Pell 方程递归解的幂型因子性质 及其在不定方程中的应用
华南师范大学附属中学: 赵玉博 指导教师: 郝保国
摘要
本文的目标是研究 Catlan 猜想(xmyn 1除 x=3,m=2,y=2,n=3 之外无正整数解) 的一个推广,即:不定方程xm2yn 1都可能有哪些正整数解。为解决这一问题,
深入研究了pell方程的解的一些数论性质,即:当最小解确定时,方程的递归解 的因子性质。并将其结论应用于对这种不定方程的讨论中,在限定n为偶数的情 况下取得了较为成功的结果。此外,文中还包含了关于pell方程解的数论性质的 结论在一些其它问题中的应用。
关键词:pell方程、Catlan猜想
Some Arithmetic Properties about the Factor of the Solution to Pell Equation
and Its Applications in Diophantine Equations
The Affiliated High School of South China Normal University Zhao Yubo Directed by: Hao Baoguo
Abstract
This Paper is aimed to study a generalization of the Catlan Conjecture (The only nontrivial solution to
xmyn 1is x=3,m=2,y=2,n=3), that is, to determine the nontrivial solutions to
xm2yn 1. To solve this problem, I study some arithmetic properties of the solutions to Pell equation, or more precisely, when the minimum solution is fixed, the properties about factors of the recursive solutions to the equation. Furthermore, I apply these properties into the study of
xm2yn 1 ,and obtain some successful results when n is restricted to be even. What’s more, some applications of these arithmetic properties in other problems are contained in the paper as well.
Key Words and Phrases:
Pell equation, Catlan Conjecture
1 . Preliminaries.
We begin with some elementary properties about the solutions of Pell equation.
(1)x2Dy2 1 ( D0 and has no divisor of form n2)is called positive Pell equation(or standard Pell equation). It always has a minimal positive solution( ,x y1 1),and all solutions (x yn, n) are fixed by xnyn D(x1y1 D)n,
(or 2
n n
xn ,
2
n n
yn
D
,where x1 y1 D, x1 y1 D)
Furthermore, the recursion formula 1 1 1 1
1 1 1 1
( 2)
n n n
n n n
x x x Dy y y x y y x n
and m n m n m n( , *)
m n m n m n
x x x Dy y
m n N
y x y y x
hold.
If we define x0 1,y0 0,xn xn,yn yn,index in the recursion formulas above can be chosen from ℤ including negative integers.
(2)x2Dy2 1,is called negative Pell equation. If it has solution, it must have infinitely many solutions. Assume( , )a b1 1 is its minimal positive solution,then the general solutions (a bn, n) are fixed by anbn D(a1b1 D)2n1,(or
2 1 2 1
2
n n
an
,
2 1 2 1
2
n n
bn
D
,where a1 b1 D, a1 b1 D)
What’s more, (a1b1 D)2 x1 y1 D(As the symbols above,(x1,y1) is the minimal solution of the corresponding positive Pell equation).
(3)x2Dy2 K, K 1,is called quasi-Pell equation. If it has solution, it must have infinitely many solutions. In particular, when K 2,if we denote the minimal solution as ( , )u v1 1 , then the general solutions ( ,u vn n) are fixed by
2 1
1 1
1
( )
2
n
n n n
u v D
u v D
(or
2 1 2 1
2
n n
n n
u ,
2 1 2 1
2
n n
n n
v
D
,where
1 1
u v D
, u1 v1 D)
(4)No matter what kind of Pell equation it is, the recursion formula always holds.
Where
( x y
n,
n)
and (x y*n, *n) stand for the nth positive solution of2 2
x Dy K and x2Dy2 1.
2. Main Theorems
In this paper, we will obtain the following two main theorems
Theorem 1 (1) For standard Pell equation x2Dy2 1,assume d is a divisor of D, d is greater than 3, and ( , ) 1y d1 ,then dk yn dk n
(
nN*,kN
.
The symbol“ ”
means: ba a ba|a,ba1|a) (2) For(I)negative Pell equation x2Dy2 1,(II)quasi-Pell equation x2Dy2 2, 4, (III)quasi-Pell equation x2Dy2 4. Similar conclusion holds, that is:
In equation of type (I), if d D| and ( ,d y1) =1, d3 ,then
2 1
k k
d yn d n .
In equation of type (II),if d D| , ( ,d y1)=1,d3,and d is odd integer,
then dk yn dk 2n1.
In equation of type (III),if d D| , ( ,d y1)=1,d3,and d is odd integer,
then dk yn dk n.
(3) In particular, for standard Pell equation x2Dy2 1,we have:assume the minimal solution is ( ,x y1 1),d D| ,d3, and d y1,then if ( , y1) 1
d d ,we have
k k
d yn d n.
Remark: In (2), among the equations of type (I),the situation that 2 |d and 0
k will never occurat all,since this will yield 2 |yn 4 |Dyn2xn2 1(mod 4), which leads to a contradiction. So in this case d must be odd.
Theorem 2 (1) Denote the general solutions of standard Pell equation
2 2
1
x Dy as (x yn, n),then t N*among the following propositions either (i) or (ii) will hold; and either (iii) or (iv) will hold:
(i)n t x, | n.
(ii) f t( )N* which is uniquely determined by t,such that |
n ( ) t x n
f t is a positive odd integer.
(iii)n t, | yn.
(iv) g t( )N* which is uniquely determined by t,such that |t yn g t( ) |n.
(2) x2Dy2 2,denote its general solutions as (x yn, n),for odd integer t,among the following propositions either (i) or (ii) will hold; and either (iii) or (iv) will hold:
(i) n N t x*, | n.
(ii) f t( )N* which is uniquely determined by t , such that
| n ( ) | 2 1 t x f t n .
(iii) n N t*, | yn.
( iv) g t( )N* which is uniquely determined by t , such that
| n ( ) | 2 1 t y g t n .
(3)More generally, we can obtain:
For fixed tN*,we denote the following statement as proposition (I):either we cannot find an xn such that t x| n ,or f t( )N* such that t x| n
(2 1) ( )
n k f t
,kN*;either we cannot find an yn such that |t yn,or ( ) *
g t N
such that |t yn n kg t( ),kN*.
In parallel, we denote the following statement as proposition (II):either we cannot find an xn such that |t xn,or f t( )N* such that |t xn 2n 1 kf t( ),
kN*;either we cannot find an yn such that t y| n,or g t( )N* such that
| n
t y 2n 1 kg t( ),kN*. Then we have:
(i)For x2Dy2 1,proposition (I) holds.
(ii)For x2Dy2 1,proposition(II) holds.
(iii)For x2Dy2 2,proposition(II) holds if we require in which t is odd.
(iv)对x2Dy2 4,proposition(I)holds if we require in which t is odd.
This paper will be organized as follows:in section 3 we will prove theorem 1,
and several applications of theorem 1 will be given in section 4;in section 5 we will prove theorem 2,and several applications of theorem 2 will be given in section 6. In the applications in section 4 and 6,we will obtain parts of the results of the generalized Catlan’s Conjecture.
3. Proof of Theorem 1
Obviously, theorem 1.(1) is the corollary of 1.(3),so we prove 1.(3) first.
The proof of 1.(3) is divided into three parts:
In standard Pell equation x2Dy2 1,assume d is divisor of D, d is greater than 3, and d y1,( , y1) 1
d d (N ).
(i)d 1|yn d n|
since 1
1 1 3 3 3
1 1 1 1
1 1 1
1 1
1 1
( ) ( )
...
2
(mod )
n n
n n
n n n
n
x D y x D y
y C x y C x y D
D nx y d
and by the original equation we have ( , ) 1x d1 ,then by ( ,d y1)d the conclusion follows.
(ii)d 1 yd If not,assume
1
2 1 3 3 3
1 1 1
| d d d d ...
d y dx y C x y D Then
1
2 1 3 3
1 1 1
( 1)( 2)
| ...
6
d d d d d
d dx y x y D
1
1 1 3 3
1 1 1
( 1)( 2)
| ...
6
d D d d d
d x y x y
If 3 |d,then 3 | (d1)(d2) ,2 | (d1)(d2) so 6 | (d1)(d2)
then
1
1 3 3 1 1
1 1 1
( 1)( 2)
| |
6
d d
d d
d D x y d x y ,which leads to a contradiction.
If 3 |d,then assume d 3d1,then d11
so 1
1 1 3 3 1 1
1 1 1 1 1 1 1
( 1)( 2)
| ... |
3 2
d D d d d d
d x y x y d x y however (x y d1 1, )d and d11,which leads to a contradiction. So (ii) holds.
(iii)Now we prove the theorem directly
By the preliminary proposition (4),
2 3 2
2 2 2 2 3
2
2 2 (mod )
(mod )
d d d d d
d d d d
y x y x y d
x x Dy x d
. Generally,
if we have proven
1 2 3
2 3
(mod )
(mod )
k
kd d d
k
kd d
y kx y d
x x d
then
(k 1)d kd d kd d
y y x x y kxdk1yd xd x ydk d (k1)x ydk d (modd23)
(k 1)d kd d kd d
x x x Dy y xdkxd D kxdk1ydyd xdk1 (modd23) so
2 3
1 2 3
(mod )
(mod )
n
nd d
n
nd d d
x x d
y nx y d
so whe d |n we have d 2 | ynd,otherwise d 2|nxdn1yd d nx| dn1.
by ( ,d xn) 1 d n| ,this is a contradiction. This deduction also proves when d n| ,
2| nd.
d y 2
2
d yd .
so d 2|yn d2|n, 2 2 d yd .
Generally, assume that we have proven d s |yn ds |n,and s
s
d
y
d , sN.then
2 2 2 2 2 1
2
2 2 1
2
(mod )
2 2 (mod )
s s s s
s s s s s
s
d d d d
s
d d d d d
x x Dy x d
y x y x y d
.If we have proven
2 2 1
1 2 2 1
(mod )
(mod )
s s
s s s
k s
k d d
k s
k d d d
x x d
y kx y d
.then
2
( 1) s s s s s s s s
k s k
k d kd d kd d d d d d
x
x x D y y x x Dx y
1 2 2 1
(mod )
s
k s
x
dd
.(k 1)ds kds ds kds ds
y
x y y x
ks s ks1 sd d d d
x y kx
x
2 2 2
( 1)
s s(mod )
k s
d d
k x y d
so
2 2 2
1 2 2 2
(mod )
, (mod )
s s
s s s
n s
nd d
n s
nd d d
x x d
n y nx y d
.So naturally we have d s 1|yn ss1|n , 1
1
s
s
d
y
d .
By induction we have k N*, d k yn dk n.
Proof of 1.(2):
Firstly,the minimal solution of x2Dy2 h is ( ,x y1 1),then the general solution (x yn, n) is given by ( 1 1 )
an
n n
n
x y D
x y D
b
.
Where when h 1,an 2n1,bn 1; when h4 ,an n,bn 2n1;
when h 4,an 2n1,bn 22n2; when h 2,an 2n1,bn 2n1.
The proof will still be divided into three parts:
(i)d y| n d a| n
since
1
1 3
1 3 3
1 1 1
1 1 1 1
1
1 1
( ) ( ) ...
2 (mod )
n n
n n
n n
n
a a
a a
a a
n
n n a
n
C x y C x y D
x D y x D y
y b D b
a x y d
By the original equation we have ( , ) 1x d1 ,then by ( ,d y1) 1 the conclusion holds.
(ii)
cd
d y
Where when h 1, 4, 2, 1
d 2 c d
,when h4,cd d. Then we have
cd
a d.
If not,assume 1
1 3 3 3
1 1 1
2 ...
|
cd cd
d cd
d
d
a a
c a
c
c
a x y C x y D d y
b
Then
1
2 1 3 3
1 1 1
( 1)( 2)
| ...
6
d d d d d
d dx y x y D
1
1 3 3
1 1 1
( 1)( 2)
| ...
6
d D d d d
d x y x y
If 3 |d,then 3 | (d1)(d2) ,2 | (d1)(d2), so 6 | (d1)(d2)
and | ( 1)( 2) | 1 1 1 6
d d d
d D d x y
,which is a contradiction.
If d2|nudn1vd d nu| dn1.3 |d,then assume d 3d1,d11 so
1
1 3 3 1
1 1 1 1 1 1 1
( 1)( 2)
| ... |
3 2
d D d d d d
d x y x y d x y
but (x y d1 1, ) 1 and d1 1,which is a contradiction. So (ii)holds.
(iii)Now we prove the theorem directly
Let un Dvn(x1 D y1)n,then actually we have n an n
x u
b
, n ann
y v
b un,
vnsatisfy m n m n m n
m n m n n m
u u u Dv v v u v u v
Thus what we have got in the previous two steps is exactly d v| n d n| ,d vd.
Then we have
3 2
2 2 2 3
2
2 2 (mod )
(mod )
d d d d d
d d d d
v u v u v d
u u Dv u d
Generally, assume we have proven
1 3
3
(mod ) (mod )
k
kd d d
k
kd d
v ku v d
u u d
Then v(k1)d v ukd d u vkd d kudk1vdud u vdk d (k1)u vdk d (modd3) u(k1)d u ukd d Dv vkd d udkud D kudk1vdvd udk1(modd3) So we have
3
1 3
(mod ) (mod )
n
nd d
n
nd d d
u u d
v nu v d
So when d |n we have d2 |vnd,otherwise d2|nudn1vd d nu| dn1.
By ( ,d un) 1 d n| ,which is a contradiction. This also proves when d n| we have d2|vnd. 2
2
d vd
So d2|vn d2|n, 2 2. d vd
Generally, assume we have proven ds|vn ds|n,and s
s
d vd , then
2 2 2 2
2
2 2
(mod )
2 2 (mod ).
s s s s
s s s s s
s
d d d d
s
d d d d d
u u Dv u d
v u v u v d
If we have proven
2
1 2
(mod ) (mod ).
s s
s s s
k s
k d d
k s
k d d d
u u d
v ku v d
Then
2
( 1) s s s s s s s s
k s k
k d kd d kd d d d d d
u
u u D v v u u Du v
1 2
(mod ).
s
k s
u
dd
(k 1)ds kds ds kds ds
v
u v v u
s s s1 sk k
d d d d
u v ku
u
( 1)
s s(mod
2)
k s
d d
k u v d
so
2
1 2
(mod )
, (mod ).
s s
s s s
n s
nd d
n s
nd d d
u u d
n v nu v d
so naturally we have
d
s1| v
n d
s1| n
, 11
s
s
d
v
d Then by induction we have k N*, dk vn dk n. That is exactly dk yn dk an
4. Applications of Theorem 1
Example 1 x2242n11
x
,n
N.This equation has an obvious solution x5,n0. Assume ( , )x n is a non-negative solution of x2242n11 such that n0,then in the Pell equation
2 2
24 1
x y ,assume ym 24 ,n n0,then 24n ym 24n m. So ym y24n. Notice that
y
24n C
124n 5 C
243n 5 24 ...
3
5 24n 24n ym This is a contradiction.
So this equation has no non-negative integral solution other than ( , )x n (5, 0). This example can be directly generalized into the following result:
2 2 2 1
( 1) n 1
x a (a2 is fixed,
x
,n
are variables)has no non-negative solutions.A more general question is: Besides n0, is there any non-negative integral solution of x2y2n11? We can only obtain parts of answer to this question here.
Example 2 x2y2n1 1(n0)when 2n 1 p is prime,we must have 2 | y,
| p x.
Proof:First we prove2 |y. Otherwise 2 | y,then 2 |x. So x1,x1 are both odd.
By (x1)(x 1) y2n1,and (x1)(x 1) y2n1 So
2 1
2 1 2 1
2 1
1 2
1
n
n n
n
x u
u v
x v
. So uv|u2n1v2n1 2.
Since u,v are either both odd or both even, we must have u v 2. So
2 1 2 1 2 1
(v2) n v n 2 n 2. This is a contradiction.
So 2 |y.
By the original equation, we have x2 (y1)(yp1yp2 ... 1).
1 2
... 1 1 ( 1) ... 1 (mod 1)
p p
y y p y .
If p y| 1 then p x| 2 p x| , and the conclusion follows.
Next we assume p| y1 so (y1,yp1yp2 ... 1) 1.
So
2
1 2 2
1 ... 1
p p
y a
y y b
since y2 and 2 | y so y8,a2. Now the original equation turns into x2(a21)2n1 1 (a2)
By the generalization of example 1 we know this equation has no non-negative integral solutions such that n0.
In fact x2y2n1 1 is a special case of Catalan’s Conjecture:the Diophantine equation xa yb 1 has only one solutionx3,y2,a2,b3 if we require the positive integers x,y,a,b are greater than 1. This conjecture has been solved by Preda Mihăilescu in 2002.
Now we deal with the generalized Catlan’s conjecture: the Diophantine euqation
2 1
m n
x y .
When m=n=2, it is just the standard Pell equationx22y2 1,we have known it has infinitely many solutions with minimal solution (3,2),general solutions can be given by xn 2yn (3 2 2)n.
When m=p or 2p,p is odd prime,the following several examples show some necessary conditions of the solutions of this equation.
Example 3 xp2y2 1(x3) (p is odd prime), its any positive integral solution must satisfy p y| .
Proof:(xp1)(xp1 ... 1) 2y2 xp1 ... 1 p(mod(x1))
Ifp x| 1,then obviously p y| .
If p x| 1,then (x1,xp1 ... 1) 1. Since x is odd,xp1 ... 1 is odd, too.
so xp1 ... 1 a2,x 1 2b2.
The equation turns to (2b21)p2y2 1 ,so
1
2 2
(2 , (2 y b 1) )
p is solution of quasi-Pell equation X22(2b21)Y2 2.The minimal solution of this equation is (X Y1, )1 (2 ,1)b ,assume
1
2 2
(2 , (2 1) ) ( , )
p
n n
y b X Y
,Then by theorem 1.(2) we have
1
2 2
(2 1)
p
b n
, thus
1
2 2
(2 1)
p
n b
Now 2
2
2 11
2 2(2 1)
2(2 1)
2
n
n n n
b b
X b Y
,
So 21
2 2
2 1
(2 1)(2 )
2 1 (2 1)
2
n p
n n n
n b
Y n n b Y
,this is a contradiction.
So p x| 1,p y| .
Completely in parallel, we can prove any positive integral solution of
2 2 1
xp y - (x3) (p is odd prime) must satisfyp y| .
Example 4 x2p 2y2 1 (x3) (pis odd prime), its any positive integral solution must satisfy p y| .
Proof:x2py4 (y21)2,and x is oddx2p 1(mod8),so y is even.
By results from Pythagoras equation,a b, with one odd and the other even,ab, such that xp a2b2,y2 2ab,y21a2b2.
so a2b2 1 2ab a b 1.
so xp 2b1,y2 2 (b b1).
When b is odd,(2(b1), )b 1,so 2(b 1) u2,bv2,yuv, 2 2 1
xp v ,by Example 3,p v| so p y| .
When b is even,(2 , (b b1))1,so 2bu2,(b 1) v2,yuv,
xp2v2 1,by the remark after Example 3,p v| so p y| .
Example 5 The Diophantine equation 32n12y2 1 has no positive integral solution when 2n1 is a composite number.
Lemma:a,m,nN*,则(am1,an 1) a( , )m n 1. Proof of lemma:For m n , by Euclidean algorithm we have,
1 1
mq n r nq r2 1r2 …
rk qk2rk1rk2 …
rs qs2rs1 thus rs1 ( , )m n .
Then we have (am1,an1)
((am 1) (an 1),an1) (ama an, n 1)
(a an( m n 1),an 1) (am n 1,an1)
…
( a
m q n 1 1, a
n 1) ( a
n 1, a
r1 1)
…
( a
r1 1, a
r2 1)
…(ars 1,ars1 1) a
rs1 1
a( , )m n 1. The lemma is proven.Now return to Example 5. Denote prime divisors of 2n1 as p1,p2,…,pk. since 2n1 is a composite number,
So for 1 i k,
2 1
pi n . Now we have
2 1
3 2 2 1
i i
n p
p y
, since
2 1
3
i3
n p
,by conclusion ofExample 3 we know for 1 i k,p yi| . So 32n1 1(modpi), 1 i k.
Without losing generality, we can assume p1 is the minimal prime divisor of 2n1,then p1 is an odd integer that is greater than 1,and we have
2 1
1| 3 n 1
p , p1| 3p111 By the lemma,p1| 3(2n1,p11)1.
Notice that all the prime divisors of p11 is smaller than p1,by our assumption ofp1, 2n1 cannot divided by them. Sop11 is relative prime to 2n1.
Thus (2n1,p1 1) 1, p1| 3 1 2 , this is a contradiction.
Thus we have proven when 2n1 is a composite number, 32n12y2 1 has no positive integral solutions.