kajihara y - euler transformation formula for multiple basic hypergeoemtric series of type a and...

Upload: renee-bravo

Post on 29-May-2018

221 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/9/2019 Kajihara Y - Euler Transformation Formula for Multiple Basic Hypergeoemtric Series of Type a and Some Application

    1/45

    http://www.elsevier.com/locate/aim

    Advances in Mathematics 187 (2004) 5397

    Euler transformation formula for multiple basic

    hypergeometric series of type A and

    some applications

    Yasushi Kajihara

    Department of Mathematics, Kobe University, Rokkodai, Kobe 657-8501, Japan

    Received 15 January 2001; accepted 13 August 2003

    Communicated by A. Lascoux

    Abstract

    A multiple generalization of the Euler transformation formula for basic hypergeometric

    series 2f1 is derived. It is obtained from the symmetry of the reproducing kernel forMacdonald polynomials by a method of multiple principal specialization. As applications,

    elementary proofs of the PfaffSaalschutz summation formula and the Gauss summation

    formula for basic hypergeometric series in Un 1 due to S.C. Milne are given. Some othermultiple transformation and summation formulas for very-well-poised 10f9 and 8f7 series,

    balanced 4f3 series and 3f2 series are also given.

    r 2003 Elsevier Inc. All rights reserved.

    MSC: 33D67; 33D52; 33D15; 33C20

    Keywords: Multiple basic hypergeometric series of type A; Euler transformation; Macdonald

    polynomials; PfaffSaalschutz summation; Watson type transformation; BaileyJackson type transforma-tionsummation formula

    1. Introduction

    In this paper, we derive a multiple generalization of the Euler transformation

    formula for basic hypergeometric series 2f1: We obtain our main result(Theorem 1.1) by specializing suitably the well-known symmetry of the reproducing

    ARTICLE IN PRESS

    Corresponding author. Present address: Department of Mathematics, Graduate School of Science,

    Osaka University, Machikaneyama 1-1, Toyonaka, Osaka 560-0043, Japan. Fax: +81-6-6850-5327.

    E-mail address: [email protected].

    0001-8708/$ - see front matterr 2003 Elsevier Inc. All rights reserved.

    doi:10.1016/j.aim.2003.08.012

  • 8/9/2019 Kajihara Y - Euler Transformation Formula for Multiple Basic Hypergeoemtric Series of Type a and Some Application

    2/45

    kernel for Macdonald polynomials [12]. As applications of our result, we give

    elementary proofs of the PfaffSaalschutz summation formula and the Gauss

    summation formula for basic hypergeometric series in U

    n

    1

    due to S.C. Milne

    [16,17] from the special case of our Euler transformation formula (Proposition 5.1).We also give some other multiple transformation and summation formulas for basic

    and ordinary hypergeometric series ( for example, very-well-poised series, Sears

    transformation and others).

    There are many well-known summation and transformation formulas in the

    theory of hypergeometric and basic hypergeometric series. One of the most

    important is the binomial theorem

    1 uz XkANzkk!

    uk; 1:1

    where we denote by

    zn zz 1?z n 1 1:2

    the (ordinary) Pochhammer symbol. A q-analogue of (1.1) is given by

    auN

    uN

    XkAN

    akqk

    uk: 1:3

    Throughout of this paper, we use the notation of q-series [5]

    aN

    : a; qN

    YnAN

    1 aqn; ak : a; qk aN

    aqkN

    for kAC; 1:4

    assuming that 0oqo1: We often omit the basis q in q-shifted factorials.On the other hand, one of the most fundamental transformation formulas for

    hypergeometric series is the Euler transformation for Gauss hypergeometric series

    2F1 (see, for example, [1])

    2F1 a; bc

    ; u 1 ucab2F1 c a; c b

    c; u

    ; 1:5

    where

    n1Fna0; a1;y; an

    b1;y; bn; u

    XkAN

    a0ka1k?ankk!b1k?bnk

    uk: 1:6

    Among the most important transformation formulas for basic hypergeometric

    series are Heines transformation formulas for 2f1 series.

    2f1a; b

    c; q; u

    bNauNc

    NuN

    2f1c=b; u

    au; q; b

    1:7

    ARTICLE IN PRESS

    Y. Kajihara / Advances in Mathematics 187 (2004) 539754

  • 8/9/2019 Kajihara Y - Euler Transformation Formula for Multiple Basic Hypergeoemtric Series of Type a and Some Application

    3/45

    c=bNbuNcN

    uN

    2f1abu=c; b

    bu; q; c=b

    1:8

    abu=cNuN

    2f1c=a; c=b

    c; q; abu=c

    ; 1:9

    where the basic hypergeometric series r1fr is defined by

    r1fra0; a1;y; ar

    c1;y; cr; q; u

    XnAN

    a0na1nyarnqnc1nycrn

    un: 1:10

    The third equality of Heines transformation formulas (1.9) is a q-analogue of Euler

    transformation formula for 2F1 (1.5). It is needless to say that these hypergeometricand basic hypergeometric identities frequently appear in many areas of mathematics

    such as in combinatorics, analytic number theory, statistics and representation

    theory of Lie algebras.

    The multiple basic hypergeometric series in Un 1 initiated by Holman et al.[7,8] have been investigated by many researchers from various viewpoints. The most

    important summation formula for basic hypergeometric series in Un 1 is themultiple q-binomial theorem due to S.C. Milne (Theorem 1.47 in [14]):

    a1?anuNuN XbANn u

    jbj DxqbDx Y1pi; jpn

    ajxi=xjbiqxi=xjbi

    ;

    1:11

    where jaj Pni1 ai: In this formula,Dx

    Y1piojpn

    xi xj and Dxqb Y

    1piojpn

    xiqbi xjqbj

    denote the Vandermonde determinants of x x1;y; xn and xqb x1qb1 ;y; xnqbn; respectively. In our previous work [10], we encountered aterminating version of (1.11) when we determine the explicit coefficients of theraising and lowering operators of row type for Macdonald polynomials. We also

    gave a proof of (1.11) based on the identity that a constant function 1 is a

    eigenfunction of the Macdonalds q-difference operator

    Dxu; t; q 1 X

    KC1;y;nujKjq

    jKj2

    YiAK; jeK

    1 qxi=xj1 xi=xj

    un: 1:12

    The terminating version of (1.11) is in fact obtained by a method of multipleprincipal specialization from (1.12). In this paper, we apply the same method to

    derive the main result of this paper.

    ARTICLE IN PRESS

    Y. Kajihara / Advances in Mathematics 187 (2004) 5397 55

  • 8/9/2019 Kajihara Y - Euler Transformation Formula for Multiple Basic Hypergeoemtric Series of Type a and Some Application

    4/45

    Theorem 1.1. Suppose that none of denominators vanish. Then we have Euler

    transformation formula between basic hypergeometric series in Un 1 and Um 1

    XgANn

    ujgjDxqgDx

    Y1pi; jpn

    ajxi=xjgiqxi=xjgi

    Y1pipn;1pkpm

    bkxiyk=xnymgicxiyk=xnymgi

    a1?anb1?bmu=cmN

    uN

    XdANm

    a1?anb1?bmu=cmjdj Dyqd

    Dy

    Y1pk;lpmc=blyk=yldk

    qyk=yldk Y1pipn;1pkpmc=aixiyk=xnymdk

    cxiyk=xnymdk 1:13

    for a11 ;y; a1n ; b1=c;y; bm=cAC:

    Remark 1.1. As discussed in [15] more general form, the radius of convergence of the

    series in (1.13) is 1.

    In this paper, we refer to Eq. (1.13) as the multiple q-Euler transformation. What

    is interesting in our multiple q-Euler transformation formula is that the dimension of

    the summation in both sides are independent of each other.

    In the case when m n 1; formula (1.13) reduces to the third Heinestransformation formula (1.9). We remark that our multiple q-Euler transformation

    (1.13) is different from the third Heines transformation for basic hypergeometric

    series in Un 1 studied in [6].We give a proof of Theorem 1.1 in Section 3. The source identity for proving our

    multiple q-Euler transformation formula for basic hypergeometric series is based on

    self-duality of the reproducing kernel for Macdonald polynomials. That is,

    Macdonald polynomials and Macdonalds q-difference operators know someproperties of the basic hypergeometric series in Un 1: Our transformationformula of Theorem 1.1 also involves in a sense fundamental properties of

    Macdonald polynomials.

    We also give some special cases and q-1 limits of (1.13) in Section 4. In

    particular, formula (1.13) for m 1 implies an interesting and useful transformationformula (4.1) between n1fn and basic hypergeometric series in Un 1: Eq. (4.1)has also an expression in terms of basic Lauricella function f

    nD obtained by

    Andrews transformation [2,3].

    In fact, formula (1.13) can be considered as a master formula from which various

    multiple hypergeometric transformation and summation formulas are derived andwhich provides elementary proofs of them as in usual basic and ordinary

    hypergeometric series case.

    ARTICLE IN PRESS

    Y. Kajihara / Advances in Mathematics 187 (2004) 539756

  • 8/9/2019 Kajihara Y - Euler Transformation Formula for Multiple Basic Hypergeoemtric Series of Type a and Some Application

    5/45

    In Section 5, we make use of (4.1) to give elementary proofs of the Pfaff

    Saalschutz summation formula and the Gauss summation formula for basic

    hypergeometric series in U

    n

    1

    by S.C. Milne [16,18].

    We also discuss some generalizations of multiple PfaffSaalschutz summationformula in Section 6. In a similar way to our proof of multiple PfaffSaalschutz

    summation, we also derive from (1.13) some new multiple transformation and

    summation formulas for very-well-poised (basic) hypergeometric series. A general-

    ization of a transformation formula between terminating 8W7 series and terminating

    balanced 4f3 series and a formula of multiple (basic) hypergeometric series which

    involves multiple Jacksons 8W7 terminating summation due to S.C. Milne [17] and

    multiple 10W9 transformation formula is included.

    In Section 7, we give two type of multiple generalizations of Sears transformation

    for balanced 4f3 basic hypergeometric series from the product of two multiple q-

    Euler transformations. Two multiple generalizations of terminating 3f2 transforma-

    tions are presented.

    2. Background information

    In this section, we summarize some basic facts on Macdonald polynomials of type

    An1 and Andrews transformation formula for basic hypergeometric series betweenf

    nD and n1fn: Definitions of very-well-poised (basic) hypergeometric series are also

    given.

    2.1. Macdonald symmetric polynomials

    Here, we define Macdonald polynomials and summarize their properties which we

    need in this paper. We follow basically the notation of Macdonalds book [12]. For

    detail on Macdonald polynomials, see [12].

    Suppose that t is non zero real number and let K Qq; t be the field of rationalfunctions in q; t; and KxSn Kx1;y; xnSn the ring of symmetric functions in nvariables over K

    :The Macdonald symmetric polynomials

    Plx;

    q; tare a family of

    homogeneous polynomials that are parametrized by partitions l l1;y; ln withllpn: Macdonalds q-difference operator Dxu; q; t is defined by

    Dxu; q; t X

    KC1;y;nujKjt

    jKj2

    YiAK; jeK

    1 txi=xj1 xi=xj

    YiAK

    Tq;xi

    Xnr0

    urDrx; 2:1

    where Tq;xi are the q-shift operators in xi

    Tq;xi fx1;y; xi;y; xn fx1;y; xiq;y; xn: 2:2

    ARTICLE IN PRESS

    Y. Kajihara / Advances in Mathematics 187 (2004) 5397 57

  • 8/9/2019 Kajihara Y - Euler Transformation Formula for Multiple Basic Hypergeoemtric Series of Type a and Some Application

    6/45

    This operator is the generating function of the commuting family of q-difference

    operators Dr; r 0;y; n: The Macdonald polynomial Plx; q; t are characterized asthe joint eigenfunctions in K

    x

    Sn of Dr: Namely, each Pl

    x; q; t

    satisfies the

    following q-difference equation

    Dxu; q; tPlx; q; t Plx; q; tYni1

    1 utniqli: 2:3

    Note that Plx; q; t has the leading term mlx under the dominance order ofpartition when it is expressed as a linear combination of monomial symmetric

    function mmx: It is known that Plx; q; t is determined uniquely by two conditionsmentioned above.

    Macdonald polynomials have a following reproducing kernelYx;y :

    Y1pipn;1pkpm

    txiykNxiykN

    X

    llpminn;mblq; tPlx; q; tPly; q; t 2:4

    for the variables x x1;y; xn and y y1;y;ym: The coefficient blq; t isdetermined by

    blq; t YsAl

    1 tls1qas1 tlsqas1: 2:5

    The following identity can be proved from (2.3) and (2.4).

    Proposition 2.1 (Kirillov and Noumi [11], Mimachi and Noumi [13]). Suppose that

    nXm: Then

    DxuY

    x;y u; tnmDyutnmY

    x;y: 2:6

    2.2. Andrews transformation formula for basic hypergeometric function

    Here, we recall on the multivariable basic hypergeometric function flD and

    Andrews transformation formula. Our fundamental reference on the notation of

    q-series and basic hypergeometric function is [5].

    The Lauricella type multivariable basic hypergeometric series flD is defined as

    follows:

    flD

    a; b1; b2;y; bl

    c

    ; q; x1; x2;y; xl XaANlajajb1a1b2a2yblalcjajqa1qa2yqal

    xa;

    2:7

    where jaj Pli1 ai is the length of multi-index a a1;y; al:

    ARTICLE IN PRESS

    Y. Kajihara / Advances in Mathematics 187 (2004) 539758

  • 8/9/2019 Kajihara Y - Euler Transformation Formula for Multiple Basic Hypergeoemtric Series of Type a and Some Application

    7/45

    Andrews derived the transformation formula between flD and l1fl ([2] for l 2;

    [3] for any positive integer l).

    Proposition 2.2 (Andrews [2,3]).

    flD

    a; b1; b2;y; bl

    c; q; x1; x2;y; xl

    aNcN

    Ylk1

    bkxkNxkN l1

    flc=a; x1;y; xl

    b1x1;y; blxl; q; a

    : 2:8

    Note that (2.8) reduces 1st Heines transformation formula when l 1: For ourpurpose, it is convenient to rewrite (2.8) into the following form:

    l1fla; b1;y; bl

    c1;y; cl; q; u

    auNuN

    Ylk1

    bkNckN

    flD

    u; c1=b1;y; cl=bl

    au; q; b1;y; bl

    : 2:9

    2.3. Very-well-poised hypergeometric series

    The hypergeometric series n1Fn is well-poised if a0 1 a1 b1 ? an bn: It is called very-well-poised if it is well-poised and if a1 a0=2 1: Namely,the very-well-poised n1Fn series is expressed as the following form:

    n1Fna0;

    a0

    2 1; a2; y; an

    a0

    2; a0 a2 1; y; a0 an 1

    ; u

    264

    375

    XkANa0 2k

    a0

    a0ka2k?ankk!a0 a2 1k?a0 an 1k

    uk:

    2:10

    The basic hypergeometric series n1fn is well-poised if

    a0q a1c1 ? ancn: It is called very-well-poised if it is well-poised and if a1 q

    ffiffiffiffiffia0

    pand a2 q ffiffiffiffiffia0p : Namely, the very-well-poised n1fn is expressed as the

    following form

    n1fna0; q

    ffiffiffiffiffia0

    p; q ffiffiffiffiffia0p ; a3; y; an

    ffiffiffiffiffia0

    p;

    ffiffiffiffiffia0

    p; a0q=a3; y; a0q=an

    ; q; u

    XkAN

    1 a0q21 a0

    a0ka3k?ankqka0q=a3k?a0q=ank

    uk: 2:11

    ARTICLE IN PRESS

    Y. Kajihara / Advances in Mathematics 187 (2004) 5397 59

  • 8/9/2019 Kajihara Y - Euler Transformation Formula for Multiple Basic Hypergeoemtric Series of Type a and Some Application

    8/45

    Hereafter, we employ the notation of very-well-poised (basic) hypergeometric

    series as

    n1Fna0; a0

    2 1; a2; y; an

    a0

    2; a0 a2 1; y; a0 an 1

    ; u

    264375

    n1Vna0; a2;y; an; u 2:12

    and

    n1fna0; q

    ffiffiffiffiffia0

    p; q ffiffiffiffiffia0p ; a3; y; an

    ffiffiffiffiffia0p

    ;

    ffiffiffiffiffia0

    p; a0q=a3; y; a0q=an

    ; q; u

    n1Wna0; a3;y; an; q; u 2:13

    for short.

    3. Proof of the multiple q-Euler transformation

    In this section, we show the main theorem of this paper, Euler transformation

    formula between basic hypergeometric series in Un 1 and Um 1 (1.13)XgANn

    ujgjDxqgDx

    Y1pi; jpn

    ajxi=xjgiqxi=xjgi

    Y1pipn;1pkpm

    bkxiyk=xnymgicxiyk=xnymgi

    a1?anb1?bmu=cmN

    uN

    XdANm

    a1?anb1?bmu=cmjdj Dyqd

    Dy

    Y1pk;lpm c=bl

    yk=yl

    dk

    qyk=yldkY

    1pipn;1pkpm c=ai

    xiyk=xnym

    dk

    cxiyk=xnymdk

    by starting from well-known property of Macdonald polynomials in An:First we shall prove

    Proposition 3.1. Formula (1.13) is valid for all a1i qai; 1pipn; bk=c qbk;1pkpm with aANn; bANm:

    To prove this proposition, we consider the following rational function (see

    [11,13]),

    Fujz; w Y

    z; w1DzuY

    z; w 3:1

    ARTICLE IN PRESS

    Y. Kajihara / Advances in Mathematics 187 (2004) 539760

  • 8/9/2019 Kajihara Y - Euler Transformation Formula for Multiple Basic Hypergeoemtric Series of Type a and Some Application

    9/45

    for the variables z z1;y; zr and w w1;y; wp: The rational function Fujz; wis of the form

    Fujz; w X

    KC1;y;rujKjt jKj2

    YiAK; jeK

    1 tzi=zj1 zi=zj

    YiAK;1pkpp

    1 ziwk1 tziwk: 3:2

    We now assume rXp: Note that the rational function Fujz; w has a followingself-duality from (2.6)

    Fujz; w u; trpFutrpjw; z: 3:3For convenience of latter discussion, replace t to q: For each multi-index aANn

    with jaj r; we define the multiple principal specialization y y1;y;yr to thepoints pav; x by

    y y1;y;yr

    -pav; x v=x1;v=x1q;y;v=x1qa11;v=x2;y;

    v=xn;y;v=xnqan1: 3:4We analyze some properties for multiple principal specialization. When we specialize

    z to pav; x; the index set 1;y; r is divided into n blocks with cardinality a1;y; an;respectively. Note also that

    YiAK; jeK

    qzi zjzi zj

    zpav;x

    is equal to zero unless the elements of K should be packed to the left in each block.

    Simple verification leads to the following lemma.

    Lemma 3.1. For such a configuration K; we denote the number of points of K sitting ini-th block for i 1;y; n as gi and set g g1;y; gn: We have

    YiAK; jeK

    qzi

    zj

    zi zj zpav;x YiAK; jeKqzi

    zj

    zi zj zpa1;x qjajjgjq

    jgj2

    DxqgDx

    Y1pi; jpn

    qajxi=xjgiqxi=xjgi

    3:5

    for any v and

    YiAK

    Ymk1

    1 ziwk1 qziwk

    zpa1;x;wpbq

    1;y qjbjjgj q

    bjxiyjgixiyjgi

    : 3:6

    Proof of Proposition 3.1. We first replace t in (3.3) by q; and then specialize z and w

    to pa1; x and pbq1;y; respectively, where x x1;y; xn; y y1;y;ym and

    ARTICLE IN PRESS

    Y. Kajihara / Advances in Mathematics 187 (2004) 5397 61

  • 8/9/2019 Kajihara Y - Euler Transformation Formula for Multiple Basic Hypergeoemtric Series of Type a and Some Application

    10/45

    aANn; bANm are multi-indices with jaj r and jbj p; respectively. Then we have

    XgANn;gpa

    qjajjbjujgj Dxqg

    Dx

    Y

    1pi; jpn

    qajxi=xjgiqxi=xjgi

    Y1pipn;1pkpm

    qbkxiykgixiykgi

    ujajjbjX

    dANm;dpb

    ujdj Dyqd

    Dy

    Y1pk;lpm

    qblyk=yldkqyk=yldk Y1pipn;1pkpm

    qaixiykdkxiykdk

    : 3:7

    Then by changing the variables and parameters u-uqjbjjaj; xi-xi=xn 1pipn;yk-cyk=ym 1pkpm to (3.7), we obtain the desired identity. Thus we complete theproof of the proposition.

    Proof of the main theorem. To prove the main theorem, by replacing u-uqjbj; werewrite (3.7) as follows:

    uqjajN XgANn

    ujgjqjgj2

    Y1pi; jpn

    qaigj1xi=xjgj

    qgigj1xi=xj

    gj Y1pipn;1pkpm

    qbkxiykgi

    xiyk

    gi

    uqjbjN

    XdANm

    ujdjqjdj2

    Y1pk;lpm

    qbkdl1yk=yldlqdkdl1yk=yldl

    Y

    1pipn;1pkpm

    qaixiykdkxiykdk

    : 3:8

    By taking the coefficient of us in (3.8), we obtain

    XgANn;jgjps

    1

    jgj

    qjgj2 Y

    1pi; jpn

    qaigj1xi=xjgjqgigj1xi=xjgj Y1pipn;1pkpm

    qbkxiykgixiykgi

    qjajsjgjqsjgj

    2

    X

    dANm;jdjps1jdjq

    jdj2

    Y1pk;lpm

    qbkdl1yk=yldlqdkdl1yk=yldl

    Y1pipn;1pkpm

    qaixiykdkxiykdk

    qjbjsjdjqsjdj

    2

    : 3:9

    Note that this is a polynomial identity of qai; 1pipn and qbk; 1pkpm: Hence (1.13)is valid for all a1i 1pipn; bk=c 1pkpmAC for formal power series of u: Thusthe main theorem is proved.

    ARTICLE IN PRESS

    Y. Kajihara / Advances in Mathematics 187 (2004) 539762

  • 8/9/2019 Kajihara Y - Euler Transformation Formula for Multiple Basic Hypergeoemtric Series of Type a and Some Application

    11/45

    Remark 3.1. The homogeneous version of the multiple q-Euler transformation (1.13)

    is expressed as

    ut1?tnNXgANn

    ut1?tnjgj DxqgDx

    Y

    1pi; jpn

    t1j xi=xjgiqxi=xjgi

    Y1pipn;1pkpm

    skxiykgixiykgi

    us1?smNXdANm

    us1?smjdj Dyqd

    Dy

    Y1pk;lpms1l yk=yldkqyk=yldk Y1pipn;1pkpm

    tixiykdkxiykdk

    :

    3:10

    4. Some special and limiting cases of multiple q-Euler transformation

    In this section, we give some formulas of special cases of our Euler transformation

    formula (1.13)

    XgANn

    ujg

    jD

    xqg

    Dx Y1pi; jpn

    ajxi=xj

    gi

    qxi=xjgiY

    1pipn;1pkpm

    bkxiyk=xnym

    gi

    cxiyk=xnymgi a1?anb1?bmu=c

    mN

    uN

    XdANm

    a1?anb1?bmu=cmjdj Dyqd

    Dy

    Y

    1pk;lpm

    c=blyk=yldkqyk=yldk

    Y1pipn;1pkpm

    c=aixiyk=xnymdkcxiyk=xnymdk

    and the corresponding q-1 limits.

    Proposition 4.1. In the case when m 1; the multiple q-Euler transformation formula(1.13) reduces to

    XgANn

    ujgjDxqgDx

    Y1pi; jpn

    ajxi=xjgiqxi=xjgi

    Y1pipn

    bxi=xngicxi=xngi

    a1?anbu=cN

    uN

    n1fnc=b; c=a1x1=xn;y; c=an1xn1=xn; c=an

    cx1=xn;y; cxn1=xn; c; q; a1yanbu=c

    : 4:1

    ARTICLE IN PRESS

    Y. Kajihara / Advances in Mathematics 187 (2004) 5397 63

  • 8/9/2019 Kajihara Y - Euler Transformation Formula for Multiple Basic Hypergeoemtric Series of Type a and Some Application

    12/45

    Furthermore, This formula has an alternative expression in terms of Lauricella basic

    hypergeometric function

    XgANn

    ujgjDxqgDx

    Y1pi; jpn

    ajxi=xjgiqxi=xjgi

    Y1pipn

    bxi=xngicxi=xngi

    a1?anuNuN

    Y1pipn

    c=aixi=xnNcxi=xnN

    fnDa1?anbu=c; a1;y; an

    a1?anu; q; c=a1x1=xn;y; c=an1xn1=xn; c=an

    : 4:2

    Formula (4.2) is obtained from (4.1) by using Andrews transformation (2.9).

    Remark 4.1. In the case when n 1; (4.2) reduces to the 2nd Heines transformationformula (1.8).

    Formula (4.1) will be a key to giving elementary proofs of some summation

    formulas for basic hypergeometric series in Un 1 in the next section of thepresent paper.

    Remark 4.2. In the case when m n and yk x1

    k ; (1.13) reduces to

    XgANn

    ujgjDxqgDx

    Y1pi; jpn

    ajxi=xjgiqxi=xjgi

    bjxi=xjgicxi=xjgi

    a1?anb1?bnu=cnN

    uN

    XdANn

    a1?anb1?bnu=cnjdj Dx1qd

    Dx1

    Y1pi; jpn

    c=bixj=xidiqxj=xidic=aixi=xjdicxi=xjdi

    : 4:3

    The informed reader might compare this formula with the third Heines

    transformation formula for basic hypergeometric series in Un 1 due toGustafson and Krattenthaler [6].

    Now, we will write down the q-1 limit of the formulas in this section.

    To this end, note that

    limq-1

    1 qz1 q z: 4:4

    ARTICLE IN PRESS

    Y. Kajihara / Advances in Mathematics 187 (2004) 539764

  • 8/9/2019 Kajihara Y - Euler Transformation Formula for Multiple Basic Hypergeoemtric Series of Type a and Some Application

    13/45

    As an application of (4.4), we have

    limq-1

    qzn1 qn

    zn:

    4:5

    Note also that

    limq-1

    qzuN

    uN

    1 uz 4:6

    and

    limq-1

    qN

    qsN

    1 qs1 Gs; 4:7

    where Gs is the Euler gamma function. Before taking the limit, we replace all theparameters z in our formulas by qz:

    Then, by taking the limit q-1 in Eq. (1.13), we have the Euler transformation for

    multiple hypergeometric series of type A:

    Proposition 4.2.

    XgANn

    ujgjDx gDx

    Y1pi; jpn

    aj xi xjgi1 xi xjgi

    Y1pipn;1pkpm

    bk xi yk xn ymgic xi yk xn ymgi 1 umc

    P1pipn

    aiP

    1pkpmbk

    XdANm

    ujdjDy dDy

    Y1pk;lpm

    c bl yk yldk1 yk yldk

    Y1pipn;1pkpm

    c ai xi yk xn ymdkc xi yk xn ymdk

    : 4:8

    Corollary 4.1. In the case when m 1; our Euler transformation formula for multiple(ordinary) hypergeometric series (4.8) reduces to

    XgANn

    ujgjDx gDx

    Y1pi; jpn

    aj xi xjgi1 xi xjgi

    Y1pipn

    b xi xngic xi xngi

    1 ucbP

    1pipnai

    n1Fn c b; c a1 x1 xn;y; c an1 xn1 xn; c an

    c x1 xn;y; c xn1 xn; c; u

    : 4:9

    ARTICLE IN PRESS

    Y. Kajihara / Advances in Mathematics 187 (2004) 5397 65

  • 8/9/2019 Kajihara Y - Euler Transformation Formula for Multiple Basic Hypergeoemtric Series of Type a and Some Application

    14/45

    Remark 4.3. The q-1 limit of (4.3) is

    XgANn uj

    g

    jD

    x

    g

    Dx Y1pi; jpnaj xi xjgi1 xi xjgi

    bj xi xjgic xi xjgi

    1 uncP

    1pipnaibi X

    dANn

    ujdjDx dDx

    Y

    1pi; jpn

    c bi xj xidi1 xj xidi

    c ai xi xjdic xi xjdi

    : 4:10

    Remark 4.4. The homogeneous version of multiple Euler transformation (4.8) is

    expressed as

    1 uP

    itiXgANn

    ujgjDx gDx

    Y

    1pi; jpn

    ti xi xjgi1 xi xjgi

    Y1pipn;1pkpm

    sk xi ykgixi ykgi

    1 uP

    ksk

    XdANm

    ujdjDy dDy

    Y1pk;lpm

    sl

    yk

    yl

    dk

    1 yk yldkY

    1pipn;1pkpmti

    xi

    yk

    dk

    xi ykdk : 4:11

    5. Simple proofs of some summation formulas for basic hypergeometric series in

    Un 1

    In this section, we give simple proofs of the PfaffSaalschutz summation formula

    and the Gauss summation formula for basic hypergeometric series in Un 1 byusing the special case of our Euler transformation formula for multiple basichypergeometric series (4.1)

    XgANn

    ujgjDxqgDx

    Y1pi; jpn

    ajxi=xjgiqxi=xjgi

    Y1pipn

    bxi=xngicxi=xngi

    a1?anbu=cNuN

    n1fnc=b; c=a1x1=xn;y; c=an1xn1=xn; c=an

    cx1=xn;y; cxn

    1=xn; c; q; a1y; anbu=c :

    By virtue of this formula, our proofs are essentially the same as in the case of basic

    hypergeometric series of one variable.

    ARTICLE IN PRESS

    Y. Kajihara / Advances in Mathematics 187 (2004) 539766

  • 8/9/2019 Kajihara Y - Euler Transformation Formula for Multiple Basic Hypergeoemtric Series of Type a and Some Application

    15/45

    Proposition 5.1 (PfaffSaalschutz summation formula for basic hypergeometric

    series in Un 1).X

    jgjANnqjgj Dxqg

    DxY

    1pi; jpn

    ajxi=xjgiqxi=xjgi

    Y1pipn

    bxi=xngicxi=xngi

    qNjgja1?anbq1N=cjgj

    c=bNc=a1?anbNY

    1pipn

    c=aixi=xnNcxi=xnN

    : 5:1

    Remark 5.1. This formula is already known by S.C. Milne ([18, Theorem 4.15]). In

    the case when n 1; (5.1) reduces to the ordinary PfaffSaalschutz summationformula for terminating balanced 3f2 series ( formula (1.7.2) in [5])

    3f2a; b; qN

    c; q1Nab=c; q; q

    c=aNc=bNcNc=abN

    : 5:2

    Proof. We rewrite Eq. (4.1) as

    cu=a1?anbN

    u

    N XgANn

    cu=a1?anbjgj Dxqg

    D

    x

    Y1pi; jpn

    ajxi=xjgiqxi=xjgi

    Y1pipn

    bxi=xngicxi=xngi

    n1fnc=b; c=a1x1=xn;y; c=an1xn1=xn; c=an

    cx1=xn;y; cxn1=xn; c; q; u

    : 5:3

    By taking the coefficient of uN of the equation above, we have

    XgANn;jgjpl

    c=a1?anb

    jgj Dxqg

    Dx Y1pi; jpnajxi=xjgiqxi=xjgi

    Y

    1pipn

    bxi=xngicxi=xngi

    c=a1?anbNjgjqNjgj

    c=bNqNY

    1pipn

    c=aixi=xnNcxi=xnN

    : 5:4

    Note that

    z

    N

    r

    qNr z

    N

    qN qN

    r

    z1q1Nrq

    z r

    : 5:5Then a simple verification leads to the desired identity (5.1). &

    ARTICLE IN PRESS

    Y. Kajihara / Advances in Mathematics 187 (2004) 5397 67

  • 8/9/2019 Kajihara Y - Euler Transformation Formula for Multiple Basic Hypergeoemtric Series of Type a and Some Application

    16/45

    By taking the limit l-N; we have

    Proposition 5.2 (Gauss summation formula for basic hypergeometric series in

    Un 1).

    XgANn

    c=a1?anbjgj Dxqg

    DxY

    1pi; jpn

    ajxi=xjgiqxi=xjgi

    Y1pipn

    bxi=xngicxi=xngi

    c=bNc=a1?anbNY

    1pipn

    c=aixi=xnNcxi=xnN

    : 5:6

    Remark 5.2. This formula is already known by S.C. Milne ([16, Theorem 3.9]). Inthe case when n 1; Eq. (5.6) reduces to the ordinary Gauss summation formula for2f1 series ( formula (1.5.1) in [5])

    2f1a; b;

    c;; q; c=ab

    c=aNc=bNc

    Nc=ab

    N

    : 5:7

    Similarly, from (4.9)

    XgANn

    ujgjDx gDx

    Y1pi; jpn

    aj xi xjgi1 xi xjgi

    Y1pipn

    b xi xngic xi xngi

    1 ucb

    P1pipn

    ai

    n1Fnc b; c a1 x1 xn;y; c an1 xn1 xn; c an

    c

    x1

    xn;y; c

    xn

    1

    xn; c

    ; u ;

    the PfaffSaalschutz summation formula

    Xjgjpl

    DxgDx

    Y1pi; jpn

    aj xi xjgi1 xi xjgi

    Y1pipn

    bxi xngic xi xngi

    Njgj

    1

    N

    a1

    ?

    an

    b

    c

    jg

    j c bNc a1 ? an bN

    Y1pipn

    c ai xi xnNc xi xnN

    5:8

    ARTICLE IN PRESS

    Y. Kajihara / Advances in Mathematics 187 (2004) 539768

  • 8/9/2019 Kajihara Y - Euler Transformation Formula for Multiple Basic Hypergeoemtric Series of Type a and Some Application

    17/45

    and the Gauss summation formula

    XgANn

    D

    x

    g

    Dx Y1pi; jpnaj xi xjgi1 xi xjgi Y1pipn

    b xi xngic xi xngi

    Gc a1 ? an bGc b

    Y1pipn

    Gc xi xnGc ai xi xn 5:9

    for hypergeometric series in Un 1 are derived directly in the same way as in thecase of basic hypergeometric series in Un 1 case.

    Remark 5.3. Eqs. (5.8) and (5.9) are the q-1 limiting cases of (5.1) and (5.6),

    respectively. In the case when n 1; (5.8) reduces to the PfaffSaalschutz summationformula for balanced terminating

    3F

    2series

    3F2N; a; b

    c; a b c N 1 ; 1

    c aNc bNcNc a bN5:10

    and (5.6) reduces to the Gauss summation formula for Gauss hypergeometric

    series 2F1

    2F1a; b

    c; 1

    GcGc a bGc aGc b: 5:11

    6. Beyond the q-PfaffSaalschutz summation formula in Un 1

    In this section, we discuss some generalizations of multiple q-PfaffSaalschutz

    formula by using the multiple q-Euler transformation formula (1.13). As a result, we

    derive some multiple transformation formulas for very-well-poised basic hypergeo-

    metric series and recover a multiple generalization of the Jacksons 8W7 summation

    theorem due to Milne [17] by using the same method as in the previous section.

    Before preceding our derivations, let us explain the essence of our discussion. Note

    that, the homogeneous part of the summation (the summand is taken by all the

    multiindices bANn1 with jbj N for any nonnegative integer N) in the multiplebasic hypergeometric series is of the form

    XbANn1;jbjN

    DxqbDx u

    b11 ?u

    bn1n1

    basic hypergeometric stuff

    XbANn

    D

    xqb

    Dx Y1pipn

    1

    aqjbjbixi=xn

    1 axi=xn zb11 ?z

    bnn

    another basic hypergeometric stuff; 6:1

    ARTICLE IN PRESS

    Y. Kajihara / Advances in Mathematics 187 (2004) 5397 69

  • 8/9/2019 Kajihara Y - Euler Transformation Formula for Multiple Basic Hypergeoemtric Series of Type a and Some Application

    18/45

    where we notice that bn1 N b1 ? bn: Furthermore, the case when n 1;that series in right-hand side of equation above is reduced to very-well-poised basic

    hypergeometric series which are introduced in Section 2. In the course of his

    derivation of multiple q-binomial theorem, Milne [14] proved a multiple general-ization of Rogers terminating very-well-poised 6W5 summation theorem. Con-

    versely, after changing n-n 1; take the coefficients ofuN in the both side of (1.11)b1?bnbn1uN

    uN

    X

    bANn1ujbj Dxq

    bDx

    Y1pi; jpn1

    bjxi=xjbiqxi=xjbi

    :

    Then by changing the parameter suitably, we obtain multiple Rogers terminating

    very-well-poised 6W5 summation formula [18],

    aq=b1?bncNaq=cN

    Y1pipn

    aqxi=xnNaq=bixi=xnN

    XbANn

    aq1N

    b1?bnc

    jbjDxqbDx

    Y1pipn

    1 aqjbjbixi=xn1 axi=xn

    Y

    1pi; jpn

    bjxi=xjbiqxi=xjbi

    Y1pipn

    cxi=xnbiaq1Nxi=xnbi

    qNjbjaq=cjbj Y1pipn

    axi=xnjbjaq=bixi=xnjbj

    :

    6:2

    Note that in the case when n 1; (6.2) reduces to the Rogers terminating very-well-poised 6W5 summation (cf. [5])

    aq=bcNaq=cN

    aqNaq=bN

    6W5 a; b; c; qN; q; aq1N

    bc

    : 6:3

    Multiple q-binomial theorem and our Euler transformation formula (1.13) can be

    interpreted as a generating functions of (multiple) very-well-poised hypergeometric

    series. We shall carry out the latter one in several cases.

    6.1. Multiple generalization of Watson type transformation formula

    In this subsection, we give a multiple transformation formula of Watson type for

    very-well-poised basic hypergeometric series (see Proposition 6.1 below). As a special

    case, it implies the following transformation formula between a 4f3 series and a very-

    well-poised 8W7 series:

    8W7 a; b; c; d; e; qN; q;

    a2qN2

    bcde a

    2q2=bcdeNeNaqNaq=bNaq=cNaq=dN4

    f3qN; aq=be; aq=ce; aq=deq1N=e; a2q2=bcde; aq=e

    ; q; q

    : 6:4

    ARTICLE IN PRESS

    Y. Kajihara / Advances in Mathematics 187 (2004) 539770

  • 8/9/2019 Kajihara Y - Euler Transformation Formula for Multiple Basic Hypergeoemtric Series of Type a and Some Application

    19/45

    Formula (6.4) is similar to a Watson transformation formula (Eq. (2.5.1) in [5])

    8W7 a; b; c; d; e; qN

    ; q;

    a2q2N

    bcde

    aqNaq=deNaq=dNaq=eN4f3

    qN; d; e; aq=bcaq=b; aq=c; deqN=a

    ; q; q

    : 6:5

    In fact, (6.4) includes the same number of free parameters as (6.5) and right-hand

    side of (6.4) and (6.5) are balanced. Formula (6.5) may be obtained by combining

    Sears transformation (see next section) from (6.4). In this paper, we call (6.4) a

    Watson type transformation formula. The informed reader might compare some

    formula for multiple Watson transformation formula in Milne [17], MilneLily [19]

    and Schlosser [22].

    Proposition 6.1

    an1qn1=b1?bmcd1?dnenNaq=cN

    Y1pkpmaqyk=ymN

    aq=bkyk=ymN Y1pipnexn=xiN

    aq=dixn=xiN

    XgANn;jgjpN

    qjgjDxqgDx

    Y1pi; jpn

    aq=djexi=xjgiqxi=xjgi

    Y

    1pipn;1pkpm

    aq=bkexiyk=xnymgiaq=exiyk=xnymgi

    Y1pipn

    aq=cexi=xngiq1Ne1xi=xngi

    qNjgj

    an1qn1=b1?bmcd1?dnenjgj X

    dANma

    n

    1

    qN

    n

    1

    =b1?bmcd1?dnen

    jdj

    Dyqd

    DyY

    1pkpm

    1 aqjdjdkyk=ym1 ayk=ym

    Y

    1pk;lpm

    blyk=yldkqyk=yldk

    Y1pkpm

    cyk=ymdkaqNyk=ymdk

    Y1pipn;1pkpm

    dixiyk=xnymdkaq=exiyk=xnymdk

    qNjdj

    aq=cjdjY

    1pkpm

    ayk=ymjdjaq=bkyk=ymjdj

    Y1pipn

    exn=xijdjaq=dixn=xijdj

    : 6:6

    ARTICLE IN PRESS

    Y. Kajihara / Advances in Mathematics 187 (2004) 5397 71

  • 8/9/2019 Kajihara Y - Euler Transformation Formula for Multiple Basic Hypergeoemtric Series of Type a and Some Application

    20/45

    Proof. By taking the coefficient of uN in

    cm1u=a1?anb1?bmbm1NuN

    XgANn

    cm1u=a1?anb1?bmbm1jgj Dxqg

    Dx

    Y

    1pi; jpn

    ajxi=xjgiqxi=xjgi

    Y1pipn;1pkpm1

    bkxiyk=xnym1gicxiyk=xnym1gi

    XdANm1

    ujdjD yqdD

    y Y1pk;lpm1

    c=blyk=yldk

    qyk=yldk

    Y

    1pipn;1pkpm1

    c=aixiyk=xnym1dkcxiyk=xnym1dk

    ; 6:7

    where y y1;y;ym;ym1; we have

    cm1=a1?anb1?bmbm1Nc=bm1N Y1pkpm

    ym1=ykNc=bkym1=ykN

    Y

    1pipn

    cxi=xnNc=aixi=xnN

    X

    gANn;jgjpNqjgj

    DxqgDx

    Y1pi;jpn

    ajxi=xjgiqxi=xjgi

    Y

    1pipn;1pkpm

    bkxiyk=xnym1gicxiyk=xnym1gi

    Y1pipn

    bm1xi=xngicxi=xngi

    qN

    jgjq1Na1?anb1?bmbm1=cm1jgj

    XdANm

    a1?anb1?bmbm1q=cm1jdj

    Dyqd

    DyY

    1pkpm

    1 qNjdjdkyk=ym11 qNyk=ym1

    Y1pk;lpm

    c=blyk=yldkqyk=yldk Y1pkpm

    c=bm1yk=ym1dkqyk=ym1dk

    Y

    1pipn;1pkpm

    c=aixiyk=xnym1dkcxiyk=xnym1dk

    ARTICLE IN PRESS

    Y. Kajihara / Advances in Mathematics 187 (2004) 539772

  • 8/9/2019 Kajihara Y - Euler Transformation Formula for Multiple Basic Hypergeoemtric Series of Type a and Some Application

    21/45

    qNjdj

    q1Nbm1=cjdjY

    1pkpm

    qNyk=ym1jdjq1Nbk=cyk=ym1jdj

    Y

    1pipn

    q1Nc1xn=xijdjq1Nai=cxn=xijdj

    : 6:8

    After some simple verification in the equation above, we obtain

    cm1=a1?anb1?bmbm1Nc=bm1N

    Y1pkpm

    ym1=ykNc=bkym1=ykN

    Y1pipn

    cxi=xnNc=aixi=xnN

    X

    gANn;jgjpNqjgj

    DxqgDx

    Y1pi; jpn

    ajxi=xjgiqxi=xjgi

    Y

    1pipn;1pkpm

    bkxiyk=xnym1gicxiyk=xnym1gi

    Y1pipn

    bm1xi=xngicxi=xngi

    qNjgj

    q1

    Na

    1?a

    nb

    1?b

    mb

    m1=cm

    1

    jgjXdANm

    a1?anb1?bmbm1q=cm1jdj

    Dyqd

    DyY

    1pkpm

    1 qNjdjdkyk=ym11 qNyk=ym1

    Y

    1pk;lpm

    c=blyk=yldkqyk=yldk

    Y1pkpm

    c=bm1yk=ym1dkqyk=ym1dk

    Y1pipn;1pkpmc=aixiyk=xnym1dk

    cxiyk=xnym1dk

    qNjdj

    q1Nbm1=cjdjY

    1pkpm

    qNyk=ym1jdjq1Nbk=cyk=ym1jdj

    Y

    1pipn

    q1Nc1xn=xijdjq1Nai=cxn=xijdj

    : 6:9

    Finally by changing the parameters suitably, we have the desired formula. &

    In the case when m 1; the right-hand side of Eq. (6.6) reduces to the specialvalue of very-well-poised basic hypergeometric series.

    ARTICLE IN PRESS

    Y. Kajihara / Advances in Mathematics 187 (2004) 5397 73

  • 8/9/2019 Kajihara Y - Euler Transformation Formula for Multiple Basic Hypergeoemtric Series of Type a and Some Application

    22/45

    Corollary 6.1.

    an1qn1=bcd1?dnen

    N

    aq=cN aq

    N

    aq=bN Y1pipn exn=xi

    N

    aq=dixn=xiN

    X

    gANn;jgjpNqjgj

    DxqgDx

    Y1pi; jpn

    aq=djexi=xjgiqxi=xjgi

    Y

    1pipn

    aq=bexi=xngiaq=cexi=xngiaq=exi=xngiq1Ne1xi=xngi

    " #

    qNjgj

    an1qn1=bcd1?dnenjgj 2n6W2n5a; b; c; d1x1=xn;?dn1xn1=xn; dn;

    exn=x1;?exn=xn1; e; qN; q; an1qNn1=bcd1?dnen: 6:10

    In the case when n 1; Eq. (6.6) reduces to

    Corollary 6.2.

    a2q2=b1?bmcdeNaq=cN

    exNaq=dN

    Y1pkpm

    aqyk=ymNaq=bkyk=ymN

    m3fm2qN; aq=b1ey1=ym;y;

    a2q2=b1?bmcde; aq=ey1=ym;y;

    aq=bm1eym1=ym; aq=bme; aq=ce; aq=de

    aq=eym1=ym; aq=e; q1N=e; q; q

    XdANm

    a2qN2=b1?bmcdejdj

    DyqdDy

    Y1pkpm

    1 aqjdjdkyk=ym1 ayk=ym

    Y

    1pk;lpm

    blyk=yldkqyk=yldk

    Y1pkpm

    cyk=ymdkdyk=ymdkaqNyk=ymdkaq=eyk=ymdk

    " #

    qNjdjejdj

    aq=cjdjaq=djdjY

    1pkpm

    ayk=ymjdjaq=bkyk=ymjdj

    : 6:11

    Remark 6.1. Eq. (6.6) is a multiple generalization of Watson type transformation.

    Namely, (6.4) is a n 1; m 1 case of (6.6).

    ARTICLE IN PRESS

    Y. Kajihara / Advances in Mathematics 187 (2004) 539774

  • 8/9/2019 Kajihara Y - Euler Transformation Formula for Multiple Basic Hypergeoemtric Series of Type a and Some Application

    23/45

    6.2. BaileyJackson type transformation-summation formula

    We give a multiple transformation-summation formula of very-well-poised basic

    hypergeometric series including, as special cases, Jacksons terminating 8W7summation formula

    8W7a; b; c; d; e; qN; q; q

    aqNaq=bcNaq=bdNaq=cdNkq=bNaq=cNaq=dNaq=bcdN6:12

    when a2qN1 bcde (Eq. (2.6.1) in [5]), and 10W9 transformation formula

    aq=bNaq=cNaq=dNaq=eNmqNmf=aNmbf=aNmcf=aNmdf=aNmef=aNaqN fN 10W9a; b; c; d; e;f; mfqN; qN; q; q

    10W9m; aq=bf; aq=cf; aq=df; aq=ef; mf=a; mfqN; qN; q; q; 6:13

    where m a3q2=bcdef2:Note that (6.13) is similar to the Bailey transformation formula for a terminating

    balanced 10W9 series (Eq. (2.9.1) in [5])

    10W9a; b; c; d; e;f; laqN1=ef; qN; q; q

    aq

    N

    aq=ef

    N

    lq=e

    N

    lq=f

    N

    aq=eNaq=fNlqNlq=efN 10W9l; lb=a; lc=a; ld=a; e;f; laqN1=ef; qN; q; q; 6:14

    where l a2q=bcd:Though it has same conditions (number of free parameters and series of both side

    are balanced) as in Baileys 10W9 transformation (6.14), formula (6.13) is different

    from known ones. One can check that both the Bailey transformation (6.14) and our

    10W9 transformation (6.13) can be obtained by iterating twice the another one. For

    comparison to other multiple 10W9 transformation formula in An; see MilneNewcomb [20] and DenisGustafson [4].

    In this paper, we call (6.13) the Bailey type transformation.

    ARTICLE IN PRESS

    Y. Kajihara / Advances in Mathematics 187 (2004) 5397 75

  • 8/9/2019 Kajihara Y - Euler Transformation Formula for Multiple Basic Hypergeoemtric Series of Type a and Some Application

    24/45

    Proposition 6.2.

    XgANn;jgjN

    DxqgDx Y1pi; jpn

    ajxi=xjgiqxi=xjgi Y1pipn;1pkpm

    bkxiyk=xnymgicxiyk=xnymgi

    X

    dANm;jdjN

    DyqdDy

    Y1pk;lpm

    c=blyk=yldkqyk=yldk

    Y

    1pipn;1pkpm

    c=aixiyk=xnymdkcxiyk=xnymdk

    6:15

    when a1?anb1?bm cm:

    Proof. In the case when a1?anb1?bm cm; (1.13) is written asXgANn

    ujgjDxqgDx

    Y1pi; jpn

    ajxi=xjgiqxi=xjgi

    Y1pipn;1pkpm

    bkxiyk=xnymgicxiyk=xnymgi

    XdANm

    ujdjDyqdDy

    Y1pk;lpm

    c=blyk=yldkqyk=yldk

    Y1pipn;1pkpm

    c=aixiyk=xnymdkcxiyk=xnymdk

    : 6:16

    By taking the coefficient of uN in (6.16), we have the desired formula. &

    In this paper, we call (6.15) the BaileyJackson type transformation-summation

    formula.

    Now we give a multiple generalization of Bailey type transformation (6.13).

    Corollary 6.3.

    aq=d

    N

    aq=e

    N

    mdf=aNmef=aNY

    1pkpmmqyk=ym

    N

    aq=ck

    ym=yk

    N

    mckf=ayk=ymN fym=ykN

    Y1pipn

    aq=bixi=xnNmf=axn=xiNaqxi=xnNmbif=axn=xiN

    XgANn

    qjgjDxqgDx

    Y1pipn

    1 aqjgjgixi=xn1 axi=xn

    Y1pi; jpn

    bjxi=xjgiqxi=xjgi Y1pipn

    dxi=xngiaqN1xi=xngi

    Y

    1pipn;1pkpm

    ckxiyk=xnymgiaq=fxiyk=xnymgi

    Y1pipn

    exi=xngiaq1N=mfxi=xngi

    ARTICLE IN PRESS

    Y. Kajihara / Advances in Mathematics 187 (2004) 539776

  • 8/9/2019 Kajihara Y - Euler Transformation Formula for Multiple Basic Hypergeoemtric Series of Type a and Some Application

    25/45

    qNjgj

    aq=djgjY

    1pipn

    axi=xnjgjaq=bixi=xnjgj

    mfqNjgjaq=ejgjY

    1pkpm

    fym=ykjgjaq=ckym=ykjgj

    XdANm

    qjdjDyqdDy

    Y1pkpm1

    1 mqjdjdkyk=ym1 myk=ym

    Y

    1pk;lpm

    aq=clfyk=yldkqyk=yldk

    Y1pkpm

    aq=efyk=ymdkmqN1yk=ymdk

    Y1pipn;1pkpmaq=bi fxiyk=xnymdkaq=fxiyk=xnymdk Y1pkpm

    aq=dfyk=ymdkq

    1

    N

    =fyk=ymdk

    qNjdj

    mef=ajdjY

    1pkpm

    myk=ymjdjmck f=ayk=ymjdj

    mfqNjdj

    mdf=ajdjY

    1pipn

    mf=axn=xijdjmbi f=axn=xijdj

    ; 6:17

    where m am2qm1=b1?bnc1?cmdefm1:

    Proof. Replace n to n 1 and m to m 1 in (6.15). Some simple but longmanipulation leads to the following in the case when a1?anan1b1?bmbm1 cm1

    an1Nc=an1N

    bm1Nc=bm1N

    Y1pkpm

    ym1=ykNc=bkym1=ykN

    bkyk=ym1Ncyk=ym1N

    Y

    1pipn

    cxi=xn1Nc=aixi=xn1N

    aixn1=xiNxn1=xiN

    XgANn qjg

    jD

    xqg

    Dx Y1pipn1

    qNjgjgixi=xn1

    1 qNxi=xn1

    Y

    1pi; jpn

    ajxi=xjgiqxi=xjgi

    Y1pipn

    an1xi=xn1giqxi=xn1gi

    Y

    1pipn;1pkpm

    bkxiyk=xn1ym1gicxiyk=xn1ym1gi

    Y1pipn

    bm1xi=xn1gicxi=xn1gi

    qNjgj

    q1Na1n1jgj Y1pipnqNxi=xn1jgj

    q1Na1i xi=xn1jgj

    Y

    1pkpm

    q1Nc1ym1=ykjgjq1Nb1k ym1=ykjgj

    q1Nc1jgjq1Nb1m1jgj

    ARTICLE IN PRESS

    Y. Kajihara / Advances in Mathematics 187 (2004) 5397 77

  • 8/9/2019 Kajihara Y - Euler Transformation Formula for Multiple Basic Hypergeoemtric Series of Type a and Some Application

    26/45

    X

    dANm;jdjpNqjdj

    DyqdDy

    Y1pkpm

    1 qNjdjdkyk=ym11 qNyk=ym1

    Y

    1pk;lpm

    c=blyk=yldkqyk=yldk

    Y1pkpm1

    c=bm1yk=ym1dkqyk=ym1dk

    Y

    1pipn;1pkpm

    c=aixiyk=xn1ym1dkcxiyk=xn1ym1dk

    Y1pkpm

    c=an1yk=ym1dkcyk=ym1dk

    qNjdj

    q1Nbm1=cjdjY

    1pkpm

    qNyk=ym1jdjq1Nbk=cyk=ym1jdj

    Y1pipnq1Nc1xn1=xijdj

    q1

    N

    ai=cxn1=xijdjq1Nc1jdj

    q1

    N

    an1=cjdj:

    6:18

    Finally by changing the parameters suitably, we have the desired identity (6.17). &

    Corollary 6.4. In the case when m 1; (6.17) reduces to

    mqNaq=cNaq=dNaq=eNfNmcf=aNmdf=aNmef=aN Y

    1pipn

    aq=bixi=xnNmf=axn=xiNaqxi=xnNmbi f=axn=xiN

    XgANn

    qjgjDxqgDx

    Y1pipn

    1 aqjgjgixi=xn1 axi=xn

    Y1pi;jpn

    bjxi=xjgiqxi=xjgi

    Y

    1pipn

    cxi=xngidxi=xngiexi=xngiaq=fxi=xngiaqN1xi=xngiaq1N=mfxi=xngi

    " #

    fjgjmfqNjgjqNjgjaq=cjgjaq=djgjaq=ejgj Y1pipn

    axi=xnjgjaq=bixi=xnjgj

    2n8W2n7m; aq=b1 fx1=xn;y; aq=bn1 fxn1=xn; aq=bn f; aq=cf;

    aq=df; aq=ef; mf=axn=x1;y; mf=axn=xn1; mf=a; mfqN; qN; q; q 6:19

    where m a3q2=b1?bncdef2:

    Remark 6.2. m n 1 case of (6.17) is (6.13).

    In the case when m 1; (6.15) reduces to the following multiple generalization ofJackson summation (6.12) by some manipulations and change of parameters as in

    multiple Bailey type transformation.

    ARTICLE IN PRESS

    Y. Kajihara / Advances in Mathematics 187 (2004) 539778

  • 8/9/2019 Kajihara Y - Euler Transformation Formula for Multiple Basic Hypergeoemtric Series of Type a and Some Application

    27/45

    Corollary 6.5.

    XgANn

    qjg

    jD

    xqg

    Dx Y1pipn

    1

    aqjgjgixi=xn

    1 axi=xn

    Y

    1pi; jpn

    bjxi=xjgiqxi=xjgi

    Y1pipn

    cxi=xngiexi=xngiaqN1xi=xngiaq=dxi=xngi

    " #

    qNjgj

    aq=cjgjdjgj

    aq=ejgjY

    1pipn

    axi=xnjgjaq=bixi=xnjgj

    aq=b1?bncNaq=cdNaq=b1?bncdNaq=cN Y1pipnaq=bidxi=xnNaqxi=xnN

    aq=bixi=xnNaq=dxi=xnN6:20

    provided a2qN1 b1?bncde:

    Remark 6.3. Formula (6.20) is already known by Milne (Theorem 4.70 in [17]). The

    n 1 case of (6.20), namely m 1; n 2 of (6.15) is (6.12).

    6.3. q-1 limits

    We can also obtain the results for ordinary hypergeometric series from (4.8) in thesame way as in basic hypergeometric case. Here, we state the q-1 limits of the

    transformation and summation formulas of this section.

    First, we give some q-1 limits of Watson type transformations.

    Proposition 6.3. The q-1 limit of (6.6) is

    n 1a n 1 b1 ? bm c d1 ? dn neNa 1 cN

    Y1pkpm

    a

    1

    yk

    ym

    N

    a 1 bk yk ymNY

    1pipne

    xn

    xi

    N

    a 1 di xn xiN

    X

    gANn;jgjpN

    Dx gDx

    Y1pi; jpn

    a 1 dj e xi xjgi1 xi xjgi

    Y

    1pipn;1pkpm

    a 1 bk e xi yk xn ymgia 1 e xi yk xn ymgi

    Y1pipn

    a 1 c e xi yk xn ymgi

    1

    N

    e

    xi

    yk

    xn

    ym

    gi

    Njgjn 1a n 1 b1 ? bm c d1 ? dn nejgj

    ARTICLE IN PRESS

    Y. Kajihara / Advances in Mathematics 187 (2004) 5397 79

  • 8/9/2019 Kajihara Y - Euler Transformation Formula for Multiple Basic Hypergeoemtric Series of Type a and Some Application

    28/45

    XdANm

    Dy dDy

    Y1pkpm

    a jdj dk yk yma yk ym

    Y1pk;lpm

    bl yk yldk1 yk yldkY

    1pkpm

    c yk ymdka N yk ymdk

    Y

    1pipn;1pkpm

    di xi yk xn ymdka 1 e xi yk xn ymdk

    Njdja 1 cjdjY

    1pkpm

    a yk ymjdja 1 bk yk ymjdj

    Y1pipn

    e xn xijdj

    a

    1

    di

    xn

    xi

    jdj: 6:21

    In the case when m 1; the right-hand side of Eq. (6.21) reduces to the specialvalue of very-well-poised hypergeometric series.

    Corollary 6.6

    n 1a n 1 b c d1 ? dn neN

    a

    1

    c

    N

    a 1Na 1 bNY

    1pipn

    e xn xiNa 1 di xn xiN

    X

    gANn;jgjpN

    Dx gDx

    Y1pi; jpn

    a 1 dj e xi xjgi1 xi xjgi

    Y

    1pipn

    a 1 b e xi xngia 1 c e xi xngia 1 e xi xngi1 N e xi xngi

    " #

    N

    jg

    jn 1a n 1 b c d1 ? dn nejgj 2n5V2n4a; b; c; d1 x1 xn;?dn1 xn1 xn; dn;

    e xn x1;?e xn xn1; e;N; 1: 6:22

    Corollary 6.7. In the case when n 1; Eq. (6.21) reduces to

    2a 2 b1 ? bm c d eN

    a

    1

    c

    N

    eNa 1 dNY

    1pkpm

    a 1 yk ymNa 1 bk yk ymN

    ARTICLE IN PRESS

    Y. Kajihara / Advances in Mathematics 187 (2004) 539780

  • 8/9/2019 Kajihara Y - Euler Transformation Formula for Multiple Basic Hypergeoemtric Series of Type a and Some Application

    29/45

    m3Fm2N; a 1 b1 e y1 ym;y;

    2a 2 b1 ? bm c d e; a 1 e y1 ym;y;

    a 1

    bm1 e ym1 ym; a 1

    bm e; a 1

    c e; a 1

    d ea 1 e ym1 ym; a 1 e; 1 N e ; 1

    XdANm

    Dy dDy

    Y1pkpm

    a jdj dk yk yma yk ym

    Y1pk;lpm

    bl yk yldk1 yk yldk

    Y

    1pkpm

    c yk ymdkd yk ymdka N yk ymdka 1 e yk ymdk

    " #

    Njdjejdj

    a

    1

    c

    jd

    ja

    1

    d

    jd

    j Y1pkpma yk ymjdj

    a

    1

    bk

    yk

    ym

    jd

    j

    : 6:23

    Remark 6.4. In the case when m n 1 of (6.21) is

    7V6a; b; c; d; e;N; 1

    2a 2 b c d eNeNa 1Na 1 bNa 1 cNa 1 dN 4F3

    N; a 1 b e; a 1 c e; a 1 d e1

    N

    e; 2a

    2

    b

    c

    d

    e; a

    1

    e

    ; 1 6:24This formula is similar to but different from the Whipple transformation formula

    between terminating very-well-poised 7F6 series and terminating 4F3 series

    (Eq. (2.4.1.1) in [23])

    a 1Na 1 d eNa 1 dNa 1 eN 4

    F3N; d; e; a 1 b c

    a 1 b; a 1 c; d e a N; 1

    7V6a; b; c; d; e;N; 1: 6:25

    Next, we present some q-1 limit of multiple Bailey-Jackson type transformation-summation formula and some special cases.

    The q-1 limit of multiple Bailey-Jackson type transformation-summation

    formula is the following.

    Proposition 6.4. When a1 ? an b1 ?bm mc; we have

    XgANn;jgjNDx gDx Y1pi; jpn

    aj xi xjgi1 xi xjgi

    Y

    1pipn;1pkpm

    bk xi yk xn ymgic xi yk xn ymgi

    ARTICLE IN PRESS

    Y. Kajihara / Advances in Mathematics 187 (2004) 5397 81

  • 8/9/2019 Kajihara Y - Euler Transformation Formula for Multiple Basic Hypergeoemtric Series of Type a and Some Application

    30/45

    X

    dANm;jdjN

    Dy dDy

    Y1pk;lpm

    c bl yk yldk1 yk yldk

    Y

    1pipn;1pkpm

    c ai xi yk xn ymdkc xi yk xn ymdk

    : 6:26

    Now we give a q-1 limit of the multiple generalization of Bailey type

    transformation (6.17).

    Corollary 6.8.

    a 1 dNa 1 eNm d f aNm e f aN

    Y1pkpm

    m 1 yk ymNa 1 ck ym ykNm ck f a yk ymN f ym ykN

    Y

    1pipn

    a 1 bi xi xnNm f a xn xiNa 1 xi xnNm bi f a xn xiN

    XgANn

    Dx gD

    x

    Y1pipna jgj gi xi xn

    a

    xi

    xn

    Y

    1pi; jpn

    bj xi xjgi1 xi xjgi

    Y1pipn

    d xi xngia N 1 xi xngi

    Y

    1pipn;1pkpm

    ck xi yk xn ymgia 1 f xi yk xn ymgi

    Y

    1pipn

    e xi xngia 1 N m f xi xngi

    N

    jgja 1 djgj

    Y1pipn

    a

    xi

    xn

    jgja 1 bi xi xnjgj

    m f Njgja 1 ejgjY

    1pkpm

    f ym ykjgja 1 ck ym ykjgj

    XdANm

    Dy dDy

    Y1pkpm

    m jdj dk yk ymm yk ym

    Y1pk;lpm

    a 1 cl f yk yldk1 yk yldk Y1pkpm

    a 1 e f yk ymdkm N 1 yk ymdk

    Y

    1pipn;1pkpm

    a 1 bi f xi yk xn ymdka 1 f xi yk xn ymdk

    ARTICLE IN PRESS

    Y. Kajihara / Advances in Mathematics 187 (2004) 539782

  • 8/9/2019 Kajihara Y - Euler Transformation Formula for Multiple Basic Hypergeoemtric Series of Type a and Some Application

    31/45

    Y

    1pkpm

    a 1 d f yk ymdk1 N f yk ymdk

    Njdjm e f ajdjY

    1pkpm

    m yk ymjdjm ck f a yk ymjdj

    m f Njdjm d f ajdjY

    1pipn

    m f a xn xijdjm bi f a xn xijdj

    ;6:27

    where m m 2a m 1 b1 ? bn c1 ? cm d e m 1f:

    Corollary 6.9. In the case when m 1; (6.27) reduces tom 1Na 1 cNa 1 dNa 1 eN

    fNm c f aNm d f aNm e f aN

    Y

    1pipn

    a 1 bi xi xnNm f a xn xiNa 1 xi xnNm bi f a xn xiN

    XgAN

    n

    Dx gD

    x Y1pipn

    a jgj gi xi xna

    x

    i x

    nY

    1pi; jpn

    bj xi xjgi1

    x

    i x

    jgi

    Y

    1pipn

    c xi xngid xi xngie xi xngia 1 f xi xngia 1 N m f xi xngia N 1 xi xngi

    " #

    fjgjm f NjgjNjgja 1 cjgja 1 djgja 1 ejgjY

    1pipn

    a xi xnjgja 1 bi xi xnjgj

    2n5V2n4m; a 1 b1 f x1 xn;y; a 1 bn1 f xn1 xn;

    a 1 bn f; a 1 c f; a 1 d f; a 1 e f;

    m f a xn x1;y; m f a xn xn1; m f a; m f N;N; 1

    6:28where m 3a 2 b1 ? bn c d e 2f:

    Remark 6.5. The m n 1 case of (6.27) is

    a 1 bNa 1 cNa 1 dNa 1 eNm b f aNm c f aNm d f aNm e f aN

    ARTICLE IN PRESS

    Y. Kajihara / Advances in Mathematics 187 (2004) 5397 83

  • 8/9/2019 Kajihara Y - Euler Transformation Formula for Multiple Basic Hypergeoemtric Series of Type a and Some Application

    32/45

    m 1Nm f aNa 1N fN

    9V8a; b; c; d; e;f; m f N;N; 1

    9V8m; a 1 b f; a 1 c f; a 1 d f; a 1 e f;

    m f a; m f N;N; 1; 6:29

    where m 3a 2 b c d e 2f:This formula is similar to the Bailey transformation formula for terminating

    balanced 9F8 series (equivalent to Eq. (2.4.3.1) in [23])

    9V8a; b; c; d; e;f; l a N 1 e f;N; 1

    a 1Na 1 e fNl 1 eNl 1 fNa 1 eNa 1 fNl 1Nl 1 e fN 9V8l; l b a; l c a; l d a; e;f; l a N 1 e f;N; 1;

    6:30

    where l 2a 1 b c d: But it is certainly different.

    Corollary 6.10. The limiting q-1 case of (6.20) is

    XgANn

    Dx gDx

    Y1pipn

    a jgj gi xi xna xi xn

    Y1pi; jpn

    bj xi xjgi1 xi xjgi

    Y1pipnc xi xngie xi xngi

    a N 1 xi xngia 1 d xi xngi" # Njgjdjgja 1 cjgja 1 ejgj

    Y1pipn

    a xi xnjgja 1 bi xi xnjgj

    a 1 b1 ? bn cNa 1 b1 ? bn c dNa 1 c dN

    a 1 cN

    Y1pipna 1 bi d xi xnNa 1 xi xnN

    a

    1

    bi

    xi

    xn

    N

    a

    1

    d

    xi

    xn

    N

    6:31

    provided 2a N 1 b1 ? bn c d e:

    ARTICLE IN PRESS

    Y. Kajihara / Advances in Mathematics 187 (2004) 539784

  • 8/9/2019 Kajihara Y - Euler Transformation Formula for Multiple Basic Hypergeoemtric Series of Type a and Some Application

    33/45

    Remark 6.6. In the case when n 1; (6.31) reduces to Dougalls summation formulafor very-well-poised 7V6 (Eq. (2.3.4.1) in [23])

    7V6a; b; c; d; e;N; 1

    a 1Na 1 b cNa 1 b dNa 1 c dNa 1 bNa 1 cNa 1 dNa 1 b c dN: 6:32

    when 2a N 1 b c d e:

    7. Multiple generalization of Sears transformation formula

    In this section, we give two types of multiple generalization of Sears

    transformation formula for 4f3 balanced basic hypergeometric series

    and 2 types of 3f2 terminating basic hypergeometric series and their q-1

    limits.

    7.1. Multiple Sears transformation

    Proposition 7.1.

    XgANn

    qjgjDxqgDx

    Y1pi; jpn

    bjxi=xjgiqxi=xjgi

    Y1pipn;1pkpm

    ckxiyk=xnymgidxiyk=xnymgi

    qNjgjajgjejgj fjgj Y1pipn;1pkpm

    ckxiyk=xnymgidxiyk=xnymgi

    e=aN f=aNeN fNaN

    XdANm

    qjdjDyqdDy

    Y1pk;lpm

    d=clyk=yldkqyk=yldk

    qNjdjajdj

    q1Na=e

    jd

    jq1Na=f

    jd

    j Y1pipn;1pkpmd=bixiyk=xnymdk

    dxiyk=xnym

    dk

    7:1

    when ab1?bnc1?cmq1N dmef:

    ARTICLE IN PRESS

    Y. Kajihara / Advances in Mathematics 187 (2004) 5397 85

  • 8/9/2019 Kajihara Y - Euler Transformation Formula for Multiple Basic Hypergeoemtric Series of Type a and Some Application

    34/45

    Proof. We consider the product of multiple q-Euler transformation (1.13) and third

    Heines transformation (1.9),

    XgANn

    ujgjDxqgDx

    Y1pi; jpn

    Ajxi=xjgiqxi=xjgi

    (

    Y

    1pipn;1pkpm

    Bkxiyk=xnymgiCxiyk=xnymgi

    )

    DEu=FNuN

    2f1F=D; F=E

    F; q; DEu=F

    A

    1?A

    nB

    1?B

    mu=

    Cm

    NuN

    XdANm

    A1?AnB1?Bmu=Cmjdj Dyqd

    Dy

    (

    Y

    1pk;lpm

    C=Blyk=yldkqyk=yldk

    Y1pipn;1pkpm

    C=Aixiyk=xnymdkCxiyk=xnymdk

    )

    2f1D; E

    F; q; u

    : 7:2

    We now suppose that a1?anb1?bm=cm DE=F: By taking the coefficients of the

    both side of uN in the equation above, we have

    XgANn

    qjgjDxqgDx

    Y1pi; jpn

    Ajxi=xjgiqxi=xjgi

    qN

    jg

    jq1N=F

    jg

    jq1ND=Fjgjq1NE=FjgjY

    1pipn;1pkpm

    Bkxiyk=xnym

    gi

    Cxiyk=xnymgi

    DNENF=DNF=ENDE

    F

    N

    XdANm

    qjdjDyqdDy

    Y1pk;lpm

    C=Blyk=yldkqyk=yldk

    qNjdjq1N=Fjdj

    q1N=Djdjq1N=Ejdj Y1pipn;1pkpmC=Aixiyk=xnymdk

    Cxiyk=xnymdk: 7:3

    After a suitable change of parameters, we obtain the desired identity. &

    ARTICLE IN PRESS

    Y. Kajihara / Advances in Mathematics 187 (2004) 539786

  • 8/9/2019 Kajihara Y - Euler Transformation Formula for Multiple Basic Hypergeoemtric Series of Type a and Some Application

    35/45

    Corollary 7.1. In the case when m 1; (7.1) reduces

    XgANn qj

    g

    jD

    xqg

    Dx Y1pi; jpnbjxi=xjgiqxi=xjgi

    qNjgjajgj

    ejgj fjgjY

    1pipn

    cxi=xngidxi=xngi

    e=aN f=aNeN fNaN

    n3fn2qN; a; d=b1x1=xn;y; d=bn1xn1=xn; d=bn; d=c

    dx1=xn;y; dxn1=xn; d; q1Na=e; q1Na=f;; q; q

    : 7:4

    Remark 7.1. m n 1 case of this transformation reduces to Sears transformationfor balanced 4f3 series (Eq. (2.10.4) in [5])

    4f3qN; a; b; c

    d; e;f; q; q

    aN e=aN f=aNeN fN 4f3

    qN; a; d=b; d=cd; aq1N=e; aq1N=f

    ; q; q

    7:5

    when abc defqN1:

    We have also another form of multiple Sears transformation.

    Proposition 7.2

    XgANn

    qjgjDxqgDx

    Y1pi; jpn

    bjxi=xjgiqxi=xjgi

    Y1pipn

    cxi=xngidxi=xngi

    qNjgj

    f

    jg

    j Y1pkpmaym=ykjgj

    ekym=yk

    jg

    j d=cNde1?em=amb1?bncN

    Y1pipn

    d=bixi=xnNdxi=xnN

    Y

    1pkpm

    aym=ykNekym=ykN

    e1?em

    am

    N

    XdANm

    qjdjDyqdDy

    Y1pk;lpm

    el=ayk=yldkqyk=yldk

    Y1pkpm

    f=ayk=ymdkq1Na1yk=ymdk

    qNjdj

    q1Nc=djdj Y1pipnq1Nd1xn=xijdj

    q1Nbi=dxn=xijdj 7:6

    when amb1?bmcq

    1N de1?em f:

    ARTICLE IN PRESS

    Y. Kajihara / Advances in Mathematics 187 (2004) 5397 87

  • 8/9/2019 Kajihara Y - Euler Transformation Formula for Multiple Basic Hypergeoemtric Series of Type a and Some Application

    36/45

    Proof. We consider the product of 2 multiple q-Euler transformations of type (4.1)

    XgANn

    ujgjDxqgDx

    Y1pi; jpn

    Ajxi=xjgiqxi=xjgi

    Y1pipn

    Bxi=xngiCxi=xngi

    ( )

    D1?DmEu=FNuN

    m1fmF=E; F=D1y1=ym;y;

    F=Dm1ym1=ym; F=Dm

    Fy1=ym;y;

    Fym

    1=ym; F; q; D1y; DmEu=F

    XdANm

    ujdjDyqdDy

    Y1pk;lpm

    Dlyk=yldkqyk=yldk

    Y1pkpm

    Eyk=ymdkFyk=ymdk

    ( )

    A1?AnBu=CNuN

    n1fnC=B; C=A1x1=xn;y;

    C=An1xn1=xn; C=An

    Cx1=xn;y;

    Cxn1=xn; C; q; A1y; AnBu=C

    :

    7:7

    We now suppose that A1?AnB=C D1?DmE=F: By taking the coefficient ofuNin both sides of (7.7), we obtain

    XgANn

    qjgjDxqgDx

    Y1pi; jpn

    Ajxi=xjgiqxi=xjgi

    Y1pipn

    Bxi=xngiCxi=xngi

    qN

    jg

    jq1NE=FjgjY

    1pkpm

    q1NF1ym=ykjg

    jq1NDk=Fym=ykjgj C=BNF=EN

    Y1pipn

    C=Aixi=xnNCxi=xnN

    Y1pkpm

    Fyk=ymNF=Dkyk=ymN

    XdANm

    qjdjDyqdDy

    Y1pk;lpm

    Dlyk=yldkqyk=yldk

    Y1pkpm

    Eyk=ymdkFyk=ymdk

    qNjdj

    q1NB=Cjdj Y1pipnq1NC1xn=xijdj

    q1NAi=Cxn=xijdj: 7:8

    By changing of parameters suitably, we get the desired identity. &

    ARTICLE IN PRESS

    Y. Kajihara / Advances in Mathematics 187 (2004) 539788

  • 8/9/2019 Kajihara Y - Euler Transformation Formula for Multiple Basic Hypergeoemtric Series of Type a and Some Application

    37/45

    Corollary 7.2. In the case when m 1; (7.6) reduces to

    XgANn

    qjgjDxqgDx

    Y1pi; jpn

    bjxi=xjgiqxi=xjgi

    qNjgjejgj

    ajgj fjgj

    Y1pipn

    cxi=xngidxi=xngi

    d=cNde1?em=amb1?bncNY

    1pipn

    d=bixi=xnNdxi=xnN

    aNeN

    e

    a

    N

    n3fn2qN; e=a;f=a; q1Nd1xn=x1;y;

    q1Na1; q1Nc=d; q1Nb1=dxn=x1;y;

    q1Nd1xn=xn1; q1Nd1

    q1Nbn1=dxn=xn1; q1Nbn=d; q; q

    : 7:9

    Note that right-hand side of (7.9) is essentially the same as that of (7.4).

    Remark 7.2. m

    n

    1 case of (7.6) is itself opposite version of Sears transforma-

    tion obtained by reversing the order of summation.

    Remark 7.3. Further properties of multiple Sears transformation of type (7.1) is

    discussed in [9].

    7.2. Multiple 3f2 transformation

    In the same way, we obtain two types of multiple transformation formula for 3f2by using the multiple q-Euler transformation formulas and (multiple) q-binomial

    theorem.

    Proposition 7.3.

    XgANnq=d1?dmjgj Dxq

    gDx Y1pi; jpn

    ajxi=xjgiqxi=xjgi

    Y

    1pipn

    qNjgjq1Ncd1?dm=a1?anbjgj

    bxi=xngicxi=xngi

    ARTICLE IN PRESS

    Y. Kajihara / Advances in Mathematics 187 (2004) 5397 89

  • 8/9/2019 Kajihara Y - Euler Transformation Formula for Multiple Basic Hypergeoemtric Series of Type a and Some Application

    38/45

    c=bNcd1?dm=a1?anbNY

    1pipn

    c=aixi=xnNcxi=xnN

    XdANm

    qjdj Dyqd

    Dy

    Y1pk;lpm

    dlyk=yldkqyk=yldk

    qNjdj

    q1Nb=cjdjY

    1pipn

    q1Nc1xn=xijdjq1Nai=cxn=xijdj

    : 7:10

    Proof. Consider the product of the multiple q-Euler transformation (1.13) and the

    multiple q-binomial theorem (1.11)

    XgANn

    ujgjDxqgDx

    Y1pi; jpn

    ajxi=xjgiqxi=xjgi

    Y1pipn

    bxi=xngicxi=xngi

    ( )

    d1?dmuNa1?anbu=cN

    XdANm

    ujdjDyqdDy

    Y1pk;lpm

    dlyk=yldkqyk=yldk

    ( )

    n1fnc=b;

    c=a1

    x1=xn;y;

    c=an

    1

    xn

    1=xn; c=an

    cx1=xn;y; cxn1=xn; c ; q; a1y; anbu=c

    : 7:11

    By taking the coefficient of uN in the equation above and by using a simple

    modification of q-shifted factorials, we have the desired formula. &

    Corollary 7.3. In the case when m 1; (7.10) reduces

    XgANn

    q=djg

    jD

    xqg

    Dx Y1pi; jpn

    ajxi=xj

    gi

    qxi=xjgi

    Y

    1pipn

    bxi=xngicxi=xngi

    " #qNjgj

    q1Ncd=a1?anbjgj

    c=bNcd=a1?anbNY

    1pipn

    c=aixi=xnNcxi=xnN

    n2fn1qN; d; q1Nc1xn=x1;y;

    q1Nb=c; q1N

    a1=c

    xn=x1;y;q1Nc1xn=xn1; q1Nc1

    q1Nan1=cxn=xn1; q1Nan=c; q; q

    : 7:12

    ARTICLE IN PRESS

    Y. Kajihara / Advances in Mathematics 187 (2004) 539790

  • 8/9/2019 Kajihara Y - Euler Transformation Formula for Multiple Basic Hypergeoemtric Series of Type a and Some Application

    39/45

    Remark 7.4. Eq. (7.10) is a multiple generalization of transformation formula for

    terminating 3f2 basic hypergeometric series.

    3f2qN; a; b

    c; q1Ncd=ab; q; q=d

    c=bNcd=abNc=aNcN 3

    f2qN; d; q1Nc1

    q1Nb=c; q1Na=c; q; q

    : 7:13

    Proposition 7.4.

    XgANn

    qjgjDxqgDx Y1pi; jpn

    ajxi=xjgiqxi=xjgi Y1pipn;1pkpm

    bkxiyk=xnymgicxiyk=xnymgi

    qNjgjq

    1

    N

    d1

    jgj a1?anb1?bm=c

    mdNdN

    XdANm

    qjdjDyqdDy

    Y1pk;lpm

    c=blyk=yldkqyk=yldk

    Y1pipn;1pkpm

    c=aixiyk=xnymdkcxiyk=xnymdk

    qNjdj

    q1Ncm=a1?anb1?bmdjdj: 7:14

    Proof. Multiply the factor uN=d1uN to the both sides of (1.13) and take thecoefficient of un in that equation. &

    Corollary 7.4. In the case when m 1; (7.14) reduces

    XgANn

    qjgjDxqgDx

    Y1pi; jpn

    ajxi=xjgiqxi=xjgi

    Y1pipn

    bxi=xngicxi=xngi

    qNjgjq1Nd1jgj

    a1?anbd=cN

    dN

    n2fn1qN; c=b; c=a1x1=xn;y; c=an1xn1=xn; c=an

    q1Nc=a1?anbd; cx1=xn;y; cxn1=xn; c; q; q

    : 7:15

    Remark 7.5. Eq. (7.14) is also a multiple generalization of transformation formula

    for terminating 3f2 basic hypergeometric series

    3f2qN; a; b

    c; q1Nd1; q; q

    abd=cNdN 3f2

    qN; c=b; c=aq1Nc=abd; c

    ; q; q

    : 7:16

    ARTICLE IN PRESS

    Y. Kajihara / Advances in Mathematics 187 (2004) 5397 91

  • 8/9/2019 Kajihara Y - Euler Transformation Formula for Multiple Basic Hypergeoemtric Series of Type a and Some Application

    40/45

    7.3. Their q-1 limits

    Now we write down q-1 limits of the formulas of this section. One can obtain

    from multiple Euler transformation (4.8), etc. in the same way as in basichypergeometric case.

    Proposition 7.5.

    XgANn

    Dx gDx

    Y1pi; jpn

    bj xi xjgi1 xi xjgi

    Y1pipn;1pkpmck xi yk xn ymgid xi yk xn ymgi

    Njgjajgjejgj fjgj e aN f aNeN fN

    XdANm

    Dy dDy

    Y1pk;lpm

    d cl yk yldk1 yk yldk

    Y1pipn;1pkpm

    d

    bi

    xi

    yk

    xn

    ym

    dk

    d xi yk xn ymdk

    Njdjajdj1 N a ejdj1 N a fjdj7:17

    when a b1 ? bn c1 ? cm 1 N md e f:

    Corollary 7.5. In the case when m 1; (7.17) reduces

    XgANn

    Dx gDx

    Y1pi; jpn

    bj xi xjgi1 xi xjgi

    Njgjajgjejgj fjgjY

    1pipn

    c xi xngid xi xngi

    e aN f aNeN fNn3Fn2

    N; a; d b1 x1 xn;y;d

    x1

    xn;y;

    d bn1 xn1 xn; d bn; d cd xn1 xn; d; 1 N a e; 1 N a f;

    ; 1

    : 7:18

    ARTICLE IN PRESS

    Y. Kajihara / Advances in Mathematics 187 (2004) 539792

  • 8/9/2019 Kajihara Y - Euler Transformation Formula for Multiple Basic Hypergeoemtric Series of Type a and Some Application

    41/45

    Remark 7.6. m n 1 case of this transformation reduces to Whipples transfor-mation for balanced 4F3 series

    4F3N; a; b; c

    d; e;f; 1

    e aN f aNeN fN 4F3

    N; a; d b; d cd; 1 N a e; 1 N a f ; 1

    7:19

    when a b c d e f N 1:

    We have also another form of the multiple Whipple transformation.

    Proposition 7.6.

    XgANn

    Dx gDx

    Y1pi; jpn

    bj xi xjgi1 xi xjgi

    Y1pipn

    c xi xngid xi xngi

    Njgj fjgjY

    1pkpm

    a ym ykjgjekym=ykjgj

    d cN

    d e1 ? em ma b1 ? bn cN

    Y1pipn

    d bi xi xnNd xi xnN

    Y1pkpm

    a ym ykNek ym ykN

    XdANm

    Dy dDy

    Y1pk;lpm

    el a yk yldk1 yk yldk

    Y

    1pkpm

    f a yk ymdk1 N a yk ymdk

    N

    jdj1 N c djdj

    Y1pipn

    1

    N

    d

    xn

    xi

    jdj1 N bi d xn xijdj 7:20

    when ma b1 ? bm c 1 N d e1 ? em f:

    Corollary 7.6. In the case when m 1; (7.20) reduces to

    XgANnDx gDx Y1pi; jpn

    bj xi xjgi1 xi xjgi Y1pipn

    c xi xngid xi xngi

    Njgj fjgjajgjejgj

    ARTICLE IN PRESS

    Y. Kajihara / Advances in Mathematics 187 (2004) 5397 93

  • 8/9/2019 Kajihara Y - Euler Transformation Formula for Multiple Basic Hypergeoemtric Series of Type a and Some Application

    42/45

    d cNd e1 ? em ma b1 ? bn cN

    aNeN

    Y1pipn

    d

    b

    i x

    i x

    nNd xi xnN

    n3Fn2N; e a;f a; 1 N d xn x1;y;

    1 N a; 1 N c d; 1 N b1 d xn x1;y;

    1 N d xn xn1; 1 N d

    1 N bn1 d xn xn1; 1 N bn d; 1

    : 7:21

    Note that right-hand side of (7.21) is essentially the same as that of (7.18).

    Remark 7.7. m n 1 case of (7.20) is itself is opposite version of the Whippletransformation (7.19) obtained by reversing the order of the summation.

    In the same way, we obtain two types of multiple transformation formula for 3F2by using the multiple Euler transformation formulas and (multiple) binomial

    theorem.

    Proposition 7.7

    XgANn

    Dx gDx

    Y1pi; jpn

    aj xi xjgi1 xi xjgi

    Y1pipn

    b xi xngic xi xngi

    Njgj1 N c d1 ? dm a1 ? an bjgj c bNc d1 ? dm a1 ? an bN

    Y1pipn

    c ai xi xnNc xi xnN

    XdANmD

    y

    d

    Dy Y1pk;lpmdl yk yldk1 yk yldk

    Njdj1 N b cjdjY

    1pipn

    1 N c xn xijdj1 N ai c xn xijdj

    : 7:22

    Corollary 7.7. In the case when m 1; (7.22) reduces

    XgANnDx gDx Y1pi; jpn

    aj xi xjgi1 xi xjgi

    Y

    1pipn

    b xi xngic xi xngi

    Njgj1 N c d a1 ? an bjgj

    ARTICLE IN PRESS

    Y. Kajihara / Advances in Mathematics 187 (2004) 539794

  • 8/9/2019 Kajihara Y - Euler Transformation Formula for Multiple Basic Hypergeoemtric Series of Type a and Some Application

    43/45

    c bNc d a1 ? an bNY

    1pipn

    c ai xi xnNc xi xnN

    n2Fn1 N; d; 1 N c xn x1;y;1 N b c; 1 N a1 c xn x1;y;

    1 N c xn xn1; 1 N c

    1 N an1 c xn xn1; 1 N an c; 1

    : 7:23

    Remark 7.8. Eq. (7.22) is a multiple generalization of transformation formula for

    terminating 3F2 basic hypergeometric series.

    3F2 N; a; b

    c; 1 N c d a b ; 1 c bNc d a bN

    c aNcN 3

    F2N; d; 1 N c

    1 N b c; 1 N a c ; 1

    : 7:24

    Proposition 7.8.

    XgANnDx gDx Y1pi; jpn

    aj xi xjgi1 xi xjgi

    Njgj1 N djgjY

    1pipn;1pkpm

    bk xi yk xn ymgic xi yk xn ymgi

    a1 ? an b1 ? bm mc dNdN

    XdANm

    Dy dDy

    Y1pk;lpm

    c bl yk yldk1 yk yldk

    Y1pipn;1pkpmc ai xi yk xn ymdk

    c xi yk xn ymdk Njdj1 N mc a1 ? an b1 ? bm djdj

    : 7:25

    Corollary 7.8. In the case when m 1; (7.25) reduces

    XgANnDx gDx Y1pi; jpn

    aj xi xjgi1 xi xjgi

    Njgj1 N djgjY

    1pipn

    b xi xngic xi xngi

    ARTICLE IN PRESS

    Y. Kajihara / Advances in Mathematics 187 (2004) 5397 95

  • 8/9/2019 Kajihara Y - Euler Transformation Formula for Multiple Basic Hypergeoemtric Series of Type a and Some Application

    44/45

    a1 ? an b d cNdN

    n2Fn1 N; c

    b; c

    a

    1 x

    1 x

    n;y;

    1 N c a1 ? an b d; c x1 xn;y;

    c an1 xn1 xn; c anc xn1 xn; c

    ; 1

    : 7:26

    Remark 7.9. Eq. (7.25) is also a multiple generalization of transformation formula

    for terminating 3F2 basic hypergeometric series

    3F2 N; a; b

    c; 1 N d; 1 a b d cNdN 3

    F2N; c b; c a

    1 N c a b d; c ; 1

    : 7:27

    Acknowledgments

    It is the authors pleasure to thank the referee for valuable comments. He would

    express his sincere gratitude to Prof. C. Krattenthaler, Prof. S.C. Milne and Dr. M.Schlosser for useful comments and especially to his adviser Prof. M. Noumi for so

    many encouragements, advises and suggestions. He also thanks Dr. K. Iohara and

    Dr. H. Rosengren for careful reading of the earlier version of this paper and pointing

    out several misprints.

    Note added

    After finishing this work, H. Rosengren kindly informed the author that he

    recovered the results of Section 6 of this paper in [21] by using his reduction formula

    of KarlssonMinton type. The author would like to thank again Dr. Rosengren for

    informing his result.

    References

    [1] G.E. Andrews, R. Askey, R. Roy, Special Functions, Cambridge University Press, Cambridge, 1999.

    [2] G.E. Andrews, Summation and transformation for basic Appell series, J. London Math. Soc. 4 (2)

    (1972) 618622.

    [3] G.E. Andrews, Problems and prospectives for basic hypergeometric functions, in: Theory and

    application of special functions, Proc. Advanced Sem. Math. Res. Center, University of Wisconsin,

    Madison, WI, 1975, pp. 191224.

    [4] R.Y. Denis, R.A. Gustafson, An SU

    n

    q-beta integral transformation and multiple hypergeometric

    series identities, SIAM J. Math. Anal. 23 (2) (1992) 552561.

    [5] G. Gasper, M. Rahman, Basic hypergeometric series, in: G.C. Rota (Ed.), Encyclopedia of

    Mathematics and Its Applications, Vol. 35. Cambridge University press, Cambridge, 1990.

    ARTICLE IN PRESS

    Y. Kajihara / Advances in Mathematics 187 (2004) 539796

  • 8/9/2019 Kajihara Y - Euler Transformation Formula for Multiple Basic Hypergeoemtric Series of Type a and Some Application

    45/45

    [6] R.A. Gustafson, C. Krattenthaler, Determinant evaluations and Un extension of Heines 2f1-transformations, in: Special functions, q-series and related topics Toronto, On, 1995, Fields Inst.

    Commun. 14 (1997) 8389.

    [7] W.J. Holman, Summation theorems for hypergeometric series in Un; SIAM J. Math. Anal. 11(1980) 523532.[8] W.J. Holman, L.C. Biedenharn, J.D. Louck, On hypergeometric series well-poised in SUn; SIAM J.

    Math. Anal. 7 (1976) 529541.

    [9] Y. Kajihara, Some remarks on multiple Sears transformations, Contemp. Math. 291 (2001) 139145.

    [10] Y. Kajihara, M. Noumi, Raising operators of row type for Macdonald polynomials, Compositio

    Math. 120 (2000) 119136.

    [11] A.N. Kirillov, M. Noumi, q-difference raising operators for Macdonald Polynomials and the

    integrality of transition coefficients, in: Algebraic Methods and q-Special functions, CRM Proc.

    Lecture Notes 22 (1999) 227243.

    [12] I.G. Macdonald, Symmetric Functions and Hall Polynomials, 2nd Edition, Oxford University Press,

    Oxford, 1995.

    [13] K. Mimachi, M. Noumi, An integral representation of eigenfunction for Macdonalds q-differenceoperators, Tohoku Math. J. 91 (1997) 517525.

    [14] S.C. Milne, An elementary proof of Macdonald identities for A1l ; Adv. in Math. 57 (1985) 3470.

    [15] S.C. Milne, A q-analogue of hypergeometric series well-poised in SUn and invariant G-functions,Adv. in Math. 58 (1985) 160.

    [16] S.C. Milne, A q-analogue of the Gauss summation theorem for hypergeometric series in Un; Adv. inMath. 72 (1988) 59131.

    [17] S.C. Milne, A q-analogue of a Whipples transformation of hypergeometric series in Un; Adv. inMath. 108 (1994) 176.

    [18] S.C. Milne, Balanced 3f2 summation formulas for Un basic hypergeometric series, Adv. in Math.131 (1997) 93187.

    [19] S.C. Milne, G.M. Lily, Consequences of the Al and Cl Bailey transform and Bailey lemma, Discrete

    Math. 139 (1995) 319346.

    [20] S.C. Milne, J.W. Newcomb, Un very-well-poised 10f9 transformations, J. Comput. Appl. Math. 68(1,2) (1996) 239285.

    [21] H. Rosengren, Reduction formula for KarlssonMinton type hypergeometric series, Constr. Approx.

    to appear.

    [22] M. Schlosser, Summation theorems for multidimensional basic hypergeometric series by determinant

    evaluations, Discrete Math. 210 (2000) 151169.

    [23] L.J. Slater, Generalized Hypergeometric Functions, Cambridge University Press, Cambridge, 1966.

    ARTICLE IN PRESS

    Y. Kajihara / Advances in Mathematics 187 (2004) 5397 97