A BINOMIAL IDENTITY ON THE LEAST PRIME FACTOR OF

arXiv:1107.5709v1 [math.NT] 28 Jul 2011

AN INTEGER

SAMUEL A. HAMBLETON

Abstra t. An identity for binomial symbols modulo an odd positive integer

relating to the least prime fa tor of

is proved.

The identity is dis ussed

within the ontext of Pell oni s.

1.

Introdu tion

Many results exist on identities relating to binomial oe ients m
r modulo n
where

n is anq odd positive integer [2. Granville [3 has given new results on erning
m
(mod p ) where p is prime, with a ni e a ount of known results. Perhaps the
r
most well known identity on fa torials modulo n is Wilson's theorem, whi h states
that a positive integer n is prime if and only if (n 1)! 1 (mod n). Granville
[3 writes that Fle k [2 has generalized Wilson's theorem to the statement that for
all positive integers r less that the least prime divisor of n, n is prime if and only if
r1  
n1r
Y r + j 
r+1 Y
r
(mod n).
(1)( 2 )
j
r
j=1
j=0

Similarly, we will onsider the residue modulo an odd positive integer n of a symbol
(n, r) dened in terms of binomial oe ients where, likewise, r is less than or
equal to the least prime divisor p of n. We will briey dis uss the ase r > p. Let
a and a respe tively denote the greatest integer A a, and the least integer
A a.

Let n be an odd positive integer, let r 2 be an integer, and let p

be the least prime divisor of n. Dene (n, r) to be the non-negative residue modulo
n of
 n1
  n1 
r
r
2
(n, r) = (1) 2 2 r 2

(1)
(2)r .
r
2

if r < p
0 (mod n)
Then (n, r) satises (n, r) =
.
n/p (mod n) if r = p

Theorem 1.1.

Eqn. (1) o urs as the leading oe ient of the dieren e modulo n of two
polynomials whi h are important in the study of the ane genus zero urves known
as Pell oni s examined in detail by Lemmermeyer [7, 8 and other authors [4, 5
in relation to the analogy between these urves and ellipti urves. Let be the
Date : July 28, 2011.
Mathemati s Subje t Classi ation. Primary 11A51, 11B65; Se ondary 11B39, 11G20.
Key words and phrases. Binomial symbols, fa torization, Pell Coni s, Di kson polynomials.

2010

S. HAMBLETON

fundamental dis riminant of a quadrati number eld K = Q( ). Pell oni s are

the urves
C : H2 U2 = 4,
with group law
 H H + U U H U + H U 
1 2
2 1
1 2
1 2
,
(2)
P1 + P2 =
2
2
dened for points P1 = (H1 , U1 ) and P2 = (H2 , U2 ) over (Z/n), Z, Q, and algebrai
numbers Q among various other rings R for whi h the binary operation + of Eqn.
(2) forms a group C(R) with identity (2, 0). See [7 for more on these urves.
We dene the polynomials n (H) by

1 = 1, 3 = H + 1, 2j+3 = H2j+1 2j1 ,

The origin of the polynomials n (H) an be tra ed to D. H. Lehmer [6 who has

ompared a Lu as fun tion to Sylvester polynomials n (x, y) appearing in Ba hmann's [1 book. The polynomials n (x, y) orrespond to the Gm (x) used by
Williams [10.
n (H) = G(n1)/2 (H)of Williams = n (H, 1)of Sylvester a ording to Lehmer.
It has been shown [4, 5 that the zeros of the polynomials n (H) are in one to
one orresponden e with the H- oordinates of the non-trivial points P 6= (2, 0) of
order dividing n in the group C(Q), non-trivial points of the n-torsion subgroup
C(Q)[n]. One simply expresses the H- oordinate of n(H, U), meaning n1 additions
(H, U) + (H, U) + . . . (H, U), as (H 2)n (H)2 + 2. In order to give a proof of
quadrati re ipro ity [5 using p-torsion on Pell oni s where p is an odd prime, it
was demonstrated that
p (H) (H 2)

p1
2

(mod p).

The leading oe ient of the polynomial n (H) (H 2) 2 evaluated modulo

n is the more general question whi h we address. The polynomials n are also
dis ussed in the ontext of Di kson polynomials of the se ond kind, En (x, a) =
Pn/2 nj

(a)j xn2j . In parti ular, the identity, p.32 of [9,
j=0
j
n1

2n+1 (H) = En (H, 1) + En1 (H, 1),

allows writing, for odd n,

n1

n (H) =

2
X

r=0

(1) 2
r

 n1
2


n1
2r
H 2 r .
r
2

This ompletes the dis ussion of the ontext of the identity for (n, r).
2.

Proof of the main result

We require the following equality whi h holds for all positive integers a.
(3)

a
Y

j=1

(a + j) = 2a

a1
Y

(2j + 1).

j=0

Eqn. (3) Q
may be proved by reordering the produ ts in the numerator and denoma
a+j
, showing that this is equal to 1. The proof of Theorem 1.1 is
inator of j=1 4j2
as follows.

A BINOMIAL IDENTITY ON THE LEAST PRIME FACTOR OF AN INTEGER

Proof. First assume that r < p. Let s = r/2 and t = r/2. Then
(n, r)

=
=

(n, p)

(1)

n1
2

n1
2

Qt1

(2)r ,

Y n1
j  t1

sj ,
s!
r!
2
j=1


Q
Q
t
t1
s
r
n1
 (1)
Y n1
j  t1


j=1 (s + j) (2)
j=0
2
sj ,
r!
2
j=1

t1
Q
Q
t
t1
s
r+t
s
 (1)
Y
2


j=1 (s + j) (1)
j=0 1 + 2j n
1 + 2s + 2j n ,
t1
r!(2)
j=1


Q
Q
t
t1
s
t1


Y
(s
+
j)

2
1
+
2j

n


j=1
j=0
2t+1 (1)r1
1 + 2s + 2j n ,
r!
j=1


Q
t
t1
s Qt1


Y


j=1 (s + j) 2
j=0 1 + 2j
2t+1 (1)r1
1 + 2s + 2j
(mod n).
r!
j=1
 (1)s

(2)r

j=0

n1
2

Sin e r is stri tly less than p, the integers r! and n are relatively prime. By Eqn.
(3), (n, r) = 0. Now let r = p = 2s + 1. Then
(n, p)

=
=

(n, p)

!
!
n1
s1
2
(1)
2p ,
+
p
s

s
Q
 (1)s
2s sj=0 n 1 2j  Y


n1
+
sj ,
s!
p!
2
j=1

s
Qs1
s n
 (1)s
2 ( p 1) j=0 n 1 2j  Y


n1
+
sj ,
s!
(p 1)!
2
j=1

Q


Qs1
s1
s n
s
1


j=0 (s + j + 1) + 2
j=0 n + 1 + 2j Y
p
n + p + 2j ,
s
(p 1)!2
j=1
s

s1
Y
j=0

n1
2

(s + j + 1) +

n
p

s
s
Y
Y




1
s + j (p 1)!1 2s
p + 2j
j=1

s
Y




n
(p 1)!1 2s
s + j p + 2j
p
j=1
p1
Y


n
(p 1)!1 2p+1
p+j
p
j=1

p1


nY
(2j)1 p + j
p j=1

Fermat's theorem shows that

(n, p) = np .

(mod n),

j=1

(mod n),

(mod n),

(mod n).

Qp1

1
j=1 (2j)

p+j

1 (mod p). It follows that



We on lude by spe ulating as to the value of (n, r) when r ex eeds the least
prime divisor of n, within some bounds. The author has only tested the following
onje ture for n < 106 .

S. HAMBLETON

Conje ture 2.1. Let p be the least prime divisor of an odd integer n and assume

that 2 n < 3p. If r is an integer bounded by p < r < n then (n, r) > 0.

If Conje ture 2.1 holds and the least prime divisor p of n satises 2 n < 3p then
the follow exponential algorithm will terminate.

 Let
j A =k(a1 , a2 ) and assume
j we wish
 fa tor n. Set A =
k to

a1 +a2
a1 +a2
(2, n). If n,
= 0, Set A =
, a2 , otherwise set A =
2
2
 j
k
a1 +a2
a1 ,
, and print A. Repeat until a2 a1 2.
2
Algorithm 2.2.

A knowledgments

The author would like to thank Vi tor S haras hkin for do toral supervision of
whi h this proje t has been a very small part of, and supported by the University
of Queensland.

Referen es
1. P. Ba hmann, Die Lehre von der Kreistheilung, Leipzig, pp. 9, 10 (1872).
2. L. Di kson, History of the theory of numbers, I, AMS Chelsea publishing, (1992)
3. A. Granville, Binomial Coe ients (mod pq ), available at:
http://www. e m.sfu. a/organi s/papers/granville/
4. S. Hambleton, Generalized Lu asLehmer tests using Pell oni s, Pro . AMS,
to appear.
5. S. Hambleton, V. S haras hkin, Pell oni s and quadrati re ipro ity, RMJM,
to appear.
6. D. H., Lehmer, An extended theory of Lu as' fun tions, The Ann. Math., 31,
(3), 419448, (1930).
7. F.
Lemmermeyer,
Coni s  A poor man's ellipti urves,
arXiv:math/0311306v1, preprint available at:
http://www.fen.bilkent.edu.tr/franz/publ/ oni s.pdf
8. F. Lemmermeyer,
Modular Pell oni s,
preprint available at:
http://www.rzuser.uni-heidelberg.de/hb3/publ-new.html
9. R. Lidl, G. Mullen, G. Turnwald, Di kson Polynomials, Longman S ienti and
Te hni al (1993).
10. H. Williams, Ee tive primality tests for some integers of the form A5n 1 and
A7n 1, Math. Comp. 48, 177, 385403 (1987).
S hool of Mathemati s and Physi s, University of Queensland, St. Lu ia, Queensland, Australia 4072

E-mail address : sahmaths.uq.edu.au