Cubic Diophantine Family

There are infinitely many integer solutions ( x , y , z ) (x,y,z) to the equation x 3 + 2 y 3 + 4 z 3 6 x y z = 1. x^3+2y^3+4z^3-6xyz = 1. There is a unique solution such that x , y , z x,y,z are positive and 500 x 4000 500 \le x \le 4000 . Find x x .


The answer is 1081.

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.

2 solutions

Patrick Corn
Feb 8, 2016

The left side of this equation looks very similar to the well-known identity a 3 + b 3 + c 3 3 a b c = ( a + b + c ) ( a 2 + b 2 + c 2 a b b c a c ) = ( a + b + c ) ( a + b ω + c ω 2 ) ( a + b ω 2 + c ω ) \begin{aligned} a^3+b^3+c^3-3abc &= (a+b+c)(a^2+b^2+c^2-ab-bc-ac) \\ &= (a+b+c)(a+b\omega+c\omega^2)(a+b\omega^2+c\omega) \end{aligned} where ω \omega is a complex primitive cube root of unity, and substituting a = x , b = y 2 3 , c = z 4 3 a = x, b = y \sqrt[3]{2}, c = z\sqrt[3]{4} turns the equation into ( x + y 2 3 + z 4 3 ) ( x + y ω 2 3 + z ω 2 4 3 ) ( x + y ω 2 2 3 + z ω 4 3 ) = 1 (x+y\sqrt[3]{2}+z\sqrt[3]{4})(x+y\omega\sqrt[3]{2}+z\omega^2\sqrt[3]{4})(x+y\omega^2\sqrt[3]{2}+z\omega\sqrt[3]{4}) = 1 This is recognizable as the product of the conjugates of the element α = x + y 2 3 + z 4 3 \alpha = x+y\sqrt[3]{2} + z\sqrt[3]{4} , also known as the norm of α \alpha . So this equation just says N ( α ) = 1 N(\alpha) = 1 .

The point is that norm is multiplicative: N ( α β ) = N ( α ) N ( β ) N(\alpha\beta) = N(\alpha)N(\beta) . (I suppose this is not exactly elementary, but it reduces to a certain extremely tedious algebraic identity that could be proved directly.)

Now note that ( 1 , 1 , 1 ) (1,1,1) is a solution, i.e. α = 1 + 2 3 + 4 3 \alpha = 1+\sqrt[3]{2} + \sqrt[3]{4} has norm 1 1 . Then so does any power of α \alpha . Then α 6 = 1081 + 858 2 3 + 681 4 3 \alpha^6 = 1081 + 858\sqrt[3]{2} + 681 \sqrt[3]{4} gives the x x -value we are looking for. So the answer is 1081 \fbox{1081} .

Notes:

(1) It is in fact also true that every solution comes from a power of α \alpha , but this requires some deep algebraic number theory.

(2) Another way to say all this is: if ( x , y , z ) (x,y,z) is a solution to the original equation, then so is ( x + 2 y + 2 z , x + y + 2 z , x + y + z ) (x+2y+2z,x+y+2z,x+y+z) . (And the inverse of this linear transformation: ( 2 z x , x y , y z ) (2z-x,x-y,y-z) .) This can be checked completely elementarily, but the motivation is that this corresponds to multiplying x + y 2 3 + z 4 3 x+y \sqrt[3]{2} + z\sqrt[3]{4} by the "fundamental unit" 1 + 2 3 + 4 3 1+\sqrt[3]{2} + \sqrt[3]{4} . You can repeatedly apply this linear transformation in Excel to compute α 6 \alpha^6 quite quickly.

If you are going to establish the statements in your problem, you would need to get even more "fancy". Certainly we are looking for units in Z [ 2 3 ] \mathbb{Z}[\sqrt[3]{2}] , and α = 1 + 2 3 + 4 3 \alpha = 1 + \sqrt[3]{2} + \sqrt[3]{4} is a unit, so that α n \alpha^n is a unit for any n Z n \in \mathbb{Z} . The solution to the question, coming from n = 6 n=6 , is the only solution with x x in the desired range of this type. If you are going to establish your claim that this is the only possible value of x x , then you are going to have to show that α \alpha is a fundamental unit in Z [ 2 3 ] \mathbb{Z}[\sqrt[3]{2}] , so that we know that all units are of this form.

Incidentally, if we drop the restriction that all three integers be positive, there are several more solutions, such as x = 521 x= 521 , y = 62 y = - 62 , z = 279 z = - 279 , coming from α 10 \alpha^{-10} .

Mark Hennings - 5 years, 4 months ago

Log in to reply

Agreed on all counts, as I indicated in note (1) to my solution.

Patrick Corn - 5 years, 4 months ago

looks pretty fancy to me ;)

Otto Bretscher - 5 years, 4 months ago
Otto Bretscher
Feb 8, 2016

This problem can be solved with some pretty fancy number theory: It turns out that the integer solutions x , y , z x,y,z satisfy an equation ( 1 + 2 3 + 4 3 ) n = x + y 2 3 + z 4 3 (1+\sqrt[3]{2}+\sqrt[3]{4})^n=x+y\sqrt[3]{2}+z\sqrt[3]{4} for some integer n n . For n = 6 n=6 we find the solution ( x , y , z ) = ( 1081 , 858 , 681 ) (x,y,z)=(1081,858,681) , the only solution with x x on the required interval. (We have x = 281 x=281 for n = 5 n=5 and x = 4159 x=4159 for n = 7 n=7 .)

Sir how did you thought of ( 1 + 2 3 + 4 3 ) n (1+\sqrt[3]{2}+\sqrt[3]{4})^{n}

Department 8 - 5 years, 4 months ago

Log in to reply

This is a Diophantine equation that is well understood; it is sometimes called the cubic analogue to Pell's equation . This one is particularly simple since x = y = z = 1 x=y=z=1 happens to be a solution... my formula is based on that observation.

Otto Bretscher - 5 years, 4 months ago

Sure, showing those are the only solutions is fancy. But it's not quite as fancy to generate those solutions given the equation. Care to elaborate on where it comes from?

Patrick Corn - 5 years, 4 months ago

Log in to reply

I don't see an elementary way to do this; I will wait for you or others to explain.

Otto Bretscher - 5 years, 4 months ago

Log in to reply

Yeah, "not quite as fancy" \ne "elementary." Let me know if my solution is helpful/illuminating.

Patrick Corn - 5 years, 4 months ago

Log in to reply

@Patrick Corn Yes, this is a great summary; thanks! I particularly like the linear transformation approach. That way we can find the answer quite quickly "by hand", even without Excel.

Otto Bretscher - 5 years, 4 months ago

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...