• Aucun résultat trouvé

Inversion techniques and combinatorial identities. Jackson’s <span class="mathjax-formula">$q$</span>-analogue of the Dougall-Dixon theorem and the dual formulae

N/A
N/A
Protected

Academic year: 2022

Partager "Inversion techniques and combinatorial identities. Jackson’s <span class="mathjax-formula">$q$</span>-analogue of the Dougall-Dixon theorem and the dual formulae"

Copied!
27
0
0

Texte intégral

(1)

C OMPOSITIO M ATHEMATICA

W ENCHANG C HU

Inversion techniques and combinatorial identities.

Jackson’s q-analogue of the Dougall-Dixon theorem and the dual formulae

Compositio Mathematica, tome 95, n

o

1 (1995), p. 43-68

<http://www.numdam.org/item?id=CM_1995__95_1_43_0>

© Foundation Compositio Mathematica, 1995, tous droits réservés.

L’accès aux archives de la revue « Compositio Mathematica » (http:

//http://www.compositio.nl/) implique l’accord avec les conditions gé- nérales d’utilisation (http://www.numdam.org/conditions). Toute utilisa- tion commerciale ou impression systématique est constitutive d’une in- fraction pénale. Toute copie ou impression de ce fichier doit conte- nir la présente mention de copyright.

Article numérisé dans le cadre du programme Numérisation de documents anciens mathématiques

http://www.numdam.org/

(2)

43

Inversion

techniques

and combinatorial identities -

Jackson’s q-analogue

of the

Dougall-Dixon

Theorem and

the dual formulae

WENCHANG CHU

Institute of Systems Science, Academia Sinica, Beijing 100080 Received 6 July 1993; accepted in final form 9 December 1993

Abstract. As the common extension of theorems due to Gould and Hsu (1973) and Carlitz (1973),

a general pair of reciprocal relations is established. By means of the embedding machinery, it is

used to demonstrate that numerous q-hypergeometric identities are the dual relations of q- Saalschutz and Dougall-Dixon formulae. This fact serves as a natural reason for the existence of strange evaluations of basic hypergeometric series.

1. Introduction As

usual,

denote

by

the binomial coefficient and the Gaussian

q-binomial

coefficient respec-

tively.

For the two

complex

sequences

{ak}

and

{bk},

define the

03C8-poly-

nomials

by

Then there exists a useful

pair

of inverse series relations

which was discovered

by

Gould and Hsu

[18]

in 1973. At the same

time,

Partially supported by Alexander von Humboldt Foundation (Germany), Italian Consiglio

Nazionale Delle Ricerche (CNR) and NSF (Chinese), Youth-Grant no. 1990-1033.

Compositio Mathematica 95: 43-68, 1995.

(Ç) 1995 Kluwer Academic Publishers. Printed in the Netherlands.

(3)

Carlitz found its

q-analogue

which may be restated as follows:

Both

(1.2)

and

(1.3)

have been

neglected,

even

though special

versions were

rediscovered

by

Andrews

([1], Bailey pair)

and Gessel and Stanton

([15,16], Lagrange inversions)

and used to prove

hypergeometric

formulas. Their

efficiency

for

confirming

combinatorial

identities, through

the

embedding machinery,

was

recognized

in

[7,9].

As a natural extension of Carlitz’

theorem,

a more

general pair

of inverse

series relations will be established in this paper. The

development

of inversion

technique

will be used to demonstrate numerous strange

q-hypergeometric identities, by properly embedding

the variants of the

q-Saalschutz

theorem and Jackson’s

q-analogue

of

Dougall-Dixon

theorem in the members of this new

pair

of inversions. For

convenience,

these two

q-series

identities may be

displayed

as below:

where the usual notation of

q-shifted-factorial,

Gaussian binomial

coefficients,

and basic

hypergeometric

series from the

monographs by Bailey [3], Gasper

and Rahman

[14]

and Slater

[24]

have been

adopted

and the fraction of

q-factorials

abbreviated as

(4)

45

The results exhibited in the present paper

together

with the

previous

one

[7]

demonstrate a remarkable fact that the most basic

hypergeometric

identities

are the dual relations of

only

three

q-formulas analogous

to those named after

Chu-Vandermonde-Gauss,

Pfaff-Saalschutz and

Dougall-Dixon.

This serves as

a natural reason for the existence of strange basic

hypergeometric

relations.

2. A new

pair

of

reciprocal

relations

As an extension of Carlitz

theorem,

this section will establish a

general pair

of

reciprocal

relations contained in the

following

THEOREM. With the

03C8-polynomials defined by (1.1),

we have the inverse

series relations:

provided

that the

sequence-transforms

involved are

non-singular,

i.e.

03C8(03BBqn;

m +

1), 03C8(q-n ; m + 1)

and

(03BBqn ; q)m+1 do

not vanish

for non-negative integers

m n.

It is obvious that this

pair

of inversions will reduce to

(1.3)

when tends to

infinity.

Similar to the role of Carlitz

inversions, (2.1)

may be

used, systemati- cally,

to deal with strange basic

hypergeometric

series

through

the so-called

embedding technique (or

creative

telescoping)

mentioned in the introduction.

The exhibition demonstrated in the next three sections will show that this

approach

can not

only certify

the known

q-hypergeometric formulas,

but also

create several new strange evaluations.

Proof of the

theorem.

Recalling

the connection

[23,

p.

44-46]

between inverse

series relations and matrix inverse

pairs,

the

reciprocity

between

(2.1a)

and

(2.1b)

is

equivalent

to either of the

following orthogonal

relations:

(5)

The former

simply

follows from

summand-splitting

and

diagonal-cancelling (but,

the dual

orthogonality (2.2b)

is more difficult to

confirm

directly).

This

completes

the

proof

of the theorem. D The rotated version of the theorem may be stated as

COROLLARY. Assume the conditions

of

the theorem. The system

of equations

is

equivalent

to the system

where N is an

arbitrary non-negative integer

or

infinity.

One

pair

of bi-basic inverse relations due to Bressoud

[4]

and

Gasper [12]

can also be

specified

from

(2.1)

and

(2.3) by defining 03C8-polynomials

to be

shifted-factorials.

(6)

47

PROPOSITION. Let a and b be

complex

numbers such that

1-apXqY, 1 - bpXq-Y

and 1 -

qxalb differ from zero for non-negative integers

x and y. Then we have the inverse series relations

and their rotated

forms

Based on the

theorem,

the

remaining

part of the paper will

display

18

examples

which may be sketched as follows. If

(2.1a)

holds with the parameter 03BB and the sequences

{ai, bj and {f(k), g(k)l

under a certain

specification (which correspond

to a known basic

hypergeometric evaluation),

then we have

(2.1b)

under the same

specification (which

can be rewritten as another

hypergeometric evaluation).

Or

conversely,

this statement is valid either

by interchanging (2.1 a)

and

(2.1b).

3.

Embedding technique

on balanced formulae The transforms

will be used to

demonstrate, by (2.4),

the dual relations between the

balanced summations

(the

Saalschutzian

series)

and the non-trivial q-

hypergeometric

evaluations. All the balanced formulae

(cf. [8])

involved

here may be

regarded

as the

q-analogue

of

Bailey’s

identities

[3,

p.

30]

and

some of them have

appeared

in a

slightly

different version

[21].

(7)

EXAMPLE 3.1. Saalschutz formula p the first non-trivial evaluation.

Rewrite the balanced formula

in the form of

(2.4b)

whose dual relation

is

equivalent

to

a very-well poised

evaluation

EXAMPLE 3.2. Balanced formula ~ the second non-trivial evaluation.

Rewrite the balanced formula

(8)

1 nversion

techniques

and combinatorial identities 49 in the form of

(2.4b)

whose dual relation

is

equivalent

to

a very-well poised

evaluation

EXAMPLE 3.3. Balanced formulas thé third non-trivial evaluation.

Rewrite the balanced formula

in the form of (2.4b)

(9)

whose dual relation

is

equivalent

to

a very-well poised

evaluation

EXAMPLE 3.4. Balanced formula ~ the fourth non-trivial evaluation.

Rewrite the balanced formulae

in the form of

(2.4b)

(10)

51

whose dual relation

is

equivalent

to

a very-well poised

evaluation

EXAMPLE 3.5. Balanced formula ~ the fifth non-trivial evaluation.

Rewrite the balanced formula

in the form of

(2.4b)

whose dual relation

(11)

is

equivalent

to

a very-well poised

evaluation

In the similar way, one can enumerate the dual relations of the so called bi-basic formulas

displayed

in

[21] (see

also

[14, §3.10]).

The details are

omitted here.

4.

Strange

évaluations associated with Jackson’s

q-Dougall-Dixon

theorem

By

means of

(3.0)

and transforms

Jackson’s

q-Dougall-Dixon

theorem

(1.5)

will be embedded in

(2.1a)

and generate,

through (2.1b),

the dual formulas. From this process, several

unexpected

strange

hypergeometric

evaluations will be found.

Among

the

examples displayed

in this

section,

some of their inverse relations

(i.e. Examples 4.2,

4.3 and

4.6-4.9)

are not balanced

by

the

q-binomial coefficient n . Instead,

it is

replaced by non-symmetric

forms

n , n ]

and

n

This is due to the

particular specification

for

sequence

g(···).

For

example,

the

specification g(2k

+

1)

= 0 in

Eq. (4.2b)

comes from the

special q-Dougall-Dixon

formula

(4.2a).

EXAMPLE 4.1.

q-Dougall-Dixon

theorem - Gessel and Stanton’s evalu- ation.

The

specialized q-Dougall-Dixon

formula

(12)

53

can be rewritten as

whose dual version

may be

expressed

as a strange

terminating

evaluation

which is due to Gessel and Stanton

[16, Eq. (1.4)].

For its

non-terminating version,

see

[12, Eq. (5.2)], [13, Eq. (5.1)]

and

[14, Eq. (3.8.12)].

EXAMPLE 4.2.

q-Dougall-Dixon

théorème another Gessel and Stanton’s evaluation.

The

specialized q-Dougall-Dixon

formula

(13)

can be rewritten as, after

replacement q - q2

whose dual version

may be

expressed

as another strange

terminating

evaluation

which is also due to Gessel and Stanton

[16, Eq. (6.14)].

EXAMPLE 4.3.

q-Dougall-Dixon

theorem p the first new strange evaluation.

The

specialized q-Dougall-Dixon

formula

(14)

55

can be rewritten as, after

replacement q - q2

whose dual version

may be

expressed

as the first new strange

terminating

evaluation in this section

EXAMPLE 4.4.

q-Dougall-Dixon

théorème the second new strange evalu- ation.

The

specialized q-Dougall-Dixon

formula

(15)

can be rewritten as, after

replacement q2 ~

q

whose dual version

may be

expressed

as the second new strange

terminating

evaluation in this section

EXAMPLE 4.5.

q-Dougall-Dixon théorème Gasper’s

strange evaluation.

The

specialized q-Dougall-Dixon

formula

(16)

57

can be rewritten as

whose dual version

may be

expressed

as a strange

terminating

evaluation

which is due to

Gasper [12, Eq. (5.22)].

See also

[13, Eq. (1.2)].

EXAMPLE 4.6.

q-Dougall-Dixon

theorem ~ the third new strange evaluation.

The

specialized q-Dougall-Dixon

formula

(17)

can be rewritten as, after

replacement q ~ q3

whose dual version

(otherwise)

may be

expressed

as the third new strange

terminating

evaluation in this section

(otherwise).

EXAMPLE 4.7.

q-Dougall-Dixon

theorem«the fourth new strange evalu- ation.

The

specialized q-Dougall-Dixon

formula

(18)

59

can be rewritten as, after

replacement q - q4

whose dual version

(otherwise)

may be

expressed

as the fourth new strange

terminating

evaluation in this section

(otherwise).

EXAMPLE 4.8.

q-Dougall-Dixon

theorem ~ the fifth new strange evaluation.

The

specialized q-Dougall-Dixon

formula

(19)

can be rewritten as, after

replacement q - q2

whose dual version

may be

expressed

as the fifth new strange

terminating

evaluation in this section

(20)

61 EXAMPLE 4.9.

q-Dougall-Dixon

theorem ~ the sixth new strange evaluati(

The

specialized q-Dougall-Dixon

formula

can be rewritten as, after

replacement

q ~

q3

whose dual version

(otherwise)

may be

expressed

as the last new strange

terminating

evaluation in this section

(21)

(otherwise).

REMARK.

Similarly,

one can show that Jackson’s

q-Dougall-Dixon

for-

mula

(1.5)

with five free parameters, is

self-reciprocal,

i.e. the dual relation has

exactly

the same formation as the

original

one under the parameter

replacement.

In

addition,

for a very

special

case of

q-Dougall-Dixon

theorem

its

resulting

dual relation is trivial.

5. Reversai

embeddings

on

q-Dougall-Dixon

theorem

In

general,

a

terminating hypergeometric

summation

Lk=O T(n, k)

=

S(n),

cannot be

telescoped

into one of the relations stated in

(2.1).

But occa-

sionally,

its reversal

03A3nk = o T(n, n - k) = S(n)

may be restated as one member

of (2.1).

In this case, the dual relation will create some

"mysterious-looking"

formulas.

By

means of

(3.0), (4.0)

and transforms

some remarkable

examples

are demonstrated as follows.

(22)

63 EXAMPLE 5.1.

q-Dougall-Dixon

theorem ~ one very strange new evaluation.

The reversal of

q-Dougall-Dixon

formula

can be

expressed

as, after

adding

some extra-zero terms

whose dual relation

can be reformulated in

q-series

which seems to be new.

(23)

EXAMPLE 5.2.

q-Dougall-Dixon theorem ~ Gasper

and Rahman’s two very strange evaluations.

For 03BB = 1 and

2,

the reversal of

q-Dougall-Dixon

formula

can be

expressed

as, after

adding

some extra-zero terms

whose dual relation

(24)

65

can be reformulated in

q-series

which

yields terminating Gasper

and Rahman’s formulas

[13, Eqs. (1.8, 4.5)], respectively,

for 03BB = 1 and 2. See also

[14, Eq. (3.8.18),

Ex.

3.32].

EXAMPLE 5.3.

q-Dougall-Dixon

theorem ~ three very strange new evalu- ations.

For

03BB = 1, 2

and

3,

the reversal of

q-Dougall-Dixon

formula

can be

expressed

as, after

adding

some extra-zero terms

whose dual relation

(25)

can be reformulated in

q-series

which is the unified version of three new

terminating

strange evaluations

corresponding

to 03BB =

1,

2 and 3.

EXAMPLE 5.4.

q-Dougall-Dixon

theorem p two very strange new evaluations.

For 03BB = 1 and

2,

the reversal of

q-Dougall-Dixon

formula

can be

expressed

as, after

adding

some extra-zero terms

(26)

67 whose dual relation

can be reformulated in

q-series

which is the unified version of two new

terminating

strange evaluations

corresponding

to 03BB = 1 and 2.

References

1. G. E. Andrews: Connection coefficient problems and partitions, Proc. Symp. Pure Math. 34 (1979) 1-24.

2. R. Askey: Variants of Clausen’s formula for the square of a special 2F1, in Number Theory and

Related Topics, Tata Institute of Fundamental Research, Bombay, 1988.

3. W. N. Bailey: Generalized Hypergeometric Series, Cambridge University Press, London, 1935.

4. D. M. Bressoud: The Bailey lattice: an introduction, 57-67 in Ramanujan Revisited (G. E.

Andrews et al. eds.), Academic Press, New York, 1988.

5. L. Carlitz: Some inverse relations, Duke Math. J. 40 (1973) 893-901.

6. W. C. Chu: A class of bivariate reciprocal formulas and applications to combinatorial identities

(Chinese, English summary), Appl. Math. 1(3) (1988) 63-66.

7. W. C. Chu: Inversion techniques and combinatorial identities: Strange evaluations of basic

hypergeometric series, Compositio Math. 9 (1994), 121-144.

8. W. C. Chu: Symmetry on q-Pfaff-Saalschutz-Sheppard series, Rivista di Mat. N11 (1992), 73-79.

9. W. C. Chu and L. C. Hsu: Some new applications of Gould-Hsu inversions, J. Combin., Informat., Systems Sci. 14(1) (1990) 1-4.

10. S. B. Ekhad: A short proof of a terminating "strange" hypergeometric summation formula of Gasper and Rahman, Preprint, 1990.

11. H. Exton: Q-Hypergeometric Series and Applications, Ellis Horwood Limited, England, 1983.

12. G. Gasper: Summation, transformation and expansion formulas for bibasic series, Trans. Amer.

Math. Soc. 312(1) (1989) 257-277.

(27)

13. G. Gasper and M. Rahman: An indefinite bibasic summation formula and some quadratic,

cubic and quartic summations and transformation formulas, Can. Math. J. 17(1) (1990) 1-27.

14. G. Gasper and M. Rahman: Basic hypergeometries series, vol. 35, in Encyclopedia of

Mathematics and Its Application (ed. G. C. Rota); Cambridge University Press, Cambridge,

1990.

15. Ira Gessel and D. Stanton: Strange evaluations of hypergeometric series, SIAM J. Math. Anal.

13 (1982) 295-308.

16. Ira Gessel and D. Stanton: Applications of q-Lagrange inversion to basic hypergeometric series, Trans. Amer. Math. Soc. 277(1) (1983) 173-201.

17. H. W. Gould: Combinatorial Identities, Morgantown, West-Virginia, 1972.

18. H. W. Gould and L. C. Hsu: Some new inverse series relations, Duke Math. J. 40 (1973)

885-891.

19. D. H. Greene and D. E. Knuth: Mathematics for the Analysis of Algorithms, Birkerhauser, 1981.

20. C. Krattenthaler: A new matrix inverse, Preprint, 1989.

21. B. Nassrallah and M. Rahman: On the q-analogues of some transformations of nearly-poised hypergeometric series, Trans. Amer. Math. Soc. 268 (1981) 211-229.

22. M. Rahman: Some cubic summation formulas for basic hypergeometric series, Utilitas Math.

36 (1989) 161-172.

23. J. Riordan: Combinatorial Identities, Wiley, New York, 1968.

24. L. J. Slater: Generalized Hypergeometric Functions, Cambridge Univ. Press, Cambridge, 1966.

25. D. Zeilberger: The method of creative telescoping, Preprint, 1990.

Références

Documents relatifs

(***) Indirizzo dell'A.: Faculty of Mathematics, Kyushu University, Hakozaki Fukuoka, 812-8581, JAPAN... We can easily show these values are obtained from the definition of F

Si p est un idéal premier de R, alors Ry est régulier si et seulement si le R -module DiflYÂ-, Ry\ = (Difî (Te, jR))p est libre (critère jacobien de simplicité, applicable, puisque

Chu, Inversion techniques and combinatorial identities: Strange evaluations of hyper- geometric series, Preprint, 1990. [8]

Toute utilisa- tion commerciale ou impression systématique est constitutive d’une in- fraction pénale.. Toute copie ou impression de ce fichier doit conte- nir la présente mention

Prendre sur la circonférence to passant par O, A, B le point C milieu de Parc OAB, décrire de ce point une circonférence passant par O et mener la bissectiice intérieure de l'angle

SCHMALZ, Solution of the ~-equation with uniform estimates on strictly q-convex domains with non-smooth boundary,

Compte tenu de la Proposition 2-5, il en résulte que par sa condition (c), le Théorème 6-2 constitue en particulier la généralisation du Théorème 9-2 de [2], dans lequel S. Mac

La complexité des calculs traditionnels d'une itération (pivotage), est celle de la modification de chacun des termes de la matrice, et donc en n 2 , si n est le nombre de