J. Math. Anal. Appl.

365 (2010) 653–658

Contents lists available at ScienceDirect

Journal of Mathematical Analysis and

Applications
www.elsevier.com/locate/jmaa

An extension of the q-beta integral with applications

Mingjin Wang
Department of Applied Mathematics, Jiangsu Polytechnic University, Changzhou city, Jiangsu province, 213164, PR China

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

Article history: In this paper, we give an extension of the q-beta integral. Applications of the extension
Received 10 April 2009 are also given, which include to derive an extension of the q-Pfaff–Saalschütz formula, an
Available online 4 December 2009 extension of the Kalnins and Miller transformations and a new identity for 3 φ2 .
Submitted by B.C. Berndt
© 2009 Elsevier Inc. All rights reserved.
Keywords:
q-beta integral
q-Chu–Vandermonde formula
q-Pfaff–Saalschütz formula
Kalnins and Miller transformations

1. Introduction and statement of result

The following is the q-beta integral [3]:

∞
(aω, bω; q)∞ 2(1 − q)(q2 ; q2 )2∞ (de , q/de , a/e , −a/d, b/e , −b/d; q)∞
dq ω = , (1.1)
(−dω, e ω; q)∞ (q; q)∞ (d2 , e 2 , q2 /d2 , q2 /e 2 ; q2 )∞ (−ab/deq; q)∞
−∞
provided that |q| < 1, |ab/deq| < 1 and there are no zero factors in the denominator of the integrals.
The q-beta integrals are very important in the theory and application of basic hypergeometric series. For example, the
present author gives some applications of the Andrews–Askey integral in [12–15]. In this paper, we give an interesting
extension of the q-beta integral (1.1). With this extension, some new identities in basic hypergeometric series can be easily
derived, which include to get an extension of the Kalnins and Miller transformations and an extension of the q-Pfaff–
Saalschütz formula. The main result is the following theorem:

Theorem 1.1. If |q| < 1, |ab/deq| < 1 and there are no zero factors in the denominator of the integrals, then for any nonnegative
integers m and n, we have
∞
(aω, bω; q)∞ (sω; q)n (t ω; q)m
dq ω
(−dω, e ω; q)∞
−∞
m+n
2
+mn+n2 t s (q ; q )∞ (de , q/de , a/eq
m n 2 2 2
, −a/dqm+n , b/e , −b/d; q)∞
= 2(1 − q)qm +
am n (q; q)∞ (d , e , q /d , q /e ; q2 )∞ (−ab/deqm+n+1 ; q)∞
2 2 2 2 2 2


n  
(q−n , a/sqn , −ab/deqm+n+1 ; q)k qk(1−m) a/tqm+n−k , −ab/deqm+n+1−k , q−m
× 3 φ2 ; q, q . (1.2)
(q, a/eqm+n , −a/dqm+n ; q)k a/eqm+n−k , −a/dqm+n−k
k =0

E-mail addresses: wmj@jpu.edu.cn, wang197913@126.com.

0022-247X/$ – see front matter © 2009 Elsevier Inc. All rights reserved.
doi:10.1016/j.jmaa.2009.11.041
654 M. Wang / J. Math. Anal. Appl. 365 (2010) 653–658

It is obvious that the case m = n = 0 of (1.2) results in (1.1).

2. Notations and known results

We first recall some definitions, notations and known results in [1,5] which will be used for the proof of Theorem 1.1.
Throughout this paper, it is supposed that 0 < |q| < 1. The q-shifted factorials are defined as

−1
n ∞

   
(a; q)0 = 1, (a; q)n = 1 − aqk , (a; q)∞ = 1 − aqk . (2.1)
k =0 k =0

We also adopt the following compact notation for multiple q-shifted factorials:

(a1 , a2 , . . . , am ; q)n = (a1 ; q)n (a2 ; q)n . . . (am ; q)n , (2.2)

where n is an integer or ∞.
In 1846, Heine introduced the r +1 φr basic hypergeometric series, which is defined by

  ∞
a 1 , a 2 , . . . , a r +1 (a1 , a2 , . . . , ar +1 ; q)n xn
r +1 φr ; q, x = . (2.3)
b 1 , b 2 , . . . , br (q, b1 , b2 , . . . , br ; q)n
n =0

The q-Chu–Vandermonde convolution formula:

 
q−n , a an (c /a; q)n
2 φ1 ; q, q = (2.4)
c (c ; q)n
and, reversing the order of summation,
 
q−n , a (c /a; q)n
2 φ1 ; q, cqn /a = . (2.5)
c (c ; q)n
The q-Pfaff–Saalschütz formula:
 
a, b, q−n (c /a, c /b; q)n
3 φ2 − −
; q , q = . (2.6)
c , abc q1 1 n (c , c /ab; q)n
F.H. Jackson defined the q-integral by [7]

d ∞
  
f (t ) dq t = d(1 − q) f dqn qn , (2.7)
0 n =0

and

d d c
f (t ) dq t = f (t ) dq t − f (t ) dq t . (2.8)
c 0 0

He also defined an integral on (0, ∞) by

∞ ∞
  
f (t ) dq t = (1 − q) f qn qn . (2.9)
n=−∞
0

On the interval (−∞, ∞) the bilateral q-integral is defined by

∞ ∞
    
f (t ) dq t = (1 − q) f qn + f −qn qn . (2.10)
−∞ n=−∞
 M. Wang / J. Math. Anal. Appl. 365 (2010) 653–658 655

3. The proof of Theorem 1.1

In this section, we use the q-beta integral (1.1) and the q-Chu–Vandermonde formula to prove Theorem 1.1.

Proof. Recall the q-Chu–Vandermonde convolution formula

 
q−n , c cn (a/c ; q)n
2 φ1 ; q, q = . (3.1)
a (a; q)n
By the following relation
(a; q)∞
(a; q)k = , (3.2)
(aqk ; q)∞
(3.1) can be written as


n
(q−n , c ; q)k qk  k   
· aq ; q ∞ = cn · aqn ; q ∞ (a/c ; q)n . (3.3)
(q; q)k
k =0

If we let a = aω in (3.3) and multiply Eq. (3.3) by

(bω; q)∞
,
(−dω, e ω; q)∞
then we obtain

n
(q−n , c ; q)k qk (aqk ω, bω; q)∞ (aqn ω, bω; q)∞ (aω/c ; q)n
· = cn · . (3.4)
(q; q)k (−dω, e ω; q)∞ (−dω, e ω; q)∞
k =0

Taking the q-integral on both sides of (3.4) with respect to variable ω, we have

n ∞ ∞
(q−n , c ; q)k qk (aqk ω, bω; q)∞ (aqn ω, bω; q)∞ (aω/c ; q)n
· dq ω = c n · dq ω . (3.5)
(q; q)k (−dω, e ω; q)∞ (−dω, e ω; q)∞
k =0 −∞ −∞
Applying the q-beta integral (1.1) to the integral on the left-hand side of (3.5), we have


n
(q−n , c ; q)k qk 2(1 − q)(q2 ; q2 )2∞ (de , q/de , aqk /e , −aqk /d, b/e , −b/d; q)∞
·
(q; q)k (q; q)∞ (d2 , e 2 , q2 /d2 , q2 /e 2 ; q2 )∞ (−abqk /deq; q)∞
k =0
∞
(aqn ω, bω; q)∞ (aω/c ; q)n
= cn · dq ω , (3.6)
(−dω, e ω; q)∞
−∞
which can be rewritten as
∞
(aqn ω, bω; q)∞ (aω/c ; q)n
dq ω
(−dω, e ω; q)∞
−∞
 
2(1 − q)(q2 ; q2 )2∞ (de , q/de , a/e , −a/d, b/e , −b/d; q)∞ c , −ab/deq, q−n
= 3 φ2 ; q, q . (3.7)
cn (q; q)∞ (d2 , e 2 , q2 /d2 , q2 /e 2 ; q2 )∞ (−ab/deq; q)∞ a/e , −a/d
On the other hand, if we multiply Eq. (3.4) by (t ω; q)m and take the q-integral on both sides of (3.4) with respect to
variable ω , we obtain


n ∞ ∞
(q−n , c ; q)k qk (aqk ω, bω; q)∞ (t ω; q)m (aqn ω, bω; q)∞ (aω/c ; q)n (t ω; q)m
· dq ω = c n · dq ω . (3.8)
(q; q)k (−dω, e ω; q)∞ (−dω, e ω; q)∞
k =0 −∞ −∞

Using (3.7) with n = m, a = aqk−m , c = aqk−m /t gives

∞
(aqk ω, bω; q)∞ (t ω; q)m 2t m (1 − q)(q2 ; q2 )2∞ (de , q/de , aqk−m /e , −aqk−m /d, b/e , −b/d; q)∞
dq ω =
(−dω, e ω; q)∞ am qm(k−m) (q; q)∞ (d2 , e 2 , q2 /d2 , q2 /e 2 ; q2 )∞ (−abqk−m /deq; q)∞
−∞
 
aqk−m /t , −abqk−m /deq, q−m
× 3 φ2 ; q, q . (3.9)
aqk−m /e , −aqk−m /d
656 M. Wang / J. Math. Anal. Appl. 365 (2010) 653–658

Substituting (3.9) into (3.8) gives

∞
(aqn ω, bω; q)∞ (aω/c ; q)n (t ω; q)m
dq ω
(−dω, e ω; q)∞
−∞

2t m (1 − q)(q2 ; q2 )2∞ (de , q/de , a/eqm , −a/dqm , b/e , −b/d; q)∞

=
cn am qm(k−m) (q; q) ∞ (d
2 , e 2 , q2 /d2 , q2 /e 2 ; q2 )
∞ (−ab /deq
m +1 ; q )
∞

n  
(q−n , c , −ab/deqm+1 ; q)k a/tqm−k
, −ab/deq m+1−k
, q−m
× 3 φ2 ; q, q . (3.10)
(q, a/eqm , −a/dqm ; q)k a/eqm−k , −a/dqm−k
k =0

First, replacing c by a/s and then a by a/qn in (3.10), respectively, after simple rearrangements, we have (1.2). 2

Letting m = 0 in (1.2) gives the following result.

Corollary 3.1. Under the conditions of the theorem, we have

∞ 2
(aω, bω; q)∞ (sω; q)n 2sn qn (1 − q)(q2 ; q2 )2∞ (de , q/de , a/eqn , −a/dqn , b/e , −b/d; q)∞
dq ω =
(−dω, e ω; q)∞ an (q; q)∞ (d2 , e 2 , q2 /d2 , q2 /e 2 ; q2 )∞ (−ab/deqn+1 ; q)∞
−∞
 
a/sqn , −ab/deqn+1 , q−n
× 3 φ2 ; q, q . (3.11)
a/eqn , −a/dqn

4. Some applications

In this section, we show some applications of the extension of the q-beta integral (1.2). We gain an extension of the
Kalnins and Miller transformations, an extension of the q-Pfaff–Saalschütz formula and a new identity for 3 φ2 .
One of the fundamental transformations in the theory of basic hypergeometric series is the following Sears 4 φ3 transfor-
mation formula [10], which is a q-extension of a well-known transformation of Whipple [11]
   
q−n , a, b, c (e /a, f /a; q)n n q−n , a, d/b, d/c
4 φ3 ; q, q = a 4 φ3 ; q, q , (4.1)
d, e , f (e , f ; q)n d, aq1−n /e , aq1−n / f

where de f = abcq1−n . It has the equivalent form [9,10],

   n  −n 
q−n , abcdqn−1 , be i ω , be −i ω (ac , cd; q)n b q , abcdqn−1 , ce i ω , ce −i ω
4 φ3 ; q, q = φ
4 3 ; q , q . (4.2)
ab, bc , bd (ab, bd; q)n c ac , bc , cd
Ismail [6] used the Askey–Wilson operator to give an interesting proof of this formula. Andrews and Bowman [2] prove this
formula by using the Bailey transformation. Liu proves the formula by using the q-exponential operator method. In [4], Askey
and Wilson show the connection between this formula and orthogonal polynomials. From the Sears 4 φ3 transformation
formula, one can get the following Kalnins and Miller’s transformation formula [8]:
   n  −n 
q−n , abcdqn−1 , bz b (ac , cd; q)n q , abcdqn−1 , cz
3 φ2 ; q, q = 3 φ2 ; q, q . (4.3)
ab, bd c (ab, bd; q)n ac , cd
The Kalnins and Miller transformation formula is also important in q-series. There have been many proofs of the Kalnins
and Miller’s transformation formula. Kalnins and Miller [8] used symmetry techniques and Liu [9] used q-differential op-
erators to prove it. In this section, first we will present an extension of the Kalnins and Miller’s transformation formula.
Using (1.2), an extension of Kalnins and Miller’s transformation can easily be derived.

Theorem 4.1 (An extension of the Kalnins and Miller transformations). We have


n  
(q−n , bz, abcdqn+m−1 ; q)k qk(1−m) byqk , abcdqn+m+k−1 , q−m
3 φ2 ; q, q
(q, ab, bd; q)k abqk , bdqk
k =0
 m+n  
(ac , cd; q)m+n  (q−n , cz, abcdqn+m−1 ; q)k qk(1−m)
n
b c yqk , abcdqn+m+k−1 , q−m
= 3 φ2 ; q, q . (4.4)
c (ab, bd; q)m+n (q, ac , cd; q)k acqk , cdqk
k =0
 M. Wang / J. Math. Anal. Appl. 365 (2010) 653–658 657

It is obvious that (4.4) leads to (4.3) when m = 0.

Proof of Theorem 4.1. The left-hand side of (1.2) is symmetric in a, b, so is right. Interchanging a and b in the right side of
(1.2), we have


n  
(q−n , a/sqn , −ab/deqm+n+1 ; q)k qk(1−m) a/tqm+n−k , −ab/deqm+n+1−k , q−m
3 φ2 ; q, q
(q, a/eqm+n , −a/dqm+n ; q)k a/eqm+n−k , −a/dqm+n−k
k =0
 m+n
(b/eqm+n , −b/dqm+n ; q)m+n  (q−n , b/sqn , −ab/deqm+n+1 ; q)k qk(1−m)
n
a
=
b (a/eqm+n , −a/dqm+n ; q)m+n (q, b/eqm+n , −b/dqm+n ; q)k
k =0
 
b/tqm+n−k , −ab/deqm+n+1−k , q−m
× 3 φ2 ; q , q . (4.5)
b/eqm+n−k , −b/dqm+n−k
After replacing (1/eqm , a/qn , b/qn , −1/dqm , 1/tqm , 1/s) by (a, b, c , d, y , z), we get (4.4). 2

Then, we give an extensions of the q-Pfaff–Saalschütz formula. The q-Pfaff–Saalschütz formula is a fundamental formula
in q-series. Many people gave different extensions of it. For example, Wang gives an extension of it by using the Andrews–
Askey integral [15]. Here we will present another extension of the q-Pfaff–Saalschütz formula by using (1.2).

Theorem 4.2 (An extension of the q-Pfaff–Saalschütz formula). We have


n  
(q−n , aqm , b; q)k qk(1−m) qk−n , bqk , q−m (c /a, c /b; q)n
3 φ2 ; q , q = . (4.6)
(q, c , abc −1 q1−n ; q)k cqk , abc −1 q1−n+k (c , c /ab; q)n
k =0

Proof. Letting a = tqm , b = sqn in (1.2) gives

∞
(tqm ω, sqn ω; q)∞ (sω; q)n (t ω; q)m
dq ω
(−dω, e ω; q)∞
−∞

2 t m sn (q2 ; q2 )2∞ (de , q/de , t /eqn , −t /dqn , sqn /e , −sqn /d; q)∞
= 2(1 − q)qn
t m+n (q; q)∞ (d2 , e 2 , q2 /d2 , q2 /e 2 ; q2 )∞ (−st /deq; q)∞

n  k−n 
(q−n , t /sqn−m , −st /deq; q)k qk(1−m) q , −st /deq1−k , q−m
× 3 φ 2 ; q , q . (4.7)
(q, t /eqn , −t /dqn ; q)k t /eqn−k , −t /dqn−k
k =0

On the other hand, using the q-beta integral (1.1), we have

∞
(tqm ω, sqn ω; q)∞ (sω; q)n (t ω; q)m
dq ω
(−dω, e ω; q)∞
−∞
∞
(at , bt ; q)∞ 2(1 − q)(q2 ; q2 )2∞ (de , q/de , s/e , −s/d, t /e , −t /d; q)∞
= dq t = . (4.8)
(−dt , et ; q)∞ (q; q)∞ (d2 , e 2 , q2 /d2 , q2 /e 2 ; q2 )∞ (−st /deq; q)∞
−∞

Combining (4.7) and (4.8) yields


n    n
(q−n , t /sqn−m , −st /deq; q)k qk(1−m) qk−n , −st /deq1−k , q−m t (s/e , −s/d; q)n
3 φ2 ; q, q = . (4.9)
(q, t /eqn , −t /dqn ; q)k t /eqn−k , −t /dqn−k sqn (t /eqn , −t /dqn ; q)n
k =0

After replacing (t /sqn , −st /deq, t /eqn ) by (a, b, c ) in (4.9), and employing the following formula
 n
a n  
(a/b; q)n = − q(2) b/aqn−1 ; q n
b
we obtain (4.6). 2

It is obvious that the case m = 0 of (4.6) results in the q-Pfaff–Saalschütz formula (2.6).
Finally, we use (1.2) to give a new identity for 3 φ2 .
658 M. Wang / J. Math. Anal. Appl. 365 (2010) 653–658

Theorem 4.3. We have


n    
(q−n , aqm , b; q)k qk(1−m) aqk−n , bqk , q−m a, b, q−(m+n)
3 φ2 ; q , q = 3 φ2 ; q, q . (4.10)
(q, c , d; q)k cqk , dqk c, d
k =0

Proof. Letting t = sqn in (1.2) gives

∞
(aω, bω; q)∞ (sω; q)n (sqn ω; q)m
dq ω
(−dω, e ω; q)∞
−∞
sm+n (q2 ; q2 )2∞ (de , q/de , a/eqm+n , −a/dqm+n , b/e , −b/d; q)∞
= 2(1 − q)q(m+n)
2

am+n (q; q)∞ (d2 , e 2 , q2 /d2 , q2 /e 2 ; q2 )∞ (−ab/deqm+n+1 ; q)∞


n  
(q−n , a/sqn , −ab/deqm+n+1 ; q)k qk(1−m) a/sqm+2n−k , −ab/deqm+n+1−k , q−m
× 3 φ2 ; q, q . (4.11)
(q, a/eqm+n , −a/dqm+n ; q)k a/eqm+n−k , −a/dqm+n−k
k =0

On the other hand, using (3.11), we have

∞
(aω, bω; q)∞ (sω; q)n (sqn ω; q)m
dq ω
(−dω, e ω; q)∞
−∞
∞
(aω, bω; q)∞ (sω; q)m+n
= dq ω
(−dω, e ω; q)∞
−∞

2sm+n q(m+n) (1 − q)(q2 ; q2 )2∞ (de , q/de , a/eqm+n , −a/dqm+n , b/e , −b/d; q)∞
2

=
am+n (q; q)∞ (d2 , e 2 , q2 /d2 , q2 /e 2 ; q2 )∞ (−ab/deqm+n+1 ; q)∞
 
a/sqm+n , −ab/deqm+n+1 , q−(m+n)
× 3 φ2 ; q, q . (4.12)
a/eqm+n , −a/dqm+n
Combining (4.11) and (4.12) yields


n  
(q−n , a/sqn , −ab/deqm+n+1 ; q)k qk(1−m) a/sqm+2n−k , −ab/deqm+n+1−k , q−m
3 φ2 ; q, q
(q, a/eqm+n , −a/dqm+n ; q)k a/eqm+n−k , −a/dqm+n−k
k =0
 
a/sqm+n , −ab/deqm+n+1 , q−(m+n)
= 3 φ2 ; q, q . (4.13)
a/eqm+n , −a/dqm+n
After replacing (a/sqm+n , −ab/deqm+n+1 , a/eqm+n , −a/dqm+n ) by (a, b, c , d) in (4.13), we obtain (4.10). 2

Acknowledgments

The author would like to express deep appreciation to the referee for the helpful suggestions. The author was supported by STF of Jiangsu Polytechnic
University.

References

[1] G.E. Andrews, q-Series: Their Development and Applications in Analysis, Number Theory, Combinatorics, Physics and Computer Algebra, CBMS Reg.
Conf. Ser. Math., vol. 66, American Mathematical Society, Providence, RI, 1986.
[2] G.E. Andrews, D. Bowman, The Bailey transform and D.B. Sears, Quaest. Math. 22 (1999) 19–26.
[3] R. Askey, q-Extension of Cauchy’s form of the beta function, Quart. J. Math. Oxford 32 (2) (1981) 255–266.
[4] R. Askey, J.A. Wilson, Some basic hypergeometric polynomials that generalize Jacobi polynomials, Mem. Amer. Math. Soc. 319 (1985) 1–55.
[5] G. Gasper, M. Rahman, Basic Hypergeometric Series, Cambridge University Press, Cambridge, MA, 1990.
[6] M.E.H. Ismail, The Askey–Wilson Operator and Summation Theorems, Contemp. Math., vol. 190, American Mathematical Society, Providence, RI, 1995,
pp. 171–178.
[7] F.H. Jackson, On q-definite integrals, Quart. J. Pure Appl. Math. 50 (1910) 101–112.
[8] E.G. Kalnins, W. Miller, q-series and orthogonal polynomials associated with Barnes’ first lemma, SIAM J. Math. Anal. 19 (1988) 1216–1231.
[9] Z. Liu, Some operator identities and q-series transformation formulas, Discrete Math. 265 (2003) 119–139.
[10] D.B. Sears, On the transformation theory of basic hypergeometric functions, Proc. London Math. Soc. (2) 53 (1951) 158–180.
[11] W.N. Bailey, Generalized Hypergeometric Series, Hafner, New York, 1972.
[12] M. Wang, A remark on Andrews–Askey integral, J. Math. Anal. Appl. 341 (2) (2008) 1487–1494.
[13] M. Wang, Generalizations of the Milne’s U (n + 1) q-binomial theorem, Comput. Math. Appl. 58 (2009) 80–87.
[14] M. Wang, A recurring q-integral formula, Appl. Math. Lett. 23 (2010) 256–260.
[15] M. Wang, On another extension of q-Pfaff–Saalschütz formula, Czechoslovak Math. J., in press.