deriving with respect to y we get: (f
)0 (0) = uy (x0 , y0 )+ivy (x0 , y0 ) = fy (x0 , y0 ). Noting that 0 (0) = i 0 (0) we get | 0 (0)| =
| 0 (0)| and arg( 0 (0)) = 2 + arg( 0 (0)). As f is conformal at z0 , then r > 0 and
[0, 2[ such that |(f )0 (0)| = r|0 (0)| and arg(f )0 (0) = + arg(0 (0) for any
differentiable curve starting at z0 , and so we have:
|fx (x0 , y0 )| = |(f )0 (0)| = r| 0 (0)| = r| 0 (0)| = |(f )0 (0)| = |fy (x0 , y0 )|
arg fx (x0 , y0 ) = arg (f )0 (0)
0
= + arg (0)
= + arg 0 (0)
2
0
= arg (f ) (0)
2
= arg fy (x0 , y0 )
2
This means that fx (x0 , y0 ) = ify (x0 , y0 ), and so:
Taking the real and imaginary parts we get that the Cauchy-Riemann equations are
verified and thus f is analytic at z0 with f 0 (z0 ) 6= 0 as a direct conclusion from the
existence of the scale factor r > 0 as f being conformal.
33
3. The Joukowski map [Olv] was first deployed to study flows around airplane
wings by the pioneering Russian aero and hydro-dynamics researcher Nikolai
Zhukovski (Joukowski). Joukowswis map is defined as follows:
1 1
(z) = (z + ).
2 z
d
Since dz = 12 (1 z12 ) = 0 if and only if z = 1, then the Joukowski map is con-
formal at all points except z = 1, as well as the singularity z = 0 where it is
not defined. Consider z = rei . It follows that (z) = 12 r(ei + ei ) and so when
z = ei is on the unit circle centered at the origin then (z) = cos and thus this
mapping maps the unit circle onto the interval [1; 1] on the real axis, and so
the points inside and outside the unit circle will be mapped to fill the rest of the
plane. Finding the inverse map is determinedpby finding z(), and a simple
calculation leads to the following result: z = 2 1, this means that every
6= 1 comes from two different points z, if / [1, 1] then came from two
points p p
z1 = + 2 1 and z2 = 2 1,
where z1 is clearly outside the unit circle and z2 is inside the unit circle whereas
if ] 1, 1[ we saw that both points z1 and z2 lie on the unit circle with same
abscissa in the z-plane.
Defining the Joukowski map over the region |z| > 1 defines a one to one conformal
mapping onto the exterior of segment [-1,1] in the -plane C\[1, 1]. Under the
Joukowski map, the concentric circles |z| = r 6= 1 are mapped into ellipses with
foci at 1 in the -plane, the effect of this mapping on circles not centered at the
origin can take wide variety of shapes depending whether the circle passes through
the singular z = 1, and the image of such circles are not smooth anymore but has
a cusp at = 1.
34
Figure 2.4: Airfoils Obtain Under the Joukowski map [Olv])
Now, let be a smooth curve on S2 through p, and let t be its tangent at p and
t be the intersection of the plane containing n and t with R2 . Clearly t is the
35
tangent to at (p), since and t are curves in R3 with the same tangent t
at p, it follows that and t = t are curves in R2 with the same tangent
at (p). Since t is its own tangent at (p), it is also tangent to there. The
lengths b are equal and so finally the angles are also equal. Similarly, taking a
second curve through p completes the proof (see Figure 2.6).
Theorem 2.3.1 (Rouches theorem). For any two complex-valued functions f and g
analytic inside some region K with closed contour K, if |g(z)| < |f (z)| on K, then f
and f + g have the same number of zeros inside K, where each zero is counted as many
times as its multiplicity.
f 0 (z) + tg 0 (z)
Z
1
(t) = dz.
2i (K) f (z) + tg(z)
As,
| |f (t)| t|g(t)| | | |f (t)| |g(t)| | > 0
36
thus the denominator in the integrand is never zero. Noting that is continuous in the
interval [0,1] and is integer valued. Clearly from the argument principle (see Theorem
1.4.3), (t) is the number of zeroes of the function f + tg inside (K). Being continuous
and integer valued on the connected set [0, 1] then it must be constant. In particular
1
R f 0 (z)
(0) = (1). But (0) = 2i (K) f (z)
is the number of zeroes of f as f is analytic and
1 f 0 (z)+g 0 (z)
(1) = 2i f (z)+g(z)
is the number of zeroes of f + g, and thus the result follows.
Lemma 2.3.1. Suppose that a non-constant function f from D to D0 is analytic, and
let z0 D such that f 0 (z0 ) = 0, then f cannot be a one-to-one mapping in any disk
containing z0 .
Proof. Let w0 = f (z0 ). Since f (z) is not constant then the function g(z) = f (z) w0
is not an identically zero function. Since f 0 (z0 ) = 0, then g(z) has a zero of finite order
at least 2 at z0 . Since zeroes of analytic functions are isolated, then we can find r > 0
such that g(z) and f 0 (z) have no zeroes in the punctured disk {0 < |z z0 | r}.
Then m = min |g(z)| for z A, where A = {z : |z z0 | = r}. Let w C such that
0 < |w w0 | < m, then it follows from Rouches theorem that the functions g(z)
and g(z) + (w0 w) have the same number of zeroes inside A and so the function
g(z) + (w0 w) = f (z) w has at least two zeroes inside A none of them is z0 as
w 6= w0 . Since f 0 (z) 6= 0 inside the punctured disk, it follows that the zeroes must be
simple and distinct, and so f cannot be one-to-one inside any disk containing z0 .
Corollary 2.3.1. Suppose that a function f is analytic and one-to-one in a domain D,
then f is conformal at every point in D.
Proof. Let z0 D, since f is one-to-one and analytic in D, we have by Lemma 2.3.1
that f 0 (z0 ) 6= 0 and thus f is conformal at z0 and thus in D.
Lemma 2.3.2. Suppose that D and D0 are domains, and let f be an analytic one-to-one
and onto mapping, then the inverse mapping f 1 from D0 to D is analytic in D.
Proof. The proof just follows from the fact that
z z0 w w0 1
lim = lim 6= 0
w w0 w w0 z z0 z z0
This limits being not zero is a consequence from f being one-to-one and onto which
guarantees f 0 (z) 6= 0 by Corollary 2.3.1.
Remark 2.3.1. If f is analytic and one-to-one on a domain D then by applying the
Cauchy-Riemann equations on the inverse map f 1 we have:
x y
= , (2.6)
u v
x y
= . (2.7)
v u
Moreover, we have
2x 2x 2y 2y
+ =0 and + =0 (2.8)
u2 v 2 u2 v 2
37
Theorem 2.3.2. Suppose that D and D0 C are domains, and let f be an analytic
one-to-one and onto mapping from D to D0 . Suppose further that for every z in D, we
write w = f (z), where z = x + iy and w = u + iv, with x, y, u, v R. Then a function
(x, y) is harmonic in D, if and only if the function (u, v)is harmonic in D0 , where
(u, v) = x(u, v)), y(u, v) .
Proof. Note that (u, v) = x(u, v)), y(u, v) can be written as (w) = f 1 (w)
and that the function f 1 is analytic on D0 . Using the chain rule we get:
x y
= +
u x u y u
Now,
2 x y 2 x x 2 y y
= + = + + 2 + .
u2 u x u y u u2 x u u x u y u u y
On the other hand, we have:
x y x 2 y 2
= + = + .
u x u x x u y x u x u xy
Similarly we get:
x 2 y 2
= + .
u y u xy u y 2
Remark here that order partial differentiating x or y does not matter due to the conti-
nuity of the functions. Combining all terms to get:
2 2 x 2 y x 2 2 x y 2 y 2 2
= + + +2 + (2.9)
u2 u2 x u2 y u x2 u u xy u y 2
By symmetry of differentiation and in a similar manner we get:
2 2 x 2 y x 2 2 x y 2 y 2 2
= 2 + + +2 + (2.10)
v 2 v x v 2 y v x2 v v xy v y 2
Now adding Equations 2.9 and 2.10 to get
2 x 2 x 2 y 2 y
4 = + + +
x u2 v 2 y u2 v 2
x y x y 2
+2 +
u u v v xy
x
2 x 2 2 y 2 y 2 2
+ + ) + + ( (2.11)
u v x2 u v y 2
Using (2.8), (2.6) and (2.7), we get:
4 = |f 0 (z)|2 4 (2.12)
38
Remark 2.3.2. This clearly shows that harmonic functions are conformally mapped
into harmonic functions, the importance of the previous theorem is that when solving
a problem, among other things when we change variables we sometimes dont want to
change the physical significance of this problem, we may be solving examples in the xy-
plane and for convenience we map the problem into the uv-plane. It may happen that
certain physical properties are present in the xy-plane (potential or force) or maybe one
family of lines intersect another family of lines at right angles and we want a mapping
into the uv plane where images of these families will meet at right angles.
39
Chapter 3
Conformal mappings can be effectively used for constructing solutions to the Laplace
equation on complicated planar domains that appear in a wide range of physical prob-
lems, including fluid mechanics, aerodynamics, thermomechanics, electrostatics, and
elasticity. In this chapter, we will develop the basic techniques and theorems of complex
analysis that impinge on the solution to boundary value problems associated with the
planar Laplace equations and will expose three applications of conformal mappings
[Saf-Sni, Rud, Hen, Ahl].
40
Consider a planar steady state fluid flow, with velocity vector field:
where (x, y) C where represents the domain occupied by the liquid, while the
vector V(x, y) represents the instantaneous velocity of the fluid at (x, y). Recall that
the flow is incompressible if and only if it has vanishing divergence, that is:
u v
OV = + = 0.
x y
Incompressibility means that the fluid volume does not change as it flows. Most liquids,
including water, are, for all practical purposes, incompressible. On the other hand, the
flow is irrotational if and only if it has vanishing curl, that is:
v u
OV = =0
x y
Irrotational flows have no vorticity, and hence no circulation. A flow that is both
incompressible and irrotational is known as an ideal fluid flow. In many physical regimes,
liquids (and, although less often, gases) behave as ideal fluids. The velocity vector field
V(x, y) induces an ideal fluid flow if and only if f (z) = u(x, y) iv(x, y) is analytic. The
corresponding complex function f is called the complex velocity of the fluid flow. Under
the flow induced by the velocity vector field V = (u(x, y), v(x, y))T , the fluid particles
follow the trajectories z(t) = x(t) + iy(t) obtained by integrating the system of ordinary
differential equations:
dx dy
= u(x, y) and = v(x, y).
dt dt
Now, suppose that the complex velocity f (z) admits a complex primitive
Then:
d
= i = u iv
dz x y
Thus,O = V, and hence the real part (x, y) of the complex function (z) defines a
velocity potential for the fluid flow. For this reason, the anti-derivative (z) is known as a
complex potential function for the given fluid velocity field. Since the complex potential
is analytic, its real part, the potential function, is harmonic, and therefore satisfies the
Laplace equation 4 = 0. Conversely, any harmonic function can be viewed as the
potential function for some fluid flow. The real fluid velocity is its gradient: V = O,
and is incompressible and irrotational. The harmonic conjugate (x, y) to the velocity
potential also plays an important role, and, in fluid mechanics, is known as the stream
function. It also satisfies the Laplace equation 4 = 0, and the potential and stream
function are related by the Cauchy- Riemann equations:
=u= and =v=
x y y x
41
Example 3.1.1. Consider a complex potential function (z) = z = x + iy, the velocity
potential is (x, y) = x and the steam function is (x, y) = y. The complex derivative
of the potential is the complex velocity
d
f (z) = =1
dz
Let () = (, ) + i(, ) be an analytic function representing the complex po-
tential function for a steady state fluid flow in a planar domain D. Composing the
complex potential () with a one to one conformal map = g(z) leads to a trans-
formed complex potential (z) = (g(z)) = (x, y) + i(x, y) on the corresponding do-
main = g 1 (D), thus we can use conformal maps to construct fluid flows in compli-
cated domains from known flows in much simpler domains. We will restrict our study
to planar fluid flows around a closed and bounded subset D R2 = C, representing a
cross-section of a cylindrical object see, Figure 3.1.
The (complex) velocity and potential are defined on the complementary domain
= C\D occupied by the fluid. The velocity potential (x, y) will satisfy the Laplace
equation 4 = 0 in the exterior domain . For a solid object, we should impose the
homogeneous Neumann boundary conditions
=0 (3.1)
n
on the boundary = D indicating that there is no fluid flux into the object. We
note that, since it preserves angles and hence the normal direction to the boundary,
a conformal map will automatically preserve the Neumann boundary conditions. We
shall assume our object is placed in a uniform horizontal flow, e.g., a wind tunnel, as
sketched in Figure 3.2.
Thus, far away, the object will not affect the flow, and so the velocity should ap-
proximate the uniform velocity field V = (1, 0)T , where, for simplicity, we choose our
physical units so that the fluid moves from left to right with an asymptotic speed equal
to 1. Equivalently, the velocity potential should satisfy:
42
Figure 3.2: Flow pass a solid object [Olv]
Think of an airplane flying through the air at constant speed. If we adopt a moving
coordinate system by sitting inside the airplane, then the effect is as if the plane is
sitting still while the air is moving towards us at unit speed.
Note that O = (1, 0)T , and hence this flow satisfies the Neumann boundary conditions
3.1 on the horizontal segment D. The corresponding complex potential is (z) = z
with complex velocity 0 (z) = f (z) = 1
43
relative to the flow. The key observation is that, while the effect of rotating a plate in
a fluid flow is not so evident, rotating a circularly symmetric disk has no effect on the
flow around it. Thus the rotation w = eiz maps the disk potential:
1 x y
(z) = z + = (x + 2 2
) + i(y 2 )
z x +y x + y2
to the complex potential:
ei
(w) = (ri w) = ei w + .
w
The streamlines of the induced flow are no longer asymptotically horizontal, but rather
at an angle , If we now apply the original Joukowski map to the rotated flow, the circle
is again squashed down to the horizontal line segment, but the stream lines continue to
be at angle at large distances. Thus, if we then rotate the resulting flow through an
angle -, the net effect will be to tilt the segment to the desired angle while rotating
the streamlines to be asymptotically horizontal giving the result:
p
(z) = ei (z cos i sin z 2 e2i )
3.1.5 Airfoils
We have seen earlier that Joukowski map to off-center disks will, in favorable con-
figurations, produce airfoil-shaped objects. The fluid motion around such airfoils can
thus be obtained from the flow past such an off-center circle. Note that the affine
transformation w = z + transforms the disk |z| 1 onto the disk |w | ||. In
particular, the boundary circle will continue to pass through the point w = 1 provided
that || = |1 |, now apply the Joukowski transformation to get:
1 1
= z + +
2 z +
The resulting complex potential for the flow past the airfoil is obtained by substituting
the inverse map: p
w + 2 1
z= =
into the disk potential whereby:
p p
+ 2 1 ( 2 1)
() = +
2 + 1 2
Finally, if the airfoils position is with an attack angle with the horizontal line, we
replace by ei to get:
i
p i
p
e + (ei )2 1 (e (ei )2 1)
(ei ) = +
2 + 1 2ei
44
3.2 Application 2: A non-coaxial cable
The goal of this section is to determine the electrostatic potential u inside a non-coaxial
cylindrical cable with prescribed constant potential values on the two bounding cylin-
ders, see Figure 3.3, where u(x, y) is the potential function (decribed in the previous
section).
Suppose that the larger cylinder has a radius of 1 and is centered at the origin,
while the smaller cylinder has a radius of 52 and centered at z = 25 . The resulting
electrostatic potential u(x, y) will be independent of the longitudinal coordinate, and
so can be viewed as a planar potential in the annular domain contained between the
two circles representing the cross section of the cylinders = {|z| < 1 and |z 25 | > 25 }.
The desired potential must satisfy the Dirichlet boundary value problem:
4u = 0 on
u = b on |z| = 1 (3.2)
2 2
u = a on |z 5 | = 5
45
concentric circles, the larger circle with radius 1 and the smaller circle with radius 25 ,
with center z = 25 , the mapping takes those non-concentric circles into concentric an-
nulus of the form A = {r < || < 1}, to see this note that we proved earlier that linear
z
transformations with form = z1 with || < 1 map the unit disk into itself, moreover,
linear fractional transformations map circles into circles, so if we seek a value of to
make the inner circles |z 25 | = 25 a circle of the form || = r which is centered at the
origin, choosing to be real and trying to map the points 0 and 54 on the inner circle
to the points r and -r on the circle || = r, we get:
1. (0) = = r
4
2. ( 45 ) = 4
5
1
= r
5
U = a on || = 21
(3.3)
U = b on || = 1
It is clear that coaxial potential energy must be a symmetric solution of Laplace equation
and thus having the form:
U (, ) = log || +
Replacing the initial conditions, we find that:
ba
= log 2 (3.4)
=b
And so:
ba ba
U (, ) =log || + b = log( 2 + 2 ) + b (3.5)
log 2 2 log 2
Equation 3.5 represents the conformal map of the potential of the concentric (see Figure
3.5), then by Theorem 2.3.2 we can find the expression of the potential of the non-coaxial
cable u = U to get:
2 2
ba 2z 1
+ b = b a log (2x 1) + y + b
u(x, y) = log
log 2 z2 2 log 2 (x 2)2 + y 2
Remark 3.2.1. The same harmonic function determines the equilibrium temperature
of an annular plate whose inner boundary is kept at temperature u = a while the outer
boundary is kept ar temperature u = b
46
Figure 3.5: Electrostatic potential between coaxial and non-coaxial cylinders [Olv]
3.3.1 Background
In the theory of heat conduction, the flux across a surface within a solid body at a point
on that surface is the quantity of heat flowing in a specified direction normal to the
surface per unit time per unit area at the point. Flux is, therefore, measured in such
units as calories per second per square centimeter. It is denoted here by , and it varies
with the normal derivative of the temperature T at the point on the surface:
dT
= K (K > 0). (3.6)
dN
Relation 3.6 is known as F ourier0 s law and the constant K is called the thermal
conductivity of the material of the solid, which is assumed to be homogeneous. The
points in the solid are assigned rectangular coordinates in three-dimensional space, and
we restrict our attention to those cases in which the temperature T varies with only
the x and y coordinates. Since T does not vary with the coordinate along the axis
perpendicular to the xy-plane, the flow of heat is, then, two-dimensional and parallel to
that plane. We agree, moreover, that the flow is in a steady state; that is, T does not
vary with time. It is assumed that no thermal energy is created or destroyed within the
solid. That is, no heat sources or sinks are present there. Also, the temperature function
T (x, y) and its partial derivatives of the first and second order are continuous at each
point interior to the solid. This statement and 3.6 for the flux of heat are postulates
in the mathematical theory of heat conduction, postulates that also apply at points
within a solid containing a continuous distribution of sources or sinks. The temperature
47
function satisfies Laplaces equation at each interior point of the solid so
Txx + Tyy = 0. (3.7)
And noting the continuity of T leads it to be a harmonic function of x and y in the
domain representing the interior of the solid body. The surfaces T (x, y) = c1 , where c1
is any real constant, are the isotherms within the solid. They can also be considered as
curves in the xy-plane; then T (x, y) can be interpreted as the temperature at a point
(x, y) in a thin sheet of material in that plane, with the faces of the sheet thermally
insulated. The isotherms are the level curves of the function T . The gradient of T is
perpendicular to the isotherm at each point, and the maximum flux at a point is in the
direction of the gradient there. If T (x, y) denotes temperatures in a thin sheet and if S
is a harmonic conjugate of the function T , then a curve S(x, y) = c2 has the gradient
of T as a tangent vector at each point where the analytic function T (x, y) + iS(x, y) is
dT
conformal. The curves S(x, y) = c2 are called lines of flow. If the normal derivative dN
is zero along any part of the boundary of the sheet, then the flux of heat across that
part is zero. That is, the part is thermally insulated and is, therefore, a line of flow.
48
r1
Figure 3.6: w = Log z1
z+1 r2
> 0, 2 < 1 2 < 3
2
[Olv]
The isotherms T (x, y) = c1 (0 < c1 < 1) are arcs of the circles with equation:
note that this result follows directly from re-writing T (x, y) = c1 in its canonical form.
These circles pass through (1, 0) and with centers on the y-axis.
1.
Txx + Tyy = 0, < x < ,y > 0
2 2
2.
T , y = T , y = 0, y > 0
2 2
3.
T (x, 0) = 1, <x<
2 2
where T (x, y) is bounded. The map needed in this case is w = u + iv = sin z which
transforms this boundary value problem into the same one as the previous application
and thus:
1 2v
T = arctan 2 (3.9)
u + v2 1
49
The change of variable in z = u + iv = sin z = sin(x + iy) leads to u = sin x cosh y and
v = cos x sinh y and thus the steady temperatures are given by the equation:
1 2 cos x sinh y
T = arctan .
sin2 x cosh2 y + cos2 x sinh2 y 1
Simplifying the desired function to get T (x, y) in its simplest form as:
2 cos x
T = arctan
sinh y
Remark 3.3.2. If the temperature on the third surface is T0 instead of 1 then the steady
temperature is given by the function:
2T0 cos x
T = arctan
sinh y
50
the existence of an inverse is ensured by the fact that the given transformation is both
one to one and onto. Since transformation w = sin z is conformal throughout the strip
except at the point w = 2 , the inverse transformation must be conformal throughout
the quadrant except at the point z = 1. This inverse transformation maps the segment
0 < x < 1 of the x-axis onto the base of the strip and the rest of the boundary onto
the sides of the strip. The required temperature function T for the new boundary value
problem is clearly: T = 2 u. Now to write T in terms of x and y, we use u = sin x cosh y
and v = cos x sinh y to get when 0 < u < 2 :
x2 y2
=1 (3.11)
sin2 u cos2 u
So for a fixed u the hyperbola 3.11 has foci the points z = 1, and the length of the
major axis, which is the line segment joining the two vertices, is 2 sin u. The required
temperature function is:
51
Bibliography
[Bak-New] J. Bak and D. J. Newman, Complex analysis, third edition, Springer, 2010.
[Blair] D. Blair, Inversion theory and conformal mapping, American Mathematical So-
ciety, 2000.
[Dit-Geo] P. Dita and V. Georgescu, Conformal invariance and string theory, Academic
Press, Inc., Boston, MA (USA), 1989, 570, ISBN 0-12-218100-X.
[Gre-Kra] R. Greene and S. Krantz, Function theory of one complex variable, third edi-
tion, Graduate studies in Mathematics, volume 40, American Mathematical Society,
2006.
[Hen] P. Henrici, Applied and computational complex analysis, vol. 1, J. Wiley and Sons,
New York, 1974.
[Neh] Z. Nehari, Conformal mapping, Dover publications, INC. New York, 1952.
[Olv] P. Olver, Complex analysis and conformal mapping, University of Minnesota, 2013.
[Rud] W. Rudin, Real and complex analysis, third edition, McGraw-Hill, New York,
1987.
52
[Sch-Lau] R. Schinzinger and P. AA Laura, Conformal mapping: methods and applica-
tions, Courier Corporation, 2003.
[Sie-Sny] R. Siegel and A. Snyder, Heat transfer in cooled porous region with curved
boundary, Journal of Heat Transfer 103 (4), 765-771, 1981.
[Thd] Th. Theodorsen, Theory of wing sections of arbitrary shape, National Advisory
Committee for Aeronautics. Langley Aeronautical Lab. Langley Field, VA, United
States, 1993.
[Wec-Zim] H. Wechsler and G. Lee Zimmerman, 2-D invariant object recognition using
distributed associative memory, IEEE Transactions on Pattern Analysis and Machine
Intelligence, Vol.10, Issue 6, 1988.
53