34
$\begingroup$

Prove that there are no positive integer solutions to $$x^2+1=y^3$$ This problem is easy if you apply Catalans conjecture and still doable talking about Gaussian integers and UFD's. However, can this problem be solved using pre-university mathematics?

I am talking about elementary number theoretical solutions. Do they exist?

$\endgroup$
  • 1
    $\begingroup$ @Alizter If ab is a square and a and b are coprime, then both are squares. $\endgroup$ – Milo Brandt Oct 2 '14 at 21:21
  • $\begingroup$ Related: math.stackexchange.com/questions/5914/… $\endgroup$ – Gamma Function Oct 2 '14 at 23:20
  • $\begingroup$ I think the answer is "YES" — see my answer/proof, below. $\endgroup$ – Kieren MacMillan Nov 9 '14 at 20:25
  • $\begingroup$ @MathGod I never said that it was not elementary? $\endgroup$ – Ali Caglayan Nov 20 '14 at 22:15
  • 2
    $\begingroup$ I just found a second, far more elegant, elementary solution. I think I'm going to write this one up and submit it for publication. Thanks for the inspiration! $\endgroup$ – Kieren MacMillan Feb 24 '16 at 20:49
14
+100
$\begingroup$

Lemma. If $a$ and $b$ are integers such that $3a^4+3a^2+1=b^2\!$, then $a=0$ and $b=\pm 1$.

Proof. We may, without loss of generality, assume that $a$ and $b$ are non-negative. Assume, contrary to the lemma, that $a \ne 0$ and $b > 1$ are integers satisfying the equation. Evidently $b$ is odd, say $b=2c+1$ for an integer $c \ge 1$. Hence \begin{align*} 3a^4 + 3a^2+1 &= (2c+1)^2 \\ 3a^2(a^2+1) &= 4c(c+1). \end{align*} Since $4 \nmid 3(a^2+1)$, regardless of the parity of $a$, we must have $2 \mid a^2$; hence $a$ is even, say $a=2d$ for an integer $d \ge 1$. Now \begin{align*} 3d^2(4d^2+1) &= c(c+1). \end{align*} We now show that this equation has no solutions with $c,d \ge 1$. Assume to the contrary that there exist integers $p,q,r,s \ge 1$ such that \begin{align*} 3d^2 &= pq, &&& c &= pr, \\ 4d^2+1 &= rs, &&& c+1 &= qs. \end{align*} Since $\gcd(p,q) \mid \gcd(pr,qs) = \gcd(c,c+1)=1$ implies $\gcd(p,q)=1$, we may write $d=uv$ with $\gcd(u,v)=1$ and consider the two possible cases.

Case 1: $p=u^2$ and $q=3v^2$. Hence $c=pr = u^2r$, and by substitution $c+1=u^2r+1 = 3v^2s$. On the other hand, $rs-1 = 4d^2=4u^2v^2$, and adding these two relations yields \begin{align*} (rs-1)+(u^2r+1) &= 4u^2v^2 + 3v^2s \\ r(s+u^2) &= v^2(4u^2+3s). \end{align*} Since $d=uv$, the equation $rs = 4d^2+1$ forces $\gcd(v,r)=\gcd(s,u)=1$. Hence $v^2 \mid (u^2+s)$, and $\gcd(s+u^2,4u^2+3s) = \gcd(s+u^2,3(s+u^2)+u^2)=\gcd(s+u^2,u^2) = \gcd(s,u^2)=1$. Thus we conclude $r=4u^2+3s$ and $v^2=s+u^2$. Substituting now gives \begin{equation*} r = 4u^2+3s = 4u^2+3(v^2-u^2) = u^2+3v^2. \end{equation*} Multiplying, $4u^2v^2+1=rs=(u^2+3v^2)(v^2-u^2)$, which implies $u^4+1=3v^2(v^2-2u^2)$. But $3 \nmid u$ (because $3 \mid q$ and $\gcd(p,q)=1)$, so $u^4+1 \equiv 2\!\pmod{3}$, a contradiction.

Case 2: $p=3u^2$ and $q=v^2$. Now $c+1=pr+1=qs$ gives $3u^2r+1=v^2s$. We have $rs-1=4d^2=4u^2v^2$, and adding the two relations yields \begin{align*} (rs-1) + (3u^2r+1) &= 4u^2v^2 + v^2s \\ r(s+3u^2) &= v^2(4u^2+s). \end{align*} As before, considering common factors leads to the conclusion $s+3u^2=v^2$ and \begin{equation*} r=4u^2+s =4u^2+(v^2-3u^2) = u^2+v^2. \end{equation*} Multiplying, $4u^2v^2+1 = rs = (u^2+v^2)(v^2-3u^2)$, which is $3u^2(u^2+2v^2)=(v^2-1)(v^2+1)$. Since $3$ can never divide $v^2+1$, we deduce $v^2-1=3u_1^2w_1$ and $v^2+1=u_2^2w_2$ where $u=u_1u_2$ and $u^2+2v^2=w_1w_2$. As $4 \nmid (v^2+1)$, we deduce that $u_2$ is odd, $w_2 \equiv 2\!\pmod{4}$, and $\gcd(u_2,w_1)=1$. Adding yields $2v^2 = (v^2-1)+(v^2+1) = 3u_1^2w_1 + u_2^2w_2$, and by substitution \begin{align*} w_1w_2 &= u^2+2v^2 \\ &= u_1^2u_2^2+ (3u_1^2w_1 + u_2^2w_2) \\ w_2(w_1-u_2^2) &= u_1^2(u_2^2+ 3w_1). \end{align*} Since $\gcd(w_1-u_2^2,u_2^2+3w_1)$ divides the sum $4w_1$, and $\gcd(w_1,u_2)=1$ and $u_2$ is odd, we deduce $w_1-u_2^2=u_1^2$. Now $w_2 = u_2^2+3w_1 = u_2^2+3(u_2^2+u_1^2) = (2u_2)^2+3u_1^2 \equiv 0\text{ or }3\!\pmod{4}$, contradicting $w_2 \equiv 2\!\pmod{4}$ and proving the lemma.

Theorem. The equation $X^2 +1 = Y^3$ has only one integer solution, namely $(x,y)=(0,1)$.

Proof. Assume $x$ and $y$ are integers such that $x^2+1 = y^3$. Considering the equation modulo $3$, we quickly deduce that $$x^2 = (3z+1)^3−1 = 9z(3z^2+3z+1)$$ for some integer $z$. Since $\gcd(z,3z^2+3z+1)=1$, both $z$ and $3z^2+3z+1$ must be squares. Write $z=a^2$ for an integer $a$, and hence for some integer $b$ we have $b^2 = 3z^2+3z+1 = 3a^4+3a^2+1$. Now the lemma forces $a=0$, so that $z=a^2=0$, and $y=3z+1=1$, as claimed.

$\endgroup$
  • 1
    $\begingroup$ +1 Nicely done. The main thing to prove the lemma was choosing factors $p$, $q$, $r$ and $s$. My approach to the equation $3a^4+3a^2+1=b^2$ was showing contradiction through infinite descent and Pell-Fermat equation but I was stuck for several days and couldn't get it. I'm glad that this question has finally been solved using elementary methods :) $\endgroup$ – MathGod Nov 18 '14 at 11:12
  • 1
    $\begingroup$ @MathGod: I think a proof by descent would be more enlightening than mine, and would probably lead to a more general method applicable to more equations than this ad hoc method. We should keep looking for that! $\endgroup$ – Kieren MacMillan Nov 18 '14 at 11:57
  • $\begingroup$ There is a small gap, which will be easy to fill (and the fix might actually give us a hint towards a proof by descent) — I'll post an updated/corrected answer soon. $\endgroup$ – Kieren MacMillan Nov 21 '14 at 15:18
  • $\begingroup$ Well, what's the flaw in this? $\endgroup$ – MathGod Nov 21 '14 at 15:33
  • $\begingroup$ It's true that $\gcd(p+s,p) = 1$, but that doesn't imply $\gcd(p,d)=1$ in a single step. That implication needs to be justified. $\endgroup$ – Kieren MacMillan Nov 21 '14 at 20:34

protected by Community Oct 20 '14 at 15:57

Thank you for your interest in this question. Because it has attracted low-quality or spam answers that had to be removed, posting an answer now requires 10 reputation on this site (the association bonus does not count).

Would you like to answer one of these unanswered questions instead?

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