Download Eulerian numbers, generalized arithmetic

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts
no text concepts found
Transcript
U.P.B. Sci. Bull., Series A, Vol. 71,Iss. 2, 2009
ISSN 1223-7027
EULERIAN NUMBERS AND GENERALIZED ARITHMETICGEOMETRIC SERIES
Mircea I CÎRNU1
Pe baza teoremei Cauchy-Mertens de înmulţire a seriilor, se obţin într-un
mod simplu unele rezultate binecunoscute, cum ar fi reprezentarea numerelor
Euleriene, o identitate combinatorială şi formula sumei seriilor aritmeticogeometrice generalizate. Se dau două aplicaţii.
Based on Cauchy-Mertens theorem of series multiplication, we obtain in a
simple manner some
well-known results, namely
the Eulerian numbers
representation, a combinatorial identity and the formula for the sum of generalized
arithmetic-geometric series. Two applications are given.
Keywords: Eulerian numbers, generalized arithmetic-geometric series.
2000 Mathematics Subject Classification: 05A19, 11B68, 65B10.
1. Eulerian numbers and the sum of generalized arithmetic-geometric series
We present the formula that express the sums of generalized
arithmetic-geometric series by the Eulerian numbers ([1]). It can to be
considered as the definition of these numbers. Are included recurrence
and symmetry relations for Eulerian numbers that give the possibility to
compute these numbers.
In entire paper z is a complex variable.
Theorem 1. Let n  0 be a natural number. The sum of a
generalized arithmetic-geometric series (the generating function of the
sequence k n , k  0,1,2, ) is given by the formula

Gn  z    k n z k 
k 1
n 1
E n, k  z
1  z  
z
n 1
k
,
z 1
(1)
k 0
where En, k  are the Eulerian numbers, given by the relations
En,0  En, n  1  1 ,
1
(2)
Prof., Department of Mathematics III, University POLITEHNICA of Bucharest, Romania
26
Mircea Cîrnu
En, k   k  1 En  1, k   n  k  En  1, k  1 , k  1,2, , n  2 . (3)

z
is uniform
1 z
k 1
convergent. By differentiation and multiplication with z , we obtain



z
z
z
k
2 k
,
,


k
z

k
z

1

z
k3zk 
1  4 z  z 2 and



2
3
4
1  z  k 1
1  z 
1  z 
k 1
k 1
so on. Hence the relations (1), (2), (3) are satisfied for n  0,1,2,3 .
We shall prove the relation (1) by mathematical induction after n .
Assuming the relation (1) is true for fixed n  0 , by differentiation and
multiplication with z , we obtain

n 1
z 1  nz  n1
z2
n 1 k
k


k
z

E
n
,
k
z

k E n, k  z k 1 

n2 
n 1 
1  z  k 0
1  z  k 1
k 1
n 1
z

E n, k k  1  n  k z  z k .
n2 
1  z  k 0
We obtain the relation (1) for n  1 if we denote
Proof. For z  r  1, the geometric series
z
k


n 1
 E n, k k  1  n  k z  z
k 0

n
k
  E n  1, k  z k .
(4)
k 0
By identification, from (4) we obtain for the coefficients of z 0 , the
relation En,0  En  1,0 . Because E1,0  1, it follows that En,0  1 . For the
coefficients of z n , we obtain the relation En, n  1  En  1, n . Because
E1,0  1, it follows that En, n  1  1 . For the coefficients of z k , we obtain the
relation
k  1 En, k   n  k  1 En, k  1  En  1, k  , k  1,2,, n  1 .
Replacing n  1 instead of n , the recurrence relation (3) follows.
Consequence. The Eulerian numbers satisfy the following symmetry
relation
(5)
En, k   En, n  k  1 , k  0,1,, n  1 , n  1,2,
Proof. We prove the relation (5) by induction after n . For n  1 the
relation (5) is obvious. By (2) it results the relation (5) is true for k  0 and
k  n  1.
We suppose (5) satisfied for a fixed natural n  1. Replacing k with n  k
in (5), it is obtain the relation
Eulerian numbers and generalized arithmetic-geometric series
En, n  k   En, k  1 , k  1, , n  2 .
27
(6)
Using the relation (3) and induction hypothesis under the forms (5) and
(6), we have
En  1, n  k   n  k  1 En, n  k   k  1 En, n  k  1 
 n  k  1 En, k  1  k  1 En, k   En  1, k  ,
hence the relation (5) is true for n  1 . By induction axiom, the relation (5) is
accomplished for every natural number n  1.
Remark. From relations (2) and (3), the values of Eulerian numbers
En, k  can be computed by recurrence. The first values of these numbers are
given below:
0
1
2
3
4
5
6
k
--------------------------------------------------------------------n
1
1
2
1
1
3
1
4
1
4
1
11
11
1
5
1
26
66
26
1
6
1
57
302
302
57
1
7
1
120 1191 2416 1191 120
1
2. Closed-form expression for the Eulerian numbers
Theorem 2. Let n  0 be a natural number. The Eulerian numbers are
given by the formula
k
j  n  1
 k  j  1n , k  0,1, , n  1 .
E n, k     1 
j 0
 j 
For k  n , we have the following identity
(7)
28
Mircea Cîrnu
n
 n  1
n  j  1n  0 ,
j


  1 
j 0
j
and, for k  n  1, the identity
n 1
n 1
 1 j   k  j  1n  0 .

j 0
 j 
(8)
(9)
Proof. One consider the relation (1) given in theorem 1, in the form

n 1
j 1
k 0
1  z n1  j n z j 1   En, k  z k ,
z  1.
Using Newton's binomial theorem and changing the index j  k  1 in the
left hand series, the above relation becomes

n 1
 n1
k  n  1 k  
n n
k







1
z
k

1
z


  E n, k z .
 k   
 k 0

   k 0
 k 0
(10)
Denoting

k  n  1
 , k  0,1, , n  1
 1 
n
ak  
, bk  k  1 , 0,1, ,
 k 

0
,
k  n 1

and using the Cauchy-Mertens theorem about the multiplications of series (see
for example [2]), the identity (10) can be rewritten as:


n 1

k 
k
k
a
z
b
z

c
z

E n, k z k ,

 k   k   k
k 0
 k 0
  k 0
 k 0
(11)
where the coefficients ck are given by the formula
k
ck   a j bk  j , k  0,1,
j 0
Identifying the coefficients in relation (11), if k  n  1 , we have
(12)
29
Eulerian numbers and generalized arithmetic-geometric series
k
j  n  1
k  j  1n  En, k  , n  0,1,  ,
ck    1 
j
j 0


n
j  n  1
n  j  1n  0 and (8)
hence (7) is obtained. If k  n , then cn    1 
j
j 0


n 1
j  n  1
k  j  1n  0 and follows
is also verified. If k  n  1, then ck    1 
j
j 0


(9).
Remarks. 1) One observe that the identity (8) follows from (9). Namely,
n 1
j  n  1
 n  j  2n  0 hence, if
from (9) for k  n  1, we have   1 
j 0
 j 
n
j n
n 1
substitute n with n  1 , we obtain   1  n  j  1  0 , therefore
j 0
 j
n
n  1
 n
n  j  1n  n  1  1 j  n  j  1n 1 0 .
j 0
 j 
 j
j
  1 
n
j 0
2) Representing the combinatorial coefficients by factorials, the identities
(8) and (9) may be written as:
 1 j n  j  1n1

n  j !
j!
j 0
n
 1 j k  j  1n  0 ,

j ! n  j  1!
j 0
n 1
 0,
k  n  1.
3. Formula for the sum of generalized arithmetic-geometric series
Replacing in formula (1) the closed-form expression of Eulerian numbers
given by the formula (7), it obtains the following result (see [3] or [1]) :
Theorem 3. Let be n  0 a natural number. The sum of the generalized
arithmetic-geometric series (the generating function of the sequence k n ,
k  0,1,) has the following formula

Gn  z    k n z k 
k 1
k
j  n  1
n k







1
k

j

1

 z , z  1. (13)


 j 
1  z n1 k 0  j 0



z
n 1
30
Mircea Cîrnu
4. Applications
4.1. Arithmetic-geometric series
Let a, r , q be complex numbers, q  1 . Using the formula (13) for n  1,
the sum of arithmetic-geometric series is



a  k  1r  q k 1  a  r  q k 1  r  k q k  a  r  r q 2 

q k 1
1  q q 1  q 
k 1
k 1

a 1  q   rq
1  q 2
.
4.2. Z-transforms
For n  1,2, , from (1) and (5), changing the index j  n  1  k , the Ztransform of the sequence of powers of natural numbers { k n : k  0,1,2, }, is
given by the formula

kn
1
n
Z k   k  Gn   
z
k 0 z
1
n 1
n 1
E n, j 
z
z

E n, j  z n 1 j 
n 1 
j
n 1 
z  1 j 0
 1  j 0 z
1  
 z
n 1
n 1
z
z
k
(14)



E
n
,
n

1

k
z

E n, k  z k , z  1.


n 1
n 1
z  1 k 0
z  1 k 0
z
z 1  z 
z 1  4z  z 2
2
3
In particular, Z k  
, Z k  
, Z k 
.
z  12
z  13
z  14
REFERENCES
 
  

[1]. L. Comtet, Advanced Combinatorics, Reidel, 1974.
[2]. W. Rudin, Principles of Mathematical Analysis, 3rd Ed., Mc Graw-Hill, New York, 1976.
[3]. R. Stalley, A Generalization of the Geometric Series, American Mathematical Monthly, 56
(1949) 5, 325-327.
Related documents