How can one find all primes $(p,q)$ such that $p^{q+1}+q^{p+1}$ is a perfect square

I considered it $\mod 2$ and found a trival solution .

Im curious about an eventual answer Diophantine equations are extremely hard. This seems harder than IMO Q2 of this year .

edit1: I think one should consider it $\mod 4$ .

  • 1
    $\begingroup$ $p=q=2 $ works. $\endgroup$ – vadim123 Jul 17 '15 at 17:33
  • $\begingroup$ Would $(1,0)$ not give $1^1 + 0^2 = 1$ - also a perfect square? Oops, I now see that those are no primes... Sorry - ignore the comment :( $\endgroup$ – johannesvalks Jul 17 '15 at 21:10
  • 1
    $\begingroup$ Full simple solution here: artofproblemsolving.com/community/c6h1156298_moroccan_mo_2014. I disagree it's more difficult than the IMO P2. $\endgroup$ – user236182 Oct 28 '15 at 16:44

Claim: $(p,q)=(2,2)$ is the only pair of prime solutions.


Consider $$ p^{q+1} + q^{p+1} = n^2.$$ With $(p,q)=(2,3),(3,2)$ the LHS equals $43$, a prime number itself.

With $(p,q)=(3,3)$ it equals $2\cdot3^4$, which obviously is not a square as the exponent of $2$ is odd.

With one equal to $2$ and the other $\ge29$, say respectively $p$ and $q$, it is not a square because $$\left(2^{\frac{q+1}{2}}\right)^2<2^{q+1}+q^3<2^{q+1}+2^{\frac{q+3}{2}}+1=\left(2^{\frac{q+1}{2}}+1\right)^2;$$ one can rule out $5\le q < 29$ by trial and error.

Suppose there exist solutions with $p,q\ge5$. They are respectively equal to $6a\pm1$ and $6b\pm1$, for some positive integers $a,b$, so substituting for each of the four cases we have $$\begin{align}(6a+1)^{6b+2}+(6b+1)^{6a+2}&=n^2 \\ \left((6a+1)^{3b+1}\right)^{2}+\left((6b+1)^{3a+1}\right)^{2}&=n^2 \tag{1}\end{align}$$ or $$ \begin{align}(6a-1)^{6b+2}+(6b+1)^{6a}&=n^2 \\ \left((6a-1)^{3b+1}\right)^{2}+\left((6b+1)^{3a}\right)^{2}&=n^2 \tag{2}\end{align}$$ or $$\begin{align} (6a+1)^{6b}+(6b-1)^{6a+2}&=n^2 \\ \left((6a+1)^{3b}\right)^{2}+\left((6b-1)^{3a+1}\right)^{2}&=n^2 \tag{3}\end{align}$$ or $$\begin{align}(6a-1)^{6b}+(6b-1)^{6a}&=n^2 \\ \left((6a-1)^{3b}\right)^{2}+\left((6b-1)^{3a}\right)^{2}&=n^2. \tag{4}\end{align}$$

All of them are pithagorean identities, and as such, one and only one of the numbers squared in their LHS is odd. But each one of them is a power of an odd number, hence an odd number itself. Thus we have a contradiction, and the claim is proved. $ \ \ \ \ \ \ \ \ \ \ \ \text{QED}$

  • 1
    $\begingroup$ You should also check $p=2,q>3$ $\endgroup$ – Empy2 Jul 17 '15 at 18:54
  • $\begingroup$ @Micheal Yeah, thanks. Damn, most importantly I should take coffee when writing an answer. Editing soon. $\endgroup$ – Vincenzo Oliva Jul 17 '15 at 18:58
  • $\begingroup$ @Micheal Check it out, please. $\endgroup$ – Vincenzo Oliva Jul 17 '15 at 19:27
  • $\begingroup$ It looks okay now. $\endgroup$ – Empy2 Jul 17 '15 at 21:10

First of all, note that $n^2\equiv 0,1\pmod 3$ for every $n\in\mathbb Z$.

Let us separate it into cases :

  • For $p\equiv q\equiv 0\pmod 3$, having $p=q=3$ leads $3^4+3^4=162$ which is not a perfect square.

  • For $p\equiv 0,q\not\equiv 0\pmod 3$, having $3^{q+1}+q^{3+1}=m^2$ leads $3^{q+1}=(m-q^2)(m+q^2)$. Thus we have $m-q^2=3^s,m+q^2=3^t$ where $0\le s\le t$, so adding these gives $2m=3^s+3^t$. Here suppose that $1\le s$, then $m\equiv 0\pmod 3$, so $q\equiv 0\pmod 3$ which contradicts $q\not\equiv 0\pmod 3$. Hence, we have $s=0,t=q+1$. Thus, we have $3^{q+1}-2q^2=1$ which has no positive integer solution.

  • For $p\equiv 1,q\not\equiv 0\pmod 3$, $p^{q+1}+q^{p+1}\equiv 2$, so this cannot be a perfect square.

  • For $p\equiv q\equiv 2\pmod 3$, if $p=q=2$, then $2^3+2^3=4^2$. For odd $p,q$, there exist $s,t$ such that $p=6s-1,q=6t-1$. Then, $p^{q+1}+q^{p+1}\equiv (-1)^{6t}+(-1)^{6s}\equiv 2\pmod 3$ which cannot be a perfect square.

Hence, $p=q=2$ is the only solution.


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.