1 Introduction and preliminaries

The Genocchi numbers G n are defined by

2 t e t + 1 = n = 0 G n t n n ! .

The relationship between the Genocchi numbers G n and the Bernoulli numbers B n and the Euler polynomials E n (x) is

G n =2 ( 1 2 n ) B n =n E n 1 (0)for n1.

They are known as follows [1].

Genocchi numbers and Genocchi polynomials are prolific in the mathematical literature, and many results on Genocchi numbers and Genocchi polynomials identities may be seen in the papers [24]. In this paper, we will mainly study the products of Genocchi numbers G n ( k ) with the following forms:

G n ( k ) n ! = i 1 + i 2 + + i k = n G i 1 G i 2 G i k i 1 ! i 2 ! i k ! (k=1,2,),

where n is a nonnegative positive integer G n ( 1 ) = G n , and G 0 = G 0 ( k ) =0 for k1, G n ( k ) =0 for n<k.

It is clear that

( 2 t e t + 1 ) k = n = k G n ( k ) t n n ! = n = 0 G n ( k ) t n n ! (k=1,2,).
(1.1)

Then we have

( 2 e t + 1 ) k = n = k G n ( k ) t n k n ! = n = 0 G n + k ( k ) t n ( n + k ) ! = n = 0 G n + k ( k ) n + 1 k t n n ! ,
(1.2)

where n j =n(n+1)(n+j1).

The paper is organized as follows. In Section 2, we obtain some properties for G n ( k ) by means of the method of generating function. In Section 3, we establish some identities involving G n ( k ) , the Stirling numbers, the generalized Stirling numbers, the higher order Bernoulli numbers and the Cauchy numbers. Finally, in Section 4, we give the asymptotic expansion of some sums involving G n ( k ) .

For convenience, we recall some definitions and notations involved in the paper. Stirling number of the first kind is denoted by s(n,k,r), let B n ( k ) , C n ( k ) , B n , k , P n ( k ) , stand for the higher order Bernoulli numbers, products of Cauchy numbers, Bell polynomials and the potential polynomials. The corresponding generating functions are

n = k s ( n , k , r ) t n n ! = ln k ( 1 + t ) ( 1 + t ) r k ! , n = k | s ( n , k , r ) | t n n ! = ( ln ( 1 t ) ) k ( 1 t ) r k ! , where  n = k , k + 1 , , k = 0 , 1 , 2 , , s ( n , k , 0 ) = s ( n , k ) ( Stirling numbers of the first kind ) ,
(1.3)
n = k S ( n , k , r ) t n n ! = e r t ( e t 1 ) k k ! , n = k , k + 1 , , k = 0 , 1 , 2 , , S ( n , k , 0 ) = S ( n , k ) ( Stirling numbers of the second kind ) ,
(1.4)
n = 0 B n ( r ) t n n ! = ( t e t 1 ) r (r is an integer),
(1.5)
n = 0 C n ( k ) t n n ! = ( t ln ( 1 + t ) ) k ,k=1,2,, C n ( k ) n ! = i 1 + i 2 + + i k = n C i 1 C i 2 C i k i 1 ! i 2 ! i k ! ,
(1.6)
n = k B n , k ( x 1 , x 2 ,) t n n ! = 1 k ! ( n = 1 x m t m m ! ) k ,k=0,1,,
(1.7)
( n = 0 g n t n n ! ) r =1+ n = 1 P n ( r ) t n n ! , g 0 =1(r is a complex number)and 
(1.8)
P n ( r ) = P n ( r ) ( g 1 , g 2 , , g n ) = j = 1 n ( k ) j B n , j ( g 1 , g 2 , ) , P 0 ( r ) = 1 , where  ( k ) j = k ( k 1 ) ( k j + 1 ) .
(1.9)

In this paper, we let [ t n n ! ]f(t) denote the coefficient of t n n ! in f(t), where f(t)= n = 0 f n t n n ! .

An exponential Riordan array is a pair (d(t),h(t)) of formal power series. Where d(t)= n = 0 d n t n n ! , h(t)= n = 1 h n t n n ! .

It defines an infinite, lower triangular array ( d n , k ) n , k N according to the rule

d n , k = [ t n n ! ] d(t) ( t h ( t ) ) k k ! ,nN.

Hence we write ( d n , k )=(d(t),h(t)).

The most important property of Riordan array is: If (d(t),h(t)) is an exponential Riordan array, and let f(t) be the exponential generating function of the sequence { f k } k N , then we have

k = 0 d n , k f k = [ t n n ! ] d(t)f ( t h ( t ) ) = [ t n n ! ] d(t) [ f ( y ) y = t h ( t ) ] ,
(1.10)

where we use the notation [f(y)y=g(t)] as a linearization of the more common one f(y) | y = g ( t ) to denote substitution f(g(t)).

A singularity of f(z) at |z|=w is called algebraic if f(z) can be written as the sum of a function analytic in a neighborhood of w and a finite number of terms of the form

( 1 z w ) α g(z),
(1.11)

where g(z) is analytic near w, g(z)0, and α{0,1,2,} (see [5]).

Lemma 1.1 (See [5])

Suppose that f(z) is analytic for |z|<R, R>0, and has only algebraic singularities of on |z|=R. Let a be the minimum of Re(α) for the terms of the form at the singularity of f(z) on |z|=R, and let w j , α j , and g j (z) be the w, α, and g(z) for those terms of form (1.11), for which Re(α)=a. Then, as n,

[ z n ] f(z)= j g j ( w j ) n α j 1 Γ ( α j ) w j n +o ( R n n α 1 ) .

2 Properties of products of Genocchi numbers

In this section, we obtain some properties for G n ( k ) by means of the method of generating functions and the Euler transformation.

Theorem 2.1 Let nk2 be any integers, then

G n ( k ) = j = k 1 n ( n j ) G j ( k 1 ) G n j ,

where G n ( 1 ) = G n .

Proof By (1.1), we have

n = k G n ( k ) t n n ! = m = k 1 G m ( k 1 ) t m m ! m = 1 G m t m m ! = n = k 1 j = k 1 n ( n j ) G j ( k 1 ) G n j t n n ! ,

and the comparison of the coefficients of t n /n! in the first and the last formulas completes the proof. □

Theorem 2.2 Let nk+1, k1 be any integers, then

k G n ( k + 1 ) =2kn G n 1 ( k ) +2(nk) G n ( k ) .

Proof Deriving both sides of (1.1) with respect to t, we get

n = k G n ( k ) t n 1 ( n 1 ) ! = k ( 2 t e t + 1 ) k 1 ( 2 t e t + 1 ) = k t ( 2 t e t + 1 ) k k ( 2 t e t + 1 ) k + k 2 t ( 2 t e t + 1 ) k + 1 .

Multiplying both sides by t, we have

n = k n G n ( k ) t n n ! =k n = k G n ( k ) t n n ! k n = k G n ( k ) t n + 1 n ! + k 2 n = k + 1 G n ( k + 1 ) t n n ! ,

and the comparison of the coefficients of t n /n! in the first and the last formulas completes the proof. □

Theorem 2.3 Let nk1 be any integers, then

G n ( k ) =k! B n , k ( G 1 , G 2 ,).

Proof By (1.1) and (1.7), we get

n = k G n ( k ) t n n ! = ( n = 1 G n t n n ! ) k = ( G 1 t + G 2 t 2 2 ! + + G n t n n ! + ) k 1 k ! k ! = k ! n = k B n , k ( G 1 , G 2 , ) t n n ! ,

and the comparison of the coefficients of t n /n! in the first and the last formulas completes the proof. □

Theorem 2.4 Let nk1 be any integers, then

G n ( k ) = i = 1 n j = 0 k ( k j ) ( j i ) ( 1 ) k j G n ( i ) .

Proof By (1.1), (1.8), (1.9) and Theorem 2.3, we have

n = 0 G n ( k ) t n n ! = ( 1 + m = 1 G n t n n ! 1 ) k = j = 0 k ( k j ) ( 1 ) k j ( 1 + m = 1 G m t m m ! ) j = j = 0 k ( k j ) ( 1 ) k j n = 0 P n ( j ) ( G 1 , G 2 , ) t n n ! .

Then

G n ( k ) = j = 0 k ( k j ) ( 1 ) k j P n ( j ) ( G 1 , G 2 , ) = j = 0 k ( k j ) ( 1 ) k j i = 1 n ( j ) i B n , i ( G 1 , G 2 , ) = i = 1 n j = 0 k ( k j ) ( j i ) ( 1 ) k j G n ( i ) ,

which completes the proof. □

Theorem 2.5 Let nk1 be any integers, then

G n + k ( k ) k ! = j = k n + k ( n + k j ) S(j,k) 2 n + k j B n + k j ( k ) .

Proof By (1.2) and (1.5), we have

n = 0 G n + k ( k ) n + 1 k t n n ! = ( 2 e t + 1 ) k = k ! t k ( 2 t e 2 t 1 ) k ( e t 1 ) k k ! = k ! t k m = 0 B m ( k ) ( 2 t ) m m ! m = k S ( m , k ) t m m ! = k ! t k n = k j = k n ( n j ) S ( j , k ) 2 n j B n j ( k ) t n n ! = n = 0 j = k n + k ( n + k j ) ( n + k k ) 1 S ( j , k ) 2 n + k j B n + k j ( k ) t n n ! ,

and the comparison of the coefficients of t n /n! in the first and the last formulas completes the proof. □

Theorem 2.6 Let nk1 be any integers, then

G n + k ( k ) n + 1 k = j = 0 n S ( n , j ) k j ( 1 ) j 2 j .

Proof Let (S(n,j))=(1, e t 1 t ), and by (1.10), we have

j = 0 n S ( n , j ) k j ( 1 ) j 2 j = [ t n n ! ] ( 1 ( 1 + y 2 ) k | y = e t 1 ) = [ t n n ! ] ( t e t + 1 ) k = G n + k ( k ) n + 1 k ,

which completes the proof. □

3 Identities involving G n ( k ) , s(n,k,r), S(n,k,r), B n ( k ) and C n ( k )

In this section, by using the Riordan and generating functions method, we explore relationships between these numbers, the G n ( k ) , s(n,k,r), S(n,k,r), B n ( k ) and C n ( k ) .

Theorem 3.1 Let nk1 be any integers, then

j = 0 n ( n j ) G n j ( k ) B j ( k ) = 2 n k B n k ( k ) ( n ) k = 2 n k B n k ( k ) j = 0 k s(k,j) n j ,

where s(n,h) are the Stirling numbers of the first kind, ( n ) k =n(n1)(nk+1).

Proof By (1.1), (1.5), we have

j = 0 n ( n j ) G n j ( k ) B j ( k ) = [ t n n ! ] ( 2 t e t + 1 ) k ( t e t 1 ) k = [ t n n ! ] ( 2 t 2 e 2 t 1 ) k = [ t n n ! ] m = 0 B m ( k ) 2 m t m + k m ! = 2 n k B n k ( k ) ( n ) k = 2 n k B n k ( k ) j = 0 k s ( k , j ) n j ,

which yields Theorem 3.1. □

Theorem 3.2 Let nk1 be any integers, then

j = 0 n h = j n ( n h ) s ( h , j ) C n j ( i ) G j ( k ) = { n ! j = 0 n k ( k + j 1 j ) G n k j ( i k ) ( n k j ) ! ( 1 2 ) j , i > k , n ! ( n 1 k 1 ) ( 1 2 ) n k , i = k , n ! ( k i ) ! j = 0 n k ( k + j 1 j ) s ( n i j , k i ) ( 1 2 ) j , i < k .

Proof Since

h = j n ( n h ) s(h,j) C n j ( i ) = [ t n n ! ] t i ln i ( 1 + t ) ln j ( 1 + t ) j ! ,

then by ( h = j n ( n h ) s(h,j) C n j ( i ) )=( t i ln i ( 1 + t ) , ln ( 1 + t ) t ).

From (1.10), we have

j = 0 n h = j n ( n h ) s ( h , j ) C n j ( i ) G j ( k ) = [ t n n ! ] t i ln i ( 1 + t ) [ ( 2 y e y + 1 ) k | y = ln ( 1 + t ) ] = [ t n n ! ] t i ln i ( 1 + t ) ln k ( 1 + t ) ( 1 + t 2 ) k = [ t n n ! ] { t i ( ln ( 1 + t ) ) i k ( 1 + t 2 ) k , i > k , t k ( 1 + t 2 ) k , i = k , t i ( ln ( 1 + t ) ) k i ( 1 + t 2 ) k , i < k ,

which completes the proof. □

By the proof of Theorem 3.2, we can get the following corollary.

Corollary 3.3 Let i1, j0 be any integer, then

h = j n ( n h ) s(h,j) C n j ( i ) ={ ( n j ) C n j ( i j ) , i > j , δ n , i , i = k , ( n i ) ( j i ) 1 s ( n i , j i ) ( 1 ) n j , i < k .

Theorem 3.4 Let nk1, be any integer, let r>0 be a real number, then

j = 0 n s(n,j,r) G j ( k ) = j = 0 n k ( n j ) ( 1 ) j s ( n j , k , r ) k j k ! 2 j .

Proof By (1.3) and (1.10), we have

j = 0 n s ( n , j , r ) G j ( k ) = [ t n n ! ] 1 ( 1 + t ) r [ ( 2 y e y + 1 ) k | y = ln ( 1 + t ) ] = [ t n n ! ] ln k ( 1 + t ) ( 1 + t ) r 1 ( 1 + t 2 ) k = k ! [ t n n ! ] m = k s ( m , k , r ) t m m ! m = 0 ( k + m 1 m ) ( 1 2 ) m t m = j = 0 n k ( n j ) ( 1 ) j s ( n j , k , r ) k j k ! 2 j .

 □

Theorem 3.5 Let nk1 be any integers, let r>0 be a real number, then

j = 0 n | s ( n , j , r ) | G j ( k ) = j = 0 n k ( n j ) | s ( n j , k , r 1 ) | k j k ! 2 j .

The proof of Theorem 3.5 is similar to that of Theorem 3.4, and is omitted here.

Theorem 3.6 Let nk1 be any integers, let r>0 be a real number, then

j = 0 n s(n,j,r) G k + j ( k ) j + 1 k =n! ( 1 ) n j = 0 n ( r + n j 1 n j ) ( k + j 1 j ) 1 2 j .

Proof For (s(n,j,r))=( 1 ( 1 + t ) r , ln ( 1 + t ) t ).

By (1.2), (1.3) and (1.10), we have

j = 0 n s ( n , j , r ) G k + j ( k ) j + 1 k = [ t n n ! ] 1 ( 1 + t ) r [ ( 2 e y + 1 ) k | y = ln ( 1 + t ) ] = [ t n n ! ] 1 ( 1 + t ) r 1 ( 1 + t 2 ) k = n ! ( 1 ) n j = 0 n ( r + n j 1 n j ) ( k + j 1 j ) 1 2 j ,

which completes the proof. □

Theorem 3.7 Let nk1 be any integers, let r0 be a real number, then

j = 0 n | s ( n , j , r ) | G k + j ( k ) j + 1 k ={ n ! j = 0 n ( r k + n j 1 n j ) ( k + j 1 j ) 2 k j , r > k , k n 2 n k , r = k , n ! j = 0 k r ( k r j ) ( k 1 + n j n j ) 1 2 n k j , r < k .

Proof Since (|s(n,j,r)|)=( 1 ( 1 t ) r , ln ( 1 t ) t ), then by (1.2) and (1.10), we have

j = 0 n | s ( n , j , r ) | G k + j ( k ) j + 1 k = [ t n n ! ] 1 ( 1 t ) r [ ( 2 e y + 1 ) k | y = ln ( 1 t ) ] = n ! [ t n ] 1 ( 1 t ) r k 1 ( 1 t 2 ) k = { n ! j = 0 n ( r k + n j 1 n j ) ( k + j 1 j ) 2 k j , r > k , k n 2 n k , r = k , n ! j = 0 k r ( k r j ) ( k 1 + n j n j ) 1 2 n k j , r < k ,

which completes the proof. □

Theorem 3.8 Let nk1 be any integers, and let r>0 be a real number, then

j = 0 n S(n,j,r) ( 1 ) j k j 2 j = j = 0 n ( n j ) G k + j ( k ) r n j j + 1 k .

Proof Since (S(n,j,r))=( e r t , e t 1 t ), then by (1.3), (1.8) and (1.10), we have

j = 0 n S ( n , j , r ) ( 1 ) j k j 2 j = [ t n n ! ] e r t [ ( 1 1 + y 2 ) k | y = e t 1 ] = n ! [ t n ] e r t ( 2 e t + 1 ) k = j = 0 n ( n j ) G k + j ( k ) r n j j + 1 k .

 □

Theorem 3.9 Let n1 be any integer, and let r>0 be a real number, then

j = 0 n s(n,j,r) 2 j G k + j ( k ) j + 1 k = 2 k n! j = 0 i = 0 n ( r + n i 1 n i ) ( k + i + j 1 i + j ) ( 2 ( i + j ) i ) ( 1 ) n i .

Proof Since (s(n,j,r))=( 1 ( 1 + t ) r , ln ( 1 + t ) t ), then by (1.2) and (1.10), we have

j = 0 n s ( n , j , r ) 2 j G k + j ( k ) j + 1 k = [ t n n ! ] 1 ( 1 + t ) r [ ( 2 e 2 y + 1 ) k | y = ln ( 1 + t ) ] = [ t n n ! ] 1 ( 1 + t ) r 2 k ( 1 + ( 1 + t ) 2 ) k = 2 k n ! [ t n ] i = 0 ( r + i 1 i ) ( 1 ) i t i i = 0 ( j = 0 ( k + i + j 1 i + j ) ( 2 ( i + j ) i ) ) x i = 2 k n ! j = 0 i = 0 n ( r + n i 1 n i ) ( k + i + j 1 i + j ) ( 2 ( i + j ) i ) ( 1 ) n i ,

which completes the proof. □

4 Asymptotics

In this section, we give the asymptotic expansion of certain sums involving G n ( k ) .

Theorem 4.1 Let r>0 be a real number, and suppose that n, then

j = 0 n s(n,j,r) G k + j ( k ) j + 1 k ( 1 ) n 2 k n n + r Γ ( r ) e n 2 π n .

Proof From the proof of Theorem 3.6, we know

j = 0 n s(n,j,r) G k + j ( k ) j + 1 k = [ t n n ! ] 1 ( 1 + t ) r 1 ( 1 + t 2 ) k .

In Lemma 1.1, let f(t)= 1 ( 1 + t ) r , g(t)= 1 ( 1 + t 2 ) k . Obviously, f(t) is analytic for |t|<1, and it has only algebraic singularity on |t|=1, namely at t=1, so by Lemma 1.1 and Stirling’s formula n! e n n n 2 π n , we have

j = 0 n s(n,j,r) G k + j ( k ) j + 1 k ( 1 ) n 2 k n n + r Γ ( r ) e n 2 π n ,

which completes the proof. □

Theorem 4.2 Let k1 be any integer, let r>0 be a real number, and suppose that n, then

j = 0 n | s ( n , j , r ) | G k + j ( k ) j + 1 k { 2 k n n + r + k Γ ( r k ) e n 2 π n , r > k , n n + k Γ ( k ) e n 2 π n , r = k , n n + k ( 1 ) k r Γ ( k ) 2 n 2 π n , r < k .

Proof By Lemma 1.1 and Stirling’s formula, we have

j = 0 n | s ( n , j , r ) | G k + j ( k ) j + 1 k = n ! 2 k [ t n ] { 1 ( 1 t ) r k 1 ( 1 t 2 ) k , r > k , 1 ( 1 t 2 ) k , r = k , ( 1 t ) k r ( 1 t 2 ) k , r < k { 2 k n n + r + k Γ ( r k ) e n 2 π n , r > k , n n + k Γ ( k ) e n 2 π n , r = k , n n + k ( 1 ) k r Γ ( k ) 2 n 2 π n , r < k ,

which completes the proof. □

Theorem 4.3 Let r>0 be a real number, as n, and suppose that

j = 0 n s(n,j,r) 2 j G k + j ( k ) j + 1 k ( 1 ) n 2 k n n + r Γ ( r ) e n 2 π n .

Proof By Lemma 1.1 and Stirling’s formula, we have

j = 0 n s(n,j,r) 2 j G k + j ( k ) j + 1 k =n! [ t n ] 1 ( 1 + t ) r 2 k ( 2 + 2 t + t 2 ) k ( 1 ) n 2 k n n + r Γ ( r ) e n 2 π n ,

which completes the proof. □