A number theory problem by Vitor Santos

x , y , z , t x,y,z,t are all integers satisfying the following system of equations: { x z 2 y t = 3 x t + y z = 1. \begin{cases} \begin{aligned} xz-2yt&=3 \\ xt+yz&=1. \end{aligned} \end{cases} Find x 2 + y 2 + z 2 + t 2 x^2+y^2+z^2+ t^2 .


The answer is 11.

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.

6 solutions

Mark Hennings
Jan 31, 2017

SInce x t + y z = 1 xt + yz = 1 , x x and y y are coprime. Now x z 2 y t = 3 ( x t + y z ) xz - 2yt = 3(xt + yz) , so we see that x ( z 3 t ) = y ( 3 z + 2 t ) x(z-3t) \; = \; y(3z+2t) and hence, since x x and y y are coprime, we have 3 z + 2 t = e x z 3 t = e y 3z + 2t \; = \; ex \hspace{2cm} z - 3t \; = \; ey for some integer e e . Since x t + y z = 1 xt + yz = 1 we deduce that e = z 2 + 2 t 2 e \; = \; z^2 + 2t^2 There are two cases to consider:

  • If 11 11 divides e e , we have e = 11 f e = 11f for some integer, and hence z = f ( 3 x + 2 y ) z = f(3x+2y) , t = f ( x 3 y ) t = f(x-3y) , and 1 = x t + y z = f ( x 2 + 2 y 2 ) 1 = xt + yz = f(x^2 + 2y^2) . This means that f = 1 f=1 , x = ± 1 x = \pm1 and y = 0 y=0 , and hence z = 3 x z = 3x and t = x t = x

  • If e e and 11 11 are coprime then, since e e divides both 3 z + 2 t 3z+2t and z 3 t z - 3t , it follows that e e divides both 11 z 11z and 11 t 11t , so we deduce that e e divides both z z and t t . But that means that e e divides 1 = x t + y z 1 = xt + yz . Since e = z 2 + 2 t 2 0 e = z^2 + 2t^2 \ge 0 , we deduce that e = 1 e=1 , and hence x = 3 z + 2 t y = z 3 t z 2 + 2 t 2 = 1 x \; = \; 3z + 2t \hspace{1.5cm} y \; = \; z - 3t \hspace{1.5cm} z^2 + 2t^2 \; = \; 1 Thus we deduce that t = 0 t = 0 and z = ± 1 z = \pm1 , so that x = 3 z x = 3z and y = z y = z .

Thus there are just four solutions in integers to these equations, namely ( x , y , z , t ) = ( 1 , 0 , 3 , 1 ) , ( 1 , 0 , 3 , 1 ) , ( 3 , 1 , 1 , 0 ) , ( 3 , 1 , 1 , 0 ) (x,y,z,t) = (1,0,3,1)\,,\,(-1,0,-3,-1)\,,\,(3,1,1,0)\,,\,(-3,-1,-1,0) . In all four cases we have x 2 + y 2 + z 2 + t 2 = 11 x^2+ y^2 + z^2 + t^2 = \boxed{11} .

Moderator note:

See Ivan's solution to understand how this problem is motivated by thinking about factorization in the Euclidean domain Z [ 2 ] \mathbb{Z}[\sqrt{-2}] .


For a rigorous solution, be careful that you don't make the assumption of "the values have to be small". The equations do not yet directly yield a bound on the possible values of the variables.

@Mark Hennings Can you please explain why x & y are co-prime. Thanks

Chirayu Bhardwaj - 4 years, 3 months ago

Log in to reply

Since x t + y z = 1 xt + yz = 1 , any integer dividing both x x and y y divides 1 1 , so is ± 1 \pm1 .

Mark Hennings - 4 years, 3 months ago

Log in to reply

co-prime are t and z as well!

A Former Brilliant Member - 3 years, 6 months ago

Why did you check if 11 divides e or not.

Sanjay Kumar - 3 years, 11 months ago

Log in to reply

Sorry, got it.

Sanjay Kumar - 3 years, 11 months ago

Apart from x x and y y , co-prime are z z and t t as well. Solving the system in x x and y y as a result yields x = 3 z + 2 t z 2 + 2 t 2 x=\dfrac{3 z + 2 t}{z^2 + 2 t^2} and y = z 3 t z 2 + 2 t 2 y=\dfrac{z - 3 t}{z^2 + 2 t^2} . For x x and y y to be integers, there has to be z 3 t = k ( z 2 + 2 t 2 ) z-3t=k(z^2+2t^2) and 3 z + 2 t = l ( z 2 + 2 t 2 ) 3z+2t=l(z^2+2t^2) , for some k , l Z k, l\in\mathbb Z . Multiplying the first relation by 2 2 , and the second by 3 3 and summing both relation, there follows 11 z = ( z 2 + 2 t 2 ) ( 2 k + 3 l ) 11z=(z^2+2t^2)(2k+3l) . Since z z and t t are co-prime, z z and z 2 + 2 t 2 z^2+2t^2 have no common divisors, we conclude z 2 + 2 t 2 z^2+2t^2 must divide 11 11 , whence z = ± 1 z=\pm1 and t = 0 t=0 or z = ± 3 z=\pm3 , t = ± 1 t=\pm1 , meaning x = 1 x=1 and y = 0 y=0 or x = 3 x=3 and y = 1 y=1 or x = 3 x=-3 and y = 1 y=-1 . So, x 2 + y 2 + z 2 + t 2 = 11 x^2+y^2+z^2+t^2=11 .

A Former Brilliant Member - 3 years, 6 months ago
Vitor Santos
Feb 1, 2017

For solving { x z 2 y t = 3 x t + y z = 1 \begin{cases} xz-2yt=3 \\ xt+yz=1 \end{cases} I'll do some algebraic manipulations involving complex numbers.

  • Let 2 = ( i 2 ) 2 -2={(i \sqrt {2}) } ^ {2} , substituting on the first equation will lead to x z + ( i 2 ) 2 y t = 3 xz+{(i \sqrt {2}) } ^ {2} yt=3 .
  • Multipling the second one by ( i 2 ) (i \sqrt {2}) , leading to ( i 2 ) x t + ( i 2 ) y z = ( i 2 ) (i \sqrt {2})xt+(i \sqrt {2})yz=(i \sqrt {2}) .
  • Aligning the two equations x z + ( i 2 ) 2 y t = 3 ( i 2 ) x t + ( i 2 ) y z = i 2 \begin{aligned} xz+{ (i\sqrt { 2 } ) }^{ { 2 } }yt=3 \\ (i\sqrt { 2 } )xt+(i\sqrt { 2 } )yz=i\sqrt { 2 } \end{aligned} and summing then up.
  • Lefting us with x z + ( i 2 ) x t + ( i 2 ) 2 y t + ( i 2 ) y z = 3 + i 2 x ( z + t 2 i ) + y 2 i ( z + t 2 i ) = 3 + 2 i ( z + t 2 i ) ( x + y 2 i ) = 3 + 2 i xz+(i\sqrt { 2 } )xt+{ (i\sqrt { 2 } ) }^{ { 2 } }yt+(i\sqrt { 2 } )yz=3+i\sqrt { 2 } \Rightarrow x(z+t\sqrt { 2 } i)+y\sqrt { 2 } i(z+t\sqrt { 2 } i)=3+\sqrt { 2 } i\Rightarrow (z+t\sqrt { 2 } i)(x+y\sqrt { 2 } i)=3+\sqrt { 2 } i .
  • Applying moduli both sides wields ( z + t 2 i ) ( x + y 2 i ) = 3 + 2 i ( z + t 2 i ) ( x + y 2 i ) = 3 + 2 i |(z+t\sqrt { 2 } i)(x+y\sqrt { 2 } i)|=|3+\sqrt { 2 } i| \Rightarrow |(z+t\sqrt { 2 } i)| \cdot |(x+y\sqrt { 2 } i)|=|3+\sqrt { 2 } i| .
  • Using a complex number modulus, we have ( z 2 + 2 t 2 ) ( x 2 + 2 y 2 ) = 11 ( z 2 + 2 t 2 ) ( x 2 + 2 y 2 ) = 11 \sqrt { (z ^ 2 +2 t ^2)\cdot (x^2+2y^2) } =\sqrt {11} \Rightarrow (z ^ 2 +2 t ^2)\cdot (x^2+2y^2) =11
  • Now for using some number theory, if z , x , y , t Z z, x, y, t \in \mathbb{Z} then their squares belong to Z \mathbb{Z } too. So we have, in the last equation, a product of two integers resulting in a integer. There is only one product of integers resulting in 11 11 which is 1 11 1 \cdot 11 .
  • z 2 + 2 t 2 = 1 x 2 + 2 y 2 = 11 z^{ 2 }+2t^{ 2 }=1 \wedge x^{ 2 }+2y^{ 2 } = 11 , and the symmetry of the equations will lead to the same answers, no need of breaking this to two or more cases.
  • Testing for small integers, e.g. 0 , 1 , 2 0,1,2 , equation z 2 + 2 t 2 = 1 z^{ 2 }+2t^{ 2 }=1 leads to t = 0 , z = ± 1 t=0, z= \pm 1 and x 2 + 2 y 2 = 11 x^{ 2 }+2y^{ 2 } = 11 to x = ± 3 , y = ± 1 x=\pm 3, y=\pm 1 .
  • Thus x 2 + y 2 + z 2 + t 2 9 + 1 + 1 + 0 = 11 x^2+y^2+z^2+ t^2 \Rightarrow \boxed { 9+1+1+0=11}

Very nice interpretation!

The first half of the proof could be simplified to the identity:

11 = ( x z 2 y t ) 2 + 2 ( x y + y z ) 2 = ( z 2 + 2 t 2 ) ( x 2 + 2 y 2 ) 11 = ( xz - 2yt) ^2 + 2 ( xy + yz ) ^2 = ( z^2 + 2t^2 ) ( x^2 + 2y^2)

This is actually equivalent to what you're saying. You are using the language of Norm in Z [ 2 ] \mathbb{Z} [ \sqrt{2} ] as the motivator, but I think it's clearer to just be explicit about the identity (since thinking in Z [ 2 ] \mathbb{Z}[\sqrt{-2}] doesn't confer any additional benefits.

Calvin Lin Staff - 4 years, 4 months ago

Log in to reply

I disagree. I think introducing the concept of Z [ i 2 ] \mathbb{Z}[i \sqrt{2}] is very fruitful! Writing u = x + y i 2 u = x+yi\sqrt{2} , v = z + t i 2 v = z + ti\sqrt{2} , the equations read u v = 3 + i 2 uv = 3+i\sqrt{2} ,and so we are trying to factorise 3 + i 2 3+i\sqrt{2} in the Euclidean domain Z [ i 2 ] \mathbb{Z}[i\sqrt{2}] . Since 3 + i 2 3+i\sqrt{2} has prime norm, it is irreducible in Z [ i 2 ] \mathbb{Z}[i\sqrt{2}] , and hence one of u , v u,v is a unit. The only units in Z [ i 2 ] \mathbb{Z}[i\sqrt{2}] are ± 1 \pm 1 , and we are done.

Looked at from this perspective, the question is more Algebra than Number Theory!

Mark Hennings - 4 years, 4 months ago

Log in to reply

Yes, that's the Z [ 2 ] \mathbb{Z}[\sqrt{-2} ] argument, which (I believe) is the motivation behind the problem.

My statement about "be explicit about the identity, and avoid talking about Z [ 2 ] \mathbb{Z}[\sqrt{-2}] stems from the perspective of a problem solver who has not seen Z [ 2 ] \mathbb{Z} [ \sqrt{-2} ] , and so would feel more resistance to understanding these proofs. Because the concepts used are very basic, and their proofs do not require understanding of Z [ 2 ] \mathbb{Z} [\sqrt{-2} ] , in such a case I think it's better to first present the proof and then second show the linkage to this other area of math.

Calvin Lin Staff - 4 years, 4 months ago

Log in to reply

@Calvin Lin Fair enough, but "where did that identity spring from?" and "how am I expected to have thought of it?" become legitimate responses if you just write the identity down!

Mark Hennings - 4 years, 4 months ago
Ivan Koswara
Feb 2, 2017

We work in Z [ 2 ] \mathbb{Z}[\sqrt{-2}] ; that is, numbers in the form a + b 2 a + b \sqrt{-2} . If p = a + b 2 Z [ 2 ] p = a + b \sqrt{-2} \in \mathbb{Z}[\sqrt{-2}] , define p = a 2 + 2 b 2 |p| = a^2 + 2b^2 ; this is the square of the usual norm in complex field. We can prove that p q = p q |pq| = |p| \cdot |q| , and if a , b a, b are integers, then so as a + b 2 |a + b\sqrt{-2}| .

Now, suppose we have x + y 2 x + y \sqrt{-2} and z + t 2 z + t \sqrt{-2} . Their product is ( x z 2 y t ) + ( x t + y z ) 2 (xz - 2yt) + (xt + yz) \sqrt{-2} , or 3 + 2 3 + \sqrt{-2} . Thus we're trying to factorize this number.

Note that 3 + 2 = 3 2 + 2 1 2 = 11 |3 + \sqrt{-2}| = 3^2 + 2 \cdot 1^2 = 11 . If 3 + 2 3 + \sqrt{-2} factors to x + y 2 x + y \sqrt{-2} and z + 2 z + \sqrt{-2} , then 11 must also factor to x + y 2 |x + y \sqrt{-2}| and z + t 2 |z + t \sqrt{-2}| . Thus one of them is 11 and the other is 1. Without loss of generality, assume x + y 2 = 11 |x + y \sqrt{-2}| = 11 and z + t 2 = 1 |z + t \sqrt{-2}| = 1 .

Since z + t 2 = z 2 + 2 t 2 |z + t \sqrt{-2}| = z^2 + 2t^2 , clearly we have t = 0 t = 0 (any other integer t t makes the sum larger than 1) and z 2 = 1 z^2 = 1 . With a little casework, we also find that x 2 = 9 , y 2 = 1 x^2 = 9, y^2 = 1 is the only integer solution to x + y 2 = x 2 + 2 y 2 = 11 |x + y \sqrt{-2}| = x^2 + 2y^2 = 11 . Thus we have x 2 + y 2 + z 2 + t 2 = 9 + 1 + 1 + 0 = 11 x^2 + y^2 + z^2 + t^2 = 9 + 1 + 1 + 0 = \boxed{11} . In other words, this means 3 + 2 3 + \sqrt{-2} is irreducible, because if we try to factor it into a product of two numbers, one of the factors must be 1 or -1, so we don't simplify it at all. (Compare with 4 + 2 = ( 1 + 2 ) ( 2 2 ) 4 + \sqrt{-2} = (1 + \sqrt{-2})(2 - \sqrt{-2}) .)

See my note to Vitor's solution...

Mark Hennings - 4 years, 4 months ago

Log in to reply

Ah, yes, I didn't see that comment.

Ivan Koswara - 4 years, 4 months ago
Jp Morin
May 9, 2017

squaring both equations: { x 2 z 2 4 x y z t + 4 y 2 t 2 = 9 x 2 t 2 + 2 x y z t + y 2 z 2 = 1 \begin{cases}x^2z^2-4xyzt+4y^2t^2&=9\\x^2t^2+2xyzt+y^2z^2&=1\end{cases}

sum up (Eq1)² and ( 2 × 2\times Eq2²): x 2 ( z 2 + 2 t 2 ) + y 2 ( 4 t 2 + 2 z 2 ) = 9 + 2 x^2(z^2+\mathbf 2t^2)+y^2(4t^2+\mathbf 2z^2)=9+\mathbf 2 .

which can be factored as ( x 2 + 2 y 2 ) ( z 2 + 2 t 2 ) = 11 (x^2+2y^2)(z^2+2t^2)=11

Then x 2 , y 2 , z 2 , t 2 x^2,y^2,z^2,t^2 are non negative integers, which yield a factorisation 11 = 1 11 11=1\cdot 11 or 11 = 11 1 11=11\cdot 1 . It is clear now that it suffices to check when x 2 , y 2 , z 2 , t 2 x^2,y^2,z^2,t^2 have values in { 0 2 , 1 2 , 2 2 , 3 2 } \{0^2,1^2,2^2,3^2\} , thus the possible sums of form a 2 + 2 b 2 a^2+2b^2 :

sum a 2 0 1 4 9 b 2 2 b 2 0 2 0 0 1 4 9 1 2 2 2 3 6 11 2 2 8 8 9 \begin{array}{cc|cccc} \text{sum} &a^2&0&1&4&9\\ b^2&2b^2 \\ \hline 0^2&0&0&\mathbf1&4&9\\ 1^2&2&2&3&6&\mathbf{11}\\ 2^2&8&8&9&-&-\\ \end{array}

Therefore, { ( x , y ) , ( z , t ) } = { ( 0 , 1 ) , ( 1 , 3 ) } \{(x,y),(z,t)\}=\{(0,1), (1,3)\} and thus x 2 + y 2 + z 2 + t 2 = 0 2 + 1 2 + 1 2 + 3 2 = 11 x^2+y^2+z^2+t^2=0^2+1^2+1^2+3^2=11 .

This is by far the easiest solution to follow; but there are two minor problems with the last line. The ordered pairs are reversed, and negative values for x,y,z, and t are acceptable.

{(x,y),(z,t)} = {(±1,0),(±3,±1)}

Going back to the original equations, the three non-zero x,y,z,t must be all positive or all negative.

Robert Beggs - 3 years, 11 months ago

Just play around till you come up with the idea to put t = 0 t=0 . Then it is easy but has many solutions. The smallest solution for which all are integers is x = 3 , y = z = 1 x=3,y=z=1 which gives answer 11.

How do you know that t t cannot be (say) 100?

Calvin Lin Staff - 4 years, 3 months ago

You will get (x^2-2y^2)(2t^2-z^2) = -7, since all the variables are integers, it would be easy to find solutions.

Vincentpaul Fadri - 4 years, 2 months ago

i did same wow

Aaditya Prakash - 2 years, 11 months ago
Ashish Gupta
Feb 8, 2017

Good solution

I Gede Arya Raditya Parameswara - 4 years, 3 months ago

What makes you say that "Deom >= Num"? Are you assuming that t t is positive?

More accurately, what we want to do is to consider cases where |Deom| > |Num|, and proceed from there. Unfortunately, I think this gets tricky to rule out in the 2 variable case (but I might be wrong).

Calvin Lin Staff - 4 years, 3 months ago

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...