Applied Mathematics and Computation 243 (2014) 33–43

Contents lists available at ScienceDirect

Applied Mathematics and Computation

journal homepage: www.elsevier.com/locate/amc

Solutions of elliptic integrals and generalizations by means

of Bessel functions
Sávio L. Bertoli ⇑, Laércio Ender, Juliano de Almeida, Pollyana M. da Silva Kalvelage
Department of Chemical Engineering, Regional University of Blumenau, FURB, São Paulo street, 3250, 89030-000 Blumenau, Brazil

a r t i c l e i n f o a b s t r a c t

Keywords: The elliptic integrals and its generalizations are applied to solve problems in various areas
Hyperelliptic integrals of science. This study aims to demonstrate a new method for the calculation of integrals
Elliptic integrals through Bessel functions. We present solutions for classes of elliptic integrals and
Epstein–Hubbell integral generalizations, the latter, refers to the hyperelliptic integrals and the integral called
Bessel functions
Epstein–Hubbell. The solutions obtained are expressed in terms of power series and/or
Lauricella function
trigonometric series; under a particular perspective, the final form of a class of
hyperelliptic integrals is presented in terms of Lauricella functions. The proposed method
allowed to obtain solutions in ways not yet found in the literature.
Ó 2014 Elsevier Inc. All rights reserved.

1. Introduction

The elliptic integrals can be viewed as generalizations of inverse trigonometric functions and can provide solutions to var-
ious problems in Electromagnetism [1,2], Fluid Mechanics [3] and Chemical Reactions Kinetics [4]. In particular, in geometry
are used to measure the perimeter of an ellipse and the area of an ellipsoid [5,6].
In general, those integrals cannot be expressed in terms of elementary functions (all attempts failed, until the nineteenth
century when it was finally proved, in fact, to be impossible to carry out this objective [7]).
The solutions obtained in this work are compared with the usual forms of the literature, which are in most cases, different
in the representation forms. In the present work, the method used to solve the elliptic integrals is to relate them to the Bessel
functions.
Under a particular perspective, the complete elliptic integrals are expressed in terms of power series, while the incom-
plete forms are expressed in terms of power and trigonometric series. The complete elliptic integral of the first kind obtained
with the proposed method is the same form found in the literature [8,9]. For the complete elliptic integrals of the second and
third kind, the solutions are given in a different form from the usual representations [10,11,8]. The incomplete elliptic inte-
grals of the first, second and third kind are also solved with the Bessel function’s method. In exception of the incomplete
elliptic integral of the first kind, the other expressions are not found in the literature.
The Epstein–Hubbell integral, called as a generalization of elliptic integrals [9], is of a significant importance and is used
in the solving of various problems [12,13]. For this integral, we propose a new solution, through the same method here used
in other elliptic integrals.

⇑ Corresponding author.
E-mail address: savio@furb.br (S.L. Bertoli).

http://dx.doi.org/10.1016/j.amc.2014.05.084
0096-3003/Ó 2014 Elsevier Inc. All rights reserved.
34 S.L. Bertoli et al. / Applied Mathematics and Computation 243 (2014) 33–43

Hyperelliptic integrals are also addressed in this work, specifically, two types of these integrals are proposed with their
respective solutions through Bessel functions. One of these integrals is expressed in its final form as a series of Lauricella
functions.

1.1. Definition of elliptic integrals

An elliptic integral is any function that can be expressed as,

Z pffiffiffiffiffiffiffiffiffi
Rðx; PðxÞÞdx; ð1:1Þ

where PðxÞ is a polynomial of third or fourth degree.

With some reductions, each elliptic integral can be represented in a form involving integrals of rational, trigonometric,
inverse trigonometric, logarithmic and exponential functions; Bronshtein et al. [14], Jeffrey and Dai [9].

2. Complete elliptic integral of the first kind

In this work, we propose that the complete elliptic integral of the first kind can be written in terms of a modified Bessel
function of the first kind zeroth order, as expressed in the following:
 p Z p2 dh
Z 1 Z p
2
F k; ¼ pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ I0 ðxk sin hÞex dx dh; ð2:1Þ
2 0 1  k sin h
2 2
0 0

 p Z 1 Z p
2
F k; ¼ ex dx I0 ðxk sin hÞdh: ð2:2Þ
2 0 0

The relationship between a Bessel function of the first kind zeroth order and its modified Bessel function is represented as
found in Bowman [15],
v
Iv ðzÞ ¼ i J v ðizÞ; ð2:3Þ

I0 ðzÞ ¼ J 0 ðizÞ: ð2:4Þ

The following representation found in Abramowitz and Stegun [10] allows to rewrite the integral form of the Bessel function
in Eq. (2.2) in a closed form expression:
Z p
2 p
J n ð2iz sin hÞdh ¼ J 2n ðizÞ; ð2:5Þ
0 2
Z p
2 p
¼ In ð2z sin hÞdh ¼ I2n ðzÞ: ð2:6Þ
0 2
After this, replacing n ¼ 0 and z ¼ kx=2, the integral forms are represented by:
Z p  
2 p kx
J 0 ðikx sin hÞdh ¼ J 20 i ; ð2:7Þ
0 2 2
Z p  
2 p kx
¼ I0 ðkx sin hÞdh ¼ I2o
: ð2:8Þ
0 2 2
Substituting the Eq. (2.8) in Eq. (2.2.), we have,
 p Z 1 Z p=2 Z  
p 1 x 2 kx
F k; ¼ ex dx I0 ðxk sin hÞdh ¼ e I0 dx: ð2:9Þ
2 0 0 2 0 2
The Bessel function in Eq. (2.9) can be represented by a series, obtained from the following expression found in Abramowitz
and Stegun [10]:
 z v þz X  r
1
ð1Þr Cðv þ l þ 2r þ 1Þ 14 z2
J v ðzÞJ l ðzÞ ¼ : ð2:10Þ
2 r¼0
Cðv þ r þ 1ÞCðl þ r þ 1ÞCðv þ l þ r þ 1Þr!
After the substitutions: v ¼ l ¼ 0, we obtain the form,
X
1
ð2rÞ!z2r
I20 ðzÞ ¼ : ð2:11Þ
r¼0 22r ðr!Þ4
 S.L. Bertoli et al. / Applied Mathematics and Computation 243 (2014) 33–43 35

Replacing the series given by Eq. (2.11) in Eq. (2.9), follows the expression,

 p p Z 1 X1 2
ð2rÞ!ðk x2 Þ
r

F k; ¼ ex dx: ð2:12Þ

2 2 0 r¼0 22r ðr!Þ4

Rearranging it,

 p p X 1 2r Z 1
ð2rÞ!ðkÞ
F k; ¼ 2r 4
x2r ex dx: ð2:13Þ
2 2 r¼0 2 ðr!Þ 0

Finally, solving the improper integral in Eq. (2.13) we find the final form, represented by:

p  pX 1 r
2
ð2rÞ!k
F ;k ¼ : ð2:14Þ
2 2 r¼0 22r r!2

The final expression obtained for the complete elliptic integral of the first kind showed equivalence with the usual form of
literature, which can be found in Olver et al. [8] and Gradshteyn and Ryzhik [16].

3. Complete elliptic integral of the second kind

The complete elliptic integral of the second kind, can be expressed through a modified Bessel function of the first kind in
the following form,

p  p  Z p 2
2 2 sin hdh
E ;k ¼ F ;k  k pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ; ð3:1Þ
2 2 0 1  k sin h
2 2

p  p  Z 1 Z p=2
2 2
E ;k ¼ F ;k  k ex dx sin hI0 ðkx sin hÞdh: ð3:2Þ
2 2 0 0

With the results of Appendix A, Eq. (A.7) and the Eqs. (2.9) and (2.14), the Eq. (3.2) can be expressed as:
!  
p  k
2 p  p Z 1 kx
2 2
E ;k ¼ 1 F ;k þ k ex dxJ1 i : ð3:3Þ
2 2 2 4 0 2

Also, from Abramowitz and Stegun [10], we have:

 2 X1
 r
1 ð2r þ 2Þ! 14 z2
J 21 ðizÞ ¼  z 3 4
: ð3:4Þ
2 r¼0 ðr þ 2Þðr þ 1Þ r!

Substituting the Eq. (3.4) in the Eq. (3.3), we obtain:

Z   2 1 r
2
1
2 kx k X ð2rÞ!k ð2r þ 1Þ2
ex dxJ1 i ¼ : ð3:5Þ
0 2 4 r¼0 22r r!2 ðr þ 1Þðr þ 2Þ

After replacing Eq. (3.5) in Eq. (3.3), it results,

!
p  2 p  k4 p X1 r
2
k ð2rÞ!k ð2r þ 1Þ2
E ;k ¼ 1 F ;k  2r 2
: ð3:6Þ
2 2 2 8 2 r¼0 2 r! ðr þ 1Þðr þ 2Þ

Substituting the Eq. (2.14) in Eq. (3.6), and grouping the similar terms we find the final representation for the complete ellip-
tic integral of the second kind,
" #
p  p X 1 r
2 2 4
ð2rÞ!k k k ð2r þ 1Þ2
E ;k ¼ 1  : ð3:7Þ
2 2 r¼0 22r r!2 2 8 ðr þ 1Þðr þ 2Þ

The final representation (3.7) is different from the usual form found in the literature, for example in Olver et al. [8], Abra-
mowitz and Stegun [10]:
36 S.L. Bertoli et al. / Applied Mathematics and Computation 243 (2014) 33–43

The results obtained with Eq. (3.7) are compared with the expression found in Abramowitz and Stegun [10]:

k Eðp2 ; kÞ Eðp2 ; kÞ
0.2 1.554968546243 1.554968546243
0.3 1.534833464923 1.534833464923
0.4 1.505941612360 1.505941612360
0.5 1.467462209339 1.467462209339
0.6 1.418083394449 1.418083394449
0.7 1.355661135572 1.355661135572
0.8 1.276349943169 1.276349943169
0.9 1.171697052782 1.171697052782
Abramowitz and Stegun [10]

4. Complete elliptic integral of the third kind

The complete elliptic integral of the third kind can also be solved by means of Bessel function.
p  Z p
2 dh
Z 1
P ; n; k ¼ 2
ex I0 ðkx sin hÞdx: ð4:1Þ
2 0 ð1  n sin hÞ 0

Also, it is possible to express the Eq. (4.1) in the form,

p  Z p
2
Z 1
2
Z 1
P ; n; k ¼ dh etð1n sin hÞ
dt ex I0 ðkx sin hÞdx; ð4:2Þ
2 0 0 0

p  Z 1 Z 1 Z p
2 2
P ; n; k ¼ et dt ex dx etn sin h I0 ðkx sin hÞdh: ð4:3Þ
2 0 0 0

Using the following series representations [10]:

X
1
zr
ez ¼ ; ð4:4Þ
r¼0
r!

X
1
z2r
I0 ðzÞ ¼ : ð4:5Þ
r¼0 22r r!2

And substituting in the previous expressions, we have:

p  Z 1 Z 1 Z p
2 X 1 s 1
ðtn sin hÞ X
2
ðkx sin hÞ
2r
P ; n; k ¼ et dt ex dx dh 2r 2
: ð4:6Þ
2 0 0 0 s¼0
s! r¼0 2 r!

Or equivalently,

p  X 1 2r Z Z 1 Z p
k ns 1 s t 2
2ðsþrÞ
P ; n; k ¼ 2r 2
t e dt x2r ex dx sin hdh: ð4:7Þ
2 r;s¼0 2 r! s! 0 0 0

The integral of the sine function is represented as:

Z p
2
2ðsþrÞ p ½2ðs þ rÞ!
sin hdh ¼ : ð4:8Þ
0 2 22ðsþrÞ ðs þ rÞ!2

Finally, substituting the Eq. (4.8) in the Eq. (4.7) and solving the integrals, we obtain:
p  pX 1
ð2rÞ! ½2ðs þ rÞ! s 2r
P ; n; k ¼ nk : ð4:9Þ
2 2 r;s¼0 24r r!2 22s ðs þ rÞ!2

The final representation is not found in the literature, the usual form can be found for example in Carlitz [11] or Olver et al. [8].
 S.L. Bertoli et al. / Applied Mathematics and Computation 243 (2014) 33–43 37

The results obtained with Eq. (4.9) are compared with the results obtained from the expression found in Carlitz [11]:

k Pðp2 ; 0:5; kÞ Pðp2 ; 0:5; kÞ

0.2 2.24810462594213 2.24810462594213
0.3 2.28335058819339 2.28335058819339
0.4 2.33674613731765 2.33674613731765
0.5 2.41367150420119 2.41367150420119
0.6 2.52390070844927 2.52390070844927
0.7 2.68680199682369 2.68680199682369
0.8 2.94787811582397 2.94787811582397
0.9 3.45910690021046 3.45910690021046
Carlitz [11]

5. Incomplete elliptic integral of the first kind

Using the Bessel function method proposed in the present work, the incomplete elliptic integral of the first kind is
expressed as:
Z 1 Z u
Fðu; kÞ ¼ ex dx I0 ðxk sin hÞdh: ð5:1Þ
0 0
From Abramowitz and Stegun [10] we also can write:
X
1
z2r
I0 ðzÞ ¼ : ð5:2Þ
r¼0 2 ðr!Þ2
2r

So replacing the Eq. (5.2) in the Eq. (5.1), we have:

X
1 2r Z 1 Z u
k 2r
Fðu; kÞ ¼ 2r
ex x2 dx sin hdh: ð5:3Þ
r¼0 2 r!2 0 0

Solving the improper integral, it follows,

X
1 2r Z u
ð2rÞ!k 2r
Fðu; kÞ ¼ 2r 2
sin hdh: ð5:4Þ
r¼0 2 r! 0

The integral of the sine function in Eq. (5.4) is represented by the relationship below from Zhao et al. [17]:
Z
" X 2mþ1
#
u
2r ð2rÞ! 2 ðm!Þ2 sin
r1 2m
u
sin hdh ¼  cos u  u : ð5:5Þ
0 22r r!2 m¼0
ð2m þ 1Þ!

Finally, the incomplete elliptic integral of the first kind is expressed as:
r
2
" #
X
1
ð2rÞ!k X
r1 2m
2 m!2 sin
2mþ1
u
Fðu; kÞ ¼  cos u u : ð5:6Þ
r¼0 22r r!2 m¼0
ð2m þ 1Þ!

The incomplete elliptic integral of the first kind was first solved by Qureshi and Chaudhary [18].
This result also shows the applicability of the proposed method to the incomplete elliptic integral. Qureshi and Chaudhary
[18] presented a solution for the incomplete elliptic of the first kind in terms of series of hypergeometric functions.
The results obtained with Eq. (5.6) are compared with the results obtained from the expression found in Qureshi and
Chaudhary [18]:

Fðp3 ; kÞ Fðp3 ; kÞ
0.2 1.053430587029 1.053430587029
0.3 1.061489706726 1.061489706726
0.4 1.073313629047 1.073313629047
0.5 1.089550670051 1.089550670051
0.6 1.111233322932 1.111233322932
0.7 1.140044752769 1.140044752769
0.8 1.178902299538 1.178902299538
0.9 1.233446325452 1.233446325452
Qureshi and Chaudhary [18]
38 S.L. Bertoli et al. / Applied Mathematics and Computation 243 (2014) 33–43

6. Incomplete elliptic integral of the second kind

Applying the proposed method in relating the incomplete elliptic integral of the second kind with the Bessel function, we
have,
Z 1 Z u
2 2
Eðu; kÞ ¼ Fðu; kÞ  k ex dx sin hI0 ðkx sin hÞdh: ð6:1Þ
0 0

After replacing the Bessel function in the series form in Eq. (6.1) and solving the improper integral, it follows the expression,
" #
X
1 2r Z u
2 ð2rÞ!k 2rþ2
Eðu; kÞ ¼ Fðu; kÞ  k 2r 2
sin hdh: ð6:2Þ
r¼0 2 r! 0

The integral of the sine function was given by Eq. (5.5), then we can write,
Z " #" #
u
2rþ2 ð2rÞ! ð2r þ 1Þ X
r
22m ðm!Þ2 sin
2mþ1
u
sin hdh ¼  2r cos u u : ð6:3Þ
0 2 ðrÞ!2 2ðr þ 1Þ m¼0
ð2m þ 1Þ!
Substituting the expression (6.3) in the Eq. (6.2), we obtain the final expression for the incomplete elliptic integral of the
second kind:
" !
#
1
X r
2
ð2r þ 1Þ 2 X u X
2mþ1 2mþ1
ð2rÞ!k r
22m ðm!Þ2 sin 2 ðm!Þ2 sin
r1 2m
u ð2r þ 1Þ 2
Eðu; kÞ ¼ cos u k  þu 1 k :
r¼0 22r r!2 2ðr þ 1Þ m¼0 ð2m þ 1Þ! m¼0
ð2m þ 1Þ! 2ðr þ 1Þ
ð6:4Þ
This representation for the incomplete elliptic integral of the second kind in terms of power series and trigonometric series is
not found in the literature. Qureshi and Chaudhary [18] proposed an expression in terms of a series of hypergeometric func-
tions for the integral analyzed above, as well as for the incomplete elliptic integral of first kind.
The results obtained from Eq. (6.4) are compared with the expression found in Qureshi and Chaudhary [18]:

k Eðp3 ; kÞ Eðp3 ; kÞ
0.2 1.04566021970563 1.04566021970563
0.3 1.03322345140654 1.03322345140654
0.4 1.02213013278769 1.02213013278769
0.5 1.00755555514447 1.00755555514447
0.6 0.989221593549377 0.989221593549377
0.7 0.966723133094528 0.966723133094528
0.8 0.93945480372495 0.93945480372495
0.9 0.906456986263154 0.906456986263154
Qureshi and Chaudhary [18]

7. Incomplete elliptic integral of the third kind

Applying the present Bessel function method, we have:

Z u Z 1
dh
Pðu; n; kÞ ¼ 2
ex I0 ðkx sin hÞdx: ð7:1Þ
0 ð1  n sin hÞ 0

It is also possible to express the last expression in the form:

Z u Z 1 Z 1
2
Pðu; n; kÞ ¼ dh etð1n sin hÞ
dt ex I0 ðkx sin hÞdx; ð7:2Þ
0 0 0

or,
Z 1 Z 1 Z u
2
Pðu; n; kÞ ¼ et dt ex dx etn sin h I0 ðkx sin hÞdh: ð7:3Þ
0 0 0
2
The functions etn sin h
and I0 ðkx sin hÞ can be represented by series, then it follows,
Z Z Z X s 1 2r
ðtn sin hÞ X
1 1 / 1 2
ðkx sin hÞ
Pð/; n; kÞ ¼ et dt ex dx dh : ð7:4Þ
0 0 0 s¼0
s! r¼0 22r r!2
Or equivalently,
X
1 2r Z 1 Z 1 Z u
k ns 2ðsþrÞ
Pð/; n; kÞ ¼ 2r 2 s!
t s et dt x2r ex dx sin hdh: ð7:5Þ
r;s¼0 2 r! 0 0 0
 S.L. Bertoli et al. / Applied Mathematics and Computation 243 (2014) 33–43 39

Solving the two improper integrals, we obtain the expression,

X
1 2r Z u
ð2rÞ!k ns 2ðsþrÞ
Pð/; n; kÞ ¼ 2r 2
sin hdh: ð7:6Þ
r;s¼0 2 r! 0

The integral in Eq. (7.6) is given by (see Eq. (5.5)):

Z " #" #
u
2ðrþsÞ ½2ðr þ sÞ! X
rþs1 2mþ1
22m ðm!Þ2 sin u
sin hdh ¼  cos u u : ð7:7Þ
0 22ðrþsÞ ðr þ sÞ!2 m¼0
ð2m þ 1Þ!

After replacing the previous expression in the Eq. (7.6), we obtain the following result:
" #" #
X1 2r X 22m ðm!Þ2 sin2mþ1 u
rþs1
ð2rÞ!k ns ½2ðr þ sÞ!
Pðu; n; kÞ ¼  4r 2
cos u u : ð7:8Þ
r;s¼0 2 r! 22s ðr þ sÞ!2 m¼0
ð2m þ 1Þ!

This form for the incomplete elliptic integral of the third kind is not found in the literature. The usual representation is
expressed in terms of a series of hypergeometric functions, or in terms of other series, for example by Karp et al. [19], using
a partial fraction decomposition. The results obtained from Eq. (7.8) are compared with the expression found in [19] Karp
et al. [19]:

k Pðp3 ; 0; 8; kÞ Pðp3 ; 0; 8; kÞ
0.2 1.48453402981057 1.48453402981057
0.3 1.49858743111330 1.49858743111330
0.4 1.5192950589409 1.5192950589409
0.5 1.5479055930718 1.5479055930718
0.6 1.5864286332503 1.5864286332503
0.7 1.63818518991736 1.63818518991736
0.8 1.70903214209174 1.70903214209174
0.9 1.81054697605315 1.81054697605315
Karp et al. [19]

8. Hyperelliptic integrals

The hyperelliptic integral or Abelian integral, in its general form is expressed as in Springer [12]:
Z
dx
pffiffiffiffiffiffiffiffiffiffi ; ð8:1Þ
QðxÞ
where Q ðxÞ is a polynomial of degree > 4. In the following, it will be show how the method used anteriorly can also solve
integrals of this class.
Consider now the integral:
Z
dx
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi : ð8:2Þ
1  xa
This integral is considered a hyperelliptic integral for a > 4.
It can be expressed in terms of a Bessel function,
Z Z Z 1
dx a
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ dx I0 ðsx2 Þes ds: ð8:3Þ
1  xa 0

Replacing the Bessel function by its power series form in Eq. (8.3), we have:
Z Z 1 Z X
1 a 2r
dx ðsx2 Þ
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ es ds dx: ð8:4Þ
1  xa 0 r¼0 r!2
Or,
Z X1 Z 1 Z
dx 1
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ 2r 2
es s2r ds xar dx: ð8:5Þ
1x a
r¼0 2 r! 0

Finally, solving the two integrals in Eq. (8.5), it is obtained the following expression for this particular class of hyperelliptic
integral,
Z X1
dx ð2rÞ! xarþ1
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ 2r 2 ðar þ 1Þ
þ C: ð8:6Þ
1x a
r¼0 2 r!
40 S.L. Bertoli et al. / Applied Mathematics and Computation 243 (2014) 33–43

The integral in Eq. (8.1) is usually solved through the binomial series expansion.
The next integral is a more general class of hyperelliptic integral:
Z 1
du
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi :
0 1  a2
It can be expressed in the following form:
Z 1 Z 1 Z 1
du
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ ex I0 ðaxÞdxdu; ð8:7Þ
0 1  a2 0 0

where,
Y
n
a2 ¼ ð1  xi uÞ: ð8:8Þ
i¼1

Following the proposed method:

Z 1 X1 Z 1 Z 1
du 1
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ 2r 2
ex x2r dx a2r du: ð8:9Þ
0 1a 2 2 r! 0 0
r¼0

Solving the improper integral, it results:

Z 1 X1 Z 1
du ð2rÞ!
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ 2r 2
a2r du: ð8:10Þ
2
0 1a r¼0 2 r! 0

Substituting the expression (8.8) in the Eq. (8.10), it shows:

Z 1 X1 Z 1 Y
n
du ð2rÞ!
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ 2r 2
ð1  xi uÞr du: ð8:11Þ
0 1a 2 2 r! 0
r¼0 i¼1

Now,
Z 1 Y
n
Cð2Þ
ð1  xi uÞr du ¼
ðnÞ
F ð1; r; . . . ; r; 2; x1 ; . . . ; xn Þ; ð8:12Þ
0 i¼1
Cð1ÞCð1Þ D
ðnÞ
where F D is a Lauricella function.
Substituting the expression (8.12) in Eq. (8.11) the following result is obtained:
Z 1 X1
du ð2rÞ! ðnÞ
pffiffiffiffiffiffiffiffiffiffiffiffiffiffi ¼ F ð1; r; . . . ; r; 2; x1 ; . . . ; xn Þ:
2r 2 D
ð8:13Þ
2
0 1a r¼0 2 r!

This expression for the hyperelliptic integral in terms of Lauricella functions can also be solved by the binomial expansion,
different from the way developed anteriorly.

9. Epstein–Hubbell integral

The Epstein–Hubbell integral contains the parameters j and k, represented in the form [16,20]:
Z p
2 j12 2
Xj ðkÞ ¼ ð1  k cos hÞ dhk < 1; j ¼ 0; 1; 2; 3 . . . ð9:1Þ
0
The solution of this integral is usually obtained through the binomial expansion.
X1 2n
2 j ðjÞn k
ð1  k cos hÞ ¼ cosn h j ¼ 1; 2; 3 . . . ; ð9:2Þ
n¼0
n!
where ðjÞn is the Pochhammer symbol, which can be found in Gradshteyn and Ryzhik [16].

9.1. Epstein–Hubbell integral for j ¼ 0

For the case of j ¼ 0, we have from the expression (9.4):

Z p
dh
X0 ðkÞ ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi : ð9:3Þ
0 2
ð1  k cos hÞ
It can be rewritten the integral (9.6) as it was done in terms of the modified Bessel function:
Z Z p  pffiffiffiffiffiffiffiffiffiffiffi X Z
1 1
ð2rÞ! 2r p
X0 ðkÞ ¼ ex dx I0 xk cos h dh ¼ 2r 2
k cosr hdh: ð9:4Þ
0 0 r¼0 2 ðr!Þ 0
 S.L. Bertoli et al. / Applied Mathematics and Computation 243 (2014) 33–43 41

The integral of the cosine function in the last equation is represented in the form [21]:
8
Z p >
> 0; if r odd
<
r
cos hdh ¼ : ð9:5Þ
>
>
0
: p 2r r!
; if r even
ð2r Þ!2
From Eq. (9.8) it is know that only the r even terms will contribute for X0 ðkÞ, then:
X
1
ð2rÞ! r! 2r
X0 ðkÞ ¼ p r
r  2 k : ð9:6Þ
r¼0 22r ðr!Þ2 2 2
!
r–odd

Now, making the following change r ! 2r:

X
1
ð4rÞ! 4r
X0 ðkÞ ¼ p k : ð9:7Þ
r¼0 26r ð2rÞ!ðr!Þ2
The final form given by the Eq. (9.10) can be found Srivastava [22].

9.2. Epstein–Hubbell integral for j ¼ 1; 2; 3 . . .

For the Epstein–Hubbell integral with argument j > 0 it follows the expression,
Z p
2 j12 2
Xj ðkÞ ¼ ð1  k cos hÞ dhk < 1; j ¼ 1; 2; 3 . . . : ð9:8Þ
0

It can be represented in the following way,

Z p 2 j
ð1  k cos hÞ
Xj ðkÞ ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi dh: ð9:9Þ
0 2
ð1  k cos hÞ

Relating it to the modified Bessel function of the first kind, it is obtained:

Z 1 Z p pffiffiffiffiffiffiffiffiffiffiffi 2 j
Xj ðkÞ ¼ ex dx I0 ðxk cos hÞð1  k cos hÞ dh: ð9:10Þ
0 0

Now, substituting the binomial expansion found in expression (9.13):

X 2n 1 2r Z p
1
ðjÞn k X ð2rÞ!k
Xj ðkÞ ¼ cosnþr hdh: ð9:11Þ
n¼0
n! r¼0 22r r!2 0

The integral of the cosine function is from Eq. (9.8):

8
Z p >
> 0; if n þ r ¼ odd
<
cosrþn hdh ¼ : ð9:12Þ
0 >
> pðrþnÞ!
: 2rþn ; if n þ r ¼ even
ðrþn
2 Þ
!2

Now, following the same procedure as it was done for j ¼ 0, it results:

X1
ðjÞn ð2rÞ! ðr þ nÞ! 2ðrþnÞ
Xj ðkÞ ¼ p n 3r 2   2 k : ð9:13Þ
n;r¼0
n!2 2 r! rþn
!
nþr¼ev en 2

There are two conditions for ðn þ r) to be an even number: (a) when ðn; rÞ are even numbers; (b) when ðn; rÞ are odd
numbers.
For case (a) is made the following substitution in Eq. (9.16): r ! 2r and n ! 2n, which results:
X
1
ðjÞ2n ð4rÞ! ð2r þ 2nÞ! 4ðrþnÞ
Xj ðkÞ ¼ p 2n
k : ð9:14Þ
n;r¼0 ð2nÞ!2 26r ð2rÞ!2 22n ðr þ nÞ!2

For case (b) is made the following substitution in Eq. (9.16): r ! 2r þ 1 and n ! 2n þ 1, then we have:
X 1
ðjÞ2nþ1 ð4r þ 2Þ! ð2r þ 2n þ 2Þ! 4ðrþnþ1Þ
Xj ðkÞ ¼ p 6r 2 2nþ4
k : ð9:15Þ
n;r¼0
ð2n þ 1Þ! 2 ð2r þ 1Þ! 2 ðr þ n þ 1Þ!2

Finally adding the expressions (9.14) and (9.15), it is obtained the solution for the Epstein–Hubbel integral:
" 4ðrþnÞ
#" " 4
##
X 1
ð4rÞ!ð2r þ 2nÞ!k ðjÞ2n 1 ðjÞ2nþ1 ð4r þ 1Þð2r þ 2n þ 1Þk
Xj ðkÞ ¼ p 2 2 6rþ2n
þ : ð9:16Þ
n;r¼0 ð2rÞ! ðr þ nÞ! 2
ð2nÞ! 4 ð2n þ 1Þ! ð2r þ 1Þðr þ n þ 1Þ
42 S.L. Bertoli et al. / Applied Mathematics and Computation 243 (2014) 33–43

The expression (9.16) is the final form for the Epstein–Hubbell integral with argument j > 0. This form of representation is
not found in the literature. The usual forms may be found, for example, in Al-Zamel and Kalla [23], Cengiz [21], Gradshteyn
and Ryzhik [16], Kalla and Montanus [24], Kalla et al. [25], Kalla and Tuan [26], Srivastava [22] and Galué [5].
The comparison with the expression in Al-Zamel and Kalla [23], it follows:

j Xj ð0; 5Þ Xj ð0; 5Þ
1 3.337745013 3.337745013
2 3.616446839 3.616446839
3 1.8105469760 1.8105469760
Al-Zamel and Kalla [23]

10. Conclusions

This paper proposed a new method for solving elliptic integrals and some of its generalizations, based on the modified
Bessel function of first kind and order zero. By the present method it was possible to obtain various solutions representations
similar to those found in the literature and also some integral representations obtained in unusual ways.
The proposed method suggests a broad applicability that may be more general than was exploited.

Acknowledgments

The authors are thankful to the Department of Chemical Engineering at the University of Blumenau. The authors also
thanks to Marta H. Caetano and Luiz H. da Silva for the English revision of the manuscript.

R p=2 2
Appendix A. Alternative representation for the integral: 0
sin hI 0 ðkx sin hÞdh

Starting from Abramowitz and Stegun [10], it is know that,

Z p
J 0 ð2z sin tÞ cos 2t dt ¼ pJ 21 ðzÞ; ðA:1Þ
0

also, the following trigonometric identity is valid,

2
cos 2t ¼ 1  2 sin t; ðA:2Þ
with the two last relations, it can be written,
Z p Z p
1 p
J 0 ð2z sin tÞsen2 tdt ¼ J 0 ð2z sin tÞdt  J 21 ðzÞ: ðA:3Þ
0 2 0 2
The functions sen2 t and J 0 ð2zsen tÞ are symmetrical to each other in the interval ½0; p=2 and ½p=2; p, then using this property,
it can be established,
Z p Z p
2 1 2 p
J 0 ð2z sin tÞsen2 tdt ¼ J 0 ð2z sin tÞdt  J 21 ðzÞ: ðA:4Þ
0 2 0 4
Now making the following substitution z ¼ iz, we have from Eq. (A.4),
Z p Z p
2
2 1 2 p
J 0 ð2iz sin tÞ sin tdt ¼ J 0 ð2iz sin tÞdt  J 21 ðizÞ; ðA:5Þ
0 2 0 4
Z p Z p
2
2 1 2 p
¼ I0 ð2z sin tÞ sin tdt ¼ I0 ð2z sin tÞdt  J 21 ðizÞ: ðA:6Þ
0 2 0 4
Finally, substituting z ¼ kx=2, it results,
Z p Z p  
2 1 2 p kx
I0 ðkx sin tÞsen2 tdt ¼ I0 ðkx sin tÞdt  J 21 i : ðA:7Þ
0 2 0 4 2

References

[1] S. Datta, Electric and magnetic fields from a circular coil using elliptic integrals, Phys. Educ. (2007) 203–212.
[2] F.L. Derek, Elliptic Functions and Applications, Springer, 1989.
[3] J.T. Sommerfeld, M.P. Stallybrassa, Elliptic integral solutions for fluid discharge rates from punctured horizontal cylindrical vessels, J. Loss Prev. Process
Ind. 6 (1993) 11–13.
 S.L. Bertoli et al. / Applied Mathematics and Computation 243 (2014) 33–43 43

[4] P.W. Hart, J.T. Sommerfeld, Elliptic integral solutions for the kinetics of autocatalytic termolecular reactions with reactive intermediate species, Chem.
Eng. J. 66 (1997) 231–234.
[5] L. Galué, Asymptotic expansions for generalized elliptic-type integrals, Radiat. Phys. Chem. 58 (2000) 307–313.
[6] I.M. Mladenov, New geometrical applications of the elliptic integrals: the Mylar balloon, J. Nonlinear Math. Phys. (2004) 55–65.
[7] R. Courant, Differential and Integral Calculus, vol. 1, Wiley-Interscience, 1988.
[8] F.W.J. Olver, D.W. Lozier, R.F. Boisvert, C.W. Clark (Eds.), NIST Handbook of Mathematical Functions, Cambridge University Press, 2010.
[9] A. Jeffrey, H. Dai, Handbook of Mathematical Formulas and Integrals, Academic Press, New York, 2003.
[10] M. Abramowitz, I. Stegun, Handbook of Mathematical Functions, Dover, New York, 1972.
[11] L. Carlitz, On the complete elliptic integral of the third kind, Bollettino dell’Unione Matematica Italiana, serie 3 (15) (1960) 373–378.
[12] G. Springer, Introduction to Riemann Surfaces, American Mathematical Society, 2002.
[13] S.L. Kalla, S. Conde, J.H. Hubbell, Some results on generalized elliptic-type integrals, Appl. Anal. 22 (1986) 273–287.
[14] I.N. Bronshtein, K.A. Semendyayev, G. Musiol, H. Muehlig, Handbook of Mathematics, Springer, Berlin, 2004.
[15] F. Bowman, Introduction to Bessel Functions, Dover Publications, New York, 1958.
[16] I.S. Gradshteyn, I.M. Ryzhik, Table of Series, Integrals and Products, Academic Press, New York, 2007.
[17] J. Zhao, J. Jia, X. He, H. Wang, Post-buckling and snap-through behavior of inclined slender beams, J. Appl. Mech. (2008) 1–7.
[18] M.I. Qureshi, M.P. Chaudhary, Ramanujan’s entries and solutions of incomplete elliptic integrals in terms of hypergeometric functions, Int. J. Open Prob.
Comput. Math. 1 (2009) 85–97.
[19] D. Karp, A. Savenkova, S.M. Sitnik, Series expansions for the third incomplete elliptic integral via partial fraction decompositions, J. Comput. Appl. Math.
207 (2007) 331–337.
[20] L.F. Epstein, J.H. Hubbell, Evaluation of generalized elliptic-type integral, J. Res. N. B. S. 67 (1963) 1–17.
[21] A. Cengiz, A study on the Epstein–Hubbell generalized elliptic-type integral using residue theory, Appl. Math. Comput. (1997) 19–26.
[22] R. Srivastava, A note on the Epstein–Hubbell generalized elliptic-type integral, Appl. Math. Comput. (1995) 255–262.
[23] A. Al-Zamel, S. Kalla, Epstein–Hubbell elliptic type integral and its generalizations, Appl. Math. Comput. (1996) 9–32.
[24] S.L. Kalla, J.M.C. Montanus, A series approximation for disk galaxies by means of the Epstein–Hubbell integral, Math. Comput. Model. 40 (2004) 611–
626.
[25] S.L. Kalla, C. Leubner, J.H. Hubbell, Further results on generalized elliptic-type integrals, Appl. Anal. 25 (1987) 269–274.
[26] S.L. Kalla, V.K. Tuan, Asymptotic formulas for generalized elliptic-type integrals, Comput. Math. Appl. 32 (1966) 49–55.