PROBLEM SET 5

MATH 115
NUMBER THEORY
PROFESSOR PAUL VOJTA

NOAH RUDERMAN

Problems 3.2.7, 3.2.8, 3.2.18, 3.3.4, 3.3.15, 3.4.1a-f, 3.3.4, 3.3.10, 3.5.1, 3.5.3, 3.5.11, and
3.5.12 from An Introduction to The Theory of Numbers, 5th edition, by Ivan Niven, Herbert
S. Zuckerman, and Hugh L. Montgomery

Date: July 28, 2014.

1
Problem (3.2.7)

Solution.

We aim to find for which primes p the congruence x2 ≡ 13 mod p has a solution.
First we note that p = 2 and p = 13 are trivial solutions because both congruence classes
of 2 are quadratic residues and 02 ≡
 0 mod 13. Next we note that the congruence will have
a solution if the Legendre symbol 13 p
is equal to 1, by definition. By theorem 3.4 we see
that
  
13 p 13−1 p−1
= (−1) 2 · 2
p 13
p−1
= (−1)6· 2

p−1
= (−1)6 2

p−1
= (1) 2

= 1.
 
13 p


Clearly, p
= 1 if and only if 13
= 1. By theorem 3.1(1), we see that
p 13−1
≡p 2 mod 13
13
≡ p6 mod 13.
Therefore, 13 is a quadratic residue modulo p if and only if p6 ≡ 1 mod 13. With Mathemat-
ica (see figure 1), we see that p6 ≡ 1 mod 13 if p ≡ 1, 3, 4, 9, 12 mod 13. So the solutions
are p = 2, p = 13, or p ≡ 1, 3, 4, 9, 12 mod 13.
In[969]:= For @i = 0, i < 13, i ++,
Module @8f, x <,
If@Mod @i ^ 6, 13D Š 1, Print@iDD;
D
D
1

10

12

Figure 1. Mathematica code prints the solutions to x6 ≡ 1 mod 13

2
Problem (3.2.8)

Solution.
 
We want to find all primes p such that 10
p
= 1.
From theorem 3.1(2) we have
    
10 2 5
= .
p p p
p2 −1
 
2
From theorem 3.3, we know that if p is an odd prime, then p
= (−1) 8 . If p is odd,
p2 −1 2
then p ≡ ±1, ±3 mod 8. if p ≡ ±1 mod 8, then 8 is even. If p ≡ ±3 mod 8, then p 8−1
is odd. From this, we see that

  
2  1 if p ≡ ±1 mod 8
= 0 if p ≡ 0 mod 2
p 
−1 if p ≡ ±3 mod 8
 
From theorem 3.4, quadratic reciprocity, we know that p5 = p5 given 5 ≡ 1 mod 4.

By theorem 3.1(1),
p 5−1
≡p 2 mod 5
5
≡ p2 mod 5.
By lemma 2.10, p2 ≡ 1 mod 5 only has solutions for p ≡ ±1 mod 5. Thus,

p   1 if p ≡ ±1 mod 5
= 0 if p ≡ 0 mod 5
5 
−1 if p ≡ ±2 mod 5
 
We see that 10 p
= 1 when p ≡ ±1 mod 5 and p ≡ ±1 mod 8 or p ≡ ±2 mod 5 and
p ≡ ±3 mod 8. By the Chinese remainder theorem, there will be 2 · 2 + 2 · 2 = 8 solutions.
To find the solutions, first we solve the set of congruences
p ≡ ±1 mod 5
p ≡ ±1 mod 8.
Clearly, if p ≡ 1 mod 5 and p ≡ 1 mod 8, then p ≡ 1 mod 40 is a solution. A similar
argument can be used to show p ≡ −1 mod 40 is a solution. To solve the set of congruences
p ≡ −1 mod 5
p ≡ 1 mod 8,
we start with the second congruence whose solution is 1 + 8k for k ∈ Z. Plugging this into
the first congruence, we see that
1 + 8k ≡ −1 mod 5
3
 8k ≡ −2 mod 5
3k ≡ 3 mod 5
k ≡ 1 mod 5,
so k = 1 + 5l for l ∈ Z. Now we have our solution
1 + 8k = 1 + 8(1 + 5l)
= 9 + 40l.
For a solution p, we see that p ≡ 9 mod 40.
Now we solve the final congruence
p ≡ 1 mod 5
p ≡ −1 mod 8,
We start with the solution to the second congruence, −1 + 8k, k ∈ Z. Plugging this into the
first congruence, we see that
−1 + 8k ≡ 1 mod 5
8k ≡ 2 mod 5
3k ≡ −3 mod 5
k ≡ −1 mod 5,
so k = −1 + 5l for l ∈ Z. Now we have our solution
−1 + 8k = −1 + 8(−1 + 5l)
= −9 + 40l
= 31 + 40l0 l0 ∈ Z
For a solution p, we see that p ≡ 31 mod 40.
At this point, our first four solutions are p ≡ ±1, ±9 mod 40. Finding the remaining four
involves solving the set of congruences
p ≡ ±2 mod 5
p ≡ ±3 mod 8.
Finding the solutions to these congrunces is equally tedious. Since finding the solutions to
the chinese remainder theorem is not the point of this exercise, I will list the remaining four
solutions: p ≡±3, ±13 mod 40.
10
Therefore, p = 1 when p ≡ ±1, ±3, ±9, ±13 mod 40.

4
Problem (3.2.18)

Solution.

For q = 1111111111111, where q is prime, we wish to find whether or not 1001 is a quadratic
residue mod q.
First we note that 111 · 1001 = 111111. Using this, we see that
q = 111 · 1001 · 107 + 111 · 1001 + 1

= (111 · 107 + 111)1001 + 1.

Since 0 ≤ 1 < 1001, by the division algorithm, 1 is the remainder when q is divided by 1001.
Thus, q ≡ 1 mod 1001. Since 1001 = 7 · 11 · 13, by theorem 2.3(3),
(1) q≡1 mod 7
(2) q≡1 mod 11
(3) q≡1 mod 13.
By theorem 3.1(2), we have
     
1001 7 11 13
= .
q q q q
Using theorem 3.1(3) and 3.4, we see that
 
7 q 
=− since q ≡ 3 mod 4 and 7 ≡ 3 mod 4
q 7
 
1
=− from equation 1
7
= −1.
Likewise, we see that
 
11 q
=− since q ≡ 3 mod 4 and 11 ≡ 3 mod 4
q 11
 
1
=− from equation 2
11
= −1,
and
 
13 q
= since 13 ≡ 1 mod 4
q 13
 
1
= from equation 3
13
=1
5
 Now we have enough information to determine whether 1001 is a quadratic residue modulo
q. We see that
     
1001 7 11 13
= .
q q q q
= −1 · −1 · 1
= 1.
By definition, 1001 is a quadratic residue modulo q.

6
Problem (3.3.4)

Solution.

We want to determine whether the congruence x4 ≡ 25 mod 1013 has solutions given that
1013 is prime.
2
We see that x4 = (x2 ) = 25 mod 1013, so it is sufficient to prove that x2 ≡ ±5 mod 1013
has solutions.
First we consider x2 ≡ 5 mod 1013. Using theorem 3.4, or quadratic reciprocity, and the
fact that 5 ≡ 1 mod 4, we see that
   
5 1013
=
1013 5
 
3
=
5
= −1,
because by inspection
(
a 1 if a ≡ ±1 mod 5
=
5 −1 if a ≡ ±3 mod 5
By definition, x2 ≡ 5 mod 1013 has no solutions.
Second we consider x2 ≡ −5 mod 1013. Using theorem 3.1, we see that
    
−5 −1 5
=
1013 1013 1013
1013−1
= (−1) 2 · (−1) (from our earlier work)
= (−1)506 · (−1)
= 1 · −1
= −1.
By definition, x2 ≡ −5 mod 1013 has no solutions.
Since x2 ≡ ±5 mod 1013 has no solutions and ±5 mod 1013 are the only solutions to
2
the congruence y 2 ≡ 25 mod 1013, we see that (x2 ) = x4 ≡ 25 mod 1013 can have no
solutions.

7
Problem (3.3.15)

Solution.

We aim to show that for any prime p ≥ 7, there is some number n ∈ N where 1 ≤ n ≤ 9 and
   
n n+1
(4) = = 1.
p p
We have three cases
1. p ≡ ±1  mod 8: p2 −1
    2
Clearly, p = 1. By theorem 3.3, p2 = (−1) 8 . Clearly, p2 = 1 if p 8−1 is even.
1

Since p ≡ ±1 mod 8, we may write p as 1 + 8k for some k ∈ Z. Now we see that

p = ±1 + 8k
p2 = 1 ± 16k + 64k
p2 − 1 = ±16k + 64k
p2 − 1
= ±2k + 8k
8
p2 − 1
≡ 0 mod 2,
8
   
so p1 = 1 and p2 = 1. Therefore, equation 4 is true for n = 1.
2. p ≡ ±1 mod  5:  
Clearly, p = 1. By theorem 3.4, or quadratic reciprocity, p5 = p5 . By inspection,
4

the only quadratic residues

 modulo 5 are the congruence classes ±1. By assumption,
5
p ≡ ±1 mod 5, so p = 1. Therefore, equation 4 is true for n = 4.
3. p ≡ ±2 mod  5 and p ≡ ±3 mod 8:  
Clearly, p = 1. Our earlier work in problem 3.2.8 shows us that 10
9
p
= 1 when p ≡ ±2
mod 5 and p ≡ ±3 mod 8. This is true by assumption, so equation 4 is true for n = 9.
Since p ≥ 7, we see that p is odd. We have covered all cases because p must be congruent
to one of ±1, ±3 mod 8 and congruent to one of ±1, ±2 mod 5. Therefore, equation 4 is
true for some n ∈ Z where 1 ≤ n ≤ 9.

8
Problem (3.4.1)

Solution.

a. For the binary quadratic form

f (x, y) = x2 + y 2 ,
we have
d = b2 − 4ac
= 02 − 4(1)(1)
= −4
< 0,
so given that a and c have the same sign and a > 0, by theorem 3.11, f (x, y) is
positive definite.
b. For the binary quadratic form
f (x, y) = −x2 − y 2 ,
Since x2 + y 2 is positive definite and f (x, y) = −(x2 + y 2 ), clearly f (x, y) is negative
definite.
c. For the binary quadratic form
f (x, y) = x2 − 2y 2 ,
we have
d = b2 − 4ac
= 02 − 4(1)(−2)
=8
> 0,
so by theorem 3.11, f (x, y) is indefinite.
d. For the binary quadratic form
f (x, y) = 10x2 − 9xy + 8y 2
we have
d = b2 − 4ac
= (−9)2 − 4(10)(8)
= 81 − 320
= −239
< 0.
We see that a = 10, c = 8 have the same sign and that a > 0, so by theorem 3.11,
f (x, y) is positive definite.
9
e. For the binary quadratic form
f (x, y) = x2 − 3xy + y 2
we have
d = b2 − 4ac
= (−3)2 − 4(1)(1)
=9−4
=5
> 0,
so by theorem 3.11, f (x, y) is indefinite.
f. For the binary quadratic form
f (x, y) = 17x2 − 26xy + 10y 2
we have
d = b2 − 4ac
= (−26)2 − 4(17)(10)
= 576 − 680
= −104
< 0.
We see that a = 17 and c = 10 have the same sign and that a > 0, so by theorem
3.11, f (x, y) is positive definite.

10
Problem (3.3.4)

Solution.
√ √
First we find a formula for positive integers xk and yk such that (3 + 2 2)k = xk + 2yk .
We see that
k  
√ k X k k−i √ i
(3 + 2 2) = 3 (2 2)
i
i=0
k   k  
X k k−i √ i X k k−j √ j
= 3 (2 2) + 3 (2 2)
i=0
i j=0
j
i even j odd
k   k  
X k k−i i i/2 X k k−j j √ j−1 √
= 3 ·2 ·2 + 3 · 2 · ( 2) · 2
i=0
i j=0
j
i even j odd
k   k  
X k k−i 3i √ X k k−j 3j−1
= 3 ·22 + 2 3 ·2 2 .
i=0
i j=0
j
i even j odd

Thus, such a representation is possible for

k  
X k k−i 3i
xk = 3 ·22
i=0
i
i even
k  
X k k−j 3j−1
yk = 3 ·2 2 .
j=0
j
j odd
√
Next we show that (3 − 2 2)k = xk − yk .
k  
√ k X k k−i √
(3 − 2 2) = 3 (−2 2)i
i
i=0
k   k  
X k k−i √ i X k k−j √
= 3 (−2 2) + 3 (−2 2)j
i=0
i j=0
j
i even j odd
k   k  
X k k−i √ i i
X k k−j √ j
= 3 (2 2) · (−1) + 3 (2 2) · (−1)j
i=0
i j=0
j
i even j odd
k   k  
X k k−i √ i X k k−j √ j
= 3 (2 2) − 3 (2 2)
i=0
i j=0
j
i even j odd
11
 k   k  
X k k−i i i/2 X k k−j j √ j−1 √
= 3 ·2 ·2 − 3 · 2 · ( 2) · 2
i=0
i j=0
j
i even j odd
k   k  
X k k−i 3i √ X k k−j 3j−1
= 3 ·22 − 2 3 ·2 2
i=0
i j=0
j
i even j odd
√
= xk − 2yk

Now we deduce that x2k − 2yk2 = 1 for k = 1, 2, 3. . . ..

We see that
√ √
x2k − 2yk2 = (xk + 2yk )(xk − 2yk )
√ √
= (3 + 2 2)k (3 − 2 2)k
h √ √ ik
= (3 + 2 2)(3 − 2 2)
= (9 − 4 · 2)k
= (9 − 8)k
= 1k
= 1 for all k = 1, 2, 3, . . .

Now we show gcd(xk , yk ) = 1. By theorem 1.3, the greatest common divisor or x2k and yk2 is
the smallest positive integer that is a linear combination of the two with integer coefficients.
Since we have showed that x2k − 2yk = 1, we see that gcd(x2k , yk2 ) = 1 because there are
no positive integers smaller than 1. By definition, two coprime numbers share no prime
factors. Since the prime factors of xk are a subset of x2k and the prime factors of yk are a
subset of yk2 , we see that xk , yk cannot share any prime factors. Thus, they are coprime and
gcd(xk , yk ) = 1 for all k ∈ N.
Next we show that xk+1 = 3xk + 4yk and yk+1 = 2xk + 3yk for all k ∈ N. We see that
k  
X k k−i 3i
xk = 3 ·22
i=0
i
i even
k+1  
X k + 1 k+1−i 3i
xk+1 = 3 ·22
i=0
i
i even
k  
X k k−j 3j−1
yk = 3 ·2 2
j=0
j
j odd
k+1  
X k + 1 k+1−j 3j−1
yk+1 = 3 ·2 2 .
j=0
j
j odd
12
We see that
k   k  
X k k−i 3i X k k−j 3j−1
3xk + 4yk = 3 3 ·22 +4 3 ·2 2
i=0
i j=0
j
i even j odd
k   k  
X k k+1−i 3i X k k−j 3j+3
= 3 ·22 + 3 ·2 2
i=0
i j=0
j
i even j odd
k+1   k+1  
X k k+1−i 3i X k 3(j−1)+3
= 3 ·2 +
2 3k−(j−1) · 2 2
i=0
i j=0
j−1
i even j even
k+1    
X k k+1−i 3i k 3(i−1)+3
= 3 ·22 + 3k−(i−1) · 2 2
i=0
i i−1
i even
k+1    
X k k 3i
= + 3k+1−i · 2 2
i=0
i i−1
i even
k+1  
X k + 1 k+1−i 3i
= 3 ·22
i=0
i
i even
= xk+1 ,

where we have made use of the equality

   
n n
= ,
k n−k
as outlined in equation 1.10 of Zuckerman et al.
Likewise, we see that
k   k  
X k k−i 3i X k k−j 3j−1
2xk + 3yk = 2 3 ·22 +3 3 ·2 2
i=0
i j=0
j
i even j odd
k   k  
X k k−i 3i+2 X k k+1−j 3j−1
= 3 ·2 2 + 3 ·2 2
i=0
i j=0
j
i even j odd
k+1   k  
X k k−(i−1)
3(i−1)+2 X k k+1−j 3j−1
= 3 ·2 2 + 3 ·2 2
i=0
i−1 j=0
j
i odd j odd
k+1    
X k k 3j−1
= + 3k+1−j 2 2
j=0
j j−1
j odd
13
 k+1  
X k + 1 k+1−j 3j−1
= 3 ·2 2
j=0
j
j odd
= yk+1
Next we show that {xk } and {yk } are strictly increasing sequences.
For i, k ∈ N, we see that
−i < 2k + 2
−i + (k + 1) < 2k + 1 + (k + 1)
k − i + 1 < 3k + 3
(k)(k − 1)(k − 2) · · · (k − i + 2) (k)(k − 1)(k − 2) · · · (k − i + 2)
(k − i + 1) < 3(k + 1)
(i)! (i)!
(k)(k − 1)(k − 2) · · · (k − i + 2)(k − i + 1) (k + 1)(k)(k − 1)(k − 2) · · · (k − i + 2)
<3
(i)! (i)!
   
k k+1
<3
i i
   
k−i k k−i k+1
3 <3 ·3·
i i
   
k k+1
· 3k−i < · 3k+1−i .
i i
Suppose i is even. From this we see that
   
k k−i k+1
·3 < · 3k+1−i
i i
   
k k−i 3i k+1 3i
·3 ·22 < · 3k+1−i · 2 2
i i
xk < xk+1 .
It is easy to see that for m < n, xm < xn . Thus, {xk } is a monotonic strictly increasing
sequence.
Suppose i is odd. From this we see that
   
k k−i k+1
·3 < · 3k+1−i
i i
   
k k−i 3i+1 k+1 3i+1
·3 ·2 2 < · 3k+1−i · 2 2
i i
yk < yk+1 .
It is easy to see that for m < n, ym < yn . Thus, {yk } is a monotonic strictly increasing
sequence.
Finally we show that 1 has infinitely many proper representations by the quadratic form
x − 2y 2 .
2

We have showed that x2k − 2yk2 = 1 for k ∈ N. Furthermore, we showed gcd(xk , yk ) = 1

for k ∈ N. Lastly, we showed that xm 6= xn for m 6= n because m < n implies xm < xn .
14
Likewise, we showed that ym 6= yn for m 6= n because m < n implies ym < yn . From this, we
can conclude that there is an infinite sequence {(xk , yk )} whose distinct members properly
represent 1.

15
Problem (3.3.10)

Solution.

We want to show that for f (x, y) = ax2 + bxy + cy 2 , a quadratic form with integral coeffi-
cients, that there exist integers x0 , y0 not both 0 such that f (x0 , y0 ) = 0, if and only if the
discriminant d of f (x, y) is a perfect square, possibly 0.

−→
Suppose there exist two integers, x0 and y0 , not both 0, such that f (x0 , y0 ) = 0. We know
that
4af (x0 , y0 ) = (2ax0 + by0 )2 − dy02 = 0.
Call v = 2ax0 + by0 . We see that
v 2 = dy02
v2
=d for y0 6= 0
y02
 2
v
= d.
y0
6 0, then d is a perfect square. If y0 = 0, then f (x0 , 0) = ax20 = 0, so a = 0.
Thus, if y0 =
Given that d = b2 − 4ac, if a = 0 then d = b2 , so again d is a perfect square.

←−
Suppose d is a perfect square. Let d = e2 for some e ∈ Z. Given
4af (x0 , y0 ) = (2ax0 + by0 )2 − dy0 ,
we see that
(2ax0 + by0 )2 − dy02 = (2ax0 + by0 )2 − e2 y02
= (2ax0 + by0 + ey0 )(2ax0 + by0 − ey0 )
= (2ax0 + (b + e)y0 )(2ax0 + (b − e)y0 ).
We see that x0 = (b − e), y0 = −2a and x0 = (b + e), y0 = −2a are two solutions. If x0
and y0 are not both 0, we are done, so we suppose that they are both 0. Then −2a = 0
implies a = 0 and b − e = 0 and b + e = 0 imply b = e = 0. Plugging in a = b = 0, we
see that f (x, y) = cy 2 . Clearly, x0 = 1, y0 = 0 is a solution to f (x, y) = 0, and they are
not both zero. Thus if d is a perfect square, there exist integers x0 , y0 not both 0, such that
f (x0 , y0 ) = 0. 

16
Problem (3.5.1)

Solution.

We want to find a reduced form equivalent to the form

f (x, y) = 7x2 + 25xy + 23y 2 .
We see that in the original form, a = 7, b = 25, and c = 23. We see that 25 = b 6≤ |a| = 7,
so we first apply the matrix  
1 −2
.
0 1
From equations 3.7a-c, we get the new form
A1 x2 + B1 xy + C1 y 2 ,
where
A1 = f (1, 0) = a
=7
B1 = 2am12 + b
= 2 · 7 · −2 + 25
= −28 + 25
= −3
C1 = am212 + bm12 + c
= 7 · (−2)2 + 25 · (−2) + 25
= 7 · 4 − 50 + 23
= 28 − 50 + 23
=1
Now we have
f1 (x, y) = 7x2 − 3xy + y 2 ,
where f ∼ f1 .
We see that |A1 | <
6 |C1 | so we apply the matrix
 
0 1
.
1 0
From equations 3.7a-c, we get the new form
A2 x2 + B2 xy + C2 y 2 ,
where
A 2 = C1
=1
B2 = −B1
17
 = −(−3)
=3
C2 = A 1
=7
Now we have
f2 (x, y) = x2 + 3xy + 7y 2 ,
where f2 ∼ f1 .
Again, we see that B2 6≤ |A2 |, so we apply the matrix
 
1 −1
0 1
to get
A3 = f (1, 0) = a
=1
B3 = 2am12 + b
= 2 · 1 · −1 + 3
= −2 + 3
=1
C3 = am212 + bm12 + c
= 1 · (−1)2 + 3 · (−1) + 7
=1−3+7
=5
Now we have
f3 (x, y) = x2 + xy + 5y 2 ,
where f3 ∼ f2 , and f3 is in reduced form.
By transitivity, f3 ∼ f .

18
Problem (3.5.3)

Solution.
 
x y
For x, y ∈ Z, we want to show that there exist u, v ∈ Z such that ∈ Γ if and only if
u v
gcd(x, y) = 1.
By theorem 1.3, gcd(x, y) = 1 if and only if we can write
xv + yu0 = 1,
for some u0 , v ∈ Z. Let u = −u0 . We have
xv − yu = 1.
Consider the matrix  
x y
M= .
u v
We see that M ∈ Γ if and only if gcd(x, y) = 1 because mij ∈ Z and det M = xv − yu = 1
if and only if gcd(x, y) = 1. 

19
Problem (3.5.11)

Solution.

Suppose that ax2 + bxy + cy 2 ∼ Ax2 + Bxy + Cy 2 . We want to show that gcd(a, b, c) =
gcd(A, B, C).
Let f (x, y) = ax2 + bxy + cy 2 and h(x, y) = Ax2 + Bxy + Cy 2 . Since f ∼ h, they represent
the same points. Let g = gcd(a, b, c) and G = gcd(A, B, C). Clearly, f (x,y)
g
∈ Z. Because f
and h represent the same points, h(x,y)
g
∈ Z as well. We see that h(1, 0) = A and h(0, 1) = C,
so g | A and g | C. Again, h(1, 1) = A + B + C and because g | h(x, y), then g | B as well.
So g is a common divisor of A,B, and C. Thus, g | G.
Likewise, G | h(x, y) G | f (x, y) for all x, y ∈ Z. We see that f (1, 0) = a and f (0, 1) = c,
so G | a and G | c. Again, f (1, 1) = a + b + c and because G | f (x, y), then G | b as well. So
G is a common divisor of a,b, and c. Thus, G | g.
Since g | G and G | g, g = ±G. But the greatest common divisor is always positive, so
g = G.

20
Problem (3.5.12)

Solution.

Suppose f (x, y) = ax2 + bxy + cy 2 is a positive semidefinite quadratic form of discriminant

0. Let g = gcd(a, b, c). We want to show that f is equivalent to the form gx2 .
We see that d = b2 − 4ac = 0, so
 2
b
(5) ac = ,
2
So the product ac is a square. Consider the prime factorization of a and c such that
Y
a= pα(p)
p
Y
c= pγ(p) ,
p

for primes p. Clearly,

(6) α(p) + γ(p) ≡ 0 mod 2.
Let g = gcd(a, b, c) be factored as
Y
g= pmin(α(p),β(p),γ(p)) .
p
 
Considering equation 5, we see that min(α(p), β(p), γ(p)) = min α(p), α(p)+γ(p)
2
, γ(p) , if
 
p 6= 2 and min α(2), α(2)+γ(2)+1
2
, γ(2) , if p = 2. Furthermore, we see that if α(p) ≤ γ(p),
then
2α(p)
α(p) =
2
α(p) + α(p)
=
2
γ(p) + α(p)
≤ ,
2
and if γ(p) ≤ α(p), then
2γ(p)
γ(p) =
2
γ(p) + γ(p)
=
2
γ(p) + α(p)
≤ .
2
Therefore, the exponents of the greatest common divisor in its prime factorization have the
form min(α(p), γ(p)).
21
 By definition of a divisor, there is some n ∈ Z such that ng = a. Consider the prime
factorization of n. We see that
Y
n = 2j pα(p)−min(α(p),γ(p)) ,
p6=2

for some j ∈ N. From equation 6, we see that α(p) and γ(p) have the same parity. That is,
they are both even or both odd.
Thus, we see that α(p) − min(α(p), γ(p)) is even, so n is a perfect square. Let m2 = n for
some m ∈ Z. We see that
4af (x, y) = (2ax + by)2 − dy 2
4af (x, y) = (2ax + by)2
(2ax + by)2
f (x, y) =
4a
 2
by
g 2 2ax
g
+ g
f (x, y) =
4ng
 2
2ax by
g g
+ g
f (x, y) =
4n
 2
2ax by
g g
+ g
f (x, y) =
4m2
 2
2ax + by
f (x, y) = g ,
2mg
so all values of f (x, y) can be represented by gx2 .
Call h(x, y) = gx2 . We see that
 
2a b
f (x, y) = g x+ y, ux + vy ,
2mg 2mg
where u and v are such that
a b
v− u = 1.
mg 2mg
 
a b a
Such numbers u, v will exist if gcd mg , 2mg = 1. We also see that mg
= m because
2
a = ng = m g.  
b

2
We aim to show that gcd m, 2mg = 1. We see that ac = m2 gc = 2b , so m2 | (b2 /4), so
m | (b/2).  
Now we note that if gcd(a, b, c) = g, then gcd ag , gb , gc = 1. We see that
 2
b
ac =
2
 2
1 b
c=
a 2
22
  2
1 b
c= 2
mg 2
1 b b
c= 2 · ·
mg 2 2
c b b
=· · .
g 2mg 2mg
Also,
a m2 g
=
g g
= m2 .
So  
2 b b b
gcd m , , · = 1.
g 2mg 2mg
  2   
2 b b
Clearly, gcd m , 2mg = 1 so we can conclude that gcd m, 2mg = 1 considering their
prime factorizations.
Therefore, there will be u, v which satisfy
a b
v− u = 1,
mg 2mg
and the matrix
b
 
m 2mg
∈Γ
u v
defines the transformation from h to f . By definition, h ∼ f .

23