4
$\begingroup$

Let $a,b,c$ be poistive integers,and such $\gcd(a,b)=\gcd(b,c)=\gcd(a,c)=1$,fine the all $a,b,c$ such $$a^2+3b^2c^2=7^c$$

I'm not sure that this question has been studied, but I've been trying for a long time$(a,b,c\le 100)$, and there's only one set of solutions:$(a,b,c)=(2,1,1)$,But I can't prove it. I may need your help. Thank you

$\endgroup$
3
  • 1
    $\begingroup$ Have you taken the first step of considering a useful ring and factoring both sides? $\endgroup$
    – user44191
    Commented Nov 17, 2020 at 6:51
  • $\begingroup$ You can get gcd(b,c)=2 with $a,b,c=-137986556241619682249090303, 9885082693522654065041378, 64$ $\endgroup$
    – joro
    Commented Nov 17, 2020 at 11:32
  • 1
    $\begingroup$ Where does this problem come from, if I may ask? $\endgroup$ Commented Nov 17, 2020 at 14:17

1 Answer 1

7
$\begingroup$

We work in $\mathbb{Z}[\omega]$ where $\omega=\frac{1+i\sqrt{3}}2$. It is a factorial ring, and we factorize both sides as $(a+i\sqrt{3}bc)(a-i\sqrt{3}bc)=(2+i\sqrt{3})^n(2-i\sqrt{3})^n$. Since $2+i\sqrt{3},2-i\sqrt{3}$ are prime (and coprime), the guy $a+i\sqrt{3}bc$ can not be divisible by both (otherwise it is divisible by $(2+i\sqrt{3})(2-i\sqrt{3})=7$ that contradicts to our assumption that $\gcd(a,b)=\gcd(a,c)=1$), we see that $a+i\sqrt{3}bc=\varepsilon(2\pm i\sqrt{3})^c$ where $\varepsilon$ is a unit in $\mathbb{Z}[\omega]$. If $\varepsilon\ne \pm 1$, we get $\varepsilon=\frac{\pm 1\pm i\sqrt{3}}2$ and therefore $$\varepsilon(2\pm i\sqrt{3})^c= \left(\varepsilon (\pm i\sqrt{3})^n+(\text{something in } \mathbb{Z}[i\sqrt{3}])\right)\notin \mathbb{Z}[i\sqrt{3}],$$ a contradiction.

So we get $\varepsilon=\pm 1$ and $a+i\sqrt{3}bc=\pm (2\pm i\sqrt{3})^c$. Consider two cases.

  1. $c$ is even, $c=2^\alpha \beta$ for odd $\beta$. Then $$2i\sqrt{3}bc=\pm\left((2+i\sqrt{3})^{c}-(2-i\sqrt{3})^{c}\right)$$ is divisible by $$ (2+i\sqrt{3})^{2^\alpha}-(2-i\sqrt{3})^{2^\alpha}=2i\sqrt{3}\cdot 4\\ \cdot((2+i\sqrt{3})^{2}+(2-i\sqrt{3})^{2}) ((2+i\sqrt{3})^{4}+(2-i\sqrt{3})^{4})\ldots ((2+i\sqrt{3})^{2^{\alpha-1}}+(2-i\sqrt{3})^{2^{\alpha-1}}) $$ which is divisible by $2i\sqrt{3}\cdot 2^{\alpha+1}$, thsu $b$ is also even, a contradiction with $\gcd(b,c)=1$.

  2. 3 divides $c$. Since $(2+i\sqrt{3})^3-(2-i\sqrt{3})^3=18i\sqrt{3}$, we analogously get that 3 divides $b$ (the guys $A^2+AB+B^2$ for $A,B=(2\pm i\sqrt{3})^{3^t}$ are divisible by 3.)

  3. $c$ is odd and not divisible by 3. Let $3<p_1<p_2<\ldots<p_k$ be distinct prime divisors of $c$ (over $\mathbb{Z}$). Then $p_1$ divides (in $\mathbb{Z}[\omega]$) the number $(2+i\sqrt{3})^c-(2-i\sqrt{3})^c=\pm 2i\sqrt{3}bc$. In particular $p_1\ne 7=(2+i\sqrt{3})(2-i\sqrt{3})$.

3.1) If $p_1$ is prime in $\mathbb{Z}[\omega]$, then by Lagrange theorem it divides $(2 \pm i\sqrt{3})^{p_1^2-1}-1$, thus it divides $(2 + i\sqrt{3})^{p_1^2-1}-(2 - i\sqrt{3})^{p_1^2-1}$. Therefore $p_1$ divides $(2 + i\sqrt{3})^{\gcd(p_1^2-1,c)}-(2 - i\sqrt{3})^{\gcd(p_1^2-1,c)}=2i\sqrt{3}$, a contradiction. (We used that $p_1^2-1=4\cdot \frac{p_1-1}2\cdot \frac{p_1+1}2$ has prime divisors less than $p_1$, thus is coprime with $c$.)

3.2) If $p_1=q\bar{q}$ for a prime $q\in \mathbb{Z}[\omega]$, then by Lagrange theorem $q$ divides $(2 + i\sqrt{3})^{p_1-1}-(2 - i\sqrt{3})^{p_1-1}$ and we analogously get a contradiction.

$\endgroup$
8
  • $\begingroup$ Ooh, the proofs in the second half are surprisingly pretty (I was working through the even case, and my proof there depended on the fact that $c \choose i$ was divisible by $2^m$ for $i$ odd, and I didn't have an answer for the odd case). I think it might be worth pointing out the exact reason why $gcd(p_1^2 - 1, c) = 1$ and writing the conclusion that $c$ has trivial prime factorization explicitly; a great answer! $\endgroup$
    – user44191
    Commented Nov 17, 2020 at 10:55
  • $\begingroup$ On a second look - I think you may need to address $c$ divisible by $3$ in the same way you addressed $c$ divisible by $2$, as your contradiction in 2.2 doesn't work when $p_1 = 3$. It should be solvable in much the same way as the even case; $3$ will divide $(x^{3^\alpha})^2 + (x^{3^\alpha})(y^{3^\alpha}) + (y^{3^\alpha})^2$ if it divides $x - y$, and so $9$ will divide $x^3 - y^3$, etc. $\endgroup$
    – user44191
    Commented Nov 17, 2020 at 11:31
  • $\begingroup$ You can get gcd(b,c)=2 with a,b,c=−137986556241619682249090303,9885082693522654065041378,64 $\endgroup$
    – joro
    Commented Nov 17, 2020 at 11:34
  • 1
    $\begingroup$ @user173628192 $x^{|G|}=e$ for every element $x$ of a finite group $G$, where $e$ is the unit element $\endgroup$ Commented Oct 28, 2021 at 9:48
  • 1
    $\begingroup$ @user914367 $2+i\sqrt{3}$ is an element, coprime residues modulo $p_1$ in $\mathbb{Z}[e^{2 \pi i/3}]$ are a multiplicative group $\endgroup$ Commented Dec 14, 2021 at 15:10

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .