Sun 2016

2nd Reading
January 17, 2017 15:15 WSPC/S1793-0421 203-IJNT 1750079
International Journal of Number Theory

(2017)
c World Scientific Publishing Company
DOI: 10.1142/S1793042117500798
Some further properties of even and odd sequences
Zhi-Hong Sun
by "THE UNIVERSITY OF NEWCASTLE, AUSTRALIA" on 03/31/17. For personal use only.
School of Mathematical Sciences

Huaiyin Normal University, Huaian
Jiangsu 223300, P. R. China
zhsun@hytc.edu.cn
Int. J. Number Theory Downloaded from www.worldscientific.com
Received 6 April 2016

Accepted 21 July 2016
Published 19 January 2017
In this paper, we continue

Pn `nto´ investigate the properties of those sequences {an } satis-
k
fying the condition k=0 k (−1) ak = ±an (n ≥ 0). As applications we deduce some
recurrence relations and congruences for Bernoulli and Euler numbers.
Keywords: Even sequence; odd sequence; congruence; Bernoulli number; Euler number;
Fibonacci number.
Mathematics Subject Classification 2010: 11B83, 11B37, 11B65, 11A07, 11B39, 11B68,
05A15, 05A19
1. Introduction

The classical binomial inversion formula states that an = nk=0 nk (−1)k bk (n =
n n
0, 1, 2, . . .) if and only if bn = k=0 k (−1)k ak (n = 0, 1, 2, . . .). Following
[10] we

continue to study those sequences {an } with the property nk=0 nk (−1)k ak = ±an
(n = 0, 1, 2, . . .).
Definition 1.1. If a sequence {an } satisfies the relation

n
n
(−1)k ak = an (n = 0, 1, 2, . . .),
k
k=0
we say that {an } is an even sequence. If {an } satisfies the relation

n
n
(−1)k ak = −an (n = 0, 1, 2, . . .),
k
k=0
we say that {an } is an odd sequence.

From [10, Theorem 3.2] we know that {an } is an even (respectively, odd)
∞ n
sequence if and only if e−x/2 n=0 an xn! is an even (respectively, odd) function.
1
2nd Reading
2 Z.-H. Sun
Throughout this paper, S + denotes the set of even sequences, and S − denotes the
set of odd sequences. In [10] the author stated that
−1
−1
1 n + 2m − 1 2n −2n n x
, , 2 , (−1) dx ∈ S + .
2n m n 0 n
n−1 n
Let {Bn } be the Bernoulli numbers given by B0 = 1 and k=0 k Bk = 0
(n ≥ 2). It is well known that B1 = − 21 and B2m+1 = 0 for m ≥ 1. Thus,
n
n
n−1
n k k
(−1) · (−1) Bk = Bn + Bk = (−1)n Bn
k k
k=0 k=0
∞ n
n
and so {(−1) Bn } ∈ S as claimed in [10]. It is also known that
+
n=0 Bn xn! = exx−1
(|x| < 2π). Thus, for |x| < 2π,

∞
xn −2x −x x
(−1)n (2n − 1)Bn = −2x − −x = −x .
n=0
n! e −1 e −1 e +1
−x/2
Since e x/(e−x +1) is an odd function, we deduce that {(−1)n (2n −1)Bn } ∈ S − .
2et
∞ n
The Euler numbers {En } is defined by e2t = En tn! (|t| < π2 ), which is
n 2nn=0
+1
equivalent to (see [4]) E0 = 1, E2n−1 = 0 and r=0 2r E2r = 0 (n ≥ 1). It is clear
that
n n
t t
∞
En − 1 tn ∞ ∞
2 2
n
· = En −
n=0
2 n! n=0
n! n=0
n!
t
2e 2 t t 1 − et
= t − e2 = e2 · (|t| < π).
e +1 1 + et
e−t et −1 − 2t
∞ En −1 tn En −1
As 1−
1+e−t = et +1 , we see that e n=0 2n · n! is an odd function. Thus { 2n }
is an odd sequence.
For two numbers b and c, let {Un (b, c)} and {Vn (b, c)} be the Lucas sequences
given by
U0 (b, c) = 0, U1 (b, c) = 1, Un+1 (b, c) = bUn (b, c) − cUn−1 (b, c) (n ≥ 1)
and
V0 (b, c) = 2, V1 (b, c) = b, Vn+1 (b, c) = bVn (b, c) − cVn−1 (b, c) (n ≥ 1).
It is well known that (see [14])
 √ n √ n

 1 b + b2 − 4c b − b2 − 4c

 √ − if b2 − 4c = 0,

 b2 − 4c 2 2
Un (b, c) =

 n−1



n
b
if b2 − 4c = 0
2
2nd Reading
Some further properties of even and odd sequences 3
and
√ n √ n
b+ b2 − 4c b− b2 − 4c
Vn (b, c) = + .
2 2
From this one can easily see that for b = 0, {Un (b, c)/bn } is an odd sequence and
{Vn (b, c)/bn } is an even sequence. We note that Fn = Un (1, −1) is the Fibonacci
sequence and n = Un (2, 1).
Let {An } be an even sequence or an odd sequence. In Sec. 2 we deduce new
recurrence formulas for {An } and give a criterion for polynomials Pm (x) with the
property Pm (1 − x) = (−1)m Pm (x), in Sec. 3 we establish a transformation for-
p−1 Ak+1 p−1 Ak
mula for nk=0 nk Ak , in Sec. 4 we give congruences for k=1 k , k=1 k and
p−2 Ak 2
modulo p , where p is an odd prime. As applications we establish some
k=0 k+1
recurrence formulas for Bernoulli and Euler numbers. Here are some typical results:

If {An } is an odd sequence, then nk=0 nk (−1)k A2n−k = 0. If {An } is an even
n n
sequence, then k=0 k (−1)k (2n − k)A2n−k−1 = 0.
If {Ak } is an even sequence and n is odd, then
n
n n
 2  (−1)k An−k = n n+k
(−1)k Ak = 0
k k k
k=0 k=0
and
n
n
n 1 n
An−k = Ak .
k 3 k
k=0 k=0
3|k
m
Let m be a positive integer and Pm (x) = k=0 ak xm−k . Then
Pm (1 − x) = (−1)m Pm (x)
n
n a a
⇔ k = (−1)n  n (n = 0, 1, . . . , m).
k m m
k=0
 
k
n
Let p be an odd prime, and let {Ak } be an odd sequence of rational p-integers.
Then
p−1
p−1

Ak+1 Ak
2Ap+1 − Ap ≡ A1 − p (mod p2 ) and ≡ 0 (mod p2 ).
k p+k
k=1 k=1
In addition to the above notation, throughout this paper we use the follow-
ing notation: [x] the greatest integer not exceeding x, N the set of positive
integers, R the set of real numbers, Zp the set of those rational numbers whose
denominator is coprime to p, ( ap ) the Legendre symbol.
2nd Reading
4 Z.-H. Sun
2. Recurrence Formulas for Even and Odd Sequences

n
Suppose that k=0 nk (−1)k ak = ±an for n = 0, 1, 2, . . . . Then clearly
n
n − 1
n−1
n k
(−1) kak−1 = −n (−1)r ar = ∓nan−1
k r=0
r
k=0
and
a0
n n
ak+1 − 1 n+1 a0
n k 2
(−1) =− (−1)k+1 ak+1 −
k k+1 n+1 k+1 2
k=0 k=0
n+1
1 n + 1 a a
(−1)r ar −
0 0
=− − a0 −
n+1 r=0
r 2 2
a0
1 a0 an+1 ∓
2 .
=− ±an+1 − =∓
n+1 2 n+1
Thus,
 a 
 an+1 − 0 
{an } ∈ S + implies {nan−1 }, 2 ∈ S−. (2.1)
 n+1 
When {an } ∈ S − , we have a0 = −a0 and so a0 = 0. Therefore, from the above we

deduce that

an+1
{an } ∈ S − implies {nan−1 }, ∈ S+. (2.2)
n+1
For x, y ∈ R and n ∈ {0, 1, 2, . . .} it is well known that (see [2, (3.1)])
n
x y x+y
= .
k n−k n
k=0
n n−m
This is called Vandermonde’s identity. Let an = k=0 k (−1)n−k bn−k (n =
0, 1, 2, . . .). Using Vandermonde’s identity, we see that
n
n−m
(−1)n−k an−k
k
k=0
n
n − k − m
n−k
n−m
= (−1)n−k (−1)n−k−j bn−k−j
k j=0
j
k=0
n s
n−m s−m
= (−1)s (−1)s−j bs−j
s=0
n − s j=0
j
2nd Reading
n
s n n
n−m m−r−1 n−m m−r−1
= br = br
s=0
n−s r=0
s−r r=0 s=r
n−s s−r
n
n−r−1
= br = bn (n = 0, 1, 2, . . .).
r=0
n−r
Thus,
n

n−m
an = (−1)n−k bn−k (n = 0, 1, 2, . . .)
k
k=0
n

n−m
⇔ bn = (−1)n−k an−k (n = 0, 1, 2, . . .). (2.3)
k
k=0
n n−m
Lemma 2.1. Let m, p ∈ R and k=0 k (−1)n−k an−k = ±an for n = 0,
1, 2, . . . . Then
n n
n−p−m n−k p
(−1) an−k = ± (−1)k an−k for n = 0, 1, 2, . . . .
k k
k=0 k=0
Proof. Using Vandermonde’s identity we see that

n n
n−p−m n−k n−p−m
(−1) an−k = (−1)k ak
k n−k
k=0 k=0
n k
n−p−m k k−m
= ± (−1) (−1)r ar
n−k r=0
k − r
k=0
n
n
n−p−m k−r k − m
= ± (−1) ar
r=0 k=r
n−k k−r
n
n n − p − mm − 1 − r
= ± ar
r=0 k=r
n−k k−r
n−r
n n − p − mm − 1 − r
= ± ar
r=0 s=0
n−r−s s
n
n−p−r−1
=± ar
r=0
n−r
n
p
=± (−1)n−r ar
r=0
n − r
n
p
=± (−1)k an−k .
k
k=0
So the lemma is proved.

2nd Reading
6 Z.-H. Sun
Theorem 2.1. If {An } is an odd sequence, then

n
n
(−1)k A2n−k = 0 for n = 0, 1, 2, . . . .
k
k=0
If {An } is an even sequence, for n = 0, 1, 2, . . . we have

n
n
(−1)k (2n − k)A2n−k−1 = 0 and
k
k=0
n
n A2n−k+1 (−1)n A0
(−1)k = .
k 2n − k + 1 2(2n + 1) 2n
n
k=0
Moreover, for given even sequence {Ak } we also have

n
n
 2  (−1)k An−k = 0 for n = 1, 3, 5, . . . .
k=0 k
Proof. We first assume that {An } is an odd sequence. Putting m = 0, p = n/2

and an = An in Lemma 2.1 we see that
n n
n n
 2  (−1)n−k An−k = −  2  (−1)k An−k .
k=0 k k=0 k
Thus, for even n we have
n n
n/2 n

 2  (−1)k An−k =  2  (−1)k An−k = 0.
k=0 k k=0 k
n n
Replacing n with 2n we get k=0 k (−1)k A2n−k = 0 for n = 0, 1, 2, . . . .
Now we assume that {An } ∈ S + . By (2.1), {nAn−1 } ∈ S − and { An+1n+1−A0 /2
}∈
−
S . Applying the above we find that
n n
A0
n n A2n−k+1 −
(−1)k (2n − k)A2n−k−1 = (−1)k 2 = 0.
k k 2n − k + 1
k=0 k=0
By [2, (1.40)],
n
n 1 1
(−1)k = .
k k+z n+z
k=0 z
n
Thus,
n
n 1 1 (−1)n
(−1)k = = .
k 2n + 1 − k n − 2n − 1 2n
k=0 (2n + 1) (2n + 1)
n n
2nd Reading
Hence
n
n A2n−k+1 (−1)n A0
(−1)k = .
k 2n − k + 1 2n
k=0 2(2n + 1)
n
If n is odd, taking m = 0, p = n/2 and an = An in Lemma 2.1 we deduce that
n n
2 (−1)k A
k=0 k n−k = 0. This completes the proof.
Since { En2n−1 } is an odd sequence and E2k−1 = 0 for k ≥ 1, from Theorem 2.1

we see that nk=0 nk (−1)k E2n−k22n−k

−1
= 0 and so
[n/2] n n
n E2n−2k n n (−1)k (−1)n
k E2n−k
= (−1) = = .
2k 22n−2k k 22n−k k 22n−k 22n
k=0 k=0 k=0

That is,
[n/2]
n 2k
2 E2n−2k = (−1)n for n = 0, 1, 2, . . . . (2.4)
2k
k=0
Since {(−1)n Bn }is an even sequence, from Theorem 2.1 we have
n n k
k=0 k (−1) (2n − k)(−1) 2n−k−1
B2n−k−1 = 0. As B2m+1 = 0 for m > 1, we
obtain
2 ]
[ n+1

n
(2n − 2r + 1)B2n−2r = 0 for n = 3, 4, 5, . . . . (2.5)
r=1
2r − 1
From Theorem 2.1 we also have
n
n

2 Bn−k = 0 for n = 1, 3, 5, . . . . (2.6)
k=0 k
Theorem 2.2. Let {ak } be a given sequence. Then
n k

n n+k n−k k
ak − (−1) as = 0 for n = 0, 1, 2, . . . .
k k s=0
s
k=0
Hence, if {An } is an even sequence and n is odd, or if {An } is an odd sequence and
n is even, then
n
n n+k
(−1)k Ak = 0.
k k
k=0
−x
Proof. Since k = (−1)k x+k−1 k , using Vandermonde’s identity we see that for
m ∈ {0, 1, . . . , n},
n n
n−m n+k n−m −n − 1
(−1)n−k = (−1)n
k−m k n−k k
k=m k=0

n −m − 1 m+n
= (−1) = .
n n
2nd Reading
8 Z.-H. Sun
n k
n n−m
Note that k m =
m k−m . Applying the above we deduce that
n k

n n+k n−k k
ak − (−1) as
k k s=0
s
k=0
n
n
n n+m n n+k k
= am − (−1)n−k
m=0
m m k k m
k=m
n
n
n n+m n n−m n−k n + k
= am − (−1)
m=0
m m m k−m k
k=m
n

= am · 0 = 0.
m=0
Putting ak = (−1)k Ak in the above formula we obtain the remaining result.
As an example, taking Ak = (−1)k Bk in Theorem 2.2 we obtain

n
n n+k
Bk = 0 for n = 1, 3, 5, . . . . (2.7)
k k
k=0
Let {Fn } be the Fibonacci sequence given by F0 = 0, F1 = 1 and Fn+1 = Fn +

Fn−1 (n ≥ 1). As {Fn } is an odd sequence, taking Ak = Fk in Theorem 2.2 we get
n
n n+k
(−1)k Fk = 0 for n = 0, 2, 4, . . . . (2.8)
k k
k=0
Lemma 2.2. Suppose that m is a non-negative integer. Then

n
n−m−1
(−1)n−k an−k = ±an (n = 0, 1, 2, . . .)
k
k=0
if and only if
n
n ak a
= ±(−1)n n (n = 0, 1, . . . , m)
k m m
k=0
k n
and
n
n
(−1)k ak+m+1 = ±(−1)m+1 an+m+1 (n = 0, 1, 2, . . .).
k
k=0
n−m−1
Proof. For n = 0, 1, . . . , m we have m n an
n = 0. Set An = (−1) (m) . As k
m
k n m
n
n−k = (−1) k n , we see that

n n

n−m−1 n−m−1 m
(−1)n−k an−k = An−k
k k n−k
k=0 k=0
2nd Reading

n
m n
= (−1)k An−k
n k
k=0

n
m n
= (−1)n (−1)k Ak .
n k
k=0
Thus,
n
n

n−m−1 n
(−1)n−k an−k = ±an ⇔ (−1)k Ak = ±An . (2.9)
k k
k=0 k=0
This together with the fact that
n + m + 1 − m − 1
n+m+1
(−1)n+m+1−k an+m+1−k
k
k=0
n
n
= (−1)n−k+m+1 an−k+m+1
k
k=0
n
n
= (−1)r+m+1 ar+m+1 (n = 0, 1, 2, . . .)
r=0
r
yields the result.

Lemma 2.3. Let {an } be a given sequence, a(x) = ∞ n
n=0 an x and m ∈ R. Then

x
(1 − x)m a = ±a(x)
x−1
n
n−m−1
⇔ (−1)n−k an−k
k
k=0
= ±an (n = 0, 1, 2, . . .).
Proof. Clearly, for |x| < 1,

∞
m x
(1 − x) a = (−1)r ar xr (1 − x)m−r
x−1 r=0
∞
∞

m−r
= (−1)r ar xr (−x)k
r=0
k
k=0
∞ n

m − (n − k)
n−k
= (−1) an−k (−1) xn
k
n=0
k
k=0
n
∞ n − m − 1
n−k
= (−1) an−k xn .
n=0
k
k=0
Thus the result follows.
2nd Reading
10 Z.-H. Sun
m m
Theorem 2.3. Let m ∈ N, Pm (x) = k=0 ak xm−k and Pm
∗
(x) = k=0 ak xk . Then
the following statements are equivalent:
x
(a) (1 − x)m Pm∗
( x−1 ) = ±Pm∗
(x).
m
(b) Pm (1 − x) = ±(−1) Pm (x).
n
(c) For n = 0, 1, . . . , m we have k=0 n−m−1 (−1)n−k an−k = ±an .
n n−m−1
k
(d) Set an = 0 for n > m. Then k=0 k (−1)n−k an−k = ±an (n = 0, 1,
2, . . .).
(e) For n = 0, 1, . . . , m we have
n
n ak an
= ±(−1)n .
k m m
k=0
k n
∗
Proof. Since Pm (x) = xm Pm ( x1 ) we see that

x 1 1
(1 − x)m Pm∗
= ±Pm ∗
(x) ⇔ (−x)m Pm 1 − = ±xm Pm
x−1 x x

1 1
⇔ (−1)m Pm 1 − = ±Pm
x x
⇔ Pm (1 − x) = ±(−1)m Pm (x).
So (a) and (b) are equivalent. By Lemma 2.3, (a) is equivalent to (d). Assume
an+m+1 = 0 for n ≥ 0. Then
n
n
an+m+1 = 0 = (−1)n−k+m+1 am+1+n−k
k
k=0

m+n+1
m+n+1−m−1

= (−1)m+n+1−k am+n+1−k .
k
k=0
So (c) is equivalent to (d). To complete the proof, we note that (d) is equivalent to
(e) by Lemma 2.2.
Remark 2.1. Let {Bn (x)} and {En (x)} be the Bernoulli polynomials and Euler
polynomials given by
n n
n 1 n
Bn (x) = Bk xn−k and En (x) = n (2x − 1)n−k Ek .
k 2 k
k=0 k=0
It is well known that (see [4]) Bn (1−x) = (−1) Bn (x) and En (1−x) = (−1)n En (x).
n
Theorem 2.4. Let {An } ∈ S + with A0 = · · · = Al−1 = 0 and Al = 0 (l ≥ 1).

Then

An+l
∈ S+.
(n + 1)(n + 2) · · · (n + l)
2nd Reading
∞ ∞
Proof. Assume an = An+l . Let a(x) = n=0 an xn and A(x) = n=0 An xn . Then
l
clearly A(x) = xl a(x). Since Al = k=0 kl (−1)k Ak = (−1)l Al we see that 2 | l.
Thus, applying Lemma 2.3 and (2.9) we see that

x x
{An } ∈ S ⇔ A
+
= (1 − x)A(x) ⇔ a = (1 − x)l+1 a(x)
x−1 x−1
n
n+l
⇔ (−1)n−k an−k = an (n = 0, 1, 2, . . .)
k
k=0
 

 


 (−1)n a  
n
⇔ ∈ S+.

 −l − 1  
 
 
n
Note that
an an An+l
(−1)n = = · l!.
−l − 1 n+l (n + 1)(n + 2) · · · (n + l)
n l
We then obtain the result.

n
Corollary 2.1. Suppose that {an } ∈ S + with a0 = 0 and An = 1
(n+1)(n+2) k=0
ak (n ≥ 0). Then {An } ∈ S + .
n−2
Proof. Let b0 = b1 = 0 and bn = k=0 ak (n ≥ 2). By [10, Corollary 3.2],
bn+2
{bn } ∈ S + . Now applying Theorem 2.4 we find that { (n+1)(n+2) } ∈ S + . That is,
{An } ∈ S + .
Theorem 2.5. Let F be a given function. If {An } is an even sequence, then

n
k
n k
k s
(−1) Ak (−1) (F (s) − F (n − s)) = 0 (n = 0, 1, 2, . . .).
k s=0
s
k=0
If {An } is an odd sequence, then

n
k
n k
k s
(−1) Ak (−1) (F (s) + F (n − s)) = 0 (n = 0, 1, 2, . . .).
k s=0
s
k=0
n
n
Proof. Suppose that k=0 k (−1)k Ak = ±An for n = 0, 1, 2, . . .. From [9,
Lemma 2.1] we have
n
k
n

n n k
k r
(−1) f (k)Ak = ± (−1) F (n − k + r) Ak ,
k k r=0
r
k=0 k=0
2nd Reading
12 Z.-H. Sun
k k
where f (k) = (−1)s F (s). Thus
s=0 s
n
k

n k
k s
(−1) Ak f (k) ∓ (−1) F (n − s) = 0.
k s=0
s
k=0
This yields the result.
Corollary 2.2. If {An } ∈ S + , then

n
n
(−1)k Ak (1 + x)k (1 − (−1)n xn−k ) = 0
for n = 0, 1, 2, . . . .
k
k=0
If {An } ∈ S − , then
n
n
(−1)k Ak (1 + x)k (1 + (−1)n xn−k ) = 0 for n = 0, 1, 2, . . . .
k
k=0
Proof. Taking F (s) = (−x)s in Theorem 2.5 and then applying the binomial the-
orem we obtain the result.
Remark 2.2. From [9, (2.5)] we know that

n m
n k m
(−1) f (m + k) = (−1)k F (n + k),
k k
k=0 k=0
r r s
where F (r) = s=0 s (−1) f (s). Hence for any non-negative integers m and n we
have
n m
n k m
(−1) Ak+m = (−1)k Ak+n for {Ak } ∈ S + , (2.10)
k k
k=0 k=0
n
m

n k m
(−1) Ak+m =− (−1)k Ak+n for {Ak } ∈ S − . (2.11)
k k
k=0 k=0
n n
3. A Transformation Formula for k=0 k An
Lemma 3.1 ([10, Theorems 4.1 and 4.2]). Let f be a given function and n ∈ N.
(i) If {An } is an even sequence, then
n
k

n k
n−k
f (k) − (−1) f (s) An−k = 0.
k s=0
s
k=0
(ii) If {An } is an odd sequence, then

n
k

n k
n−k
f (k) + (−1) f (s) An−k = 0.
k s=0
s
k=0
We remark that a simple proof of Lemma 3.1 was given by Wang [13].
2nd Reading
Theorem 3.1. Let n ∈ N. If {Am } is an even sequence and n is odd, or if {Am }

is an odd sequence and n is even, then
n n n
n n 1 n
An−k = Ak = Ak .
k k 3 k
k=0 k=0 k=0
3|k 3 | n−k
√
Proof. Set ω = (−1 + −3)/2. If {Am } is an even sequence and n is odd, or if
{Am } is an odd sequence and n is even, putting f (k) = ω k in Lemma 3.1 we obtain
n
n
(ω k + (−1)k (1 + ω)k )An−k = 0.
k
k=0

As 1 + ω = −ω 2 , we have nk=0 nk (ω k + ω 2k )An−k = 0. Therefore,

n
n n n
n n n n
3 An−k − Ak = (1 + ω k + ω 2k )An−k − An−k
k k k k
k=0 k=0 k=0 k=0
3|k
n
n
= (ω k + ω 2k )An−k = 0.
k
k=0
This proves the theorem.
Corollary 3.1 (Ramanujan [7]). For n = 3, 5, 7 . . . we have

 n
n 
− 6 if n ≡ 1 (mod 6),
n
Bn−k =
k 
n
k=0 if n ≡ 3, 5 (mod 6).
6 | k−3 3
Proof. As {(−1)n Bn } ∈ S + , B1 = − 12 and B2m+1 = 0 for m ≥ 1, taking An =

(−1)n Bn in Theorem 3.1 we obtain
n n
n
n 1 n 1 n
n−k k
(−1) Bn−k = (−1) Bk = Bk + n
k 3 k 3 k
k=0 k=0 k=0
3|k
1 n
= (n + Bn ) = .
3 3
To see the result, we note that

 n
n n −nB1 = if n ≡ 1 (mod 6),
n n 2
(−1)n−k Bn−k − Bn−k =
k k 

k=0
3|k
k=0
6 | k−3
0 if n ≡ 3, 5 (mod 6).
2nd Reading
14 Z.-H. Sun
Corollary 3.2 (Ramanujan [7]). For n = 0, 2, 4 . . . we have

 n
[n/6] 
− 6 if n ≡ 4 (mod 6),
1 n n
n−6k
(2 − 1)Bn + (2 − 1)Bn−6k =
3 6k n

k=0 if n ≡ 0, 2 (mod 6).
3
Proof. Since {(−1)n (2n − 1)Bn } is an odd sequence, B1 = − 12 and B2m+1 = 0 for
m ≥ 1, applying Theorem 3.1 we see that for even n,
n
n
(−1)n−k (2n−k − 1)Bn−k
k
k=0
3|k
n
1 n
= (−1)k (2k − 1)Bk

3 k
k=0
n
1 n
= (2k − 1)Bk + n
3 k
k=0
1 n 1
= (−(−1)n (2n − 1)Bn + n) = − (2n − 1)Bn .
3 3 3
On the other hand, for even n,
n n

n n
(−1)n−k (2n−k − 1)Bn−k − (2n−k − 1)Bn−k
k k
k=0 k=0
3|k 6|k
n
−nB1 = if 6 | n − 4,
= 2
0 if 6 n − 4.
Now combining all the above yields the result.
Corollary 3.3 (Lehmer [3]). For n = 0, 2, 4 . . . we have

[n/6]
n 1 + (−3)n/2
En + 3 26k−2 En−6k = .
6k 2
k=1
Proof. Since {(En − 1)/2n } is an odd sequence and E2k+1 = 0, from Theorem 3.1
and the binomial theorem we see that for even n,
n n
n En−k n 1
n−k
− n−k
k 2 k 2
k=0 k=0
6|k 3|k
n
n
n En−k − 1 1 n Ek − 1
= =
k 2n−k 3 k 2k
k=0 k=0
3|k
2nd Reading

n n n

1 n k Ek − 1 1 1
= (−1) + 1− − 1+
3 k 2k 2 2
k=0

1 En − 1 1 − 3n 2 − 3n − En
= − + = .
3 2n 2n 3 · 2n
For even n, we also have
n n
n k n k 1
2 = 2 · (1 + ω k + ω 2k )
k k 3
k=0 k=0
3|k
1
= ((1 + 2)n + (1 + 2ω)n + (1 + 2ω 2 )n )
3
√ √
1 1 n
= (3n + ( −3)n + (− −3)n ) = (3n + 2 · (−3) 2 ).
3 3
Hence
n
n
1 n k 2 − 3n 3n + 2 · (−3) 2 2 n
En + 2 En−k = + = (1 + (−3) 2 ).
3 k 3 3 3
k=0
6|k
Remark 3.1. Compared with known proofs of Corollaries 3.1–3.3 (see [1, 3, 7]),
our proofs are simple and natural.
4. Congruences Involving Even and Odd Sequences

Let p be an odd prime. For k ∈ {1, 2, . . . , p − 1}, we see that

p p(p − 1) · · · (p − (k − 1)) p
= ≡ (−1)k−1 (mod p2 ). (4.1)
k k! k
If {An } is an even sequence and A0 , A1 , . . . , Ap−2 , pAp−1 , Ap ∈ Zp , applying
(4.1) we see that
p p
p−2
p
Ap = (−1)k Ak = A0 + pAp−1 − Ap + (−1)k Ak
k k
k=0 k=1
p−2
Ak
≡ A0 + pAp−1 − Ap − p (mod p2 ).
k
k=1
Hence
p−2
Ak
2Ap − pAp−1 ≡ A0 − p (mod p2 ) for {An } ∈ S + . (4.2)
k
k=1
If {An } is an odd sequence and A0 , A1 , . . . , Ap ∈ Zp , using (4.1) we see that
p p−1

p Ak
−Ap = (−1)k Ak ≡ A0 − Ap − p (mod p2 ).
k k
k=0 k=1
2nd Reading
16 Z.-H. Sun
Since A0 = −A0 we have A0 = 0 and so

p−1
Ak
≡0 (mod p) for {An } ∈ S − . (4.3)
k
k=1
We note that (4.3) was first obtained by Mattarei and Tauraso [5] via a complicated
method.
For an odd prime p and a ∈ Zp let a p ∈ {0, 1, . . . p − 1} be given by a p ≡ a
(mod p). Let p be an odd prime, a ∈ Zp and A0 , A1 , . . . , Ap−1 ∈ Zp . By [11,
Theorem 2.4], if a p is odd and {An } is an even sequence, or if a p is even and

{An } is an odd sequence, then
a−1 − a
p−1
Ak ≡ 0 (mod p2 ). (4.4)
k k
k=0
In the case a = − 21 ,
(4.4) was given by the author in an earlier unpublished preprint.
Inspired by the author, Sun deduced (4.4) in the cases a = − 13 , − 41 , − 16 . See [12,
Theorem 1.4].
Now we establish new congruences for sums involving even or odd sequences.
Theorem 4.1. Let p be an odd prime. If {An } ∈ S + and A1 , . . . , Ap−2 , pAp−1 , Ap ∈

Zp , then
Ak+1p−3
pAp−1
Ap − ≡ (p + 1)A1 − p (mod p2 ).
2 k
k=1
−
If {An } ∈ S and A1 , A2 , . . . , Ap+1 ∈ Zp , then
p−1
Ak+1
2Ap+1 − Ap ≡ A1 − p (mod p2 ).
k
k=1
Proof. If {An } ∈ S + , from [10, Corollary 3.1(a)] we have {2An+1 − An } ∈ S − .

Thus,
p p
p k p
Ap − 2Ap+1 = (−1) (2Ak+1 − Ak ) = 2 (−1)k Ak+1 − Ap
k k
k=0 k=0
p−1
p
= 2 A1 − Ap+1 + (−1)k Ak+1 − Ap .
k
k=1
Hence applying (4.1) we deduce that
p(p − 1) p
p−3
Ap = A1 + pAp − Ap−1 + (−1)k Ak+1
2 k
k=1
Ak+1 p−3
p(p − 1)
≡ A1 + pAp − Ap−1 − p (mod p2 ).
2 k
k=1
2nd Reading
This yields the first part. If {An } ∈ S − , from [10, Corollary 3.1(a)] we have {2An+1 −
An } ∈ S + . Thus,
p p
p k p
2Ap+1 − Ap = (−1) (2Ak+1 − Ak ) = 2 (−1)k Ak+1 + Ap
k k
k=0 k=0

p−1
p

k
= 2 A1 − Ap+1 + (−1) Ak+1 + Ap .
k
k=1
Hence applying (4.1) we obtain

p−1
p
Ap+1 − Ap = A1 − Ap+1 + (−1)k Ak+1
k
k=1
p−1
Ak+1
≡ A1 − Ap+1 − p (mod p2 ).
k
k=1
This yields the remaining result. The proof is now complete.
Corollary 4.1. Let p be an odd prime. Then

(p−3)/2
B2k
pBp−1 ≡ −p − 1 + 2p (mod p2 ).
2k − 1
k=1
Proof. Since {(−1)n Bn } ∈ S + , B1 = − 12 and B2m+1 = 0 for m > 1, taking

An = (−1)n Bn in Theorem 4.1 yields the result.
Corollary 4.2. Let p be an odd prime and b, c ∈ Zp with b ≡ 0 (mod p). Then
p−1

Uk+1 (b, c)
p
Vp (b, c) ≡ b 1 − p (mod p2 ).
kbk
k=1
Proof. Since { Unb(b,c)

n } ∈ S − and Vp (b, c) = 2Up+1 (b, c) − bUp (b, c), from Theo-
rem 4.1 we deduce the result.
Theorem 4.2. Let p be an odd prime. If {An } ∈ S + and A0 , A1 , . . . , Ap−2 ∈ Zp ,

then
p−2
p−2
Ak
Ak ≡ p (mod p2 ).
k+1
k=0 k=0
−
If {An } ∈ S and A0 , A1 , . . . , Ap−1 ∈ Zp , then
p−2
p−2
Ak
Ak ≡ −2Ap−1 − p (mod p2 ).
k+1
k=0 k=0
2nd Reading
18 Z.-H. Sun
Proof. Since

p−1
p−1−p
p−1
p−1

p−1−k
(−1) Ap−1−k = Ap−1−k = Ak
k
k=0 k=0 k=0
and

p−1
p
p−2
p

(−1)k Ap−1−k = Ap−1 + (−1)k Ak ,
k k+1
k=0 k=0
putting m = 0 and n = p − 1 in Lemma 2.1 and then applying (4.1) we see that if
{An } ∈ S + , then
p−2
p
p−2 p−2
Ak
Ak = (−1)k Ak ≡ p (mod p2 );
k+1 k+1
k=0 k=0 k=0
−
if {An } ∈ S , then
p−1

p−2
p
p−2
Ak
Ak = −Ap−1 − (−1)k Ak ≡ −Ap−1 − p (mod p2 ).
k+1 k+1
k=0 k=0 k=0
Corollary 4.3. Let p > 3 be a prime and b, c ∈ Zp with bc ≡ 0 (mod p). Then
p−2
p pc Vk (b, c)
Vp (b, c) ≡ b − (mod p2 ).
b (k + 1)bk
k=0
√2 k √ 2 k
Proof. It is well known that Vk (b, c) = b+ b2 −4c + b− b2 −4c . Thus,
p−2 p−2
√ k p−2 √ k
Vk (b, c) b + b2 − 4c b − b2 − 4c
= +
bk 2b 2b
k=0 k=0 k=0
√ p−1 √ p−1
b + b2 − 4c b − b2 − 4c
1− 1−
2b 2b
= √ + √
b + b2 − 4c b − b2 − 4c
1− 1−
2b 2b
p
4b 2
Vp (b, c) b − Vp (b, c)
= 1− = .
4c bp bp−2 c
Since { Vnb(b,c)
n } ∈ S + , from the above and Theorem 4.2 we deduce the result.
Theorem 4.3. Let p be an odd prime, and let {An } be an odd sequence. Suppose
A1 , A2 , . . . , Ap−1 ∈ Zp . Then
p−1
p−1
Ak
Ak
≡p (mod p2 ).
k k2
k=1 k=1
2nd Reading
p−1 p−1+k
Proof. Taking n = p − 1 in Theorem 2.2 we get k=0 p−1 k k (−1)k Ak = 0.
For k = 1, 2, . . . , p − 1 we see that

p−1 p−1+k (p − 1)(p − 2) · · · (p − k) p(p + 1) · · · (p + k − 1)
= ·
k k k! k!
p (p2 − 12 )(p2 − 22 ) · · · (p2 − k 2 )
= ·
p+k k!2
p
≡ (−1)k (mod p3 ).
p+k
Since A0 = −A0 we have A0 = 0. Now, from all the above we deduce that
p−1
Ak
≡ 0 (mod p2 ) for {An } ∈ S − . (4.5)
p+k
k=1
k−p k−p p
For k = 1, 2, . . . , p − 1 we have 1
k+p = k2 −p2 ≡ k2 = 1
k − k2 (mod p2 ). Thus the
result follows.
Let Fn = Un (1, −1) and Ln = Vn (1, −1) be the Fibonacci and Lucas sequences,
respectively. From Theorem 4.3 we have the following corollary.
Corollary 4.4. Let p > 5 be a prime. Then

p−1
Fk p 5p Fp−( p ) 2
≡− 5
(mod p2 ).
k 5 4 p
k=1
Proof. Recently Pan and Sun [6, Remark 3.3] proved that
2
1 p Lp − 1
p−1
Fk
≡ − (mod p).
k2 5 5 p
k=1
It is known that (see [8]) Fp−( p5 ) ≡ 0 (mod p) and Lp−( p5 ) ≡ 2( p5 ) (mod p2 ). Also,
5Fn = 2Ln+1 − Ln = Ln + 2Ln−1 . Thus
p
5Fp−( p5 ) = 2Lp − Lp−( p5 ) ≡ 2(Lp − 1) (mod p2 )
5
and so
2
1 p 5Fp−( p5 )
p−1
Fk
≡− (mod p2 ).
k2 5 5 2p
k=1
Since {Fk } is an odd sequence, applying Theorem 4.3 we deduce the result.
Theorem 4.4. Let p be a prime greater than 3, and let {An } be an even sequence.
Suppose that A0 , A1 , . . . , Ap−2 , Ap , pAp−1 ∈ Zp . Then
p−2
p−2
Ak
Ak A0 + pAp−1 − 2Ap
≡ −p 2
+ (mod p2 ).
k k p
k=1 k=1
2nd Reading
20 Z.-H. Sun
p pp+k k
Proof. Taking n = p in Theorem 2.2 we get k=0 k k (−1) Ak = 0. For
k = 1, 2, . . . , p − 1 we see that

p p+k p(p − 1) · · · (p − k + 1) (p + 1) · · · (p + k)
= ·
k k k! k!
p (p2 − 12 )(p2 − 22 ) · · · (p2 − k 2 ) p
= · ≡ (−1)k (mod p3 ).
p−k k!2 p−k
Thus,
p−2

2p p 2p − 1 p
A0 − Ap + Ap−1 + Ak
p p−1 p−1 p−k
k=1
p
p p+k
≡ (−1)k Ak = 0 (mod p3 ).
k k
k=0
Hence

2p − 1 2p − 1
p−2
2 Ap − p Ap−1 − A0
Ak p−1 p−1
≡ (mod p2 ).
p−k p
k=1
2p−1
The famous Wolstenholme’s congruence (see [15]) states that p−1 ≡ 1 (mod p3 ).
Thus
p−2
Ak 2Ap − A0 − pAp−1
≡ (mod p2 ) for {An } ∈ S + . (4.6)
p−k p
k=1
k+p k+p p
For k = 1, 2, . . . , p − 2 we have 1
k−p = k2 −p2 ≡ k2 = 1
k + k2 (mod p2 ). Hence the
result follows.
Corollary 4.5. Let p > 5 be a prime. Then

p−1
Lk 2(1 − Lp )
≡ (mod p2 ).
k p
k=1
p−1
Proof. In [6] Pan and Sun proved that k=1 Lk2k ≡ 0 (mod p), which was con-
jectured by Tauraso. Since {Ln } ∈ S + , taking Ak = Lk in Theorem 4.4 we see
that
p−2
p−1
Lk Lk Lp−1 2 + pLp−1 − 2Lp
≡ −p − +
k k2 (p − 1)2 p
k=1 k=1
2(1 − Lp ) Lp−1 2(1 − Lp )

≡ (p + 1)Lp−1 + ≡− + (mod p2 ).
p p−1 p
2nd Reading
Corollary 4.6. Let p be a prime greater than 3. Then

(p−3)/2
B2k p + 1 pBp−1 + 1
≡ − (mod p2 ).
p − 2k 2 p
k=1
Proof. It is well known that B0 , B1 , . . . , Bp−2 , Bp , pBp−1 ∈ Zp . Taking Ak =

(−1)k Bk in (4.6) and applying the fact B2k+1 = 0 for k ≥ 1 we deduce the
result.
Corollary 4.7. Let p > 3 be a prime, b, c ∈ Zp and b ≡ 0 (mod p). Then

p−1
Vk (b, c) 2(Vp (b, c) − bp )
≡ (mod p2 ).
(p − k)bk pbp
k=1
Proof. Taking Ak = Vk (b, c)/bk in (4.6) yields the result.
Theorem 4.5. Let p be an odd prime and A0 , A1 , . . . , A p−1 ∈ Zp . If {An } ∈ S +

2
and p ≡ 3 (mod 4), or if {An } ∈ S − and p ≡ 1 (mod 4), then
(p−1)/2
2k Ak
≡0 (mod p).
k 2k
k=0
Proof. Since { 21n } ∈ S + , by Lemma 3.1(i) we have

p − 1

(p−1)/2
2
 2  (−1)k Ak p−1
k=0 k 2 −k
2
p − 1

(p−1)/2
k
  k p−1
−k k s 1
= 2 (−1) Ak − (−1) 2 (−1) As p−1 = 0.
k=0 k s=0
s 2 2 −k
p−1 1 2k
Note that k2 ≡ −k2 = (−4)1
k k (mod p) by [2, p. 90]. From the above we
deduce the result.
Theorem 4.6. Let p be an odd prime and A0 , A1 , . . . , Ap ∈ Zp . If {An } is an odd

sequence, then
(p−1)/2
2k Ap−1−k
≡ 0 (mod p).
k 4k
k=0
If {An } is an even sequence, then
(p−1)/2 2k
A0
Ap − p−1
−1
2 ≡ −A 2 + k
Ap−k (mod p).
0
p p 4k · k
k=1
2nd Reading
22 Z.-H. Sun
p−1 − 1
Proof. Note that k
2 ≡ k
2 = 1 2k
(−4)k k
(mod p). If {An } is an odd sequ-
p−1
ence, taking n = in Theorem 2.1 and applying the above we deduce that
(p−1)/2 2k Ap−1−k 2
k ≡ 0 (mod p). Now assume that {An} is an even sequence.

k=0 4k p
Since p | kp for k = 1, 2, . . . , p − 1, we see that Ap = k=0 kp (−1)k Ak ≡ A0 − Ap
(mod p) and so Ap ≡ A0 /2 (mod p). By Theorem 2.1,
p − 1 A0
Ap
(p−1)/2
Ap−k p−1
+  2  (−1) k
= (−1) 2  2  .
p p−k p−1
k=1 k
p p − 1
2
(p−1)/2 2k k
Since ≡ k /(−4) (mod p), we deduce that
k
 
p−1
A0
Ap  p − 1  − (−1)(p−1)/2
2 (p−1)/2 2k

2  k Ap−k
 ≡ (mod p).
p−1 4k · k
k=1
p p − 1
2
(p−1)/2 1 2p −2
It is well known that (see [8, Corollary 1.2] or [9, Theorem 5.2]) k=1 k ≡ − p
(mod p). Thus,

  p−1  
p−1 (p − 1)(p − 2) · · · p −
(p−1)/2
p − 1 = 2 p−1 1
≡ (−1) 2 1 − p
p−1 k
2 ! k=1
2
p−1
≡ (−1) 2 (2p − 1) (mod p2 ).
Hence,
 
p−1
A0
Ap  p − 1  − (−1)(p−1)/2 A0
2 Ap (1 + 2p − 2) −
2   ≡ 2
p−1 p(1 + 2p − 2)
p p − 1
2
A0
Ap − p
≡ 2 + 2 − 2A
p
p p
A0
Ap − p−1
−1
≡ 2 +A 2 (mod p).
0
p p
Combining all the above proves the theorem.
2nd Reading
Added Remark. Let p be an odd prime and A0 , A1 , . . . , A p−1 ∈ Zp . Observe that

− 1 2k −k
p/2
p p/2−1

p −1
2
(−1)k
k
2 = k (−4) and k = 2k k−1 ≡ 2k k−1 = − 2
2k p (mod p ) for
p−1
k ∈ N. Putting m = 0, p = − 2, n = 2 in Lemma 2.1 and then applying the
1
n
above we deduce that if k=0 nk (−1)k Ak = ±An for n ≥ 0, then

2k  
p

(p−1)/2
k
(p−1)/2
p−1
A p−1 −k = ±  2  (−1) 2 −k A p−1
4k 2 2 −k
k=0 k=0 k
 
p
(p−1)/2
A p−1
2 −k 
p−1
≡ ±(−1) 2 A p−1 − (mod p2 ).
2 2 k
k=1
Acknowledgments
The author is supported by the National Natural Science Foundation of China
(Grant No. 11371163).
References
[1] M. Chellali, Accélération de calcul de nombres de Bernoulli, J. Number Theory 28
(1988) 347–362.
[2] H. W. Gould, Combinatorial Identities: A Standardized Set of Tables Listing 500
Binomial Coeﬃcient Summations (West Virginia University, Morgantown, WV,
1972).
[3] D. H. Lehmer, Lacunary recurrence formulas for the numbers of Bernoulli and Euler,
Ann. of Math. 36 (1935) 637–649.
[4] W. Magnus, F. Oberhettinger and R. P. Soni, Formulas and Theorems for the Special
Functions of Mathematical Physics, 3rd edn. (Springer, New York, 1966), pp. 25–32.
[5] S. Mattarei and R. Tauraso, Congruences of multiple sums involving sequences invari-
ant under the binomial transform, J. Integer Seq. 13 (2010), Art. 10.5.1, 12 pp.
[6] H. Pan and Z. W. Sun, Proof of three conjectures on congruences, Sci. China Math.
57 (2014) 2091–2102.
[7] S. Ramanujan, Some properties of Bernoulli’s numbers, J. Indian Math. Soc. 3 (1911)
219–234.
Xn
`n´
[8] Z. H. Sun, Combinatorial sum k and its applications in number theory
k=0
k≡r (mod m)
I, J. Nanjing Univ. Math. Biquarterly 9 (1992) 227–240.
[9] , Congruences concerning Bernoulli numbers and Bernoulli polynomials, Dis-
crete Appl. Math. 105 (2000) 193–223.
[10] , Invariant sequences under binomial transformation, Fibonacci Quart. 39
(2001) 324–333.
[11] , Generalized Legendre polynomials and related supercongruences, J. Number
Theory 143 (2014) 293–319.
[12] Z. W. Sun, Supercongruences involving products of two binomial coeﬃcients, Finite
Fields Appl. 22 (2013) 24–44.
[13] Y. Wang, Self-inverse sequences related to a binomial inverse pair, Fibonacci Quart.
43 (2005) 46–52.
2nd Reading
24 Z.-H. Sun
[14] H. C. Williams, The Lucas functions, in Édouard Lucas and Primality Testing, Cana-
dian Mathematical Society Series of Monographs and Advanced Texts, Vol. 22 (Wiley,
New York, 1998), pp. 73–93.
[15] J. Wolstenholme, On certain properties of prime numbers, Quart. J. Pure Appl. Math.
5 (1862) 35–39.

Sun 2016

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Sun 2016

Uploaded by

Copyright:

Available Formats

2nd Reading

January 17, 2017 15:15 WSPC/S1793-0421 203-IJNT 1750079

International Journal of Number Theory

Some further properties of even and odd sequences

School of Mathematical Sciences

Received 6 April 2016

In this paper, we continue

Definition 1.1. If a sequence {an } satisfies the relation

we say that {an } is an even sequence. If {an } satisfies the relation

we say that {an } is an odd sequence.

(|x| < 2π). Thus, for |x| < 2π,

Some further properties of even and odd sequences 3

2. Recurrence Formulas for Even and Odd Sequences

When {an } ∈ S − , we have a0 = −a0 and so a0 = 0. Therefore, from the above we

Some further properties of even and odd sequences 5

Proof. Using Vandermonde’s identity we see that

So the lemma is proved.

Theorem 2.1. If {An } is an odd sequence, then

If {An } is an even sequence, for n = 0, 1, 2, . . . we have

Moreover, for given even sequence {Ak } we also have

Proof. We first assume that {An } is an odd sequence. Putting m = 0, p = n/2

Some further properties of even and odd sequences 7

we see that nk=0 nk (−1)k E2n−k22n−k

k=0 k=0 k=0

Putting ak = (−1)k Ak in the above formula we obtain the remaining result.

As an example, taking Ak = (−1)k Bk in Theorem 2.2 we obtain

Let {Fn } be the Fibonacci sequence given by F0 = 0, F1 = 1 and Fn+1 = Fn +

Lemma 2.2. Suppose that m is a non-negative integer. Then

n−k = (−1) k n , we see that

Some further properties of even and odd sequences 9

Proof. Clearly, for |x| < 1,

Theorem 2.4. Let {An } ∈ S + with A0 = · · · = Al−1 = 0 and Al = 0 (l ≥ 1).

Some further properties of even and odd sequences 11

We then obtain the result.

Theorem 2.5. Let F be a given function. If {An } is an even sequence, then

If {An } is an odd sequence, then

This yields the result.

Corollary 2.2. If {An } ∈ S + , then

Remark 2.2. From [9, (2.5)] we know that

(ii) If {An } is an odd sequence, then

Some further properties of even and odd sequences 13

Theorem 3.1. Let n ∈ N. If {Am } is an even sequence and n is odd, or if {Am }

As 1 + ω = −ω 2 , we have nk=0 nk (ω k + ω 2k )An−k = 0. Therefore,

This proves the theorem.

Corollary 3.1 (Ramanujan [7]). For n = 3, 5, 7 . . . we have

Proof. As {(−1)n Bn } ∈ S + , B1 = − 12 and B2m+1 = 0 for m ≥ 1, taking An =

Corollary 3.2 (Ramanujan [7]). For n = 0, 2, 4 . . . we have

= (−1)k (2k − 1)Bk

Corollary 3.3 (Lehmer [3]). For n = 0, 2, 4 . . . we have

Some further properties of even and odd sequences 15

This yields the result.

4. Congruences Involving Even and Odd Sequences

Since A0 = −A0 we have A0 = 0 and so

Theorem 2.4], if a p is odd and {An } is an even sequence, or if a p is even and

Theorem 4.1. Let p be an odd prime. If {An } ∈ S + and A1 , . . . , Ap−2 , pAp−1 , Ap ∈

Proof. If {An } ∈ S + , from [10, Corollary 3.1(a)] we have {2An+1 − An } ∈ S − .

Hence applying (4.1) we deduce that

Some further properties of even and odd sequences 17

Hence applying (4.1) we obtain

This yields the remaining result. The proof is now complete.

Corollary 4.1. Let p be an odd prime. Then

Proof. Since {(−1)n Bn } ∈ S + , B1 = − 12 and B2m+1 = 0 for m > 1, taking

Proof. Since { Unb(b,c)

Theorem 2.4], if a p is odd and {An } is an even sequence, or if a p is even and