Appendix A

Special functions

M. Orefice: Radiating Electromagnetic Systems (2013), Appendix A

A.1

The Gamma Function

The function is a single value function, analytical in all the complex plane, except
that for real integer negative number, or zero. It is defined as:
(z) =

Z
0

tz1 et dt

(A.1)

which is also called the Euler integral of the second kind and is an interpolation for
non integer values of the factorial function: for z integer it is
(n + 1) = n!

(A.2)

In the non-analytical points the function diverges, with simple poles having a
residue (1)n /n!, and it has alternate signs in the successive intervals separated by
the singular points.
40

30

20

(x)

10

-1 0

-2 0

-3 0
-3

-2

-1

1
x

Figure A.1: The function in linear scale.

The recursive formula holds
(z + 1) = z(z)

(A.3)

M. Orefice: Radiating Electromagnetic Systems (2013), Appendix A

so that the function can be intended as the factorial of a non integer number, so
that the notation (z + 1) = z! is used.

20

10

15

10

10

(x)

10

10

10

-5

10

10
x

12

14

16

18

Figure A.2: The function in logarithmic scale for a wider range of values.

20

M. Orefice: Radiating Electromagnetic Systems (2013), Appendix A

A.2

Bessel Functions

The Bessel equation is a differential equation of the 2nd order, with non constant
coefficients, that is often found in cylindrical coordinates problems, and which is
expressed as
d2 y
dy
x2 2 + x + (x2 + 2 )y = 0
(A.4)
dx
dx
where is a parameter, which may be complex, that however here we will assume
as real. The Bessel functions, that are indicated with J (x), are solutions of such
equation. If is real non integer, J (x) and J (x) are linearly independent and
their linear combination represents all the possible solutions of the Bessel equation.
The parameter is called the order of the Bessel function.
Bessel functions are expressed in power series as
J (x) =

(x2 /4)k
k=0 k!( + k + 1)

(A.5)

For small arguments, from eq.(A.5) we get

(x/2)
J (x)
=
( + 1)

(A.6)

which for integer becomes

xn
Jn (x)
(A.7)
= n
2 n!
The Bessel function of order increases with x up to reach the first maximum
for a value of x near to , and then decreases with damped oscillations of period 2.
For large arguments (x ) the Bessel functions are expressed asymptotically
as
s
2

cos(x )
(A.8)
J (x)
=
x
2
4
If = n, integer, the Bessel functions are no longer independent, and they are
odd or even depending on n, so that
Jn (x) = (1)n Jn (x)

(A.9)

An example of the Bessel functions of order 0,1,2,4 is shown in fig.A.3 and, for
larger orders, in fig.A.4.
Another independent solution of the Bessel equation is the Neumann function
Y (x) (sometimes indicated as N (x)), defined as
Y (x) =

J (x) cos J (x)

sin

where the RHS becomes a limit for integer or null.

(A.10)

M. Orefice: Radiating Electromagnetic Systems (2013), Appendix A

1
n=0

0.8

n=1

0.6

n=2
n=4

0.4

Jn(x)

0.2
0
-0.2
-0.4
-0.6
-0.8
-1

10

12

14

Figure A.3: Bessel functions for = 0, 1, 2 e 4.

0.4
0.3

=10

=20

0.2

=40

J (x)

0.1

0
-0.1
-0.2
-0.3
-0.4
0

10

20

30

40
x

50

60

Figure A.4: Bessel functions for = 10, 20 e 40.

70

80

M. Orefice: Radiating Electromagnetic Systems (2013), Appendix A

The expression of the Neumann function, of which a few examples are reported
in fig. A.5, is more complicated than that of the Bessel function, and for small
values of the argument the function diverges. If is integer, we have
2n (n 1)!
Yn (x)
(A.11)
=
xn

1
0.8
0.6

n=0
n=1

0.4

n=2

n=4

Yn(x)

0.2
0
-0.2
-0.4
-0.6
-0.8
-1

10

12

14

Figure A.5: Neumann functions for = 0, 1, 2 e 4.

For large arguments (x ) the Neumann functions have the asymptotic expression
s

2
sin(x )
(A.12)
Y (x)
=
x
2
4
which is coincident with that of the Bessel functions, except for a phase shift by
/2.
Another couple of independent solutions of the Bessel equation are the Hankel
functions (of the first and second kind), also called Bessel functions of the third kind,
defined as
H(1) (x) = J (x) + jY (x)
(A.13)
H(2) (x) = J (x) jY (x)
whose asymptotic expression, which may be derived from eqs.(A.8) and (A.12), is
s

H(1,2) (x)
=

2 j(x )
2
4
e
x

(A.14)

M. Orefice: Radiating Electromagnetic Systems (2013), Appendix A

It may be observed that such expressions represent two cylindrical waves (converging and diverging from the axis).
The Bessel functions can also be expressed through integral representations,
which are useful because many expression appearing in radiation problems can be
reduced to them.
Among the various expression we may note:
j n Z jz cos
e
cos(n)d
Jn (z) =
0
Z z
0

t J (t)dt =

++1
2

+1

( + 2k + 1)
X

k=0

++3
2

+1
2

+k

(A.15)

+k

J+2k+1 (z)

(A.16)

valid for <( + + 1) > 0. A particularly interesting (and simple) case is for
= 0 and = 1:
Z z
0

A.2.1

tJ0 (t)dt = zJ1 (z)

(A.17)

Spherical Bessel functions

The Spherical Bessel functions are strictly related to the ordinary Bessel functions,
and are found as solutions of problems in spherical coordinates (while the ordinary functions solve cylindrical problems). In particular, they are the independent
solutions of the differential equation (spherical Bessel equation)
x2

d2 y
dy
+ 2x + [x2 + ( + 1)]y = 0
2
dx
dx

(A.18)

where, similarly to eq.(A.4), is a parameter which may be complex.

The spherical Bessel functions of the first and second kind (more commonly
known as spherical Bessel and Neumann functions) are indicated as j (x) and y (x),
and are related to the ordinary functions by the relationships
r

j (x) =
r

y (x) =

J+1/2 (x)
2x

Y+1/2 (x)
2x

(A.19)

In general, for these functions the properties of the ordinary Bessel functions also
apply. In particular, we may define also here the spherical Hankel functions (of the
1st and 2nd kind), also called spherical Bessel functions of the third kind, defined

M. Orefice: Radiating Electromagnetic Systems (2013), Appendix A

obviously as

h(1)
(x) =

(1)
H
2x +1/2

(A.20)

(2)
h(2)
H
(x) =
2x +1/2
The behavior of the spherical functions is similar to that of the ordinary functions, i.e. they increase proportionally to x for values of the argument up to around
, then they have damped oscillations, but with amplitude proportional to x1 instead of x1/2 , thus with a faster decrease. A few examples of Bessel and Neumann
sperical functions, corresponding to figs.A.3, A.4 and A.5, are shown in figs.A.6, A.7
and A.8.
r

1
n=0
0.8
0.6
n=1
0.4

n=2
n=4

jn(x)

0.2
0
-0.2
-0.4
-0.6
-0.8
-1
0

10

12

14

Figure A.6: Spherical Bessel functions for = 0, 1, 2 and 4.

For particular values of , the spherical Bessel and Neumann functions can be
reduced to combinations of other already known functions. For example:
sin x
j0 (x) =
x
sin x cos x
j1 (x) = 2
x
x
cos x
y0 (x) = j1 (x) =
x
cos x sin x
y1 (x) = j2 (x) = 2
x
x

(A.21)

Note also that the asymptotic expression of the spherical Hankel function of the
2nd kind is simply an outward travelling spherical wave.

M. Orefice: Radiating Electromagnetic Systems (2013), Appendix A

0.2
0.15
n=10

0.1

n=20

jn(x)

0.05

n=40

0
-0.05
-0.1
-0.15
-0.2
0

10

20

30

40
x

50

60

70

80

Figure A.7: Spherical Bessel functions for = 10, 20 and 40.

0.5
n=0
n=1

n=2

n=4

yn(x)

-0.5

-1

-1.5
0

10

12

Figure A.8: Spherical Neumann functions for = 0, 1, 2 and 4.

14

M. Orefice: Radiating Electromagnetic Systems (2013), Appendix A

A.3

10

Associated Legendre functions

In the solution of the Helmholtz equation in spherical coordinates , , with the

method of the separation of variables, for the coordinate we obtain the Legendre equation, which is a differential equation of the 2nd order with non constant
coefficients. Such equation has the form
d2 y
dy
2
(1 x ) 2 2x + [( + 1)
]y = 0
dx
dx
1 x2
2

(A.22)

where and are respectively the degree and the order of the solutions. The two
independent solutions of this equation, whose linear combination represents all the
possible solutions of the Legendre equation, are called Associated Legendre Functions
of the first and second kind, and are indicated respectively with P (x) and Q (x).
Although the equation makes sense (except for singular points) also for x complex, for the use in a spherical coordinate system there is interest only for the case
with 1 < x < 1 real, because the relationship with is:
x = cos

(A.23)

Similarly to the case of cylindrical coordinates, is the eigenvalue of the equation

in the coordinate: it is therefore an integer if the domain of existence of is
0 < < 2, and it has the meaning of azimuthal variation index of the modes
solving the Helmholz equation.
The function of the second kind Q (x) has singularity for x = 1, and therefore
it is not interesting if the polar axis ( = 0) is in the domain of existence of the
equation (e.g. a conical waveguide). Consequently, in this section we will consider
only the function of the first kind with = m, integer.
If = 0 and is an integer, the functions P (x) can be reduced to the Legendre
polynomials Pn (x), which can be obviously become combinations of trigonometrical
functions through eq.(A.23).
If, still with integer, = m 6= 0 (integer), Pnm (x) are still combinations of
trigonometrical functions but they contain irrational of fractional terms in x.
The general expression of the Associated Legendre functions of the first kind is

P (x) =

1
x+1
(1 ) x 1

/2

F (, + 1; 1 ;

1x
)
2

(A.24)

where is the Gamma function, and F (a, b, c; z) is the Gauss hypergeometric series
defined as:

(c) X
(a + n)(b + n) z n
(A.25)
F (a, b, c; z) =
(a)(b) n=0
(c + n)
n!
If the order is negative, the following relationship holds:

P (x)

( + 1)
2
=
P (x) ej sin()Q (x)
( + + 1)

(A.26)

M. Orefice: Radiating Electromagnetic Systems (2013), Appendix A

11

which is not a simple expression, but, in the case of integer, becomes

Pm (x) =

2m1
k=0 (

1
P m (x)
+ m k)

(A.27)

They are therefore even functions, except for a multiplicative constant.

The associated functions of higher order or degree can be derived from some
recurrence relationships:

(x)}
P+1 (x) = (x2 1)1/2 {( )xP (x) ( + )P1

( + 1)P+1
(x) = (2 + 1)xP (x) ( + )P1
(x)

(A.28)
(A.29)

Similarly, the derivatives can be obtained from the relationship:

1
(x)
dP+1
= ( + )( + 1)(x2 1) 2 P1 (x) xP (x)
(A.30)
dx
In the case of conical electromagnetic structures, the most interesting orders are
0 and 1. Here, for integer degree, we have, for m = 0, the Legendre polynomials
(see fig.A.9):
P0 (x) = 1
P1 (x) = x
(A.31)
P2 (x) = (3x2 1)/2
1
Pn+1 (x) = n+1
[(2n + 1)xPn (x) nPn1 (x)]

(x2 1)

which, expressed in the variable, become:

P0 () = 1
P1 () = cos
P2 () = (3 cos 2 + 1)/4

(A.32)

P01 (x) = 0
1
P1 (x) = (x2 1)1/2
P21 (x) = (x2 1)1/2 3x
P31 (x) = 32 (x2 1)1/2 (5x2 1)
P41 (x) = 25 (x2 1)1/2 x(7x2 3)
1
1
(x)]
(x) = n1 [(2n + 1)xPn1 (x) (n + 1)Pn1
Pn+1

(A.33)

and, for m = 1:

(see fig.A.10), or, except for a factor j:

P01 () = 0
P11 () = sin
P21 () = 3 cos sin
P31 () = 23 sin [5 cos2 1]
1
P4 () = 52 cos sin [7 cos2 3]

(A.34)

M. Orefice: Radiating Electromagnetic Systems (2013), Appendix A

+
x
+
+x
+x
+
+ x
+
+ x

0.8
0.6

+ x
+
+ x
+
x
0.4 xxxxxxx
+
xxxx
+
x
xxx
xxx
+
xx
x
+
xx
+
xx
x
0.2
xx
+
xx
x
+
xx
+
x
xx
+
xx
+
x
xx
+
x
0 +++
+
xx
+
++
x
xx
+
++
x
xx
+
++
+
xx
x
++
+
++
x
xx
++
++
x
xx
-0.2
+
++
x
+
xx
++
x
++
++
xx
x
++
xx
++
+++
xx
xx
++
+++
+
x
xxx ++
x
+++
x
++
xx
++++
-0.4
++++ xxxxxxxxxxxxx
++++++++
+++++++++

-0.6
0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

Figure A.9: Legendre polynomials for n = 1(1)4.

3
xxxxxx
xx
x
x
x
x
x
x
x+++++
+
+
++
+
x
+
2
++
++
++ xx
+
+
+
+ x
x
++
+
++
x
+
+
1.5
x
++
+ x
+
x
+
x
++
+
+
x
1
++
x
+
+
+
x
++
x
++
x
+
0.5
+
x
++
x
++
x
+
x
++
+
x
0 xx
x
++
x
x
++
x
x
+
x
x
x
++
x
x
++
x
-0.5
x
xx
++
+
x
xx
++
x
++
x
xx
++
x
xx
++
x
-1
+
x
xx ++++
x+
xx
++
x
xx
++++ xxx
+
+
x
+
x
xx
-1.5 ++++++++++
xxx
xx
xxxx
xxx
xxxxxxxxxxx

2.5

-2
0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

Figure A.10: Associated Legendre functions for = 1 and n = 1(1)4.

12

M. Orefice: Radiating Electromagnetic Systems (2013), Appendix A

A.4

13

Fresnel Integral

The Fresnel Integral is a function, defined in integral form, which often appears in
diffraction problems. It is defined as
Z z

F (z) =

ej 2 x dx

(A.35)

which may be separated in its real and imaginary parts

C(z) =
S(z) =

Z z
0

Z z
0

2
cos
x dx
2
2
sin
x dx
2

(A.36)

The asymptotic value for z is

Z
0

ej 2 x dx =

1+j
2

(A.37)

Other alternative definitions exist, using changes of variable in in the integral or

other multiplicative constants, or the sign at the exponent: for example
s

F1 (z) =

2 Z z jx2
e dx
0

from which

(A.38)

z
(A.39)
2
This integral can be computed by using series of polynomial expansions1 .
This function has a typical behavior: it increases monotonically to the asymptotic
value, then has damped oscillations about it. The behavior of the real and imaginary
parts of eq.(A.35) is shown in fig.A.11. The polar diagram obtained by plotting the
values of the real and imaginary parts C and S in abscissas and ordinates is said
Cornu Spiral (fig.A.12).
F (z) = F1

M.Abramowitz, I.Stegun: Handbook of Mathematical functions, New York: Dover Publ., Inc.,
1965, Ch.7.

M. Orefice: Radiating Electromagnetic Systems (2013), Appendix A

14

0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0

0.5

1.5

2.5

3.5

4.5

Figure A.11: Real (solid) and imaginary (dashed) parts of the Fresnel Integral.

0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

Figure A.12: Cornu Spiral.

0.8

0.9