I have to find all real solutions of the following equation:

$x^n + y^n = (x+y)^n$

Clearly for $n = 1$, the equation holds for every $x,y$ real numbers.

If $n$ is greater or equal to $2$, we do binomial expansion on RHS and from $x^n + y^n - (x+y)^n = 0 $ it follows that the roots are either $x=0$ or $y=0$.

Am I missing out on something or these are the only solutions?

  • 1
    $\begingroup$ What if $x=-y$, and $n$ is odd? $\endgroup$ – Workaholic Jan 28 '15 at 21:15
  • $\begingroup$ Sorry, I forgot to mention that $n$ is a natural number. $\endgroup$ – Lior Jan 28 '15 at 21:19
  • 1
    $\begingroup$ We need to deal with the cases when one of $x$ or $y$ is positive and the other negative. $\endgroup$ – André Nicolas Jan 28 '15 at 21:21
  • $\begingroup$ just taking $n=3$, it has many more solutions, like $(-1,1)$. You cannot conclude from the binomial that $x=0$ or $y=0$ is your error $\endgroup$ – user45150 Jan 28 '15 at 21:23
  • $\begingroup$ I understand what you mean but how can we find roots if $x < 0$ and $y>0$. @user45150 Now I covered that case as well. $\endgroup$ – Lior Jan 28 '15 at 21:24

First, if $x=0$ or $y=0$ then the equation is satisfied.

Second, if $x>0$ and $y>0$ then because by binomial expansion $$ (x+y)^n = x^n + y^n + \text{strictly positive terms}$$ the equation can never be satisfied. On the other hand, if $x<0$ and $y<0$ then the equation is equivalent to $(-x)^n+(-y)^n=(-x-y)^n$ where all terms are positive so this has no solutions by the previous case.

So we are left with the case where, without loss of generality $x>0>y$, or if we use $-y$ instead of $y$ as a variable, the case $$ x^n + (-y)^n = (x-y)^n, \text{ where } x,y>0.$$ If $n$ is even then all terms are positive and clearly $x^n + (-y)^n = x^n+y^n > (x-y)^n = |x-y|^n$, because either $x>|x-y|$ or $y>|x-y|$. If $n$ is odd then, depending on the sign of $x-y$ the equation becomes $$ x^n + (y-x)^n = y^n \text{ or } y^n + (x-y)^n = x^n,$$ with all three terms positive. This is itself of the form $X^n+Y^n = (X+Y)^n$ so the only solutions are given by $y=x$ (which means $x=-y$ in the original equation), $x=0$ and $y=0$.

  • $\begingroup$ $x=y=-1$ and $n=1$, then $-2$ is definitely not positive. If $n=3$ then you get terms of different signs as well. $\endgroup$ – Asaf Karagila Jan 28 '15 at 22:23
  • 1
    $\begingroup$ @Asaf If $x=y=-1$ then $-x-y=2$ is positive... $\endgroup$ – Myself Jan 28 '15 at 23:07
  • $\begingroup$ Uhm, no, $(-x)+(-y)=(-1)+(-1)=-2$. $\endgroup$ – Asaf Karagila Jan 28 '15 at 23:08
  • 1
    $\begingroup$ @Asaf $x=-1\implies -x = -(-1) = 1$? $\endgroup$ – Myself Jan 28 '15 at 23:09
  • 1
    $\begingroup$ @Asaf good thing you didn't make that sound derogatory at all... Just bear with me for 5 seconds: if $x^n+y^n=z^n$ then $(-1)^nz^n=(-1)^nx^n+(-1)^ny^n$ so $(-x)^n+(-y)^n=(-z)^n$. Now put $z=x+y$ and observe that $-(x+y)=-x-y$ and obtain $(-x)^n+(-y)^n=(-x-y)^n$. Finally denote $X=-x$ and $Y=-y$ and obtain $X^n+Y^n=(X+Y)^n$. $\endgroup$ – Myself Jan 28 '15 at 23:25

You can do this purely geometrically, without much algebra. Let's get rid of $n=1$ case since every $x,y$ is a solution then.

Note that since both sides are of degree $n$, if $(a,b)$ is a solution then so is $(\lambda a, \lambda b)$ for any $\lambda \in \mathbb{R}$.

We know that $(0,y), (x,0)$ are solutions, so now assume that $x,y\neq 0$. Then by what we noted above, WLOG assume $y=1$, then we get $x^n +1 = (x+1)^n$. Now, draw two graphs of $x^n+1$ and $(x+1)^n$ and you see that: (i) when $n$ is even, the two graphs meet only at $(0,1)$, so we don't get any newer solutions; (ii) when $n$ is odd and $n>1$, the two graphs meet at $(0,1)$ and $(-1, 0)$, meaning that $x=-1$ is an additional solution.

In conclusion, we have that $x=0$, $y=0$, and if $n$ is odd additionally $(-a, a)$ as solutions.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.