World Academy of Science, Engineering and Technology

International Journal of Physical and Mathematical Sciences

Vol:4, No:7, 2010

The Pell Equation x2 − P y 2 = Q

Ahmet Tekcan, Arzu Özkoç, Canan Kocapınar, Hatice Alkan

Abstract—Let p be a prime number such that p ≡ 1(mod 4), say equations of this type in the seventeenth century. Ironically,
p = 1 + 4k for a positive integer k. Let P = 2k + 1 and Q = k2 . Pell was not the first to work on this problem, nor did
In this paper, we consider the integer solutions of the Pell equation he contribute to our knowledge for solving it. Euler (1707-
x2 − P y 2 = Q over Z and also over finite fields Fp . Also we deduce
some relations on the integer solutions (xn , yn ) of it. 1783), who brought us the ψ-function, accidentally named the
equation after Pell, and the name stuck. For N = 1, the Pell
Keywords—Pell equation, solutions of Pell equation. equation
x2 − Dy 2 = ±1 (3)
I. P RELIMINARY FACTS ON D IOPHANTINE AND P ELL
E QUATIONS . is known as the classical Pell equation and was first studied by
International Science Index, Physical and Mathematical Sciences Vol:4, No:7, 2010 waset.org/Publication/7822

A Diophantine equation is an indeterminate polynomial Brahmagupta (598-670) and Bhaskara (1114-1185), (see [1]).
equation that allows the variables to be integers only. Dio- Its complete theory was worked out by Lagrange (1736-1813),
phantine problems have fewer equations than unknown vari- not Pell. It is often said that Euler (1707-1783) mistakenly
ables and involve finding integers that work correctly for all attributed Brouncker’s (1620-1684) work on this equation to
equations. In more technical language, they define an algebraic Pell. However the equation appears in a book by Rahn (1622-
curve, algebraic surface or more general object, and ask about 1676) which was certainly written with Pell’s help: some
the lattice points on it. The word Diophantine refers to the say entirely written by Pell. Perhaps Euler knew what he
Hellenistic mathematician of the 3rd century, Diophantus of was doing in naming the equation. Baltus [2], Kaplan and
Alexandria, who made a study of such equations and was Williams [5], Lenstra [6], Matthews [7], Mollin (also Poorten
one of the first mathematicians to introduce symbolism into and Williams) [8], Stevenhagen [10] considered some specific
algebra. The mathematical study of Diophantine problems Pell (and Diophantine) equations and their integer solutions
Diophantus initiated is now called Diophantine analysis. A lin- (Further details on Pell equations can be found in [2], [3],
ear Diophantine equation is an equation between two sums of [9]).
monomials of degree zero or one. While individual equations The Pell equation in (3) has infinitely many integer solutions
present a kind of puzzle and have been considered throughout (xn , yn ) for n ≥ 1. The first non-trivial positive
√ integer so-
history, the formulation of general theories of Diophantine lution (x1 , y1 ) (in this case x1 or x1 + y1 D is minimum)
equations was an achievement of the twentieth century. For of this equation is called the fundamental solution, because
example, the equation ax + by = 1 is known the linear all other solutions can be (easily) derived from it. In fact, if
Diophantine equation. In general, the Diophantine equation (x1 , y1 ) is the fundamental solution of x2 − Dy 2 = 1, then
is the equation given by the n-th positive solution of it, say (xn , yn ), is defined by the
equality
ax2 + bxy + cy 2 + dx + ey + f = 0. (1) √ √
xn + yn D = (x1 + y1 D)n (4)
Also for n = 2, there are infinitely many solutions (x, y, z)
of the Diophantine equation xn + y n = z n . For larger values for integer n ≥ 2. (Furthermore, all nontrivial solutions
of n, Fermat’s last theorem (see [4]) states that no positive can be obtained considering the four cases (±xn , ±yn ) for
integer solutions x, y, z satisfying the equation exist. In [18], n ≥ 1). There are several methods for finding the fundamental
[19], [21], we considered some specific Diophantine equations solution of Pell’s equation x2 − Dy 2 = 1 for a positive
and their integer solutions. non-square integer D, e.g., the cyclic method known in India
Let D = 1 be a positive non-square integer and N be any (12-th century), or the slightly less efficient but more regular
fixed positive integer. Then the Diophantine equation English method (17-th century) which produce all solutions of
x2 − Dy 2 = 1. But the most efficient method for finding the
x2 − Dy 2 = ±N (2) fundamental solution √ is based on the simple finite continued
fraction expansion of D. We can describe it as follows: Let
is known as Pell equation and is named after John Pell (1611-
1685), a mathematician who searched for integer solutions to [a0 ; a1 , a2 , · · · , ar , 2a0 ]
Ahmet Tekcan, Arzu Özkoç and Hatice Alkan are with the Uludag √ √
University, Department of Mathematics, Faculty of Science, Bursa-TURKEY, be the simple continued fraction of D, where a0 =  D.
emails: [email protected], [email protected], [email protected]. Let p0 = a0 , p1 = 1 + a0 a1 , q0 = 1 and q1 = a1 . In general
http://matematik.uludag.edu.tr/AhmetTekcan.htm.
Canan Kocapınar is with the Balikesir University, Department of Mathemat- pn = an pn−1 + pn−2 (5)
ics, Faculty of Arts and Science, Balıkesir-TURKEY, email: canankocap-
[email protected]. qn = an qn−1 + qn−2

International Scholarly and Scientific Research & Innovation 4(7) 2010 795 scholar.waset.org/1307-6892/7822
 World Academy of Science, Engineering and Technology
International Journal of Physical and Mathematical Sciences
Vol:4, No:7, 2010

for n ≥ 2. Then the fundamental solution of x2 − Dy 2 = 1 is 6) The n−th solution (xn , yn ) can be given by
⎧
⎨ (pr , qr ) if r is odd xn  
= 1; (1, 2)n−2 , 1, 3 (10)
(x1 , y1 ) = yn
⎩
(p2r+1 , q2r+1 ) if r is even.
for n ≥ 2, where (1, 2)n−2 means that there are n − 2
On the other hand, in connection with (2) and (3), it is well successive terms “1, 2”.
known that if (X1 , Y1 ) and (xn−1 , yn−1 ) are integer solutions √
of x2 − Dy 2 = ±N and x2 − dy 2 = 1, respectively, then Proof: 1) It is easily seen that 3 = [1; 1, 2].
(Xn , Yn ) is also a positive solution of x2 − Dy 2 = ±N , 2) The fundamental (minimal) solution of x2 − 3y 2 = 1 is
where (x1 , y1 ) = (2, 1) since 22 − 3.11 = 1.
3) We prove the theorem by induction on n. Let n = 1.
Xn + DYn = (xn−1 + Dyn−1 )(X1 + DY1 ) (6) Then
      
for n ≥ 2. x1 2 3 1 2
= =
y1 1 2 0 1
II. T HE P ELL E QUATION x2 − P y 2 = Q.
which is the fundamental solution. Let us assume that this
In [11], [12], [13], [14], [15], [16] and [17], we considered
equation is satisfied for n − 1, that is, x2n−1 − 3yn−1
2
= 1. We
some specific Pell equations and their integer solutions. Also
International Science Index, Physical and Mathematical Sciences Vol:4, No:7, 2010 waset.org/Publication/7822

will show that it is also satisfied for n. Clearly we deduce that

we deduced some recurrence relations on the integer solutions
   n  
(xn , yn ) of these Pell equations. xn 2 3 1
=
In the present paper, we consider the very specific Pell yn 1 2 0
equation and its integer solutions. P and Q be two non-zero   n−1  
integers and let D = P 2 − 4Q be the discriminant such that 2 3 2 3 1
=
D = 0. In [22], we defined a new sequence A = An (P, Q) 1 2 1 2 0
  
with parameters P and Q defined by A0 = 0, A1 = 1 and 2 3 xn−1
=
An = P An−1 −QAn−2 for n ≥ 2 and derived some algebraic 1 2 yn−1
 
identities on it. Also we showed that every prime number 2xn−1 + 3yn−1
p ≡ 1(mod 4) can be written of the form P 2 − 4Q. Indeed, = . (11)
xn−1 + 2yn−1
let p = 1 + 4k. Then the quadratic equation P 2 − 4D = p has
a solution for Hence

P = 2k + 1 and Q = k 2 . (7) x2n − 3yn2 = (2xn−1 + 3yn−1 )2 − 3(xn−1 + 2yn−1 )2

In this work, we will consider the Pell equation = 4x2n−1 + 12xn−1 yn−1 + 9yn−1
2

−3x2n−1 − 12xn−1 yn−1 − 12yn−1

2
x2 − P y 2 = Q (8)
= x2n−1 − 3yn−1
2

over Z and also over finite fields Fp , where P and Q is defined = 1.

in (7). Note that for p = 5, we have k = 1 and hence P = 3
and Q = 1. So (8) becomes x2 −3y 2 = 1 which is the classical So (xn , yn ) is also a solution of x2n − 3yn2 = 1.
Pell equation. For the other values of p > 5, the Pell equation 4) It is clear from (11) that xn = 2xn−1 + 3yn−1 and
x2 − P y 2 = Q is not a classical Pell equation. So we have to yn = xn−1 + 2yn−1 for n ≥ 2.
consider these two conditions. 5) We only prove by induction that xn = 3(xn−1 +xn−2 )−
xn−3 . Let n = 4. Then from (9) we get x1 = 2, x2 = 7, x3 =
Theorem 2.1: Let p = 5, then for the classical Pell equation 26 and x4 = 97. So
x2 − 3y 2 = 1, we have
√
1) The continued fraction expansion of 3 is [1; 1, 2]. x4 = 3(x3 + x2 ) − x1 = 3(26 + 7) − 2 = 97,
2) The fundamental solution is (x1 , y1 ) = (2, 1).
3) Define the sequence {(xn , yn )}, where that is, xn = 3(xn−1 + xn−2 ) − xn−3 is true for n = 4. Let
   n   us assume that this relation is satisfied for n − 1, that is,
xn 2 3 1
= (9)
yn 1 2 0 xn−1 = 3(xn−2 + xn−3 ) − xn−4 . (12)
for n ≥ 1. Then (xn , xn ) is a solution of x2 − 3y 2 = 1. Then from (11) and (12), we obtain xn = 3(xn−1 + xn−2 ) −
4) The solutions (xn , yn ) satisfy xn = 2xn−1 + 3yn−1 and xn−3 for n ≥ 4.
yn = xn−1 + 2yn−1 for n ≥ 2. 6) Similarly it can be shown by induction on n that the
5) The solutions (xn , yn ) satisfy the recurrence relations n−th solution (xn , yn ) can be given by
xn = 3(xn−1 + xn−2 ) − xn−3 xn  
= 1; (1, 2)n−2 , 1, 3 (13)
yn = 3(yn−1 + yn−2 ) − yn−3 yn
for n ≥ 4. for n ≥ 2.

International Scholarly and Scientific Research & Innovation 4(7) 2010 796 scholar.waset.org/1307-6892/7822
 World Academy of Science, Engineering and Technology
International Journal of Physical and Mathematical Sciences
Vol:4, No:7, 2010

Example 2.1: Some integer solutions of x2 − 3y 2 = 1 are and hence P = 2k + 1 = t2 + 1. It is easily seen that
       1
x1 2 3 1 2 t2 + 1 = t + ( t2 + 1 − t) = t +
= = √
y1 1 2 0 1 ( t2 +1−t)
   2     1 1
x2 2 3 1 7 = t+ √ =t+ √ .
= =
y2 1 2 0 4 t2 +1+t
1
2t + ( t2 + 1 − t)
   3     √
x3 2 3 1 26 So √ p = 2t2 +3,
P = [t; 2t]. Similarly it can be shown that if √
= =
y3 1 2 0 15 2
then P = [t; t, 2t] and if p = 2t − 5, then P = [t −
   4    
x4 2 3 1 97 1; 1, t − 2, 1, 2t − 2].
= = 3) We prove it by induction on n. Let n = 1. Then
y4 1 2 0 56
   5           
x5 2 3 1 362 x1 k + 1 2k + 1 1 k+1
= = = =
y5 1 2 0 209 y1 1 k+1 0 1
   6    
x6 2 3 1 1351 which is the fundamental solution. Let us assume that the
= = . equation x2 − P y 2 = Qn is satisfied for n − 1, that is,
y6 1 2 0 780
x2n−1 − P yn−1
2
= Qn−1 . For n, we obtain
Also    n  
International Science Index, Physical and Mathematical Sciences Vol:4, No:7, 2010 waset.org/Publication/7822

xn k + 1 2k + 1 1
7 =
= [1; 1, 3] yn 1 k+1 0
4   
26 k + 1 2k + 1 xn−1
= [1; 1, 2, 1, 3] =
15 1 k+1 yn−1
 
97 (k + 1)xn−1 + (2k + 1)yn−1
= [1; 1, 2, 1, 2, 1, 3] = . (15)
56 xn−1 + (k + 1)yn−1
362
= [1; 1, 2, 1, 2, 1, 2, 1, 3] So
209
1351 x2n − P yn2 = [(k + 1)xn−1 + (2k + 1)yn−1 ]2
= [1; 1, 2, 1, 2, 1, 2, 1, 2, 1, 3].
780
−(2k + 1)[xn−1 + (k + 1)yn−1 ]2
2 2
Now we consider the the Pell equation x − P y = Q for = (k + 1)2 x2n−1 + 2(k + 1)(2k + 1)xn−1 yn−1
p > 5.
+(2k + 1)2 yn−1
2

Theorem 2.2: Let p > 5, then for the Pell equation x2 − −(2k + 1)x2n−1 − 2(2k + 1)(k + 1)xn−1 yn−1
P y 2 = Q, we get −(2k + 1)(k + 1)2 yn−1
2

1) The fundamental solution is (x1 , y1 ) √ = (k + 1, 1). = x2n−1 [(k + 1)2 − (2k + 1)]
2) The continued fraction expansion of P is 2
+yn−1 [(2k + 1)2 − (2k + 1)(k + 1)2 ]
⎧
√ ⎨ [t; 2t] if p = 2t2 + 1 = k 2 [x2n−1 − (2k + 1)yn−1
2
]
P = [t; t, 2t] if p = 2t2 + 3 = k 2 (Qn−1 )
⎩
[t − 1; 1, t − 2, 1, 2t − 2] if p = 2t2 − 5.
= k 2 (k 2 )n−1
3) Define the sequence {(xn , yn )}, where = k 2n
   n   = Qn .
xn k + 1 2k + 1 1
= (14)
yn 1 k+1 0 Therefore x2n − P yn2 = Qn as we claimed.
for n ≥ 1. Then x2n − P yn2 = Qn for n ≥ 1.
Example 2.2: 1) Let t = 6. Then p = 2t2 + 1 = 73 is a
4) The solutions (xn , yn ) satisfy 2 2
prime and k = 18. √ So we have the Pell equation x − 37y =
xn = (k + 1)xn−1 + (2k + 1)yn−1 324. Note that 37 = [6, 12]. The fundamental solution is
(x1 , y1 ) = (19, 1) and for
yn = xn−1 + (k + 1)yn−1
   2    
x2 19 37 1 398
for n ≥ 2. = =
y2 1 19 0 38
Proof: Recall that P = 2k + 1 and Q = k 2 . So we have    3    
x2 − (2k + 1)y 2 = k 2 . x3 19 37 1 8968
= =
1) The fundamental solution is (x1 , y1 ) = (k + 1, 1) since y3 1 19 0 1120
   4    
(k + 1)2 − (2k + 1).12 = k 2 + 2k + 1 − 2k − 1 = k 2 . x4 19 37 1 211832
= =
2) Let p = 2t2 + 1 for some positive integer t. Then we get y4 1 19 0 30243
   5    
t2 x5 19 37 1 5143984
4k + 1 = 2t2 + 1 ⇔ k = = =
2 y5 1 19 0 786544

International Scholarly and Scientific Research & Innovation 4(7) 2010 797 scholar.waset.org/1307-6892/7822
 World Academy of Science, Engineering and Technology
International Journal of Physical and Mathematical Sciences
Vol:4, No:7, 2010

we have x2n − 37yn2 = 324n , and etc. has also two solutions (k, 0) and (p − k, 0). So x2 − P y 2 = Q
has p − 1 + 2 = p + 1 integer solutions.
2) Let t = 23. Then p = 2t2 + 3 = 1061 is a prime and Similarly it can be shown that if p ≡ 1(mod 8), then x2 −
k = 265. So 2 2
√ we have the Pell equation x − 531y = 70225. P y 2 = Q has p − 1 integer solutions.
Note that 531 = [23; 23, 46]. The fundamental solution is
(x1 , y1 ) = (265, 1) and for R EFERENCES
   
x2 71287 [1] Arya S.P. On the Brahmagupta-Bhaskara Equation. Math. Ed. 8(1)
= (1991), 23–27.
y2 532
    [2] Baltus C. Continued Fractions and the Pell Equations:The work of Euler
x3 19244834 and Lagrange. Comm. Anal. Theory Contin. Fractions 3(1994), 4–31.
= [3] Barbeau E. Pell’s Equation. Springer Verlag, 2003.
y3 212799
    [4] Edwards, H.M. Fermat’s Last Theorem. A Genetic Introduction to Alge-
x4 5232122113 braic Number Theory. Corrected reprint of the 1977 original. Graduate
= Texts in Mathematics, 50. Springer-Verlag, New York, 1996.
y4 75849368 [5] Kaplan P. and Williams K.S. Pell’s Equations x2 −my 2 = −1, −4 and
   
x5 1432020496466 Continued Fractions. Journal of Number Theory. 23(1986), 169–182.
= [6] Lenstra H.W. Solving The Pell Equation. Notices of the AMS. 49(2)
y5 25408054001 (2002), 182–192.
[7] Matthews, K. The Diophantine Equation x2 − Dy 2 = N, D > 0.
we have x2n − 531yn2 = 70225n , and etc. Expositiones Math.18 (2000), 323–331.
International Science Index, Physical and Mathematical Sciences Vol:4, No:7, 2010 waset.org/Publication/7822

[8] Mollin R.A., Poorten A.J. and Williams H.C. Halfway to a Solution of
x2 − Dy 2 = −3. Journal de Theorie des Nombres Bordeaux, 6(1994),
3) Let t = 9. Then p = 2t2 − 5 = 157 is a prime and 421–457.
2 2
k = 39. So √ we have the Pell equation x − 79y = 1521. [9] Niven I., Zuckerman H.S. and Montgomery H.L. An Introduction to the
Note that 79 = [8, 1, 7, 1, 16]. The fundamental solution is Theory of Numbers. Fifth Edition, John Wiley&Sons, Inc., New York,
1991.
(x1 , y1 ) = (40, 1) and for [10] Stevenhagen P. A Density Conjecture for the Negative Pell Equation.
   2     Computational Algebra and Number Theory, Math. Appl. 325(1992),
x2 40 79 1 1679 187–200.
= = [11] Tekcan A. Pell Equation x2 − Dy 2 = 2, II. Bulletin of the Irish
y2 1 40 0 80
   3     Mathematical Society 54 (2004), 73–89.
x3 40 79 1 73480 [12] Tekcan A., Bizim O. and Bayraktar M. Solving
√ the Pell Equation Using
= = the Fundamental Element of the Field Q( Δ). South East Asian Bull.
y3 1 40 0 4879
   4     of Maths. 30(2006), 355–366.
x4 40 79 1 3324641 [13] Tekcan A. The Pell Equation x2 − Dy 2 = ±4. Applied Mathematical
= = Sciences, 1(8)(2007), 363–369.
y4 1 40 0 268640 [14] Tekcan A., Gezer, B. and Bizim, O. On the Integer Solutions of the
   5     Pell Equation x2 − dy 2 = 2t . Int. Journal of Computational and
x5 40 79 1 154208200
= = Mathematical Sciences 1(3)(2007), 204–208.
y5 1 40 0 14070241 [15] Tekcan A. On the Pell Equation x2 − (k2 − 2)y 2 = 2t . Crux
Mathematicorum with Mathematical Mayhem 33(6)(2007), 361–365.
we have x2n − 79yn2 = 1521n , and etc. [16] Tekcan A. The Pell Equation x2 − (k2 − k)y 2 = 2t . International Jour.
of Comp. and Mathematical Sci. 2(1)(2008), 5–9.
[17] Tekcan A. and Bizim O. The Pell Equation x2 + xy − ky 2 = ±1.
III. T HE P ELL E QUATION x2 − P y 2 = Q OVER Fp Global Journal of Pure and Applied Mathematics 4(2)(2008),
[18] Tekcan A., Özkoç A. and Alkan H. The Diophantine Equation y 2 −
In [20], we considered the Pell Equations x2 − ky 2 = N 2yx − 3 = 0 and Corresponding Curves over Fp . International Jour.
and x2 + xy − ky 2 = N over finite fields Fp . In this section, of Math.and Statis. Sci. 1 (2)(2009), 66–69.
[19] Tekcan A. and Özkoç A. The Diophantine Equation x2 − (t2 + t)y 2 −
we will consider the integer solutions of x2 − P y 2 = Q over (4t + 2)x + (4t2 + 4t)y = 0. Revista Matemática Complutense
finite fields Fp . Let 23(1)(2010), 251–260.
[20] Tekcan A. The Number of Solutions of Pell Equations x2 − ky 2 = N
Dp = {(x, y) ∈ Fp × Fp : x2 − P y 2 ≡ Q(mod p)}. and x2 + xy − ky 2 = N over Fp . Accepted for publication to Ars
Combinatoria.
Then we can give the following theorem. [21] Özkoç A. and Tekcan A. Quadratic Diophantine Equation x2 − (t2 −
t)y 2 − (4t − 2)x + (4t2 − 4t)y = 0. Bull. of the Malaysian Math. Sci.
Soc. 33(2)(2010), 273–280.
Theorem 3.1: For the Pell equation x2 −P y 2 = Q, we have [22] Özkoç A., Kocapınar C. and Tekcan A. Some Algebraic Identities on

the Sequence A = An (P, Q) with Parameters P and Q. Submitted for
p + 1 if p ≡ 5(mod 8) publication.
#Dp =
p − 1 if p ≡ 1(mod 8).
Proof: Let p ≡ 5(mod 8). If y = 0, the the quadratic
equation x2 ≡ k 2 (mod p) has two solutions x = k and x =
p − k. If x = 0, then the quadratic equation −(2k + 1)y 2 ≡
k 2 (mod p) has no solution y. Now let Sp = Fp − {k, p − k}.
2
Then there are p−12 elements x in Sp such that x P−Q is a
2
square. Let x P−Q = u2 for some u = 0. Then we get y 2 ≡
u2 (mod p) and hence y = u and y = −u, that is, there
are two integer solutions (x, u) and (x, p − u). So there are
2( p−1
2 ) = p − 1 solutions. We see as above that this equation

International Scholarly and Scientific Research & Innovation 4(7) 2010 798 scholar.waset.org/1307-6892/7822