Quick! Bound it!

y 2 ( x 2 + y 2 2 x y x y ) = ( x + y ) 2 ( x y ) y^2(x^2+y^2-2xy-x-y) = (x+y)^2 (x-y)

Find the number of integer pairs ( x , y ) (x, y) that satisfy the above expression.


The answer is 3.

This section requires Javascript.
You are seeing this because something didn't load right. We suggest you, (a) try refreshing the page, (b) enabling javascript if it is disabled on your browser and, finally, (c) loading the non-javascript version of this page . We're sorry about the hassle.

4 solutions

Sharky Kesa
Mar 13, 2016

Note the following

y 2 ( x 2 + y 2 2 x y x y ) = ( x + y ) 2 ( x y ) y 2 ( x y ) 2 y 2 ( x + y ) = ( x + y ) ( x 2 y 2 ) y 2 ( x y ) 2 = ( x + y ) ( x 2 y 2 ) + ( x + y ) y 2 y 2 ( x y ) 2 = x 2 ( x + y ) \begin{aligned} y^2(x^2+y^2-2xy-x-y)&=(x+y)^2(x-y)\\ y^2(x-y)^2-y^2(x+y)&= (x+y)(x^2-y^2)\\ y^2(x-y)^2&= (x+y)(x^2-y^2)+(x+y)y^2\\ y^2(x-y)^2 &= x^2 (x+y) \end{aligned}

Note that because the LHS is a perfect square, the RHS must also be a perfect square. Thus, x + y x+y is a perfect square. Let x + y = k 2 x+y=k^2 or k 2 x = y k^2-x=y . (We will first check when x y x \geq y )

y 2 ( x y ) 2 = x 2 ( x + y ) y ( x y ) = k x ( k 2 x ) ( 2 x k 2 ) = k x 2 x 2 ( 3 k 1 ) k x + k 4 = 0 ( 1 ) \begin{aligned} y^2(x-y)^2&=x^2(x+y)\\ y(x-y)&=kx\\ (k^2-x)(2x-k^2)&=kx\\ 2x^2 - (3k-1)kx + k^4 &= 0 \quad \quad \quad (1) \end{aligned}

If x y x \leq y , then we have

y ( y x ) = k x ( k 2 x ) ( k 2 2 x ) = k x 2 x 2 ( 3 k + 1 ) k x + k 4 = 0 ( 2 ) \begin{aligned} y(y-x) &= kx\\ (k^2-x)(k^2-2x) &= kx\\ 2x^2 - (3k+1)kx + k^4 &= 0 \quad \quad \quad (2) \end{aligned}

Let Δ 1 \Delta_1 be the discriminant of ( 1 ) (1) and Δ 2 \Delta_2 be the discriminant of ( 2 ) (2) . We know that both these discriminants have to be perfect squares for us to have integer solutions.

Δ 1 = k 2 ( 3 k 1 ) 2 8 k 4 = k 4 6 k 3 + k 2 = k 2 ( ( k 3 ) 2 8 ) \begin{aligned} \Delta_1 &= k^2(3k-1)^2 - 8k^4\\ &= k^4-6k^3+k^2\\ &= k^2((k-3)^2-8) \end{aligned}

The only values of k k which satisfy the discriminant being a perfect square are when k = 0 , 6 k=0, 6 .

Δ 2 = k 2 ( 3 k + 1 ) 2 8 k 4 = k 4 + 6 k 3 + k 2 = k 2 ( ( k + 3 ) 2 8 ) \begin{aligned} \Delta_2 &= k^2 (3k+1)^2 - 8k^4\\ &= k^4+6k^3+k^2\\ &=k^2((k+3)^2-8) \end{aligned}

The only values of k k which satisfy the discriminant being a perfect square are when k = 6 , 0 k=-6,0 .

Thus, k 2 = 0 , 36 k^2=0, 36 , which implies x + y = 0 , 36 x+y=0, 36 .

If x + y = 0 x+y=0 , x = y x=-y . Substituting, we have

x 2 ( 2 x ) 2 = 0 4 x 4 = 0 x = 0 , y = 0 \begin{aligned} x^2 (2x)^2 &= 0\\ 4x^4&=0\\ x&=0, y=0 \end{aligned}

Thus, ( 0 , 0 ) (0,0) is a solution in ( x , y ) (x, y) . If x + y = 36 x+y=36 , y = 36 x y=36-x . If x x is greater than or equal to 18,

( 36 x ) 2 ( 2 x 36 ) 2 = 36 x 2 2 x 2 + 108 x 1296 = 6 x x 2 51 x + 648 = 0 ( x 24 ) ( x 27 ) = 0 x = 24 , 27 \begin{aligned} (36-x)^2 (2x-36)^2 &= 36x^2\\ -2x^2+108x-1296 &= 6x\\ x^2-51x+648 &= 0\\ (x-24)(x-27) &= 0\\ x &= 24,27 \end{aligned}

Thus, the solutions ( x , y ) (x, y) when x x is greater than or equal to 18 are ( 24 , 12 ) (24, 12) and ( 27 , 9 ) (27, 9) . If x x is less than or equal to 18,

( 36 x ) 2 ( 2 x 36 ) 2 = 36 x 2 2 x 2 108 x + 1296 = 6 x x 2 57 x + 648 = 0 \begin{aligned} (36-x)^2 (2x-36)^2 &= 36x^2\\ 2x^2-108x+1296 &= 6x\\ x^2-57x+648=0 \end{aligned}

This quadratic is unfactorable for integers. Thus, the only solutions or ( x , y ) (x, y) are ( 0 , 0 ) (0, 0) , ( 24 , 12 ) (24, 12) and ( 27 , 9 ) (27, 9) .

Moderator note:

It is best to split up the solution into cases of x y x \geq y and x < y x < y , instead of progressing the cases in parallel.

For Δ 1 \Delta_1 , shouldn't we require k = 2 , 4 k = 2,4 instead?

Calvin Lin Staff - 5 years, 2 months ago

Log in to reply

Sorry, typo.

Sharky Kesa - 5 years, 2 months ago

Is it possible to solve it this way?

Let z be the gcd(x, y), x = mz, y = nz, and m and n are coprime. Now we see that

z^4 * n^2 * (m^2 + n^2 - 2mn) = z^3 * m^2 * (m + n)

z * n^2 * (m^2 + n^2 - 2mn) = m^2 * (m + n) if z is not 0

First thing we can notice is that one solution is clear if we set z = 0. Now, since m and n are coprime, n^2 must divide (m + n), but that means n divides m even though they are mutually prime. That means n = 1 and m divides z. We now see that m is divisible by n. If we put x = ym into the equation, we get

y^4 * (m - 1)^2 = m^2 * y^3 * (m+1)

y(m - 1)^2 = (m + 1)m^2

(m - 1)^2/m^2 = (m + 1)/y

On the LHS, (m - 1) and m are coprime, which means that (m + 1) is a multiple of (m - 1)^2 (or vice versa). This is true for two integers m: 2 and 3, the last of which I sadly missed (since (m - 1)^2 = ((m + 1) - 2)^2, (m + 1) has to divide 4, proving 2 and 3 are the only solutions).

So we have two pairs (y, m): (12, 2) and (9, 3). x = my, so the solutions (x, y) are (24, 12), (27, 9) and the previously mentioned (0, 0). My solution seems too simplistic. Am I making any leap of faith logic?

Scrub Lord - 4 years ago

Log in to reply

That's a nice way to make use of n 2 m + n n^2 \mid m+n .

Can you clean it up and I can post it as a separate solution?

Calvin Lin Staff - 4 years ago
Dong kwan Yoo
Apr 18, 2018

First, if x = 0 x=0 then, y 2 ( y 2 y ) = y 3 y^2 ( y^2 - y) = - y^3 , so y = 0 y=0 if y = 0 y=0 then, 0 = x 3 0= x^3 , so x = 0 x=0
And (0,0) satisfies the above expression

let x = g a x = ga and y = g b y = gb ( G C D ( a , b ) = 1 a n d l e t g a b 0 a n d b > 0 ) GCD(a,b) = 1 ~~and ~~let ~~gab ≠0~~and ~b>0 )

then, g 3 b 2 ( g ( a b ) 2 ( a + b ) ) = g 3 ( a + b ) 2 ( a b ) g^3 b^2 ( g(a-b)^2 - (a+b) ) = g^3 (a+b)^2 (a-b) , so we get b 2 ( g ( a b ) 2 ( a + b ) ) = ( a + b ) 2 ( a b ) b^2 ( g(a-b)^2 - (a+b) ) = (a+b)^2 (a-b)

Because b 2 b^2 divides LHS , b 2 b^2 divides RHS and G C D ( a , b ) = 1 GCD(a,b)= 1 , g c d ( b 2 , ( a + b ) 2 ) = g c d ( b 2 , ( a b ) ) = 1 gcd ( b^2 , (a+b)^2 ) = gcd( b^2 , (a-b) ) = 1

so b 2 = 1 b^2 = 1 and b = 1 b=1

By " b = 1 b=1 "

g ( a 1 ) 2 ( a + 1 ) = ( a + 1 ) 2 ( a 1 ) g(a-1)^2 - (a+1) = (a+1)^2 (a-1)

g ( a 1 ) 2 = ( a + 1 ) a 2 g(a-1)^2 = (a+1) a^2

therefore ( a = 2 a n d g = 12 o r a = 3 a n d g = 9 ) , ( x , y ) = ( 24 , 12 ) , ( 27 , 9 ) (a= 2 ~~and~~ g=12 ~~~~or~~~a=3 ~~and ~~g=9~~) , (x,y)= (24,12) , ~(27, 9)~
( g c d ( ( a 1 ) 2 , a 2 ) = 1 s o ( a 1 ) 2 a + 1 , s o a = 2 o r a = 3 ) ( ∵ gcd( (a-1)^2 , a^2 ) = 1 so (a-1)^2 | a+1 , ~ so ~ a=2 ~or ~ a=3 )

By above, ( x , y ) = ( 0 , 0 ) , ( 24 , 12 ) , ( 27 , 9 ) (x,y ) = (0,0)~,~(24,12)~,~(27,9) ~

The task allows negative numbers, so one has to check b 2 = 1 b = ± 1 b^2=1\:\Rightarrow\:b=\pm 1 . Luckily, in this task both cases lead to the same result.

Carsten Meyer - 1 year, 8 months ago

Log in to reply

thx, but b was already defined by a number greater than 0 in 3rd line

Dong kwan Yoo - 11 months ago

Log in to reply

My mistake, I had missed that all sign cases of x , y x,\:y are already covered by a , g 0 a,\:g\neq 0 in line 3, so we may assume b > 0 b>0 without loss of generality. I'm sorry about the confusion!

Carsten Meyer - 2 months, 3 weeks ago
Laurent Shorts
Feb 21, 2017

As already given by Sharky Kesa:

y 2 ( x 2 + y 2 2 x y x y ) = ( x + y ) 2 ( x y ) y 2 ( x y ) 2 y 2 ( x + y ) = ( x + y ) ( x 2 y 2 ) y 2 ( x y ) 2 = ( x + y ) ( x 2 y 2 ) + ( x + y ) y 2 y 2 ( x y ) 2 = x 2 ( x + y ) \begin{aligned} y^2(x^2+y^2-2xy-x-y)&=(x+y)^2(x-y)\\y^2(x-y)^2-y^2(x+y)&= (x+y)(x^2-y^2)\\y^2(x-y)^2&= (x+y)(x^2-y^2)+(x+y)y^2\\y^2(x-y)^2 &= x^2 (x+y)\end{aligned}

We see that x + y x+y must be square.

Substitute a 2 = x + y b = x y x = a 2 + b 2 y = a 2 b 2 \boxed{\begin{aligned}a^2=x+y\\b=x-y\\x=\frac{a^2+b}2\\y=\frac{a^2-b}2\end{aligned}} to get ( a 2 b ) 2 4 b 2 = ( a 2 + b ) 2 4 a 2 ( a 2 b ) b = ± ( a 2 + b ) a a 2 b b a = b 2 ± a 3 a 1 = b 2 ± a 3 a b \begin{aligned}\dfrac{(a^2-b)^2}{4}b^2&=\dfrac{(a^2+b)^2}4a^2 \\ (a^2-b)b&=\pm(a^2+b)a \\ a^2b\mp ba&=b^2\pm a^3 \\ a\mp1&=\dfrac{b^2\pm a^3}{ab} \end{aligned}

Case a = 0 a=0 or b = 0 b=0 : if a = 0 a=0 , then b = 0 b=0 and inversely. And then x = y = 0 x=y=0 . It is a solution.

Case a , b , x , y 0 a,b,x,y\neq0 :

We see that b b divides a 3 a^3 and a a divides b 2 b^2 , which means that each primes dividing a a also divides b b and inversely: a = p i m i , b = p i n i with m i , n i 1 \displaystyle a=\prod p_i^{m_i},~b=\prod p_i^{n_i}\text{ with }m_i,n_i\geq1

Let consider a prime divisor p p so that a = p m a a=p^m·a' and b = p n b b=p^n·b' and ( p , a ) = ( p , b ) = 1 (p,a')=(p,b')=1 . As p p doesn't divide the LHS, it doesn't divide the RHS: p 2 n b 2 ± p 3 m a 3 p m + n a b is an integer not divisible by p \dfrac{p^{2n}·b'^2\pm p^{3m}·a'^3}{p^{m+n}·a'b'} \text{ is an integer not divisible by }p

If n < m n<m , then p m + n p^{m+n} divides p 3 m a 3 p^{3m}·a'^3 but not p 2 n b 2 p^{2n}·b'^2 , that cannot be an integer.

If 2 m < n 2m<n , then p m + n p^{m+n} divides p 2 n b 2 p^{2n}·b'^2 but not p 3 m a 3 p^{3m}·a'^3 , that cannot be an integer.

If m < n < 2 m m<n<2m , then p 2 n b 2 p^{2n}·b'^2 and p 3 m a 3 p^{3m}·a'^3 both have at least a factor p p after the division: the RHS would then be divisible by p p .

Therefore, n = m n=m or n = 2 m n=2m . We can write a = q i m i r j m j , b = q i m i r j 2 m j a = e f and b = e f 2 for some integers e , f \begin{aligned}\displaystyle a=\prod q_i^{m_i}·\prod r_j^{m_j},~b=\prod q_i^{m_i}·\prod r_j^{2m_j} \\ \Longleftrightarrow a=e·f\text{ and }b=e·f^2\text{ for some integers }e,f\end{aligned}

Put it again into ( a 2 b ) b = ± ( a 2 + b ) a (a^2-b)b=\pm(a^2+b)a to get ( e 2 f 2 e f 2 ) e f 2 = ± ( e 2 f 2 + e f 2 ) e f ( e 1 ) f = ± ( e + 1 ) \begin{aligned}(e^2·f^2-e·f^2)e·f^2&=\pm(e^2·f^2+e·f^2)e·f \\ (e-1)·f&=\pm(e+1) \end{aligned} f = ± e + 1 e 1 = ± ( 1 + 2 e 1 ) f=\pm\dfrac{e+1}{e-1}=\pm\left(1+\dfrac{2}{e-1}\right) .

e 1 e-1 divides 2 and thus can only be e 1 = ± 1 e-1=\pm1 or e 1 = ± 2 e-1=\pm2 , which gives ( e , f ) = ( 2 , 3 ) , ( 0 , 1 ) , ( 3 , 2 ) , ( 1 , 0 ) (e,f)=(2,3),(0,-1),(3,2),(-1,0) .

Case e = 2 , f = 3 e=2,f=3 : a = 6 , b = 18 , x = 27 , y = 9 a=6,b=18,x=27,y=9

Case e = 3 , f = 2 e=3,f=2 : a = 6 , b = 12 , x = 24 , y = 12 a=6,b=12,x=24,y=12

There are three possibilites : ( 0 , 0 ) , ( 27 , 9 ) , ( 24 , 12 ) \boxed{(0,0),~(27,9),~(24,12)}

Kevin Tong
Apr 17, 2019

We begin by simplifying y 2 ( x 2 + y 2 2 x y x y ) = ( x + y ) 2 ( x y ) = ( x 2 y 2 ) ( x + y ) y^2(x^2+y^2-2xy-x-y)=(x+y)^2(x-y)=(x^2-y^2)(x+y) y 2 ( x y ) 2 y 2 ( x + y ) = x 2 ( x + y ) y 2 ( x + y ) y^2(x-y)^2-y^2(x+y)=x^2(x+y)-y^2(x+y) y 2 ( x y ) 2 = x 2 ( x + y ) y^2(x-y)^2=x^2(x+y) At this point we break the problem up into 2 2 cases: x y x\ge y and x < y x<y . We tackle x y x\ge y first by taking both sides modulus y y , which gives 0 x 2 ( x + y ) y 3 ( m o d y ) 0\equiv x^2(x+y)\equiv y^3\pmod{y} Hence, x = n y x=ny , for some integer n n . Plugging this into the equation gives y 2 ( y ( n 1 ) ) 2 = y 4 ( n 1 ) 2 = n 2 y 2 ( y ( n + 1 ) ) = n 2 y 3 ( n + 1 ) y^2(y(n-1))^2=y^4(n-1)^2=n^2y^2(y(n+1))=n^2y^3(n+1) y = ( n + 1 ) n 2 ( n 1 ) 2 y=\frac{(n+1)n^2}{(n-1)^2} This means that ( n 1 ) 2 ( ( n + 1 ) n 2 ) (n-1)^2\mid ((n+1)n^2) . Since these factors differ by 1 1 or 2 2 , these factors should be relatively small (otherwise they cannot divide each other). Checking the possible values reveals that n { 1 , 0 , 2 , 3 } n\in\{-1,0,2,3\} which gives y { 0 , 0 , 12 , 9 } y\in\{0,0,12,9\} , respectively. Since x = n y x=ny , this gives the solutions ( x , y ) { ( 0 , 0 ) , ( 24 , 12 ) , ( 27 , 9 ) } (x,y)\in\{(0,0),(24,12),(27,9)\} .

Now, if x < y x<y , then taking both sides modulus x x gives y 2 ( x y ) 2 y 2 ( y ) 2 y 4 0 ( m o d x ) y^2(x-y)^2\equiv y^2(-y)^2\equiv y^4\equiv 0\pmod{x} so y = n x y=nx , for some integer n n . Plugging this into the equations gives n 2 x 2 ( x ( 1 n ) ) 2 = x 4 n 2 ( 1 n ) 2 = x 2 ( x ( 1 + n ) ) = x 3 ( 1 + n ) n^2x^2(x(1-n))^2=x^4n^2(1-n)^2=x^2(x(1+n))=x^3(1+n) x = 1 + n n 2 ( 1 n ) 2 x=\frac{1+n}{n^2(1-n)^2} This means that n 2 ( 1 n ) 2 ) ( 1 + n ) n^2(1-n)^2)\mid (1+n) . Since these factors once again differ by 1 1 or 2 2 , we find that the only possible value of n n is n = 1 n=-1 , which gives x = 0 x=0 , which was already determined in the previous case.

Hence, the only solutions are ( x , y ) { ( 0 , 0 ) , ( 24 , 12 ) , ( 27 , 9 ) } \boxed{(x,y)\in\{(0,0),(24,12),(27,9)\}} , which means our desired answer is 3 \boxed{3} .

There may be an error in the fourth equation: 0 x 2 ( x + y ) x 3 m o d y 0\equiv x^2(x+y)\equiv \boxed{x}^3 \mod y .


Another problem is the implication "Hence x = n y x=ny [..]" immediately afterwards:

x 3 0 m o d y ⇏ x = n y Counter-example: x = p prime , y = p 3 , x n y \begin{aligned} x^3 & \equiv 0\mod y & \boxed{\not\Rightarrow} && x&=ny &&&&&\left|\text{Counter-example:}\right. && x&= p\text{ prime}, & y &= p^3, & x &\neq ny\end{aligned}

The same occurs within the case x < y x<y when the implication "so y = n x y=nx [..]" is used.

Carsten Meyer - 2 months, 3 weeks ago

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...