Chapter 4

Special Functions

4.1 Legendre Functions

Legendre’s function (polynomial) nm (  ) is a solution of Legendre’s

differential equation (1.72). The subscript n denotes the degree and the
subscript m the order of nm . For the solution to Laplace’s equation in
spherical polar coordinates,   cos  , so that nm ( )  nm (cos ) .

The simplest relation which defines the Legendre polynomial of the

Associated Legendre Function expressed in (1.72) is given as,

d nm
nm (  ) 
1
n
2 n!

1 2 
m/2

d nm

2 1
n
 (4. 1)

The special case of nm (  ) where m0 gives rise to the Legendre
polynomial n 0 (  ) to be determined is called Rodrigues’ formula. It is
simply denoted by n (  ) .

1 dn
n (  )  n 0 (  )  n
2 n! d n
2 1 
n
 (4. 2)

Because m0, there is no square root, that is, no sin  . Therefore, the n (  )

are simply polynomials in . The first few Legender polynomials are

1
 0 (cos  )  1
(4. 3)
1 (cos  )  cos 

The polynomials can be obtained by means of (4.2) or more simply by

the recursion formula
n 1 2n  1
n (  )   n  2 ( )   n 1 ( ) (4. 4)
n n

by which 2 can be calculated from 0 and 1; 3 from 1 and 2 , etc.

Applications: one of the fundamental applications of Legendre

polynomial is its use to expand an inverse distance into functions of
Legendre’s polynomials
n
1 1   r1 
    n (cos ) (4. 5)
r1  r2 r2 n  0  r2 

Where the cosine value of the geocentric angle  between the two field
points r1 and r2 is given by

r1  r2
cos  and r2  r1 (4. 6)
r1 r2

  cos1 cos 2 sin 1 sin 2 cos2  1  (4. 7)

Note that in practice the Legendre’s polynomials are expanded in terms

of latitudes and longitudes of the running point r1 and computation point
r2 .

2
4.2 Dirac Delta Function

One of the most frequently encountered functions is the Dirac delta

function  (t ) . This function is defined to be zero if t0 and to be infinite
at t 0 in such a way that



  (t )dt  1 (4. 8)

This is a concise manner of expressing a function that is very large

in a very small region and zero outside this region and has a unit integral.
Let us consider the function  (t ) that has the following values,

0 t0
1

 (t )   0t  (4. 9)

0 t 0

where  may be made as small as we please. This function possesses the

property (4.8).

For further understanding we now prove the following,




 F (t ) (t  a)dt  F (a) (4. 10)

Provided that F (t ) is continuous. So that we have

 a 
1


F (t ) (t  a)dt 
 
a
F (t )dt
(4. 11)
 F (a   ) where 0    1

Now, since by hypothesis F (t ) is continuous, we have

3
 lim F (a   )  F (a) (4. 12)
 0

The form of Dirac delta function in different coordinate systems,

Cartesian coordinate system:

 
 ( x  x)   ( x  x) ( y  y) ( z  z) (4. 13)
 
where  ( x  x)  0 for x, y, z   ( x, y, z) and  for x, y, z   ( x, y, z)

Having the definition of Dirac delta function in Cartesian coordinates, one

can find the form of delta function in any other coordinate system using
the following procedure.
If set of coordinates u is related to Cartesian coordinates via
ui  ui x1, x2 ,..., xn  i  1,..., n (4. 14)

then  ( x  x)  1  
   (u  u)
det xi / u j 

In spherical coordinate:
x  r sin  cos  , y  r sin  sin  , z  r cos (4. 15)
det xi / u j   r 2 sin 
 
So the determinant, , hence the Dirac delta function
has the following form,
  1
 ( x  x)   (r  r) (   ) (   ) (4. 16)
r sin 
2

4.3 Functions of complex variable theory

Differentiation, integration and solution of differential equation, etc., are

usually an operation on functions of continuous variables. When we
carryout actual computation for such operations in a computer or other

4
digital equipment we must first approximately replace the continuous
operand with finite and discrete numerical parameters. The results of
numerical computation give information about the consequence of
differentiation, integration, etc. from the result of computation (which
again consists of finite number of parameters).
For such finite number of parameters we use, in many cases, function
values (sample values) at discrete values of the independent variable
(sample points). Sometimes we also use coefficients of an expansion in
terms of some suitable set of functions (e.g., power series or the Fourier
series). However, in order to express a function in sufficiently high
precision with a relatively small number of parameters it is required that
the function is moderately smooth (or natural).
Although in many books the requirement on the smoothness is not so
explicitly mentioned, it is implicitly included in the error representation
of the numerical formula. In fact, in many cases the error representation
of numerical formulas includes the maximum absolute value of the nth
order derivative of the function in question (n is determined depending on
the formula), and hence it is implicitly assumed that the function is n-
times differentiable and that its n-th order derivative is not very large.

In actual applications, it is usually assumed implicitly that the function

under consideration is infinite times differentiable, or more strongly, is
analytic in a region which includes the real axis. In addition, there are

5
many cases in which the analyticity can actually be proved in some
explicitly given region. This is obviously the case if the function is defined
explicitly in an analytical form.
Also, to investigate what kind of analyticity the function has, or in
other words, where and what kind of singularities exist in the complex
plane, is an important subject in the field of geomatics, geophysics and
engineering. In general, it is more realistic to study in what region the
function is regular rather than to establish the n times differentiability. If
we base our argument on the analyticity of the function in this way, the
complex function theory, in particular Cauchy’s integral formula will
play a fundamental role.

Complex Algebra

A complex number z is defined as z = (x,y) = x + iy, where i  1 . The

ordering of two real numbers (x,y) is significant, i.e. in general x + iy  y
+ ix. Where x is the real part and it is labeled by Re(z); y is the imaginary
part and it is denoted by Im(z).

Representation of complex number in Cartesian and polar

coordinate system:

(1) Cartesian representation: z  x  iy

(2) polar representation: the complex number z may be write as

z  r cos  i sin   or z  rei


r  z  x2  y 2  - the modulus or magnitude of z
1/ 2

6
  tan1 ( y / x) - the argument or phase of z

Fig. 4.1

In mathematical manipulations of complex numbers or functions the

choice of the coordinate system has to be based on its convenience. For
instance, addition and subtraction of complex variables are easier in the
Cartesian representation while multiplication, division, powers and roots
are easier to handle in polar form,

z1  z2  x1  x2   i  y1  y2 
(4. 17)
z1 z2  x1 x2  y1 y2   i x1 y2  x2 y2 

z1 z2  r1r2 ei 1  2 
z n  r n ei n (4. 18)
z1 / z2  r1 / r2 ei 1  2 

The operation of replacing i by i in z  x  iy or z  rei is called complex

conjugation. This is denoted by,

7
 z*  x  iy (4. 19)

z* z  x 2  y 2  z
2
Note that

Functions of complex variable

All the elementary functions of real variables may be extended into the
complex plane by replacing the real variable by the complex variable z.
f ( x)  f ( z )
ex  ez
(4. 20)
sin x  sin z
ln x  ln z

Note that

eiz  e iz eiz  e iz

sin z  , cos z 
2i 2
z
(4. 21)
e e
z
e  e z
z
sinh z  , cosh z 
2 2

If to each value which a complex variable z can assume if there exists their
corresponding one or more values of a complex variable w, we say that w
is a function of z and it may be written as

w  f (z ) (4. 22)

If w(z) is a function of complex variables, i.e

w  f ( z)  u( x, y)  i v( x, y) (4. 23)

in which u and v are real functions

For example, if f ( z)  z 2 , we have  

f ( z)  x 2  y 2  i 2 xy . The relationship
between z and f(z) best pictured as a mapping operation.
8
 Mapping operation: the function w( x, y)  u( x, y)  i v( x, y) maps points in
the xy plane into points in the uv plane.

v
y Z-plane

u
x

Fig. 4.2

Exercise: write sin( x  iy ) and cos( x  iy ) in the form of u  iv

Cauchy – Riemann conditions

Having established complex functions, we now proceed to differentiate

them. The derivative of f(z), like that of a real function, is defined by
f ( z  z )  f ( z ) f ( z ) df
lim  lim   f ( z ) (4. 24)
z 0 z z 0 z dz

provided that the limit is independent of the particular approach to the

point z. For real variable, we require that lim f ( x)  lim f ( x)  f ( x0 ) . Now,
x  x0  x  x0

with z (or z0 ) some point in a plane, our requirement that the limit be
independent of the direction of approach is very restrictive.

Consider

9
 z  x  iy
f  u  iv (4. 25)
f u  iv

z x  iy

Let us take limit by the two different approaches as in the figure. First,
with y = 0, we let x0,

f  u v 
lim  lim   i 
 z  0 z  x  0 x x 
 (4. 26)
u v
 i
x x

Fig.4.3

Assuming the partial derivatives exist. For a second approach, we set x

= 0 and then let y0. This leads to

f  u v 
lim  lim   i  
 z  0 z  y 0
 y y  (4. 27)
u v
 i 
y y

If the derivative of the complex function f(z) exists, the real and imaginary
parts of the derivatives taken in the horizontal and vertical directions must
be equal. That is equations (4.26) and (4.27) must be identical.
10
 u v
 (4. 28)
y x

u v
 (4. 29)
x y

Such a function f(z) is said to be analytic at the point z. Equations (4.28)

and (4.29) are called the Cauchy-Riemann differential equations. These
Cauchy-Riemann conditions are necessary for the existence of a
derivative, that is, if f (z ) exists, the Cauchy-Riemann conditions must
hold. Conversely, it is easily verified that, if the real and imaginary parts
have continuous first partial derivatives satisfying Cauchy-Riemann
differential equations, then, the function is analytic.

Exercise: show that,

1
1) f ( z)  is analytic
z

f ( z )  z  zz*
2
2) is nowhere analytic

3) Show whether f ( z)  e z is analytic function or not

Harmonic Functions

By differentiating the Cauchy-Riemann differential equations and adding

the results we obtain two-dimensional Laplace equations for the real and
imaginary parts an analytic function w( z)  u( x, y)  i v( x, y) . If we differentiate

equation (4.28) with respect to y and equation (4.29) with respect to x and
add the results, we obtain

11
  2u  2 u
 0 (4. 30)
x 2 y 2

Differentiating (4.29) with respect to y and (4.28) with respect to x and

subtracting the results, we obtain

 2v  2v
 0 (4. 31)
x 2 y 2

Thus we see that the real and imaginary parts of an analytic function w(z)
of a complex variable are solutions of the two-dimensional Laplace’s
equation. Equally, w(z) itself is a solution to Laplace’s equation. These
are commonly used in potential field geodesy, geophysics and fluid
dynamic.

The functions u and v that satisfy the Cauchy-Riemann equations

(4.28) and (4.29) are called conjugate functions. If one of these functions
is known, the other may be found to within an arbitrary constant by
integrating the Cauchy-Riemann equations. For instance, if u is given and
it is desired to find the conjugate function v.
u v
From the Cauchy-Riemann equation  , we have
x y

u
v( x, y)   dy  C ( x) (4. 32)
x

Exercise: find the harmonic conjugate of the function

a) u( x, y)  y 3  3x 2 y and

b) u( x, y)  e y sin x

12