Cycling powers

Algebra Level 4

Find the number of ordered triples of distinct positive real numbers ( x , y , z ) (x,y,z) that satisfy x y = y z = z x x^y=y^z=z^x

Clarification: counting "ordered triples" means that, if more than one permutation of the triple ( a , b , c ) (a,b,c) gave a solution, you would count each permutation separately.

6 6 Infinitely many More than 6 6 , but finitely many 3 3 0 0 1 1

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

Chris Lewis
Sep 22, 2020

We'll prove there are no such triples.

Let y = u x y=ux . Then from x y = z x x^y=z^x , we find x u x = z x x u = z \begin{aligned} x^{ux} &= z^x \\ x^u &= z \end{aligned}

Now from x y = y z x^y=y^z , x u x = ( u x ) x u u x log x = x u log ( u x ) u log x = x u 1 log ( u x ) u log x = x u 1 log u + x u 1 log x log x = x u 1 log u u x u 1 \begin{aligned} x^{ux} &= (ux)^{x^u} \\ ux \log x &= x^u \log (ux) \\ u \log x &= x^{u-1} \log (ux) \\ u \log x &= x^{u-1} \log u + x^{u-1} \log x \\ \log x &= \frac{x^{u-1} \log u}{u-x^{u-1}} \end{aligned}

Let g ( x ) = x u 1 log u u x u 1 g(x)=\frac{x^{u-1} \log u}{u-x^{u-1}} . Differentiating, we have g ( x ) = u ( u 1 ) x u log u ( x u u x ) 2 g'(x)=\frac{u(u-1)x^u \log u }{\left(x^u-ux\right)^2}

There are two cases to consider: u > 1 u>1 or u < 1 u<1 . Let's start with...


Case u > 1 u>1 :

In this case, each of the quantities that multiply to make the numerator of g ( x ) g'(x) is positive: we have u > 0 , u 1 > 0 , log u > 0 , x u > 0 u>0,u-1>0,\log u>0,x^u>0 . The denominator is a square, so also positive; hence g ( x ) g'(x) is positive everywhere.

g ( x ) g(x) has a vertical asymptote when u x u 1 = 0 u -x^{u-1}=0 - that is at x = u u 1 x=\sqrt[u-1]{u} . Call this critical value c c . Note that c > 1 c>1 .

When x < c x<c , g ( x ) > 0 g(x)>0 . When x > c x>c , g ( x ) < 0 g(x)<0 .

Now let's look at the equation. When 0 < x < 1 0<x<1 , log x < 0 \log x<0 and g ( x ) > 0 g(x)>0 . So there are no solutions in this region.

When c < x c<x , log x > 0 \log x>0 and g ( x ) < 0 g(x)<0 . So there are no solutions here either.

The only place we might find a solution then is when 1 < x < c 1<x<c . Taking logs of this, we have 0 < log x < log c = log u u 1 0<\log x<\log c = \frac{\log u}{u-1} in this interval.

But g ( 1 ) = log u u 1 g(1)=\frac{\log u}{u-1} , and g ( x ) g(x) is increasing; so there are no solutions for 1 < x < c 1<x<c either.

Therefore there are no solutions when u > 1 u>1 .


Case u < 1 u<1 :

If ( x , y , z ) = ( a , b , c ) (x,y,z)=(a,b,c) is a solution, then so are both ( b , c , a ) (b,c,a) and ( c , a , b ) (c,a,b) .

In this case, we have a > b a>b . Now, either b < c b<c , or b > c b>c and c < a c<a .

If b < c b<c , then the triple ( x , y , z ) = ( b , c , a ) (x,y,z)=(b,c,a) satisfies x < y x<y , and we have a u > 1 u>1 case.

If b > c b>c , then c < a c<a and the triple ( x , y , z ) = ( c , a , b ) (x,y,z)=(c,a,b) satisfies x < y x<y , and again we have a u > 1 u>1 case.


In conclusion, there are no solutions in positive reals to the original equation.

Surely there is an easier way!! I'd be interested in alternative solutions or even ideas for a better way to get the result.

I'm also interested in what happens if we remove the restriction to positive reals - are there complex solutions?

Chris Lewis - 8 months, 3 weeks ago

Wowzers that was a long read but an understandable solution in the end! Good explanation dude!

Aatmoshru Goswami - 8 months, 2 weeks ago

Log in to reply

Thanks! As I said, it feels like there must be an easier way; the related problem of solving x y = y x x^y=y^x is much simpler, so I'm sure I'm missing something.

Chris Lewis - 8 months, 2 weeks ago
Kushal Dey
Dec 2, 2020

Let x^y=y^z=z^x=e^k(sup) =>ln(x)=k/y, ln(y)=k/z, ln(z)=k/x. On eliminating y and z we get, ln(ln(x))=ln(k)-ke^(-k/x). Now observe that this equation must have 3 real roots(for correspondence to 3 variables x,y,z in the question). However both the functions are monotonic in nature for x>0, thus only 1 solution is possible which does not serve our purpose for 3 solution

Great solution! Really elegant way to avoid the case bashing I went through.

Chris Lewis - 6 months, 1 week ago

Thanks chris

Kushal Dey - 6 months, 1 week ago

I must have mentioned the function on lhs is monotonic increasing and that of rhs is monotonic decreasing to be more exact

Kushal Dey - 6 months, 1 week ago

We may also generalise that for the system, a1^a2=a2^a3=...=an^a1 has solution only if n=2

Kushal Dey - 6 months, 1 week ago

0 pending reports

×

Problem Loading...

Note Loading...

Set Loading...