Consider the quartic system in four variables $a,b,c,d\in\mathbb R$: $$-(c^2-d^2)(a^2-b^2)=2(ad-bc)(bd+ac).$$

Does this system admit rational solution with $$abcd(c^2-d^2)(a^2-b^2)(a^2-c^2)(b^2-d^2)\neq0?$$

Is there any easy way to compute for rational solution?

  • 5
    $\begingroup$ Nice question. In the $\neq 0$ condition, you can omit the factor $(ad-bc)(bd+ac)$, since this is automatically nonzero if the equation holds and the other factors are nonzero. $\endgroup$
    – GH from MO
    May 24 at 3:33

1 Answer 1


There is no such solution. Let $$ Q(a,b,c,d) = 2(ad-bc)(bd+ac) + (a^2-b^2)(c^2-d^2) $$ be the difference between the two sides of the equation, so we seek to solve $Q(a,b,c,d) = 0$. This is a quadratic equation in each variable, so in a rational solution the discriminant of $Q$ with respect to each variable is a square. We choose $d$ (the others work the same way), and find $$ {\rm disc}_d(Q) = (2c)^2 (2a^4 - 4a^3b + 4ab^3 + 2b^4). $$ Thus either $c=0$ or the second factor is a square. The former is possible (with $d=0$ as well) but the problem statement forbids it. The latter gives rise to an elliptic curve, which turns out to be curve 40.a3 with rank $0$ over the rational numbers; its four torsion points correspond to $a=\pm b$ (each of which gives rise to two points on the curve). Since the problem statement also forbids $a^2=b^2$ we're done.

How did this problem arise?

  • $\begingroup$ Not a very profound reason: I was looking for real solutions to $l a^2+ m c^2=l b^2+ m d^2=2(lab+mcd)$. $\endgroup$
    – Turbo
    May 25 at 3:49

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.