For how many integers $n$ the expression $$(3n)^4-(n-10)^4$$ becomes a perfect square?

$$$$One way is: Let $x=3n$ and $y=n-10$ to get $$x^4-y^4=z^2.$$ This equation does not have non-trivial solutions in integers (I do not want to discuss about this).

I was trying to solve this problem with simpler elementary methods. For example we can write the equation as $$40(n+5)(2n-5)(n^2-2n+10)=m^2.$$ Observe that $40 \mid m^2$. Thus $20 \mid m$. Write $m=20k$ to get $$(n+5)(2n-5)(n^2-2n+10)=10k^2,$$ Where $k$ is a non-negative integer. Look mod $5$. We get $n^3(n-2) \equiv 0 \pmod{5}$. So we have either $n \equiv 0 \pmod{5}$ or $n \equiv 2 \pmod{5}$.

If $n \equiv 0 \pmod{5}$ then let $n=5N$ to get $$125(N+1)(2N-1)(5N^2-2N+2)=10k^2$$ It follows that $25 \mid k^2$ and $k=5K$. Now we have the following equation $$(N+1)(2N-1)(5N^2-2N+2)=2K^2$$

Is this getting simpler or harder this way? Do you have a suggestion?


2 Answers 2


$$(3n)^4-(n-10)^4=m^2\Rightarrow\begin{cases} 9n^2=x^2+y^2\\n^2-20n+100=x^2-y^2\text{ or}\space 2xy\end{cases}$$ WLG we can consider $n,x,y$ without common factors then $n$ cannot be even because the well known parametrization of Pythagorean triples we used above.

$$n\text{ odd}\Rightarrow 10n^2-20n+100=2x^2\Rightarrow 5n^2-10n+50=x^2$$ Then $$5n^2-10n+50=25x_1^2\Rightarrow n^2-2n+10=5x_1^2\iff(n-1)^2+3^2=5x_1^2$$ This implies $5x_1^2=25$ which is absurde.


Using $$\gcd(N+1,2N-1)=\gcd(N+1,-3)\mid 3$$ $$\gcd(N+1,5N^2-2N+2)=\gcd(N+1,-7N+2)=\gcd(N+1,9)\mid 9$$ $$\gcd(2N-1,5N^2-2N+2)=\gcd(2N-1,5)\mid 5 $$ we see that the factors are almost coprime, hence must almost (i.e., up to small factors involving at most a $2$, a $3$, and/or a $5$) be square. Then from $\frac{2N-1}{N+1}\approx 2$ you might arrive at a contradiction.

  • $\begingroup$ How $\frac{2N-1}{N+1}\approx 2$ gives a contradiction? $\endgroup$
    – Ghartal
    Jun 29, 2017 at 12:01
  • $\begingroup$ @Ghartal If $2N-1=a^2$ and $N+1=b^2$ are coprime squares, for example, we arrive at $\frac ab\approx \sqrt 2$; with good estimates for the goodness of this approximation, it might turn out to be too good to be true. I didn't check this in detail though $\endgroup$ Jun 30, 2017 at 12:22

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.