3
$\begingroup$

Let $k$ be a perfect field of characteristic $p \neq 2,3$ such that $\omega := \sqrt[3]{1} \in k$, where $\omega \neq 1$. Consider an absolutely irreducible (not necessarily homogenous) quadratic polynomial $Q \in k[s_1, s_2]$ in two variables $s_1, s_2$. Under what conditions is the polynomial $R(t_1,t_2) := Q(t_1^3, t_2^3)$ (of degree $6$) absolutely irreducible (or at least irreducible over $k$) ?

Thank you in advance.

$\endgroup$
2
  • $\begingroup$ What is $Q'$? I assume the prime doesn't mean a derivative, since $Q$ has two variables. Maybe it should just be $Q(t_1.t_2)$? $\endgroup$ Feb 22, 2021 at 18:25
  • $\begingroup$ You are right. I changed the notation. $\endgroup$ Feb 22, 2021 at 18:30

1 Answer 1

1
$\begingroup$

The following argument might help reduce the problem to one in elimination of variables, which can be solved using a computer algebra system.

Write $S = k[s_1,s_2]$, $T = k[t_1, t_2]$ and $\phi : S \rightarrow T$, $s_i \mapsto t_i^3$. The ramification locus of $\phi$ is defined by $t_1t_2$. Since $Q(s_1, s_2)$ is irreducible and of degree $2$, it does not ramify in $T$; hence $R(t_1, t_2) = Q(t_1^3, t_2^3)$ is a product of distinct irreducible polynomials in $T$. Since the extension $k(s_1,s_2) \subset k(t_1,t_2)$ of fraction fields is Galois with Galois group $G := \mathbb{Z}/3 \oplus \mathbb{Z}/3$ and $S$ and $T$ are integrally closed, the going-down theorem for integral extensions applies. Hence $G$ acts transitively on the irreducible factors of $R(t_1, t_2)$. By degree considerations, the stabilizer of an irreducible factor of $R(t_1, t_2)$ cannot be $\langle 1 \rangle$. If the stabilizer is $G$, then $R(t_1, t_2)$ is irreducible.

Therefore assume that the stabilizer is isomorphic to $\mathbb{Z}/3$, and that $\sigma \in G$ generates the quotient group. Again, by degree considerations, the irreducible factors of $R(t_1, t_2)$ are of degree two.

Write $Q(s_1,s_2) = as_1^2 + bs_1s_2 + cs_2^2 + ds_1 + es_2 + f$. Let $f(t_1, t_2) = a_1t_1^2 + b_1t_1t_2 + c_1t_2^2 + d_1t_1 + e_1t_2 + f_1$ be an irreducible factor of $R(t_1, t_2) = Q(t_1^3, t_2^3)$. As an example, suppose that $\sigma (t_1) = \omega t_1$ and $\sigma(t_2) = t_2$. Since $$ R(t_1, t_2) = f(t_1, t_2)\cdot \sigma(f(t_1, t_2)) \cdot \sigma(\sigma(f(t_1, t_2)) = f(t_1,t_2) \cdot f(\omega t_1,t_2) \cdot f(\omega^2 t_1,t_2) $$ we will get six polynomials of the form $a-g_1(a_1, \ldots, f_1), b-g_2(a_1, \ldots, f_1), \ldots$, equating the coefficients of the monomials in $t_1,t_2$. Now a computer algebra system can be used to eliminate the variables $a_1, \ldots, f_1$, and get the relations between $a, \ldots, f$. Let $Z$ be the algebraic set defined by these relations. Now impose the condition that there exists $\tau \in G$ such that $\tau$ fixes $f(t_1, t_2)\cdot \sigma(f(t_1, t_2)) \cdot \sigma(\sigma(f(t_1, t_2))$. Remove the corresponding points from $Z$. This will give a set of $Q(s_1, s_2)$ for which $R(t_1, t_2)$ is irreducible. (One might have to repeat this for other possible choices of the stabilizer of $R(t_1, t_2)$, to get the set of all such $Q(s_1, s_2)$.)

$\endgroup$
1
  • $\begingroup$ Thank you for the help! $\endgroup$ Mar 2, 2021 at 9:13

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.