ENGR 9420 - Engineering Analysis: Special Functions and Differential Equations 28

4 Special Functions and Solutions to

Differential Equations
4.1 Introduction
Special functions arise in the solution of ordinary and partial differential equations. These
functions are defined in terms of integrals and/or infinite series. Of particular interest to engi-
neers are the Gamma and Beta functions, the Gaussian Error function, the Bessel functions,
Legendre functions, and Elliptic Integrals. Other functions which appear in mathematical
physics which we will not consider include those related to the Bessel functions: Spherical
Bessel, Hankel, Struve, Airy, and Kelvin functions, those related to the Elliptic integrals: the
Jacobi elliptic functions, other integral functions such as the Fresnel integrals, the family of
Orthogonal polynomials: Hermite, Laguerre, Jacobi, and Chebyshev, and the Hypergeomet-
ric and Confluent Hypergeometric functions. The latter being quite useful for defining many
special functions in terms of a generalized series. There are many others not included in
the above list. The vast majority can be found discussed in the Handbook of Mathematical
Functions, by Abramowitz and Stegun (1961) and other references included in the notes.

4.2 Gamma and Beta Function

We begin our examination of special functions by considering the Gamma and Beta functions
given their close relationship to one another.

4.2.1 Gamma Function

The Gamma function is one of the simplest special functions and also quite important, since
it appears in the definitions of several other special functions and can be used to define many
definite integrals.
The Gamma function is defined by the following limit:
n!nx
Γ(x) = lim (1)
n→∞ x(x + 1)(x + 2) · · · (x + n)

From the above definition, it is clear that Γ(x) cannot be defined at x = 0, −1, −2, . . .,
since the limit becomes infinite for any of these values, see Fig. 1. It also has the special
result Γ(1) = 1, which can be deduced from the definition above. We may also define the
Gamma function as an infinite product series:
∞
z
1 Y 1 + n1
Γ(z) = (2)
z n=1 1 + nz

where z = x + iy is a complex variable. This is Euler’s product for Γ(z). The function
converges everywhere except at z = 0, −1, −2, −3, . . .. The Gamma function is one of the
few mathematical functions which is not a direct solution to a differential equation. However,
it plays a significant role in the solution to many equations and special functions.
ENGR 9420 - Engineering Analysis: Special Functions and Differential Equations 29

The Gamma function also has the following recurrence relationship which is useful for
determining values of Γ(x) for values greater than one, that is:

Γ(x + 1) = xΓ(x) x>0 (3)

Another useful relation relationship used for simplifying expressions is:

π
Γ(x)Γ(1 − x) = (4)
sin πx
The Gamma function can also be related to the factorial function n! since:
Γ(1) = 1 = 0!
Γ(2) = 1 · Γ(1) = 1 = 1!
Γ(3) = 2 · Γ(2) = 2 · 1! = 2!
Γ(4) = 3 · Γ(3) = 3 · 2! = 3! (5)
..
.
Γ(n + 1) = n! n = 0, 1, 2, . . .

which can be easily deduced, making it a generalization of the factorial function n!.

Fig. 1 - The Gamma Function, Andrews (1998).

The Gamma function can also be defined as an infinite integral:

Z ∞
Γ(x) = e−t tx−1 dt x>0 (6)
0
although it is limited to only positive values of x, the integral form is the most common way
in which Γ(x) is defined.
ENGR 9420 - Engineering Analysis: Special Functions and Differential Equations 30

The integral form also allows us to define the derivative of the Gamma function, since
we may differentiate under the integral sign and obtain:
Z ∞
0
Γ (x) = e−t tx−1 ln tdt x>0 (7)
0

and Z ∞
00
Γ (x) = e−t tx−1 (ln t)2 dt x>0 (8)
0
By and large the Gamma function is most useful for solving definite integrals. For
example, if the following change of variables are employed we obtain the following alternate
forms: for example, if t = u2 , then we obtain:
Z ∞
2
Γ(x) = 2 e−u u2x−1 du (9)
0

or if t = ln(1/u), then
Z 1  x−1
1
Γ(x) = ln du (10)
0 u
A more complicated but very useful form can be obtained by using Eq. (9) in the following
form:
Z ∞ Z ∞
−u2 2x−1 2
Γ(x)Γ(y) = 2 e u du · 2 e−v v 2y−1 dv (11)
0 0
or Z ∞ Z ∞
2 +v 2 )
Γ(x)Γ(y) = 4 e−(u u2x−1 v 2y−1 dudv (12)
0 0

Further if we introduce u = r cos θ and v = r sin θ we can show (left as an exercise) that:
Z π/2
Γ(x)Γ(y) = 2Γ(x + y) cos2x−1 θ sin2y−1 θdθ (13)
0
or Z π/2
Γ(x)Γ(y)
= cos2x−1 θ sin2y−1 θdθ x > 0, y > 0 (14)
2Γ(x + y) 0

Equation (14) proves quite useful in evaluating a host of trigonometric integrals, as we

shall see from several examples.
Finally we conclude with the incomplete Gamma function and complementary incomplete
Gamma functions. These are defined as:
Z x
γ(a, x) = e−t ta−1 dt a>0 (15)
0

and Z ∞
Γ(a, x) = e−t ta−1 dt a>0 (16)
x
ENGR 9420 - Engineering Analysis: Special Functions and Differential Equations 31

such that
γ(a, x) + Γ(a, x) = Γ(a) (17)
These functions find uses in probability theory and some fundamental transport problems
heat and mass transfer.

4.2.2 Beta Function

The Beta function is defined by the following integral relationship:
Z 1
B(x, y) = tx−1 (1 − t)y−1 dt x > 0, y > 0 (18)
0
It is related to the Gamma function, but due to its importance and frequency of ap-
pearance in applied mathematics problems, it is given its own designation. Alternate forms
of the Beta function integral are obtained with following change of variables. First, if we
introduce t = u/(1 + u) in Eq. (18), we obtain:
Z ∞
ux−1
B(x, y) = du x > 0, y > 0 (19)
0 (1 + u)x+y
Further, if introduce the change of variable t = cos2 θ in Eq. (18), we obtain:
Z π/2
B(x, y) = 2 cos2x−1 θ sin2y−1 θdθ x > 0, y > 0 (20)
0

which when compared with Eq. (14) yields:

Γ(x)Γ(y)
B(x, y) = x > 0, y > 0 (21)
Γ(x + y)
from which we also observe the following symmetry property:

B(x, y) = B(y, x) (22)

The Beta function (and hence the Gamma function) is quite useful in representing definite
integrals. One class of definite integrals which arises in engineering analysis, are those
involving, areas, moments, and volumes. These integrals can be generalized through the
definition of the Dirichlet integrals. If V denotes the closed region in the first octant bounded
by the planes x = 0, y = 0, and z = 0 and the surface:
 x α  y β  z γ
+ + =1 (23)
a b c
and if all constants are positive we define the Dirichlet integrals by:

ap bq cr Γ(p/α)Γ(q/β)Γ(r/γ)
Z Z Z
I= xp−1 y q−1 z r−1 dxdydz = (24)
V αβγ Γ(1 + p/α + q/β + r/γ)
ENGR 9420 - Engineering Analysis: Special Functions and Differential Equations 32

These integrals as we shall see, are useful in evaluating the multiple integrals associated
with the evaluation of areas, volumes, and moments of many simple and complex geometries.
In the case of a area integrals, bounded by the axes x = 0 and y = 0 and the curve:
 x α  y β
+ =1 (25)
a b
one would simplify Eq. (24) as:

ap bq Γ(p/α)Γ(q/β)
Z Z
I= xp−1 y q−1 dxdy = (26)
A αβ Γ(1 + p/α + q/β)

Example 1
Show that the integral
Z π/2
1 3 Γ(2/3)Γ(5/6)
sin 3 θdθ = √ (27)
0 2 π

Example 2
Given the curve:
 x α  y β
+=1 (28)
a b
in the first quadrant, prove the integral defined by Eq. (26).
Example 3
Consider the curve referred to as a hyper-ellipse defined by the function:
 x n  y n
+ =1 (29)
a b
where a, b, n > 0. Find the area under the curve. Verify that the solutions for a = b and
6 b and n = 2 (ellipse) are obtained from the general solution.
n = 2 (circle) and for a =
Example 4
Show that the solution for the following differential equation:

d2 φ dφ
2
+ 3η 2 =0 (30)
dη dη
subject to the conditions:
φ(0) = φ0
φ(η → ∞) = φ∞
can be obtained on terms of the incomplete gamma function.
ENGR 9420 - Engineering Analysis: Special Functions and Differential Equations 33

4.3 Gaussian Error Function

The Gaussian error function appears in many physical and mathematical applications, such
as heat conduction, mass diffusion, probability theory, and the theory of errors. It is defined
by the integral:
Z x
2 2
erf x = √ e−t dt (31)
π 0
By means of a series expansion of the exponential function we may obtain the following
infinite series formulation:
∞
2 X (−1)n x2n+1
erf x = √ |x| < ∞ (32)
π n=0 n!(2n + 1)
The error function is plotted in Fig. 2. From Eq. (32) we may deduce that the error
function is an odd function having the property:

erf(−x) = −erf(x) (33)

Additional properties include erf(0) = 0 and erf(∞) = 1. The complementary error function
denoted as erfc x occurs frequently in diffusion problems. It is defined as:
Z ∞
2 2
erfc x = √ e−t dt (34)
π x
such that
erfc x = 1 − erf x (35)
The error function and complementary error function can also be expressed in terms of
the incomplete Gamma functions. It will be left as an exercise for you to verify that:
Z x
2 2 1
erf x = √ e−t dt = √ γ(1/2, x2 ) (36)
π 0 π
and Z ∞
2 2 1
erfc x = √ e−t dt = √ Γ(1/2, x2 ) (37)
π x π

Fig. 2 - The Error Function, Andrews (1998).

ENGR 9420 - Engineering Analysis: Special Functions and Differential Equations 34

As mentioned earlier, the error function or complementary error function appears in

the solution of many diffusion problems involving semi-infinite domains. One such problem
which we shall examine involves the following ordinary differential equation:

d2 φ dφ
2
+ 2η =0 (38)
dη dη
subject to φ = 1 when η = 0 and φ → 0 when η → ∞. The solution can be found by means
of the method of reduction of order. That is we will define w = dφ/dη to obtain:
dw
+ 2ηw = 0 (39)
dη
Separating variables allows us to write:
dw
= −2ηdη (40)
w
or after integrating
ln w = −η 2 + ln C1 (41)
or
w
ln = −η 2 (42)
C1
which gives
dφ 2
w= = C1 e−η (43)
dη
Integrating once more, yields:
Z η
2
φ = C1 e−η dη + C2 (44)
0
Finally application of the two boundary conditions yields:
Z 0
2
1 = C1 e−η dη + C2 = C2 (45)
0
and Z ∞
2
0 = C1 e−η dη + C2 (46)
0
or
−1 1
C1 = R ∞ −η 2 = −√ (47)
0
e dη π/2
The above equation may be evaluated by means of Eq. (9) and Eq. (14). Again, it will
be left as an exercise for you to show that:
Z ∞ √
−η 2 π
e dη = (48)
0 2
Returning to our solution we obtain:
Z η
2 2
φ=1− √ e−η dη = 1 − erf η = erfc η (49)
π 0
ENGR 9420 - Engineering Analysis: Special Functions and Differential Equations 35

4.4 Bessel Functions

The family of functions denoted as Bessel functions, have many applications in applied
physics and engineering. We are specifically interested the functions denoted as Jn (x), Yn (x),
In (x), and Kn (x). These appear frequently in engineering problems related to heat conduc-
tion, electric fields, buckling, and dynamics. Bessel functions are also commonly referred to
as cylinder functions due to their association with solutions for cylindrical domains.

4.4.1 Bessel Functions of the First and Second Kind

We begin our examination of the Bessel functions by examining Bessel’s equation:

d2 y dy
x2 2
+ x + (x2 − n2 )y = 0 n≥0 (50)
dx dx
The general solution to this equation can be found using the series form through the
application of the Method of Frobenius. Provided that n is not an integer, the general
solution is of the form:

y(x) = C1 Jn (x) + C2 J−n (x) n 6= 0, 1, 2... (51)

where ∞
X (−1)k (x/2)2k+n
Jn (x) = (52)
k=0
k!Γ(k + n + 1)
and ∞
X (−1)k (x/2)2k−n
J−n (x) = (53)
k=0
k!Γ(k − n + 1)
Jn (x) is referred to as a Bessel function of the first kind. It can be shown that Jn (x) and
J−n (x) are not linearly independent solutions when n is an integer. If n is an integer, we
find that:
J−n (x) = (−1)n Jn (x) n = 0, 1, 2... (54)
For cases when n is equal to an integer, only Jn (x) can be assumed to be the solution to
Eq. (50). A second linearly independent solution can be found of the form:

 cos(nπ)Jn (x) − J−n (x)

n 6= 0, 1, 2, ...
sin(nπ)

Yn (x) = (55)
cos(pπ)Jp (x) − J−p (x)
lim n = 0, 1, 2, ...


sin(pπ)

p→n

This is referred to as a Bessel function of the second kind. Since both Jn (x) and Yn (x)
are linearly independent solutions, they are valid for all values of n, and hence the general
solution to Eq. (50) is taken as:

y(x) = C1 Jn (x) + C2 Yn (x) (56)

ENGR 9420 - Engineering Analysis: Special Functions and Differential Equations 36

The functions Jn (x) and Yn (x) are shown in Figs. 3 and 4 for the values of n = 0, 1 and
2. In problems where x = 0 is part of the solution domain, the Yn (x) must be excluded since
they are undefined at x = 0.
The Bessel functions of the first and second kind have several special characteristics,
particularly when the the order is half integral. In these cases the functions reduce to
elementary forms of sine and cosine functions. For example if n = 1/2 we have:
r
2
J1/2 (x) = sin(x) (57)
πx
r
2
J−1/2 (x) = cos(x) (58)
πx
r
2
Y1/2 (x) = − cos(x) (59)
πx
r
2
Y−1/2 (x) = sin(x) (60)
πx
Some useful recurrence relationships are:
2n
Jn+1 (x) = Jn (x) − Jn−1 (x) (61)
x
d 1
Jn (x) = [Jn−1 (x) − Jn+1 (x)] (62)
dx 2
d
x Jn (x) = xJn−1 (x) − nJn+1 (x) = nJn (x) − xJn+1 (x) (63)
dx
d n
x Jn (x) = xn Jn−1 (x) (64)
dx
d −n
x Jn (x) = −x−n Jn+1 (x) (65)
dx
The above equations are also valid if Jn (x) is replaced by Yn (x).

Fig. 3 - The Bessel Function Jn (x), Andrews (1998).

ENGR 9420 - Engineering Analysis: Special Functions and Differential Equations 37

Fig. 4 - The Bessel Function Yn (x), Andrews (1998).

4.4.2 Zeroes of the Bessel Functions

The Bessel functions of the First and Second kind are harmonic and as such contain repeated
zeroes or roots. The roots of these functions are tabulated in the various mathematical hand-
books and are also easily computed using root finding solvers in the various mathematical
packages. Maple and Mathematica also have special functions to represent the zeroes of the
Bessel functions. As an example, the first five roots of the Bessel functions of the first and
second kind are given in the Table below. Additional roots may be approximated from:

xi+1
0 = xi0 + π (66)

Roots of the Bessel Functions

i J0 (x) = 0 J1 (x) = 0 Y0 (x) = 0 Y1 (x) = 0
1 2.4048 3.8317 0.8936 2.1971
2 5.5201 7.0156 3.9577 5.4297
3 8.6537 10.1735 7.0861 8.5960
4 11.7915 13.3237 10.2223 11.7492
5 14.9309 16.4706 13.3611 14.8974

4.4.3 Generalized Bessel Equation

Since many problems have solutions containing Bessel functions of the First and Second
kind, a generalization of the Bessel equation is possible. One such equation is defined by the
following differential equation:

d2 y dy
x2 2
+ (1 − 2a)x + [b2 c2 x2c + (a2 − c2 n2 )]y = 0 n≥0 (67)
dx dx
which has the general solution:
y(x) = xa [C1 Jn (bxc ) + C2 Yn (bxc )] (68)
ENGR 9420 - Engineering Analysis: Special Functions and Differential Equations 38

The above equation and others like are found by applying multiple changes in variables.
We will discuss a few others in class and or in assigned problems. Use of equations such as
Eq. (67), is quite straight forward as one merely writes a differential equation in standard
form and then proceeds to develop a system of equations to define all of the coefficients in
terms of the variable a, b, c and n.
Example 5
Given the differential equation:

d2 y dy
x2 2
+ 5x + (4x2 + 3)y = 0 (69)
dx dx
find the solution in terms of Bessel functions.
Example 6
Airy’s differential equation is given by:
2
y 2d
x + xy = 0 (70)
dx2
find the solution in terms of Bessel functions.

4.4.4 Modified Bessel Functions of the First and Second Kind

Another differential equation related to Bessel’s equation is the modified Bessel equation
which takes the form:

d2 y dy
x2 2
+ x − (x2 + n2 )y = 0 n≥0 (71)
dx dx
The solution to the above equation is in the form of:

y(x) = C1 In (x) + C2 I−n (x) n 6= 0, 1, 2, ... (72)

where ∞
X (x/2)2k+n
In (x) = i−n Jn (ix) = (73)
k=0
k!Γ(k + n + 1)
and ∞
n
X (x/2)2k−n
I−n (x) = i J−n (ix) = (74)
k=0
k!Γ(k − n + 1)
where In (x) is the modified Bessel function of the first kind.
However, it can be shown that In (x) = I−n (x) when n = 0, 1, 2, ..., otherwise In (x) and
I−n (x) are linearly independent. As with the Bessel’s equation, a second linearly independent
solution can be developed which satisfies Bessel’s modified equation. The function is defined
analogously in a manner similar to the Bessel function of the second kind:
ENGR 9420 - Engineering Analysis: Special Functions and Differential Equations 39

 π
 [I−n (x) − In (x)] n 6= 0, 1, 2, ...
2 sin(nπ)

Kn (x) = π (75)
 lim
 [I−p (x) − Ip (x)] n = 0, 1, 2, ...
p→n 2 sin(pπ)

This is referred to as a modified Bessel function of the second kind. Since both In (x)
and Kn (x) are linearly independent solutions, they are valid for all values of n, and hence
the general solution to Eq. (71) is taken as:

y(x) = C1 In (x) + C2 Kn (x) (76)

The functions In (x) and Kn (x) are shown in Figs. 5 and 6 for the values of n = 0, 1 and
2. In problems where x = 0 is part of the solution domain, the Kn (x) must be excluded
since they are undefined at x = 0.
The modified Bessel functions of the first and second kind have several special character-
istics, particularly when the the order is half integral. In these cases the functions reduce to
elementary forms of hyperbolic sine and cosine functions. For example, if n = 1/2 we have:
r
2
I1/2 (x) = sinh(x) (77)
πx
r
2
I−1/2 (x) = cosh(x) (78)
πx
r
π
K1/2 (x) = K−1/2 (x) = exp(−x) (79)
2x
Some useful recurrence relationships for the modified Bessel functions are:
2n
In+1 (x) = In−1 (x) − In (x) (80)
x
d 1
In (x) = [In−1 (x) + In+1 (x)] (81)
dx 2
d
x In (x) = xIn−1 (x) − nIn (x) = xIn+1 (x) + nIn (x) (82)
dx
d n
x In (x) = xn In−1 (x) (83)
dx
d −n
x In (x) = x−n In+1 (x) (84)
dx
and
2n
Kn+1 (x) = Kn−1 (x) + Kn (x) (85)
x
d 1
Kn (x) = − [Kn−1 (x) + Kn+1 (x)] (86)
dx 2
d
x Kn (x) = −xKn−1 (x) − nKn (x) = nKn (x) − xKn+1 (x) (87)
dx
ENGR 9420 - Engineering Analysis: Special Functions and Differential Equations 40

d n
x Kn (x) = −xn Kn−1 (x) (88)
dx
d −n
x Kn (x) = −x−n Kn+1 (x) (89)
dx
These expressions only cover a number of useful relationships involving the Bessel func-
tions and modified Bessel functions. The student is referred to the various references and
handbooks for additional expressions involving integrals of Bessel functions, namely the
book by Gradshteyn and Ryzhik (2007). Of course the mathematical packages Maple and
Mathematica can also assist you quite well.

Fig. 5 - The Bessel Function In (x), Andrews (1998).

Fig. 6 - The Bessel Function Kn (x), Andrews (1998).

ENGR 9420 - Engineering Analysis: Special Functions and Differential Equations 41

4.5 Elliptic Integrals

Elliptic integrals are another class of functions defined by integrals which have no simple alge-
braic form. There are three principal forms and each may be defined as either an incomplete
or complete elliptic integral.
We begin by examining the elliptic integral of the first kind which takes the form:
Z x
dx
F (x, k) = p k2 < 1 (90)
2 2 2
(1 − x )(1 − k x )
0

An alternate form which is more frequently used is obtained by introducing the change
of variable x = sin φ:
Z φ
dφ
F (φ, k) = p 2
k2 < 1 (91)
0 2
(1 − k sin φ)
0
√ The parameter k is called the modulus of the elliptic integral and the parameter k =
1 − k 2 , the complementary modulus with respect to k. When the variable x = 1 or
φ = π/2, the elliptic integral is referred to as the complete elliptic integral of the first kind,
and is denoted as:
Z π/2
dφ
K = K(k) = p (92)
0 (1 − k 2 sin2 φ)
When the complementary modulus k 0 is the argument of the complete elliptic integral,
it is frequently denoted as:
Z π/2
0 0 dφ
K = K(k ) = p (93)
0 (1 − k 02 sin2 φ)
The elliptic integral of the second kind is defined as:
Z xr
1 − k 2 x2
E(x, k) = dx k2 < 1 (94)
0 1 − x2
or, more frequently as:
Z φ q
E(φ, k) = 1 − k 2 sin2 φdφ k2 < 1 (95)
0

Once again, if the variable x = 1 or φ = π/2, the elliptic integral is referred to as the
complete elliptic integral of the second kind, and is denoted as:
Z π/2 q
E = E(k) = 1 − k 2 sin2 φdφ (96)
0

When the complementary modulus k 0 is the argument of the complete elliptic integral,
it is frequently denoted as:
ENGR 9420 - Engineering Analysis: Special Functions and Differential Equations 42

Z π/2 q
0 0
E = E(k ) = 1 − k 02 sin2 φdφ (97)
0
Finally, we define the elliptic integral of the third kind which has the form:
Z x
dx
Π(x, n, k) = p k2 < 1 (98)
2 2 2 2
0 (1 + nx ) (1 − x )(1 − k x )

or, more frequently as:

Z φ
dφ
Π(φ, n, k) = p k2 < 1 (99)
0 (1 + n sin φ) 1 − k 2 sin2 φ
2

Once again, if the variable x = 1 or φ = π/2, the elliptic integral is referred to as the
complete elliptic integral of the third kind, and is denoted as:

Π = Π(n, k) = Π(1, n, k) = Π(π/2, n, k) (100)

Care must be taken to ensure which form is used, e.g. the algebraic form (x) or the trig
form (φ) when using computer algebra systems such as Maple or Mathematica. In Maple
the default form is the trig formulation.
Some useful identities related to complete elliptic integrals of the first and second kind,
are:
dE 1
= (E − K) (101)
dk k
and
dK 1
= 02 (E − k 02 K) (102)
dk kk
Related to the elliptic integral of the first kind are the Jacobi elliptic functions. These
are inverse forms. We will limit ourselves to the elliptic integrals for the time being.
Example 7
Find the perimeter of an ellipse defined by:
 x 2  y 2
+ =1 (103)
a b
in terms of elliptic integrals.

4.6 Legendre Functions

The Legendre functions arise in the solution of many problems in spherical coordinates.
They are sometimes referred to as spherical harmonics. Legendre’s equation is represented
by the following differential equation:

d2 y dy
(1 − x2 ) 2
− 2x + n(n + 1)y = 0 n ≥ 0, |x| < 1 (104)
dx dx
ENGR 9420 - Engineering Analysis: Special Functions and Differential Equations 43

or  
d 2 dy
(1 − x ) + n(n + 1)y = 0 n ≥ 0, |x| < 1 (105)
dx dx
The solution of Eq. (104) is found in terms of:

y(x) = C1 Pn (x) + C2 Qn (x) (106)

where Pn (x) and Qn (x) are Legendre functions of the first and second kind of order n.
When n = 0, 1, 2, 3, . . ., the solutions are obtained in terms of the Legendre polynomials.
The Legendre polynomial Pn (x) is defined as:
1 dn 2
Pn (x) = (x − 1)n (107)
2n n! dxn
or
P0 (x) = 1
P1 (x) = x
P2 (x) = 21 (3x2 − 1) (108)
P3 (x) = 12 (5x3 − 3x)
..
.
The following recurrence relationship is obtained:
2n + 1 n
Pn+1 (x) = xPn (x) − Pn−1 (x) n≥1 (109)
n+1 n+1

Fig. 7 - The Legendre Function Pn (x), Andrews (1998).

The Legendre function of the second kind for the case when n = 0, 1, 2, 3, . . ., is defined
by:
ENGR 9420 - Engineering Analysis: Special Functions and Differential Equations 44

n
1 1+x X 1
Qn (x) = Pn (x) ln − Pm−1 (x)Pn−m (x) (110)
2 1 − x m=1 m
or
Q0 (x) = 12 ln 1+x
1−x
Q1 (x) = P1 (x)Q0 (x) − 1
Q2 (x) = P2 (x)Q0 (x) − 23 x (111)
Q3 (x) = P3 (x)Q0 (x) − 52 x2 + 3
2
..
.
It is clear that that singularities at x = ±1 exist in the first term of the Qn (x) and
hence must be avoided if these points exist in the solution domain of an application. The
first 5 Legendre polynomials are plotted in Figs. 7 and 8 for Pn (x) and Qn (x). For the
general case of n not being an integer, the Legendre functions may be defined in terms of
the Hypergeometric function. We will discuss these at the conclusion of the next section.

Fig. 8 - The Legendre Function Qn (x), Andrews (1998).

4.7 Hypergeometric Function

We conclude with a short discussion of the Hypergeometric function. This product series
can be used to define many of the special functions we have considered thus far. The
Hypergeometric function is defined by the product series:

∞
Γ(a)Γ(b) X Γ(a + n)Γ(b + n) xn
F (a, b; c; x) = |x| < 1, c 6= 0, −1, −2, . . . (112)
Γ(c) n=0 Γ(c + n) n!

or ∞
X (a)n (b)n xn
F (a, b; c; x) = |x| < 1, c 6= 0, −1, −2, . . . (113)
n=0
(c)n n!
ENGR 9420 - Engineering Analysis: Special Functions and Differential Equations 45

where the operation defined by (δ)n is used to represent in a shorthand fashion the following:

Γ(δ + n)
(δ)n = (114)
Γ(δ)

It is referred to as the Pochammer symbol.

The Hypergeometric function has the following property:

F (a, b; c; x) = F (b, a; c; x) (115)

Also note that some books utilize a notation that denotes F (a, b; c; x) = 2 F1 (a, b; c; x) =
F ([a, b]; [c]; x), where the 2 and 1 denote the number of numerator and denominator variables.
Clearly this is indicative that the Hypergeometric function itself is a special case of a more
generalized hypergeometric series.
The Hypergeometric function is one solution to the following differential equation:

d2 y dy
x(1 − x) 2
+ [c − (a + b + 1)x] − aby = 0 (116)
dx dx
Another linearly independent solution, can be found yielding the following general solution:

y(x) = C1 F (a, b; c; x) + C2 x1−c F (a − c + 1, b − c + 1; 2 − c; x) (117)

A general property of the hypergeometric function is:

dk (a)k (b)k
k
F (a, b; c; x) = F (a + k, b + k; c + k; x) (118)
dx (c)k
Finally, the Hypergeometric function also has a useful integral representation:
Z 1
Γ(c)
F (a, b; c; x) = tb−1 (1 − t)c−b−1 (1 − xt)−a dt (119)
Γ(b)Γ(c − b) 0

4.7.1 Relationship to Other Functions

The Hypergeometric is most useful at representing other mathematical functions. The fol-
lowing is a short list of functions that may be evaluated using the Hypergeometric function:
1
B(p, q) = F (p, 1 − q; 1 + p; 1) (120)
p
(x/2)ν x2
Jν (x) = F ([ ]; ν + 1; − ) (121)
Γ(ν + 1) 4
π 1 1
K(k) = F ( , ; 1; k 2 ) (122)
2 2 2
π 1 1
E(k) = F (− , ; 1; k 2 ) (123)
2 2 2
ENGR 9420 - Engineering Analysis: Special Functions and Differential Equations 46

Γ(2n + 1) n n 1−n 1 1
Pn (x) = x F (− , ; − n; 2 ) (124)
2n [Γ(n
+ 1)] 2 2 2 2 x
Γ( 21 )Γ(n + 1) −n−1 n 1+n 3 1
Qn (x) = n+1 3 x F (1 + , ;n + ; 2) (125)
2 Γ(n + 2 ) 2 2 2 x
This is just a small sampling of such relationships. Additional functions may be defined on
the basis of other hypergeometric functions such as the Confluent Hypergeometric functions
of the first and second kind.

References
Abramowitz, M. and Stegun, I, Handbook of Mathematical Functions, Dover Publications,
1961.

Andrews, L.C., Special Functions of Mathematics for Engineers, SPIE Press, 1998.

Bowman, F., Introduction to Bessel Functions, Dover Publications, 1958.

Davis, H.T., Introduction to Non-Linear Differential and Integral Equations, Dover, 1962.

Farrell, O.J. and Bertram, R., Solved Problems in Analysis as Applied to Gamma, Beta,
Legendre, and Bessel Functions, Dover, 1963.

Gradshteyn, I.S. and Ryzhik, I.M., Tables of Integrals, Series, and Products, Academic
Press, 2007.

Gray, A. and Mathews, G.B., A Treatise on Bessel Functions and Their Applications to
Physics, Dover Publications, 1966.

Hildebrand, F.B., Advanced Calculus for Applications, Prentice-Hall, 1976.

Hochstadt, H. The Functions of Mathematical Physics, Dover, 1986.

Lebedev, N.N., Special Functions and Their Applications, Dover Publications, 1972.

MacMillan, W.D., The Theory of the Potential, Dover, 1958.

Sneddon, I.N., Special Functions of Mathematical Physics and Chemistry, Oliver and Boyd,
1955.