• Non ci sono risultati.

The (exponential) multipartitional polynomials and polynomial sequences of multinomial type, Part II

N/A
N/A
Protected

Academic year: 2024

Condividi "The (exponential) multipartitional polynomials and polynomial sequences of multinomial type, Part II"

Copied!
13
0
0

Testo completo

(1)

The (exponential) multipartitional polynomials and polynomial sequences of multinomial type, Part II

HACE`NEBELBACHIR*, MILOUDMIHOUBI

University of Science and Technology Houari Boumediene, USTHB, Faculty of Mathematics, P.B. 32, El Alia, 16111 Algiers, Algeria

Received 15 December 2012; revised 29 May 2013; accepted 10 June 2013 Available online 29 June 2013

Abstract. We establish recursiveness properties for multipartitional polynomials and their connection with the derivatives of polynomials of multinomial type. Various com- prehensive examples are illustrated.

Mathematis Subject Classification: 11B83; 05A10; 11B65

Keywords: Multipartitional polynomials; Derivatives of polynomial sequences of multinomial type; Recursiveness relations

1. INTRODUCTION

In a precedent work[6], the authors established identities on multipartitional polyno- mials, generalizing[1]. They give a result concerning the preservation of sequences of multinomial type. These results appear as natural extension of Bell polynomials and of polynomials of binomial type (see for instance [2–4]). Finally, they give some connection between these two concepts.

Let us introduce, as in[6], some definitions and notations.

Form,m1,. . .,mr integers such thatPr

i¼1mi¼m;we define m

m1;. . .;mr

¼

m!

m1!mr! if miP0; i¼1;. . .;r;

0 otherwise:

LetN be the set of nonnegative integers and Rthe set of real numbers. We use, the following notations, fora¼ ða1;. . .;arÞ 2Rr;b¼ ðb1;. . .;brÞ 2Rrwe set

* Corresponding author. Tel.: +213 771425883; fax: +213 21276446.

E-mail addresses:[email protected](H. Belbachir),[email protected](M. Mihoubi).

Peer review under responsibility of King Saud University.

Production and hosting by Elsevier

1319-5166ª2013 Production and hosting by Elsevier B.V. on behalf of King Saud University.

http://dx.doi.org/10.1016/j.ajmsc.2013.06.003

Arab J Math Sci21(1) (2015), 2–14

ARABJOURNAL OF

MATHEMATICALSCIENCES

(2)

a.b=a1b1+ +arbr,a+b = (a1+b1,. . .,ar+br),ka= (ka1,. . .,kar), (aPb) () (a1Pb1,. . .,arP br), (a>b) () (a1>b1,. . .,ar>br), 1ðaPbÞ¼ 1 if aPb;

0 otherwise;

forn¼ ðn1;. . .;nrÞ 2Nr we set

tn¼tn11 tnrr;tn!n¼t

n1 1 n1! tnnrr

r!;n! =n1! nr!,ŒnŒ=n1+ +nr,1= (1,. . ., 1), n

i p ¼ n1 i1

nr

ir ; n

i;j;. . .;k

p

¼ n1 i1;j1;. . .;k1

nr ir;jr;. . .;kr

;

Bn;kðxiÞ ¼Bn1;...;nr;kðx0;...;0;1;. . .;xn1;...;nrÞ,nP 0,kP0 and =0 otherwise, AnðxiÞ ¼An1;...;nrðx0;...;0;1;. . .;xn1;...;nrÞ,nP0 and =0 otherwise,

fnðxÞ ¼fn1;...;nrðxÞifnP 0,f0(x) = 1 and =0 otherwise, for ðm;nÞ 2N2 we set Bm,n;k(xi,j) =Bm,n;k(x0,1,x1,0 ,. . .,xm,n),

Am,n(xi,j) =Am,n(x0,1,x1,0,. . .,xm,n) and we set alsoDz¼0¼dzd

z¼0.

The complete (exponential) multipartitional polynomials An in the variables (xi, i„0), are defined by the sum

AnðxiÞ:¼X n!

Yn

i¼0;i–0

ki! Yn

i¼0;i–0

xi

i!

ki

; ð1Þ

where the summation is extended over all partitions of the multipartite number n= (n1,. . .,nr), that is, over all nonnegative integers (ki,06i6n) solution of the equations

Xn

i¼0ijki¼nj; j¼1;. . .;r; with convention k0¼0: ð2Þ

The partial (exponential) multipartitional polynomialsBn,kin the variables (xi,i„0), of degreek, are defined by the sum

Bn;kðxiÞ:¼X n!

Qn

i¼0;i–0ðki!Þ Yn

i¼0;i–0

xi

i!

ki

; ð3Þ

where the summation is extended over all partitions of the multipartite number n= (n1,. . .,nr) intokparts, that is, over all nonnegative integers (ki,06i6n) solution of the equations

Xn

i¼0ki¼k; Xn

i¼0ijki¼nj; j¼1;. . .;r; with conventionk0¼0: ð4Þ

Also, we can verify that the exponential generating function forAn1;...;nr is given by X

nP0AnðxjÞtn

n!¼exp X

jijP1xi

ti i!

ð5Þ

and the exponential generating function forBn; k is given by X

jnjPkBn;kðxjÞtn n!¼ 1

k!

X

jijP1xiti i!

k

: ð6Þ

(3)

The polynomials of multinomial type (fn(x)) are defined byf0(x):¼1 and X

iP0fið1Þti i!

x

¼X

nP0fnðxÞtn

n!: ð7Þ

In the followingSdesign a vector ofNrn f0g.

We will use the two following results, given in[6]:

Theorem 1. Let rP1 be an integer anda2Rr. If (fn(x)) is a multinomial type sequence of polynomials, then the polynomials (hn(x)) and (pn(x)) given by

hnðxÞ:¼ x

a:nþxfnða:nþxÞ ð8Þ

pnðxÞ:¼fjnjðxÞ ð9Þ

are of multinomial type.

Theorem 2. Leta2R;a2Rr and (fn(x)) be a sequence of polynomials of multinomial type. We have

An afiða:iÞ a:i

¼afnða:nþaÞ

a:nþa : ð10Þ

In this paper, we start with two light extensions concerning multipartitional polyno- mials given in[6]. The main results of the paper concern the recursiveness properties for multipartitional polynomials and their connections with the derivatives of polynomials of multinomial type. Each result is followed by various applications.

2. IDENTITIES RELATED TO MULTIPARTITIONAL POLYNOMIALS

In[6], Theorems 3 and 4 are established when S2 {0,1}r,S„0. Here, we extend the results to any vectorS2Nrsuch thatŒSŒP1. These extensions give an other applica- tion illustrated by vectorial multinomial coefficient.

Theorem 3. Let (xn) be a sequence of real numbers. Then jnj þk

jnj

!

BnþkS;k j

S !

p

xjjj

!

¼ 1 k!

nþkS n;S;. . .;S

!

p

Bjnjþk;kðjxjþjSj1Þ: ð11Þ

Proof. Identity(11)follows from the following expansion:

(4)

1 k!

P

jijP1

i S pxiti!i k

¼k!1 P

jP1xjP

jij¼j

i S p

ti i!

k

¼ tkS

k!ðS!Þk

P

jP1xjP

jij¼j tiS ðiSÞ!

k

¼ tkS

k!ðS!Þk

P

jP1xjþjSj1P

jij¼j1ti i!

k

¼ tkS

k!ðS!Þk

P

jP1jxjþjSj1ðjtjÞj!j1 k

¼ tkS

ðS!Þk

P

mPkBm;kðjxjþjSj1ÞðjtjÞm!mk

¼ðS!ÞtkSk

P

mPk ðmkÞ!

m! Bm;kðjxjþjSj1ÞP

jmj¼mktm m!

¼ tkS

ðS!Þk

P

mP0 ðjmjÞ!

ðjmjþkÞ!Bjmjþk;kðjxjþjSj1Þtm!m

¼k!ðS!Þ1 k

P

mP0 n!

ðnkSÞ!

BjnjkjSjþk;kðjxjþjSj1Þ

jnj ðjSj 1Þk k

tn!n

¼k!1P

nPkS

n nkS;S;. . .;S

p

BjnjðjSj1Þk;kðjxjþjSj1Þ

jnj ðjSj 1Þk k

tn!n:

Corollary 4. Let (xn) be a sequence of real numbers. Then, for r = 1 andS= u, we get nþk

n

Bnþku;k j u xj

¼ 1 k!

nþku n;u;. . .;u

Bnþk;kðjxjþu1Þ:

Corollary 5. Let (xn) be a sequence of real numbers. Then, for r = 2 andS= (u,0), we get

mþnþk mþn

Bmþku;n;k

i u xiþj

¼ 1 k!

mþku m;u;. . .;u

Bmþnþk;kðjxjþu1Þ;

for r = 2 and S= (0,v), we get mþnþk

mþn

Bm;nþkv;k j v xiþj

¼ 1 k!

nþkv n;v;. . .;v

Bmþnþk;kðjxjþv1Þ;

for r = 2 and S= (u,v), we get mþnþk

mþn

Bmþku;nþkv;k

i u

j v xiþj

¼ 1 k!

mþku m;u;. . .;u

nþkv n;v;. . .;v

Bmþnþk;kðjxjþuþv1Þ:

Theorem 6. Leta2R;a2Rrand (fn(x)) be a multinomial type sequence of polynomi- als. We have

(5)

Bn;k a i S p

fiSða:ðiSÞ þaÞ a:ðiSÞ þa

¼ a ðk1Þ!

n nkS;S;. . .;S

p

fnkSða:ðnkSÞ þakÞ

a:ðnkSÞ þak : ð12Þ

Proof. We have P

jnjPkBn;k

i

S pfiSðaÞ

tn

n!¼k!1 P

jijP1

i

S pfiSðaÞti!i k

¼ tkS

k!ðS!Þk

P

iP0fiðaÞti!i k

¼ tkS

k!ðS!Þk

P

nP0fnðakÞtn!n

¼ 1

k!ðS!Þk

P

nPkS n!

ðnkSÞ!fnkSðakÞtn!n; i.e.

Bn;k

i

S pfiSðaÞ

¼ 1 k!

n nkS;S;. . .;S

p

fnkSðakÞ: ð13Þ It suffices to replace (fn(x)) by (hn(x)) as in(8). h

Corollary 7. Leta;a2Rand (fn(x)) be a binomial type sequence of polynomials. Then, for r = 1 andS= u, we have

Bn;k a i u

fiuðaðiuÞ þaÞ aðiuÞ þa

¼ a ðk1Þ!

n nku;u;. . .;u

fnkuðaðnkuÞ þakÞ aðnkuÞ þak : Corollary 8. Let a2R;a¼ ða;bÞ 2R2 and (fm,n(x)) be a trinomial type sequence of polynomials. Then, for r = 2 andS= (u,0), we have

Bm;n;k a i u

fiu;jðaðiuÞ þbjþaÞ aðiuÞ þbjþa

¼ a ðk1Þ!

m mku;u;. . .;u

fmku;nðaðmkuÞ þbnþakÞ aðmkuÞ þbnþak ; for r = 2 andS= (0,v), we have

Bm;n;k a j v

fi;jvðaiþbðjvÞ þaÞ aiþbðjvÞ þa

¼ a ðk1Þ!

n nkv;v;. . .;v

fm;nkvðamþbðnkvÞ þakÞ amþbðnkvÞ þak ; for r = 2 andS= (u,v), we have

Bm;n;k a i u

j v

fiu;jvðaðiuÞ þbðjvÞ þaÞ aðiuÞ þbðjvÞ þa

¼ a ðk1Þ!

m!n!

ðmkuÞ!ðnkvÞ!ðu!v!Þk

fmku;nkvðaðmkuÞ þbðnkvÞ þakÞ aðmkuÞ þbðnkvÞ þak :

(6)

3. RECURSIVENESS IN MULTIPARTITIONAL POLYNOMIALS

This section is devoted to the twice composition of multipartitional polynomials.

Theorem 9. Let (xn) be a sequence of real numbers with xS¼1;d2Nr and d2NI. We have

Bn;k 1ðiPSÞðt1Þ!d iþ ðt1ÞS i;S;. . .;S 1

p

Biþðt1ÞS;t

j S pxj

!

¼1ðnPkSÞðT1Þ!d ðk1Þ!

nþ ðTkÞS n;S;. . .;S 1

p

BnþðTkÞS;T j S pxj

; ð14Þ

where t =dÆ(iS) + d T =dÆ(nkS) + kd.

Proof. Let (fn(x)) be a multinomial type sequence of polynomials with fn(1):¼xn+S. Fora=0 anda= 1 in(12), we get

fnðkÞ

k ¼ ðk1Þ! nþkS n;S;. . .;S 1

p

BnþkS;k i S pxi

: ð15Þ

Fora=d anda=d, Formula(12)becomes Bn;k d i

S p fiSðtÞ

t

¼ d ðk1Þ!

n nkS;S;. . .;S

p

fnkSðTÞ

T : ð16Þ

Therefore, form(15), we obtain fiSðtÞ

t ¼ ðt1Þ! iþ ðt1ÞS iS;S;. . .;S 1

p

Biþðt1ÞS;t

i S pxi

and fnkSðTÞ

T ¼ ðT1Þ! nþ ðTkÞS nkS;S;. . .;S 1

p

BnþðT1ÞS;T i S pxi

:

Now, replacing in(16)fiS(t) andfnkS(T) by their above expressions, we get Bn;k ðt1Þ!d i

S p

iþ ðt1ÞS iS;S;. . .;S 1

p

Biþðt1ÞS;t

i S pxi

!

¼ðT1Þ!d ðk1Þ!

n nkS;S;. . .;S

p

nþ ðTkÞS nkS;S;. . .;S 1

p

BnþðTkÞS;T

i S pxi

:

Then, to obtain(14), it suffices to observe that i

S p

iþ ðt1ÞS iS;S;. . .;S 1

p

¼1ðiPSÞ iþ ðt1ÞS i;S;. . .;S 1

p

;

n nkS;S;. . .;S

p

nþ ðTkÞS nkS;S;. . .;S 1

p

¼1ðnPkSÞ nþ ðTkÞS n;S;. . .;S 1

p

:

(7)

Corollary 10. Let (fn(x)) be a binomial type sequence of polynomials and d2NI. Then, for r = 1,S= u andd¼p2N;and Bn,k(xi) be the partial Bell polynomials, we get

Bn;k 1ðiPuÞðt11Þ!d iþ ðt11Þu i;u;. . .;u 1

Biþðt11Þu;t1 j u xj

!

¼1ðnPkuÞðT11Þ!d ðk1Þ!

nþ ðT1kÞu n;u;. . .;u 1

BnþðT1kÞu;T1 j u xj

;

where t1= p(iu) + d, T1= p(nku) + kd.

Corollary 11. Let p, q, dP1 be nonnegative integers and (fm,n(x)) be a trinomial type sequence of polynomials. Then, for r = 2,d= (p, q) andS= (u, 0), we get

Bm;n;k 1ðiPuÞðt101Þ!d iþ ðt101Þu i;u;. . .;u 1

Biþðt101Þu;j;t10 i0 u xi0;j0 !

¼1ðmPkuÞðT101Þ!d ðk1Þ!

mþ ðT10kÞu m;u;. . .;u 1

BmþðT10kÞu;n;T10 i0 u xi0;j0

;

where t10= p(iu) + qj + d and T10= p(mku) + qn + kd; for r = 2,d= (p,q) andS= (0,v), we get

Bm;n;k 1ðjPvÞðt011Þ!d jþ ðt011Þv j;v;. . .;v 1

Bi;jþðt011Þv;t01 j0 v xi0;j0

!

¼1ðnPkvÞðT011Þ!d ðk1Þ!

nþ ðT01kÞv n;v;. . .;v 1

Bm;nþðT01kÞv;T01 j0 v xi0;j0

;

where t01= pi + q(jv) + d and T01= pm + q(nkv) + kd; for r = 2,d= (p,q) andS= (u,v), we get

Bm;n;k 1ðiPuÞ1ðjPvÞðt111Þ!d

Biþðt111Þu;jþðt111Þv;t11 i0 u

j0 v xi0;j0

iþ ðt111Þu i;u;. . .;u

jþ ðt111Þv j;v;. . .;v

0 BB B@

1 CC CA

¼1ðmPkuÞ1ðnPkvÞðT111Þ!d ðk1Þ!

BmþðT11kÞu;nþðT11kÞv;T11

i0 u

j0 v xi0;j0

mþ ðT11kÞu m;u;. . .;u

nþ ðT11kÞv n;v;. . .;v ;

where t11= p(iu) + q(jv) + d and T11= p(mku) + q(nkv) + dk.

Foru= 1 in Corollary 10 we obtain Proposition 4 of[3]and foru=v= 1 in Cor- ollary 11 we obtain Theorems 7 and 10 of[5].

Using some particular cases related to Bell polynomials. Indeed, for the choice xn:¼xŒnŒ in Theorem 9 and using relation(11), we obtain:

(8)

Corollary 12. Let (xn) be a sequence of real numbers with xS¼1;d2Nrand d2NI: We have

Bn;k

d t

i S p

BjijjSjþt;tðxjÞ jij jSj þt

t

0 BB B@

1 CC CA¼ d

ðk1Þ!T

n nkS;S;. . .;S

p

BjnjkjSjþT;TðxjÞ jnj kjSj þT

T

; ð17Þ

where t =dÆ(iS) + d, T =dÆ(nkS) + kd.

Example 13. Leta,xbe real numbers and (fn(x)) be a binomial type sequence of poly- nomials. Then, forxn¼anþxnx fn1ðanþxÞ;we obtain from the identity (see Proposition given in [3])

Bn;k

nx

anþxfn1ðanþxÞ

¼ n k

kx

aðnkÞ þkxfnkðaðnkÞ þkxÞ the following identity

Bn;k dx i S p

fnkðaðjij jSjÞ þtxÞ aðjij jSjÞ þtx

¼ dx ðk1Þ!

n nkS;S;. . .;S

p

fnkðaðjnj kjSjÞ þTxÞ aðjnj kjSjÞ þTx ; wheret=dÆ(iS) +d,T=dÆ(nkS) +kd.

Theorem 14. Let (xn) be a sequence of real numbers with xS= 1, d be an integer and d2 ðNIÞr. Then, ford.n+ dP1, we have

An ðd:i1Þ!d iþ ðd:iÞS i;S;. . .;S 1

p

Biþðd:iÞS;d:i

j S pxj !

¼ ðd:nþd1Þ!d nþ ðd:nþdÞS n;S;. . .;S 1

p

Bnþðd:nþdÞS;d:nþd

j S pxj

: ð18Þ

Proof. Seta=d andx=d, Identity(10)becomes An dfiðd:iÞ

d:i

¼dfnðd:nþdÞ

d:nþd : ð19Þ

From (15)we obtain fiðd:iÞ

d:i ¼ðd:i1Þ! iþ ðd:iÞS i;S;. . .;S 1

p

Biþðd:iÞS;d:i

j S pxj

fnðd:nþdÞ

d:nþd ¼ ðd:nþd1Þ! nþ ðd:nþdÞS n;S;. . .;S 1

p

Bnþðd:nþdÞS;d:nþd

j S pxj

: Now, to obtain (18), replace in (19) fi(d.i) andfn(d.n+d) by their above expressions. h

(9)

Corollary 15. Let (xn) be a sequence of real numbers with xS= 1. Then, for r = 1, S= u andd= pP1, with p, d are integers and pn + dP1, and An(xi) be the complete Bell polynomials and we get

An ðpi1Þ!d ðpuþ1Þi i;u;. . .;u 1

Bðpuþ1Þi;pi j u xj

!

¼ ðpnþd1Þ!d ðpuþdþ1Þn n;u;. . .;u 1

Bðpuþdþ1Þn;pnþd

j u xj

:

Corollary 16. Let (xm,n) be a sequence of real numbers with xS= 1 and pP1, qP1, d be integers such that pm + qn + dP 1. Then, for r = 2,d= (p,q) andS= (u,0), we get

Am;n ðpiþqj1Þ!d iþ ðpiþqjÞu i;u;. . .;u 1

BiþðpiþqjÞu;j;piþqj

i0 u xi0;j0 !

¼ ðpmþqnþd1Þ!d mþ ðpmþqnÞu m;u;. . .;u 1

BmþðpmþqnþdÞu;n;pmþqnþd

i0 u xi0;j0

;

for r = 2,d= (p,q) andS= (0,v), we get Am;n ðpiþqj1Þ!d jþ ðpiþqjÞv

j;v;. . .;v 1

Bi;jþðpiþqjÞv;piþqj

j0 v xi0;j0

!

¼ ðpmþqnþd1Þ!d nþ ðpmþqnÞv n;v;. . .;v 1

Bm;nþðpmþqnþdÞv;pmþqnþd

j0 v xi0;j0

; for r = 2,d= (p,q) andS= (u,v), we get

Am;n ðpiþqj1Þ!d

BiþðpiþqjÞu;jþðpiþqjÞv;piþqj

i0 u

j0 v xi0;j0

iþ ðpiþqjÞu i;u;. . .;u

jþ ðpiþqjÞv j;v;. . .;v

0 BB B@

1 CC CA

¼ ðpmþqnþd1Þ!d

BmþðpmþqnþdÞu;nþðpmþqnþdÞv;pmþqnþd

i0 u

j0 v xi0;j0

mþ ðpmþqnÞu m;u;. . .;u

nþ ðpmþqnÞv

n;v;. . .;v

:

Foru= 1 in Corollary 15 we obtain Proposition 4 of[3]and foru=v= 1 in Cor- ollary 16 we obtain Theorem 13 and 16 of[7].

For the choicexn:¼xŒnŒin Theorem 14 and using relation(11), we obtain:

Corollary 17. Let (xn) be a sequence of real numbers with xS= 1,d be an integer and d2Nr:Then, ford.n + dP1, we have

(10)

An

d d:i

Bðdþ1Þ:i;d:iðxjÞ ðdþ1Þ:i

d:i 0

BB B@

1 CC CA¼ d

d:nþd

Bðdþ1Þ:nþd;d:nþdðxjÞ ðdþ1Þ:nþd

d:nþd :

Example 18. Leta,xbe real numbers and (fn(x)) be a binomial type sequence of poly- nomials. For the (xn) given in Example 13, we obtain

An dxfjijðajij þ ðd:iÞxÞ ajij þ ðd:iÞx

¼dxfjnjðajnj þ ðd:nþdÞxÞ ajnj þ ðd:nþdÞx :

4. MULTIPARTITIONAL POLYNOMIALS AND DERIVATIVES OF POLYNOMIALS OF MULTINOMIAL TYPE

Some identities related to the derivatives of polynomial sequences of multinomial type are given. We use the convention D1z¼0gðzÞ ¼0:

Theorem 19. Leta2Rbe a real number and (fn(x)) is of multinomial type sequence of polynomials. We have

Bn;k

i

S pDz¼0ðeazfiSðxþzÞÞ

¼ 1 k!

n nkS;S;. . .;S

p

Dkz¼0ðeazfnkSðkxþzÞÞ: ð20Þ

Proof. Letxn¼ n

S pDz¼0ðeazfnSðxþzÞÞ. Then

1 k!

P

jijP1xiti i!

k

¼k!1 P

iPS

i

S pDz¼0ðeazfiSðxþzÞÞti!i

k

¼ tkS

k!ðS!Þk

P

iP0Dz¼0ðeazfiðxþzÞÞti!i k

¼ tkS

k!ðS!Þk Dz¼0eazP

iP0fiðxþzÞti!i k

¼ tkS

k!ðS!ÞkðDz¼0ðeazFðtÞxþzÞÞk

¼ tkS

k!ðS!ÞkðFðtÞxDz¼0eðaþlnFðtÞÞzÞk

¼ tkS

k!ðS!ÞkFðtÞkxðaþlnFðtÞÞk

¼ tkS

k!ðS!ÞkFðtÞkxDkz¼0eðaþlnFðtÞÞz

¼ tkS

k!ðS!ÞkDkz¼0ðeazFðtÞkxþzÞ

¼ tkS

k!ðS!Þk

P

iP0Dkz¼0ðeazfiðkxþzÞÞti!i

¼k!1P

nPkS

n nkS;S;. . .;S

p

Dkz¼0ðeazfnkSðkxþzÞÞtn!n;

(11)

so, we obtain the desired identity by identification. h

Theorem 20. Let d2Nr, d2NI,a2Rr,a, x be real numbers and (fn(x)) be a multi- nomial type sequence of polynomials. Then we have

Bn;k d i

S pDd:ðiSÞþd1ðxDþ1Þ fiSðtþzÞ

tþz eaz

z¼0

¼ d ðk1Þ!

n nkS;S;. . .;S

p

Dd:ðnkSÞþkd1ðxDþ1Þ fnkSðTþzÞ ðTþzÞ eaz

z¼0

;

ð21Þ wheret¼a:ðiSÞ þdx and T¼a:ðnkSÞ þkdx.

Proof. Set in(14)xn:¼Dz=0(eazfnS(x+z)) and use Theorem 19, we obtain Bn;k

d t

i

S pDtz¼0ðeazfiSðtxþzÞÞ

¼ d

ðk1Þ!T

n nkS;S;. . .;S

p

DTz¼0ðeazfnkSðTxþzÞÞ; witht=dÆ(iS) +d,T=dÆ(nkS) +kd.

To obtain (21), use in this identity the multinomial type sequence (hn(x)) instead (fn(x)), wherehn(x) is defined, as in(8), byhnðxÞ:¼ðaxdÞ:nþxx fnððaxdÞ:nþxÞ. h Corollary 21. Let a, x, a be real numbers, pP0, dP1 be integers and (fn(x)) be a binomial type sequence of polynomials. For r = 1,d= p,a= a andS= u, we have

Bn;k d i

u DpðiuÞþd1ðxDþ1Þ fiuðt1þzÞ t1þz eaz

z¼0

¼ d ðk1Þ!

n ku

DpðnkuÞþkd1ðxDþ1Þ fnkuðT1þzÞ T1þz eaz

z¼0

;

wheret1¼aðiuÞ þdx and T1¼aðnkuÞ þkdx.

Corollary 22. Let a,x,a, b be real numbers, pP0, qP0, dP1 be integers and (fm,n(x)) be a trinomial type sequence of polynomials. Then, for r = 2, d= (p, q), a= (a, b) andS= (u, 0), we get

Bm;n;k d i

u DpðiuÞþqjþd1ðxDþ1Þ fiu;jðt10þzÞ t10þz eaz

z¼0

¼ d ðk1Þ!

m ku

DpðmkuÞþqnþkd1ðxDþ1Þ fmku;nðT10þzÞ T10þz eaz

z¼0

;

where t10¼aðiuÞ þbjþdx and T10¼aðmkuÞ þbnþkdx; for r = 2,d= (p,q), a= (a,b) andS= (0,v), we get

(12)

Bm;n;k d j

v DpiþqðjvÞþd1ðxDþ1Þ fi;jvðt01þzÞ t01þz eaz

z¼0

¼ d ðk1Þ!

n kv

DpmþqðnkvÞþkd1ðxDþ1Þ fm;nkvðT01þzÞ T01þz eaz

z¼0

;

where t01¼aiþbðjvÞ þdx and T01¼amþbðnkvÞ þkdx; for r = 2, d= (p,q), a= (a,b) andS= (u,v), we get

Bm;n;k d i u

j

v DpðiuÞþqðjvÞþd1ðxDþ1Þ fiu;jvðt11þzÞ

t11þz eaz

z¼0

¼ d ðk1Þ!

m ku n

kv

DpðmkuÞþqðnkvÞþkd1ðxDþ1Þ fmku;nkvðT11þzÞ ðT11þzÞ eaz

z¼0

; where t11¼aðiuÞ þbðjvtÞ þdx and T11¼aðmkuÞ þbðnkvÞ þkdx.

Foru= 1 in Corollary 21 we obtain Corollary 19 of[5]and foru=v= 1 in Cor- ollary 22 we obtain Theorems 16 and 17 of [5].

Now, when we use a multinomial type sequence as in (9), Theorem 20 becomes:

Corollary 23. Let d2Nr, d2NI, 2Rr, a, x be real numbers and (fn(x)) be a multinomial type sequence of polynomials. Then we have

Bn;k d i

S pDd:ðiSÞþd1ðxDþ1Þ fjiSjðtþzÞ

tþz eaz

jz¼0

¼ d ðk1Þ!

n nkS;S;. . .;S

p

Dd:ðnkSÞþkd1ðxDþ1Þ fjnkSjðTþzÞ ðTþzÞ eaz

z¼0

;

ð22Þ where t¼a:ðiSÞ þdx and T¼a:ðnkSÞ þkdx.

Theorem 24. Letd2 ðNIÞr;d be an integer,a2Rr;a;x be real numbers and (fn(x)) be a multinomial type sequence of polynomials. Then, ford.n + dP1, we have

An dDd:i1ðxDþ1Þ fiða:iþzÞ a:iþz eaz

z¼0

¼dDd:nþd1ðxDþ1Þ fnða:nþdxþzÞ a:nþdxþz eaz

z¼0

: ð23Þ

Proof. Set in(18)xn:¼Dz=0(eazfnS(x+z)) and use Lemma 19, we obtain An

d

d:iDd:iz¼0ðeazfiððd:iÞxþzÞÞ

¼ d

d:nþdDd:nþdz¼0 ðeazfnððd:nþdÞxþzÞÞ:

To obtain (23), use in this identity the multinomial type sequence (hn(x)) instead of (fn(x)), wherehn(x) is defined, as in(8), byhnðxÞ:¼ðaxdÞ:nþxx fnððaxdÞ:nþxÞ. h

(13)

Corollary 25. Leta, x, a be real numbers, pP1, d be integers such that pn + dP1 and (fn(x)) be a binomial type sequence of polynomials. For r = 1,d= p anda= a, we get

An dDpi1ðxDþ1Þ fiðaiþ aiþz eaz

z¼0

¼dDpnþd1ðxDþ1Þ fnðanþdxþzÞ

anþdxþz eaz

z¼0

:

Corollary 26. Let a, x, a, b be real numbers, pP1, qP1, d be integers such that pm + qn + dP1 and (fm,n(x)) be a trinomial type sequence of polynomials. For r = 2,d= (p,q) and a= (a,b), we get

Am;n dDpiþqj1ðxDþ1Þ fi;jðaiþbjþzÞ aiþbjþz eaz

z¼0

¼dDpmþqnþd1ðxDþ1Þ fm;nðamþbnþdxþzÞ amþbnþdxþz eaz

z¼0

:

Corollary 25 is Corollary 20 of[5]and Corollary 26 is Theorem 18 of[5].

Now, when we use a multinomial type sequence as in(9), Theorem 24 becomes:

Corollary 27. Letd2 ðNIÞr;d be an integer,a2Rr;a, x be real numbers and (fn(x)) be a multinomial type sequence of polynomials. Then, ford.n + dP1, we have

An dDd:i1ðxDþ1Þ fjijða:iþzÞ a:iþz eaz

z¼0

¼dDd:nþd1ðxDþ1Þ fjnjða:nþdxþzÞ a:nþdxþz eaz

z¼0

:

ACKNOWLEDGMENTS

The authors thank the referee for valuable advice and comments which helped to im- prove the quality of this paper.

REFERENCES

[1]H. Belbachir, M. Mihoubi, The (exponential) bipartitional polynomials and polynomial sequences of trinomial type, Part II, Integers 11 (A29) (2011) 12.

[2]E.T. Bell, Exponential polynomials, Ann. Math. 35 (1934) 258–277.

[3]M. Mihoubi, Bell polynomials and binomial type sequences, Discrete Math. 308 (2008) 2450–2459.

[4] M. Mihoubi, The role of binomial type sequences in determination identities for Bell polynomials 111 (2013) 323–337.

[5]M. Mihoubi, H. Belbachir, The (exponential) bipartitional polynomials and polynomial sequences of trinomial type, Part I, Integers 11 (A18) (2011) 17.

[6] M. Mihoubi, H. Belbachir, The (exponential) multipartitional polynomials and polynomial sequences of multinomial type, Part I, Arab J. Math. Sci. 20 (2) (2014) 233–245.

[7]S. Roman, The Umbral Calculus, Academic Press, New York, 1984.

Riferimenti

Documenti correlati

Using polyconvex and compressible stored energy functions of polynomial-type, in the present paper the equilibrium problem of slender beams in the fully nonlinear context of

We analyze the problem of carrying out an efficient iteration to approximate the eigenvalues of some rank structured pencils obtained as linearization of sums of polynomials

L.: C onstruction of alg eb raic cub ature rules using polynomial ideal theory.. P.: Some softw are implementations of the functions sine

We deal with increasing sequences of sets: by considering sequences of sets in a given class, we study when the closure of the union belongs to the same class.. We consider here

Extending the Calculus of Looping Sequences to Model Protein Interaction at the Domain Level. A Calculus of Looping Sequences for Modelling

In the framework of finite order variational sequences generalized Jacobi morphisms for General Relativity are explicitly represented by means of variational vertical derivatives

The used method, even if simple and straightforward, is very general and allows to approach also polynomial Pell equations with higher degrees by means of a generalization of the

In this paper we prove some results, and construct some counterexamples, concerning three questions that are strictly related to Sard’s theorem: structure of the level sets of maps