20
$\begingroup$

Does there exist a positive integral solution $(x, y, n)$ to $(xy+1)(xy+x+2)=n^2$? If there doesn't, how does one prove that?

$\endgroup$
5
  • 2
    $\begingroup$ Do you mean, does there exist an $n \in \mathbb N$ such that...? $\endgroup$ Oct 21, 2018 at 9:22
  • $\begingroup$ @MichaelStoll Yes. Or just simply, does there exist (x, y) such that the expression can become a square number $\endgroup$
    – Benkerd22
    Oct 21, 2018 at 9:54
  • 3
    $\begingroup$ Alternatively, you could state it as: does there exist a positive integral solution (x, y, n) to...? $\endgroup$ Oct 21, 2018 at 21:06
  • $\begingroup$ No solutions below $10^4$. $\endgroup$
    – Wojowu
    Oct 21, 2018 at 21:27
  • 9
    $\begingroup$ Where did this problem originate? $\endgroup$ Oct 22, 2018 at 4:18

3 Answers 3

22
$\begingroup$

It looks that Vieta jumping helps.

For fixed positive integer $y$ choose a minimal positive integer $x$ for which $(xy+1)(xy+x+2)$ is a perfect square.

Denote $4(xy+1)(xy+x+2)=4n^2=(2xy+x+3-z)^2$ for some integer $z=2n-2xy-x-3$, this yields $0<z<x+3$ and rewrites as $z^2-2z(2xy+x+3)+x^2+2x+1=0$. Note that $x$ must divide $z^2-6z+1$, for each $z\leqslant 5$ this gives several variants for $x$ for which it is straightforward to check that $y$ does not appear to be a positive integer. If $z\geqslant 6$, we may replace $x$ to $x'=(z^2-6z+1)/x>0$ (which is another root of the same quadratic equation in $x$.) This contradicts to the minimality since $z^2-6z+1<(z-3)^2<x^2$.

Remark: for the new pair $(x',y)$ we have different value of $z$, as $2x'y+x'+3-z$ becomes negative, but it is still true that $(x'y+1)(x'y+x'+2)$ is a perfect square.

$\endgroup$
2
  • $\begingroup$ Do you mean $z \geq 6$ instead of $z \geq 3$? (The cases $z \in \{3,4,5\}$ are also easily ruled out.) $\endgroup$ Oct 21, 2018 at 22:36
  • $\begingroup$ ops, indeed. Fixed $\endgroup$ Oct 21, 2018 at 22:39
1
$\begingroup$

This is not really an answer, but you can at least observe that $xy+1 \neq xy + x + 2$ if $x$ is positive.

Then consider what kind of common prime factors they could have, since $(xy + 1)(xy + x + 2)$ is a square if and only if the square-free parts of $xy + 1$ and $xy + x + 2$ are the same. Perhaps it is relevant that since their difference is $x + 1$, we know that $\gcd(xy + 1, xy + x + 2)$ divides $x + 1$?

$\endgroup$
1
  • 2
    $\begingroup$ Since $xy+1 = (x+1)y -(y-1),$ you can continue to say that ${\rm gcd}(xy+1,xy+x+2)$ divides ${\rm gcd}(x+1,y-1)$ ( and then note that those two gcd's are in fact equal). $\endgroup$ Oct 21, 2018 at 21:28
0
$\begingroup$

$4n^2=4(xy+1)(xy+x+2)=(3+x+2xy)^2-(1+x)^2$.

For positive even $k$ and $ka^2+b^2=c^2$, when $a,b,c$ is positive integers, and $a=2m,b=km^2-1,c=km^2+1$, when $m$ is any positive integer. Then $n=2m,1+x=4m^2-1,3+x+2xy=4m^2+1$, and then $2m^2=1$, i.e. $m$ is not integer - contradiction. Means equation $n^2=(xy+1)(xy+x+2)$ has no positive integer solution.

$\endgroup$
3
  • 3
    $\begingroup$ So if $ka^2=c^2 - b^2 $, we must have $c-b=2$? But why? $\endgroup$ Mar 13, 2019 at 16:07
  • $\begingroup$ $4\times15^2=900=1156-256=34^2-16^2$ and $34-16\ne2$. $\endgroup$ Mar 14, 2019 at 0:49
  • $\begingroup$ Yes, it is not full parametrization. Answer not true, sorry. $\endgroup$ Mar 14, 2019 at 3:33

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

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