Volume 2010, Article ID 309503,26pages doi:10.1155/2010/309503
Research Article
Extensions of Certain Classical Summation Theorems for the Series
2F
1,
3F
2, and
4F
3with Applications in Ramanujan’s Summations
Yong Sup Kim,
1Medhat A. Rakha,
2, 3and Arjun K. Rathie
41Department of Mathematics Education, Wonkwang University, Iksan 570-749, Republic of Korea
2Mathematics Department, College of Science, Suez Canal University, Ismailia 41522, Egypt
3Department of Mathematics and Statistics, College of Science, Sultan Qaboos University, P.O. Box 36, Muscat, Alkhod 123, Oman
4Vedant College of Engineering and Technology, Village-Tulsi, Post-Jakhmund, Bundi, Rajasthan State 323021, India
Correspondence should be addressed to Medhat A. Rakha,[email protected] Received 20 May 2010; Revised 7 September 2010; Accepted 23 September 2010 Academic Editor: Teodor Bulboac˘a
Copyrightq2010 Yong Sup Kim et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
Motivated by the extension of classical Gauss’s summation theorem for the series2F1given in the literature, the authors aim at presenting the extensions of various other classical summation theorems such as those of Kummer, Gauss’s second, and Bailey for the series2F1, Watson, Dixon and Whipple for the series3F2, and a few other hypergeometric identities for the series3F2and4F3. As applications, certain very interesting summations due to Ramanujan have been generalized.
The results derived in this paper are simple, interesting, easily established, and may be useful.
1. Introduction
In 1812, Gauss1systematically discussed the series ∞
n0
anbn cn
zn
n! 1a·b
1·czaa1bb1
1·2cc1 z2· · ·, 1.1 whereλndenotes the Pochhammer symbol definedforλ∈Cby
λn:
⎧⎨
⎩
1 n0
λλ1· · ·λn−1 n∈N:{1,2,3, . . .}. 1.2
It is noted that the series 1.1 and its natural generalization PFq in 1.6 are of great importance to mathematicians and physicists. This series1.1has been known as the Gauss series or the ordinary hypergeometric series and may be regarded as a generalization of the elementary geometric series. In fact1.1reduces to the geometric series in two cases, when a candb 1 also whenb canda 1. The series1.1is represented by the notation
2F1a, b;c;zor
2F1
⎡
⎢⎢
⎣ a, b
; z c
⎤
⎥⎥
⎦, 1.3
which is usually referred to as Gauss hypergeometric function. In1.1, the three elementsa, b, andcare described as the parameters of the series, andzis called the variable of the series.
All four of these quantities may be real or complex with an exception thatcis neither zero nor a negative integer. Also, in1.1, it is easy to see that if any one of the numerator parameters aorbor both is a negative integer, then the series reduces to a polynomials, that is, the series terminates.
The series1.1is absolutely convergent within the unit circle when|z|< 1 provided thatc /0,−1,−2, . . .. Also when|z|1, the series is absolutely convergent ifRc−a−b>0, conditionally convergent if−1<Rc−a−b≤0,z /1 and divergent ifRc−a−b≤ −1.
Further, if in1.1, we replacezbyz/band letb → ∞, thenbnzn/bn → zn, and we arrive to the following Kummer’s series
∞ n0
an cn
zn
n! 1 a
1·cz aa1
1·2cc1z2· · ·. 1.4 This series is absolutely convergent for all values of a, c, and z, real or complex, excludingc0,−1,−2, . . .and is represented by the notation1F1a;c;zor
1F1
⎡
⎢⎢
⎣ a,
; z c
⎤
⎥⎥
⎦, 1.5
which is called a confluent hypergeometric function.
Gauss hypergeometric function 2F1 and its confluent case 1F1 form the core special functions and include, as their special cases, most of the commonly used functions. Thus2F1
includes, as its special cases, Legendre function, the incomplete beta function, the complete elliptic functions of first and second kinds, and most of the classical orthogonal polynomials.
On the other hand, the confluent hypergeometric function includes, as its special cases, Bessel functions, parabolic cylindrical functions, and Coulomb wave function.
Also, the Whittaker functions are slightly modified forms of confluent hypergeometric functions. On account of their usefulness, the functions 2F1 and 1F1 have already been explored to considerable extent by a number of eminent mathematicians, for example, C. F.
Gauss, E. E. Kummer, S. Pincherle, H. Mellin, E. W. Barnes, L. J. Slater, Y. L. Luke, A. Erd´elyi, and H. Exton.
A natural generalization of2F1 is the generalized hypergeometric seriespFq defined by
pFq
⎡
⎢⎢
⎣
a1 · · · ap
; z b1 · · · bq
⎤
⎥⎥
⎦∞
n0
a1n· · · ap
n
b1n· · · bq
n
zn
n!. 1.6
The series1.6is convergent for all|z|<∞ifp≤qand for|z|<1 ifpq1 while it is divergent for allz,z /0 ifp > q1. When|z|1 withpq1, the series1.6converges absolutely if
R
⎛
⎝q
j1
bj− p j1
aj
⎞
⎠>0, 1.7
conditionally convergent if
−1<R
⎛
⎝q
j1
bj− p j1
aj
⎞
⎠≤0, z /1 1.8
and divergent if
R
⎛
⎝q
j1
bj− p j1
aj
⎞
⎠≤ −1. 1.9
It should be remarked here that whenever hypergeometric and generalized hyper- geometric functions can be summed to be expressed in terms of Gamma functions, the results are very important from a theoretical and an applicable point of view. Only a few summation theorems are available in the literature and it is well known that the classical summation theorems such as of Gauss, Gauss’s second, Kummer, and Bailey for the series
2F1, and Watson, Dixon, and Whipple for the series 3F2 play an important role in the theory of generalized hypergeometric series. It has been pointed out by Berndt2, that very interesting summations due to Ramanujan can be obtained quite simply by employing the above mentioned classical summation theorems. Also, in a well-known paper by Bailey3, a large number of very interesting results involving products of generalized hypergeometric series have been developed. In4a generalization of Kummer’s second theorem was given from which the well-known Preece identity and a well-known quadratic transformation due to Kummer were derived.
2. Known Classical Summation Theorems
As already mentioned that the classical summation theorems such as those of Gauss, Kummer, Gauss’s second, and Bailey for the series2F1 and Watson, Dixon, and Whipple for the series3F2 play an important role in the theory of hypergeometric series. These theorems are included in this section so that the paper may be self-contained.
In this section, we will mention classical summation theorems for the series2F1and
3F2. These are the following.
Gauss theorem5:
2F1
⎡
⎢⎢
⎣ a, b
; 1 c
⎤
⎥⎥
⎦ ΓcΓc−a−b
Γc−aΓc−b 2.1
providedRc−a−b>0.
Kummer theorem5:
2F1
⎡
⎢⎢
⎣
a, b
; −1 1a−b
⎤
⎥⎥
⎦ Γ1a−bΓ1 1/2a
Γ1 1/2a−bΓ1a. 2.2
Gauss’s second theorem5:
2F1
⎡
⎢⎢
⎢⎢
⎣
a, b
; 1 1 2
2ab1
⎤
⎥⎥
⎥⎥
⎦ Γ1/2Γ1/2a 1/2b 1/2
Γ1/2a 1/2Γ1/2b 1/2. 2.3
Bailey theorem5:
2F1
⎡
⎢⎢
⎢⎣
a, 1−a
; 1 2 c
⎤
⎥⎥
⎥⎦ Γ1/2cΓ1/2c 1/2
Γ1/2c 1/2aΓ1/2c−1/2a 1/2. 2.4
Watson theorem5:
3F2
⎡
⎢⎢
⎢⎣
a, b, c
; 1 1
2ab1, 2c
⎤
⎥⎥
⎥⎦
Γ1/2Γc 1/2Γ1/2a 1/2b 1/2Γc−1/2a−1/2b 1/2
Γ1/2a 1/2Γ1/2b 1/2Γc−1/2a 1/2Γc−1/2b 1/2 2.5
providedR2c−a−b>−1.
Dixon theorem5:
3F2
⎡
⎢⎢
⎣
a, b, c
; 1 1a−b, 1a−c
⎤
⎥⎥
⎦
Γ1 1/2aΓ1a−bΓ1a−cΓ1 1/2a−b−c Γ1aΓ1 1/2a−bΓ1 1/2a−cΓ1a−b−c
2.6
providedRa−2b−2c>−2.
Whipple theorem5:
3F2
⎡
⎢⎢
⎣
a, b, c
; 1 e, f
⎤
⎥⎥
⎦
πΓeΓ f 22c−1Γ1/2a 1/2eΓ
1/2a 1/2f
Γ1/2b 1/2eΓ
1/2b 1/2f 2.7
providedRc>0 andRef−a−b−c>0 withab1 andef2c1.
Other hypergeometric identities5:
3F2
⎡
⎢⎢
⎢⎢
⎣
a, 11
2a, b
; 1 1
2a, 1a−b
⎤
⎥⎥
⎥⎥
⎦ Γ1a−bΓ1/2a 1/2
Γ1aΓ1/2a−b 1/2, 2.8
4F3
⎡
⎢⎢
⎢⎢
⎣
a, 11
2a, b, c
; 1 1
2a, 1a−b, 1a−c
⎤
⎥⎥
⎥⎥
⎦
Γ1a−bΓ1a−cΓ1/2a 1/2Γ1/2a−b−c 1/2 Γ1aΓ1a−b−cΓ1/2a−b 1/2Γ1/2a−c 1/2
2.9
providedRa−2b−2c>−1.
It is not out of place to mention here that Ramanujan independently discovered a great number of the primary classical summation theorems in the theory of hypergeometric series. In particular, he rediscovered well-known summation theorems of Gauss, Kummer, Dougall, Dixon, Saalsch ¨utz, and Thomae as well as special cases of the well-known Whipple’s transformation. Unfortunately, Ramanujan left us little knowledge as to know how he made his beautiful discoveries about hypergeometric series.
3. Ramanujan’s Summations
The classical summation theorems mentioned in Section 2 have wide applications in the theory of generalized hypergeometric series and other connected areas. It has been pointed out by Berndt2that a large number of very interesting summations due to Ramanujan can be obtained quite simply by employing the above mentioned theorems.
We now mention here certain very interesting summations by Ramanujan2.
iForRx>1/2,
1−x−1
x1x−1x−2
x1x2− · · · x
2x−1, 3.1
1− 1
2 2
1·3
2·4 2
−
1·3·5 2·4·6
2 · · ·
√π
√2Γ23/4. 3.2
iiForRx>0,
1x−1
x1x−1x−2
x1x2· · · 22x−1Γ2x1
Γ2x1 , 3.3
1 1 2
1 2
2 1
22 1·3
2·4 2
1 23
1·3·5 2·4·6
2 · · ·
√π
Γ23/4. 3.4
iiiForRx>0,
1−1 3
x−1 x1 1
5
x−1x−2
x1x2− · · · 24xΓ4x1
4xΓ22x1. 3.5
ivForRx>1/4,
1 x−12 x12
x−1x−2 x1x2
2
· · · 2x 4x−1
Γ4x1Γ4x1
Γ42x1 . 3.6
vForRx>1,
1−3x−1
x15x−1x−2
x1x2− · · ·0, 3.7
11 5
1 2
2 1
9 1·3
2·4 2
· · · π2 4Γ43/4, 1 1
52 1
2
1 92
1·3 2·4
· · · π5/2 8√
2Γ23/4, 1
1 2
2
1·3 2·4
2
· · · π Γ43/4.
3.8
viForRx<2/3,
1x 1!
3
xx1 2!
3
· · · 6 sinπx/2sinπxΓ31/2x1
π2x2Γ3/2x112 cosπx . 3.9
viiForRx>1/2,
13x−1
x15x−1x−2
x1x2· · ·x. 3.10
viiiForRx>1/2,
13
x−1 x1
2 5
x−1x−2 x1x2
2
· · · x2
2x−1. 3.11
We now come to the derivations of these summation in brief.
It is easy to see that the series3.1corresponds to
2F1
⎡
⎢⎢
⎣
1, 1−x
; 1 1x
⎤
⎥⎥
⎦ 3.12
which is a special case of Gauss’s summation theorem2.1fora1,b1−xandc1x.
The series3.2corresponds to
2F1
⎡
⎢⎢
⎢⎣ 1 2, 1
2
; −1 1
⎤
⎥⎥
⎥⎦ 3.13
which is a special case of Kummer’s summation theorem2.2forab1/2. Similarly the series3.3corresponds to
2F1
⎡
⎢⎢
⎣
1, 1−x
; −1 1x
⎤
⎥⎥
⎦ 3.14
which is a special case of Kummer’s summation theorem2.2fora1,b1−x.
The series3.4corresponds to
2F1
⎡
⎢⎢
⎢⎢
⎣ 1 2, 1
2
; 1 2 1
⎤
⎥⎥
⎥⎥
⎦ 3.15
which is a special case of Gauss’s second summation theorem2.3forab1/2 or Bailey’s summation theorem2.4fora1/2 andc1.
Also, it can easily be seen that the series 3.5 to 3.9 correspond to each of the following series:
3F2
⎡
⎢⎢
⎢⎢
⎣ 1, 1
2, 1−x
; 1 3
2, 1x
⎤
⎥⎥
⎥⎥
⎦, 3F2
⎡
⎢⎣
1, 1−x, 1−x
; 1 1x, 1x
⎤
⎥⎦, 3F2
⎡
⎢⎢
⎢⎢
⎣ 1, 3
2, 1−x
; 1 1
2, 1x
⎤
⎥⎥
⎥⎥
⎦,
3F2
⎡
⎢⎢
⎢⎢
⎣ 1 2, 1
2, 1 4
; 1 1, 5
4
⎤
⎥⎥
⎥⎥
⎦, 3F2
⎡
⎢⎢
⎢⎢
⎣ 1 2, 1
4, 1 4
; 1 5
4, 5 4
⎤
⎥⎥
⎥⎥
⎦,
3F2
⎡
⎢⎢
⎢⎣ 1 2, 1
2, 1 2
; 1 1, 1
⎤
⎥⎥
⎥⎦, 3F2
⎡
⎢⎢
⎣
x, x, x
; 1 1, 1
⎤
⎥⎥
⎦,
3.16 which are special cases of classical Dixon’s theorem2.6foria1,b1/2,c1−x,ii a 1,b c 1−x,iiia 1,b 3/2,c 1−x,iva b 1/2,c 1/4,va 1/2, bc1/4,viabc1/2, andviiabcx, respectively.
The series3.10which corresponds to
3F2
⎡
⎢⎢
⎢⎢
⎣ 1, 3
2, 1−x
; −1 1
2, 1x
⎤
⎥⎥
⎥⎥
⎦ 3.17
is a special case of2.8fora1,b1−x, and the series3.11which corresponds to
4F3
⎡
⎢⎢
⎢⎢
⎣ 1, 3
2, 1−x, 1−x
; 1 1
2, 1x, 1x
⎤
⎥⎥
⎥⎥
⎦ 3.18
is a special case of2.9fora1,bc1−x.
Thus by evaluating the hypergeometric series by respective summation theorems, we easily obtain the right hand side of the Ramanujan’s summations.
Recently good progress has been done in the direction of generalizing the above- mentioned classical summation theorems 2.2–2.7 see 6. In fact, in a series of three papers by Lavoie et al. 7–9, a large number of very interesting contiguous results of the above mentioned classical summation theorems2.2–2.7 are given. In these papers, the authors have obtained explicit expressions of
2F1
⎡
⎢⎢
⎣
a, b
; −1 1a−bi
⎤
⎥⎥
⎦, 3.19
2F1
⎡
⎢⎢
⎢⎢
⎣
a, b
; 1 1 2
2abi1
⎤
⎥⎥
⎥⎥
⎦, 3.20
2F1
⎡
⎢⎢
⎢⎣
a, 1−ai
; 1 2 c
⎤
⎥⎥
⎥⎦ 3.21
each fori0,±1,±2,±3,±4,±5, and
3F2
⎡
⎢⎢
⎢⎣
a, b, c
; 1 1
2abi1, 2cj
⎤
⎥⎥
⎥⎦ 3.22
fori, j0,±1,±2
3F2
⎡
⎢⎣
a, b, c
; 1 1a−bi, 1a−cij
⎤
⎥⎦ 3.23
fori−3,−2,−1,0,1,2;j 0,1,2,3, and
3F2
⎡
⎢⎣ a, b, c
; 1 e, f
⎤
⎥⎦ 3.24
forab1ij,ef 2c1ifori, j0,±1,±2,±3.
Notice that, if we denote3.23byfi,j, the natural symmetry
fi,ja, b, c fij,−ja, c, b 3.25 makes it possible to extend the result toj −1,−2,−3.
It is very interesting to mention here that, in order to complete the results3.23of 7×7 matrix, very recently Choi10obtained the remaining ten results.
For i 0, the results 3.19, 3.20, and 3.21 reduce to 2.2, 2.3, and 2.4, respectively, and for i j 0, the results 3.22, 3.23, and3.24 reduce to 2.5,2.6, and2.7, respectively.
On the other hand the following very interesting result for the series3F2written here in a slightly different formis given in the literaturee.g., see11
3F2
⎡
⎢⎣
a, b, d1
; 1 c1, d
⎤
⎥⎦ Γc1Γc−a−b Γc−a1Γc−b1
c−a−b ab d
3.26
providedRc−a−b>0 andRd>0.
Ford c, we get Gauss’s summation theorem2.1. Thus3.26may be regarded as the extension of Gauss’s summation theorem2.1.
Miller12very recently rederived the result3.26and obtained a reduction formula for the Kamp´e de F´eriet function. For comment of Miller’s paper12, see a recent paper by Kim and Rathie13.
The aim of this research paper is to establish the extensions of the above mentioned classical summation theorem 2.2 to 2.9. In the end, as an application, certain very interesting summations, which generalize summations due to Ramanujan have been obtained.
The results are derived with the help of contiguous results of the above mentioned classical summation theorems obtained in a series of three research papers by Lavoie et al.
7–9.
The results derived in this paper are simple, interesting, easily established, and may be useful.
4. Results Required
The following summation formulas which are special cases of the results 2.2 to 2.7 obtained earlier by Lavoie et al.7–9will be required in our present investigations.
iContiguous Kummer’s theorem9:
2F1
⎡
⎢⎢
⎣
a, b
; −1 2a−b
⎤
⎥⎥
⎦
Γ1/2Γ2a−b 2a1−b
×
1
Γ1/2a 1/2Γ1/2a−b1− 1
Γ1/2aΓ1/2a−b 3/2
,
2F1
⎡
⎢⎢
⎣
a, b
; −1 3a−b
⎤
⎥⎥
⎦
Γ1/2Γ3a−b 2a1−b2−b
×
1a−b
Γ1/2a 1/2Γ1/2a−b2− 2
Γ1/2aΓ1/2a−b 3/2
.
4.1
iiContiguous Gauss’s Second theorem9:
2F1
⎡
⎢⎢
⎢⎢
⎣
a, b
; 1 1 2
2ab3
⎤
⎥⎥
⎥⎥
⎦
Γ1/2Γ1/2a 1/2b 3/2Γ1/2a−1/2b−1/2
Γ1/2a−1/2b 3/2
×
1/2ab−1
Γ1/2a 1/2Γ1/2b 1/2 − 2
Γ1/2aΓ1/2a
.
4.2
iiiContiguous Bailey’s theorem9:
2F1
⎡
⎢⎢
⎢⎣
a, 3−a
; 1 2 c
⎤
⎥⎥
⎥⎦
Γ1/2ΓcΓ1−a 2c−3Γ3−a ×
c−2
Γ1/2c−1/2a 1/2Γ1/2c 1/2a−1
− 2
Γ1/2c−1/2aΓ1/2c 1/2a−3/2
. 4.3
ivContiguous Watson’s theorem7:
3F2
⎡
⎢⎢
⎢⎣
a, b, c
; 1 1
2ab1, 2c1
⎤
⎥⎥
⎥⎦
2ab−2Γc 1/2Γ1/2a 1/2b 1/2Γc−1/2a−1/2b 1/2
Γ1/2ΓaΓb
×
Γ1/2aΓ1/2b
Γc−1/2a 1/2Γc−1/2b 1/2− Γ1/2a 1/2Γ1/2b 1/2
Γc−1/2a1Γc−1/2b1 4.4
provided thatR2c−a−b>−1.
3F2
⎡
⎢⎢
⎢⎣
a, b, c
; 1 1
2ab3, 2c
⎤
⎥⎥
⎥⎦
2ab1Γc 1/2Γ1/2a 1/2b 3/2Γc−1/2a−1/2b−1/2 a−b−1a−b1Γ1/2ΓaΓb
×
a2c−a b2c−b−2c1 8
Γ1/2aΓ1/2b
Γc−1/2a 1/2Γc−1/2b 1/2
−Γ1/2a 1/2Γ1/2b 1/2
Γc−1/2aΓc−1/2b
4.5
provided thatR2c−a−b>1.
3F2
⎡
⎢⎢
⎢⎣
a, b, c
; 1 1
2ab3, 2c−1
⎤
⎥⎥
⎥⎦
2ab−1Γc−1/2Γ1/2a 1/2b 3/2Γc−1/2a−1/2b−1/2 a−b−1a−b1Γ1/2ΓaΓb
×
ab−1Γ1/2aΓ1/2b
Γc−1/2a−1/2Γc−1/2b−1/2
−4c−a−b−3Γ1/2a 1/2Γ1/2b 1/2 Γc−1/2aΓc−1/2b
4.6
provided thatR2c−a−b>1.
vContiguous Dixon’s theorem8:
3F2
⎡
⎢⎢
⎣
a, b, c
; 1 2a−b, 2a−c
⎤
⎥⎥
⎦ 2−2c1Γ2a−bΓ2a−c b−1c−1Γa−2c2Γa−b−c2
×
Γ1/2a−c 3/2Γ1/2a−b−c2 Γ1/2a 1/2Γ1/2a−b1
−Γ1/2a−c1Γ1/2a−b−c 5/2 Γ1/2aΓ1/2a−b 3/2
4.7
provided thatRa−2b−2c>−4.
3F2
⎡
⎢⎢
⎣
a, b, c
; 1 2a−b, 1a−c
⎤
⎥⎥
⎦ 2−2b1Γ1a−cΓ2a−b b−1Γa−2b2Γa−b−c2
×
Γ1/2a−b1Γ1/2a−b−c 3/2
Γ1/2aΓ1/2a−c 1/2
−Γ1/2a−b 3/2Γ1/2a−b−c2 Γ1/2a 1/2Γ1/2a−c1
4.8
provided thatRa−2b−2c>−3.
3F2
⎡
⎢⎢
⎣
a, b, c
; 1 2a−c, 3a−b
⎤
⎥⎥
⎦
2−2b2Γ2a−cΓ3a−b
b−1b−2c−1Γa−2b3Γa−b−c3
×
a−2c−b3Γ1/2a−b2Γ1/2a−b−c 5/2
Γ1/2aΓ1/2a−c 3/2
−a−b1Γ1/2a−b 3/2Γ1/2a−b−c3 Γ1/2a 1/2Γ1/2a−c1
4.9
provided thatRa−2b−2c>−3.
viContiguous Whipple’s theorem9:
3F2
⎡
⎢⎢
⎣
a, 1−a, c
; 1 e, 2c2−e
⎤
⎥⎥
⎦ ΓeΓ2c2−eΓe−c−1 22aΓe−aΓe−cΓ2c−e−a2
×
Γ1/2e−1/2a 1/2Γc−1/2e−1/2a1 Γ1/2e 1/2a−1/2Γc−1/2e 1/2a1
−Γ1/2e−1/2aΓc−1/2e−1/2a 3/2
Γ1/2e 1/2aΓc−1/2e 1/2a 1/2
4.10
provided thatRc>0.
3F2
⎡
⎢⎢
⎣
a, 3−a, c
; 1 e, 2c2−e
⎤
⎥⎥
⎦ ΓeΓ2c−e2Γe−c−1
22a−2c−1a−1a−2Γe−aΓe−cΓ2c−e−a2
×
2c−eΓ1/2e−1/2a 1/2Γc−1/2e−1/2a1 Γ1/2e 1/2a−3/2Γc−1/2e 1/2a
−e−2Γ1/2e−1/2aΓc−1/2e−1/2a 3/2 Γ1/2e 1/2a−1Γc−1/2e 1/2a−1/2
4.11
provided thatRc>0.
5. Main Summation Formulas
In this section, the following extensions of the classical summation theorems will be established. In all these theorems we haveRd>0.
iExtension of Kummer’s theorem:
3F2
⎡
⎢⎢
⎣
a, b, d1
; −1 2a−b, d
⎤
⎥⎥
⎦ Γ1/2Γ2a−b
2a1−b
1a−b/d−1
Γ1/2aΓ1/2a−b 3/2 1−a/d
Γ1/2a 1/2Γ1/2a−b1
. 5.1
iiExtension of Gauss’s second theorem:
3F2
⎡
⎢⎢
⎢⎢
⎣
a, b, d1
; 1 1 2
2ab3, d
⎤
⎥⎥
⎥⎥
⎦
Γ1/2Γ1/2a 1/2b 3/2Γ1/2a−1/2b−1/2
Γ1/2a−1/2b 3/2
×
1/2ab1−ab/d
Γ1/2a 1/2Γ1/2b 1/2ab1/d−2
Γ1/2aΓ1/2b
.
5.2
iiiExtension of Bailey’s theorem:
3F2
⎡
⎢⎢
⎢⎣
a, 1−a, d1
; 1 2 c1, d
⎤
⎥⎥
⎥⎦
Γ1/2Γc1 2c
2/d
Γ1/2c 1/2aΓ1/2c−1/2a 1/2
1−c/d
Γ1/2c−1/2a1Γ1/2c 1/2a 1/2
.
5.3
ivExtension of Watson’s theorem:
First Extension:
4F3
⎡
⎢⎢
⎣
a, b, c, d1
; 1 1
2ab1, 2c1, d
⎤
⎥⎥
⎦
2ab−2Γc 1/2Γ1/2a 1/2b 1/2Γc−1/2a−1/2b 1/2
Γ1/2ΓaΓb
×
Γ1/2aΓ1/2b
Γc−1/2a 1/2Γc−1/2b 1/2 2c−d/dΓ1/2a 1/2Γ1/2b 1/2
Γ1/2c−1/2a1Γc−1/2b1
5.4
providedR2c−a−b>−1.
Second Extension:
4F3
⎡
⎢⎢
⎣
a, b, c, d1
; 1 1
2ab3, 2c, d
⎤
⎥⎥
⎦
2ab−2Γc 1/2Γ1/2a 1/2b 3/2Γc−1/2a−1/2b−1/2 a−b−1a−b1Γ1/2ΓaΓb
×
α Γ1/2aΓ1/2b
Γc−1/2a 1/2Γc−1/2b 1/2βΓ1/2a 1/2Γ1/2b 1/2
Γc−1/2aΓc−1/2b
5.5
providedR2c−a−b>1,αandβare given by
αa2c−a b2c−b−2c1−ab
d4c−a−b−1, β8
1
2dab1−1
.
5.6
vExtension of Dixon’s theorem:
4F3
⎡
⎢⎣
a, b, c, d1
; 1 2a−b, 1a−c, d
⎤
⎥⎦
α
b−1 Γ1a−cΓ2a−bΓ3/2 1/2a−b−cΓ1/2
2aΓ1/2aΓ1/2a−c 1/2Γ2a−b−cΓ1/2a−b 3/2 β
b−1
2−a−1Γ1/2Γ1a−cΓ1a−bΓ1 1/2a−b−c Γ1/2a 1/2Γ1 1/2a−bΓ1 1/2a−cΓ1a−b−c
5.7
providedRa−2b−2c>−2,αandβare given by α1− 1
d1a−b, β 1a−b
1a−b−c a
d1a−b−2c−2 1
2a−b−c1
.
5.8
viExtension of Whipple’s theorem:
4F3
⎡
⎢⎣
a, 1−a, c, d1
; 1 e1, 2c−e1, d
⎤
⎥⎦
2−2aΓe1Γe−cΓ2c−e1 Γe−a1Γe−c1Γ2c−e−a1
×
1−2c−e d
Γ1/2e−1/2a1Γc−1/2e−1/2a 1/2
Γ1/2e 1/2aΓc−1/2e 1/2a 1/2
e
d−1Γ1/2e−1/2a 1/2Γc−1/2e−1/2a1
Γ1/2e 1/2a 1/2Γc−1/2e 1/2a
5.9
providedRc>0.
viiExtension of2.8:
3F2
⎡
⎢⎣
a, b, 1d
; −1 1a−b, d
⎤
⎥⎦
1− a
2d
Γ1a−bΓ1 1/2a Γ1aΓ1 1/2a−b a
2d
Γ1a−bΓ1/2a 1/2 Γ1aΓ1/2a−b 1/2.
5.10
viiiExtension of2.9:
4F3
⎡
⎢⎣
a, b, c, d1
; 1 1a−b, 1a−c, d
⎤
⎥⎦
1− a
2d
Γ1 1/2aΓ1a−bΓ1a−cΓ1 1/2a−b−c Γ1aΓ1a−b−cΓ1 1/2a−bΓ1 1/2a−c a
2d
Γ1/2 1/2aΓ1a−bΓ1a−cΓ1/2 1/2a−b−c Γ1aΓ1a−b−cΓ1/2 1/2a−bΓ1/2 1/2a−c
5.11
providedRa−2b−2c>−1.
5.1. Derivations
In order to derive5.1, it is just a simple exercise to prove the following relation:
3F2
⎡
⎢⎣
a, b, d1
; −1 2a−b, d
⎤
⎥⎦
2F1
⎡
⎢⎣
a, b
; −1 2a−b
⎤
⎥⎦− ab
d2a−b2F1
⎡
⎢⎣
a1, b1
; −1 3a−b
⎤
⎥⎦.
5.12
Now, it is easy to see that the first and second2F1 on the right-hand side of 5.12 can be evaluated with the help of contiguous Kummer’s theorems4.1, and after a little simplification, we arrive at the desired result5.1.
In the exactly same manner, the results5.2to5.11can be established with the help of the following relations:
3F2
⎡
⎢⎢
⎢⎢
⎣
a, b, d1
; 1 1 2
2ab3, d
⎤
⎥⎥
⎥⎥
⎦
2F1
⎡
⎢⎢
⎢⎢
⎣
a, b
; 1 1 2
2ab3
⎤
⎥⎥
⎥⎥
⎦ ab
dab32F1
⎡
⎢⎢
⎢⎢
⎣
a1, b1
; 1 1 2
2ab5
⎤
⎥⎥
⎥⎥
⎦,
3F2
⎡
⎢⎢
⎣
a, 1−a, d1
; 1 c1, d 2
⎤
⎥⎥
⎦
2F1
⎡
⎢⎢
⎣
a, 1−a
; 1 c1 2
⎤
⎥⎥
⎦ a1−a 2d1c2F1
⎡
⎢⎢
⎣
a1, 2−a
; 1 2 c2
⎤
⎥⎥
⎦,
4F3
⎡
⎢⎢
⎢⎣
a, b, c, d1
; 1 1
2ab1, 2c1, d
⎤
⎥⎥
⎥⎦
3F2
⎡
⎢⎢
⎢⎣
a, b, c
; 1 1
2ab1, 2c1
⎤
⎥⎥
⎥⎦ 2abc
d2c1ab13F2
⎡
⎢⎢
⎢⎣
a1, b1, c1
; 1 1
2ab3, 2c2
⎤
⎥⎥
⎥⎦,
4F3
⎡
⎢⎢
⎢⎣
a, b, c, d1
; 1 1
2ab3, 2c, d
⎤
⎥⎥
⎥⎦
3F2
⎡
⎢⎢
⎢⎣
a, b, c
; 1 1
2ab3, 2c
⎤
⎥⎥
⎥⎦ ab
dab33F2
⎡
⎢⎢
⎢⎣
a1, b1, c1
; 1 1
2ab5, 2c1
⎤
⎥⎥
⎥⎦,
4F3
⎡
⎢⎣
a, b, c, d1
; 1 2a−b, 1a−c, d
⎤
⎥⎦
3F2
⎡
⎢⎣
a, b, c
; 1 2a−b, 1a−c
⎤
⎥⎦ abc
d2a−b1a−c3F2
⎡
⎢⎣
a1, b1, c1
; 1 3a−b, 2a−c
⎤
⎥⎦,
4F3
⎡
⎢⎣
a, 1−a, c, d1
; 1 e1, 2c−e1, d
⎤
⎥⎦
3F2
⎡
⎢⎣
a, 1−a, c
; 1 e1, 2c−e1
⎤
⎥⎦ ac1−a
de12c−e13F2
⎡
⎢⎣
a1, 2−a, c1
; 1 e2, 2c−e2
⎤
⎥⎦,
3F2
⎡
⎢⎣
a, b, d1
; −1 1a−b, d
⎤
⎥⎦
2F1
⎡
⎢⎣
a, b
; −1 1a−b
⎤
⎥⎦− ab
d1a−b2F1
⎡
⎢⎣
a1, b1
; −1 2a−b
⎤
⎥⎦,
4F3
⎡
⎢⎣
a, b, c, d1
; 1 1a−b, 1a−c, d
⎤
⎥⎦
3F2
⎡
⎢⎣
a, b, c
; 1 1a−b, 1a−c
⎤
⎥⎦ abc
d1a−b1a−c3F2
⎡
⎢⎣
a1, b1, c1
; 1 2a−b, 2a−c
⎤
⎥⎦
5.13
and using the results4.2;2.4,4.3;2.5,4.4;4.5,4.6;4.8,4.9;4.10,4.11;2.2, 4.1, and2.6,4.7, respectively.
5.2. Special Cases
1In5.1, if we taked1a−b, we get Kummer’s theorem2.2.
2In5.2, if we taked 1/2ab1, we get Gauss’s second theorem2.3.
3In5.3, if we takedc, we get Bailey’s theorem2.4.
4In5.4, if we taked2c, we get Watson’s theorem2.5.
5In5.5, if we taked 1/2ab1, we again get Watson’s theorem2.5.
6In5.7, if we taked1a−b, we get Dixon’s theorem2.6.
7In5.9, if we takede, we get Whipple’s theorem2.7.
8In5.10, if we taked 1/2a, we get2.8.
9In5.11, if we taked 1/2a, we get2.9.
6. Generalizations of Summations Due to Ramanujan
In this section, the following summations, which generalize Ramanujan’s summations3.1 to3.11, will be established.
In all the summations, we haved >0.
iForRx>1/2:
1− x−1
x2
d1 d
x−1x−2 x2x3
d2 d
− · · · x1
2dx2x−1{d2x−1 1−x}, 6.1 1−
1 2
2 d1
2d
1·3
2·4 2
d2 3d
− · · ·√ 2π
1 d−1
4 Γ21/4
1− 1
2d 1
Γ23/4
. 6.2
iiForRx>0:
1− x−1
x2
d1 d
x−1x−2 x2x3
d2 d
− · · · Γ1/2Γ2x
2x
1x d −1
1
Γ1/2Γ1x
1− 1 d
1
Γx 1/2
,
6.3
11 2
1 2
2 d1
2d
1 22
1·3 2·4
2 d2
3d
− · · ·√ π
1 dΓ23/4
1− 1
d 8
Γ21/4
. 6.4
iiiForRx>0:
1−1 3
x−1 x2
d1 d
1
5
x−1x−2 x2x3
d2 d
− · · ·
1x d −1
x1 2x2x1π
4
2− 1 d
ΓxΓx2 2x1Γ2x 1/2.
6.5
ivForRx>1/4:
1 x−12 x1x2
d1 d
x−12x−22 x1x22x3
d2 d
· · ·
1x
d −1
22x−1Γx√2Γx 3/2
πΓ2x1 −
√π 8
x1Γ2xΓ2x1 Γ2xΓ2x 1/2
1
d3x−1−4x1
. 6.6
vForRx>1:
1−3x−1 x2
d1 d
5x−1x−2 x2x3
d2 d
· · · 1
4x
Γx2Γx3 Γ2x 1/2
1−1x d
,
11 5
1 2
2 d1
2d
1 9
1·3 2·4
2 d2
3d
· · ·
3 4π
1 d−1
π2 3Γ43/4
1− 1
4d
,
11 5
d1 9d
1 2
1
92
5d2 13d
1·3 2·4
· · ·
5π3/2 48√
2Γ23/4 5
4d −1
− 5 48√
2 π5/2 Γ23/4
3 8d−3
2
,
1 1
2 3
d1 2d
1·3 2·4
3 d2
3d
· · ·
π Γ43/4 −3
2
1− 1 d
Γ23/4 π3 .
6.7
viForRx<2/3:
1n3 2
d1 d
1 1!
nn13 3.4
d2 d
1 2!· · · 1−1/d
n−1 Γ1/2Γ3/2−2n/2
2nΓn/2Γ1/2−n/2Γ3/2−n/2Γ2−n 2−n−1
n−12
n/d1−2n−2−3nΓ1/2Γ1−3n/2 Γ21−n/2Γn/2 1/2Γ1−n .
6.8
viiForRx>1/2:
1−3x−1 x2
d1 d
5x−1x−2 x2x3
d2 d
− · · · x 2d
1− 1
2d
Γ1x√ π
2Γx 1/2. 6.9 viiiForRx>1/2:
1x−12 x12
d1 d
x−12x−22 x12x22
d2 d
· · ·
1− 1 2d
√
πΓ21xΓ2x−1/2 2Γ2xΓ2x 1/2 1
2d
Γ21xΓ2x−1 Γ2xΓ2x .
6.10
6.1. Derivations
The series6.1corresponds to
3F2
⎡
⎣ 1, 1−x, 1d
; 1 2x, d
⎤
⎦ 6.11
which is a special case of extended Gauss’s summation theorem3.26fora1, b1−xand c1x.
The series6.2corresponds to
3F2
⎡
⎢⎢
⎣ 1 2, 1
2, d1
; −1 2, d
⎤
⎥⎥
⎦ 6.12
which is a special case of extended Kummer’s summation theorem 5.1for a b 1/2.
Similarly the series6.3corresponds to
3F2
⎡
⎢⎢
⎣ 1
2, 1−x, d1
; −1 2x, d
⎤
⎥⎥
⎦ 6.13
which is a special case of extended Kummer’s theorem5.1fora1 andb1−x.
The series6.4corresponds to
3F2
⎡
⎢⎢
⎢⎣ 1 2, 1
2, d1
; 1 2, d 2
⎤
⎥⎥
⎥⎦ 6.14
which is a special case of extended Gauss’s second summation theorem5.2forab1/2 or extended Bailey’s summation theorem5.3fora1/2,c1.
Also, it can be easily seen that the series6.5to6.8which correspond to
4F3
⎡
⎢⎢
⎢⎣ 1, 1
2, 1−x, 1d
; 1 3
2, 2x, d
⎤
⎥⎥
⎥⎦,
4F3
⎡
⎢⎣
1, 1−x, 1x, 1d
; 1 1x, 2x, d
⎤
⎥⎦,
4F3
⎡
⎢⎢
⎢⎣ 1, 3
2, 1−x, 1d
; 1 1
2, 1x, d
⎤
⎥⎥
⎥⎦,
4F3
⎡
⎢⎢
⎢⎣ 1 2, 1
2, 1 4, 1d
; 1 2, 5
4, d
⎤
⎥⎥
⎥⎦,
4F3
⎡
⎢⎢
⎢⎣ 1 2, 1
4, 1 4, 1d
; 1 5
4, 5 4, d
⎤
⎥⎥
⎥⎦,
4F3
⎡
⎢⎢
⎣ 1 2, 1
2, 1 2, 1d
; 1 2, 1, d
⎤
⎥⎥
⎦,
4F3
⎡
⎢⎣
n, n, n, 1d
; 1 1, 2, d
⎤
⎥⎦
6.15
are special cases of extended Dixon’s theorem5.7.
The series6.9corresponds to
3F2
⎡
⎢⎢
⎢⎢
⎣
1, 1−x, d1
; −1 1x, d
⎤
⎥⎥
⎥⎥
⎦ 6.16
which is a special case of5.10fora1 andb1−x. And the series6.10corresponds to
4F3
⎡
⎢⎢
⎢⎢
⎣
1, 1−x, 1−x, 1d
; 1 1x, 1x, d
⎤
⎥⎥
⎥⎥
⎦ 6.17
which is a special case of5.11fora1,b1−xc.
7. Concluding Remarks
1Various other applications of these results are under investigations and will be published later.
2Further generalizations of the extended summation theorem 5.1to 5.9in the forms
3F2
⎡
⎢⎢
⎢⎢
⎣
a, b, d1
; −1 2a−bi, d
⎤
⎥⎥
⎥⎥
⎦,
3F2
⎡
⎢⎢
⎢⎢
⎢⎢
⎣
a, b, d1
; 1 2 1
2ab3i, d
⎤
⎥⎥
⎥⎥
⎥⎥
⎦ ,
3F2
⎡
⎢⎢
⎢⎢
⎣
a, 1−ai, d1
; 1 c1, d
⎤
⎥⎥
⎥⎥
⎦
7.1
each fori0,±1,±2,±3,±4,±5, and
4F3
⎡
⎢⎢
⎢⎣
a, b, c, d1
; 1 1
2abi1, 2cj, d
⎤
⎥⎥
⎥⎦,
4F3
⎡
⎢⎢
⎣
a, b, c, d1
; 1 2a−bi, 1a−cij, d
⎤
⎥⎥
⎦,
7.2
each fori, j 0,±1,±2,±3, and
4F3
⎡
⎢⎢
⎣
a, b, c, 1d
; 1 e, f, d
⎤
⎥⎥
⎦, 7.3
whereab1ij,ef 2cj fori, j 0,±1,±2,±3 are also under investigations and will be published later.
Acknowledgments
The authors are highly grateful to the referees for carefully reading the manuscript and providing certain very useful suggestions which led to a better presentation of this research article. They are so much appreciated to the College of Science, Sultan Qaboos University, Muscat - Oman for supporting the publication charges of this paper. The first author is supported by the Research Fund of Wonkwang University 2011 and the second author is supported by the research grant IG/SCI/DOMS/10/03 of Sultan Qaboos University, OMAN.
References
1 C. F. Gauss, “Disquisitiones generales circa seriem infinitamαβ/1.γx αα1ββ1/1.2.γγ 1x2αα1α2ββ1β2/1.2.3.γγ1γ2x3etc. pars prior,” Commentationes Societiones Regiae Scientiarum Gottingensis Recentiores, vol. 2, 1812, reprinted in Gesammelte Werke, vol. 3, pp.
123–163 and 207–229, 1866.
2 B. C. Berndt, Ramanujan’s Note Books, Vol. II, Springer, New York, NY, USA, 1987.
3 W. N. Bailey, “Products of generalized hypergeometric series,” Proceedings of the London Mathematical Society, vol. 2, no. 28, pp. 242–254, 1928.
4 Y. S. Kim, M. A. Rakha, and A. K. Rathie, “Generalization of Kummer’s second theorem with application,” Computational Mathematics and Mathematical Physics, vol. 50, no. 3, pp. 387–402, 2010.
5 W. N. Bailey, Generalized Hypergeometric Series, Cambridge Tracts in Mathematics and Mathematical Physics, No. 32, Stechert-Hafner, New York, NY, USA, 1964.
6 M. A. Rakha and A. K. Rathie, “Generalizations of classical summation theorems for the series2F1
and3F2with applications,” to appear in Integral Transform and Special Functions.
7 J.-L. Lavoie, F. Grondin, and A. K. Rathie, “Generalizations of Watson’s theorem on the sum of a3F2,”
Indian Journal of Mathematics, vol. 32, no. 1, pp. 23–32, 1992.
8 J.-L. Lavoie, F. Grondin, A. K. Rathie, and K. Arora, “Generalizations of Dixon’s theorem on the sum of a3F2,” Mathematics of Computation, vol. 62, no. 205, pp. 267–276, 1994.
9 J. L. Lavoie, F. Grondin, and A. K. Rathie, “Generalizations of Whipple’s theorem on the sum of a
3F2,” Journal of Computational and Applied Mathematics, vol. 72, no. 2, pp. 293–300, 1996.
10 J. Choi, “Contiguous extensions of Dixon’s theorem on the sum of a3F2,” Journal of Inequalities and Applications, vol. 2010, Article ID 589618, 17 pages, 2010.
11 A. P. Prudnikov, Y. A. Brychkov, and O. I. Marichev, Integrals and Series, Gordon and Breach Science, New York, NY, USA, 1986.
12 A. R. Miller, “A summation formula for Clausen’s series3F21 with an application to Goursat’s function2F2x,” Journal of Physics A, vol. 38, no. 16, pp. 3541–3545, 2005.
13 Y. S. Kim and A. K. Rathie, “Comment on: “a summation formula for Clausen’s series3F21with an application to Goursat’s function2F2x,” Journal of Physics A, vol. 41, no. 7, Article ID 078001, 2008.