0
$\begingroup$

I tried decomposing the polynomial, but after taking out $(x^{2}+1)$ you have to break the remainder into polynomials of degree 4, which is manually hard. Perhaps this is solved by using Frobenius automorphism, I have never been able to find a way to use it

$\endgroup$
2
  • $\begingroup$ It’s $(x^2 + 1) (x^8 + 1) (x^{16} - x^{14} + x^{12} - x^{10} + 1)$ $\endgroup$ Commented May 12 at 4:49
  • $\begingroup$ Yes, I know that, but I am not at all clear what to do with a 16th degree polynomial. n general, if we show that this polynomial decomposes into some polynomials of degree 4, it is a victory, because $x^{8}+1$ is the product of irreducible polynomials of degree 4, so their decomposition field is GF(81). But decomposition of degree 16 is problematic $\endgroup$
    – mackenzie
    Commented May 12 at 4:54

1 Answer 1

6
$\begingroup$

If $p(x) = x^{26}+x^8+x^2+1$, then $q(x) = xp(x) = x^{27}+x^9+x^3+x$ is invariant under the Frobenius $\sigma(t)=t^3$ (good instincts Alvaro!) for every $x\in GF(81)$, since \begin{multline*} \sigma(q(x)) = \sigma(x^{27}+x^9+x^3+x) = \sigma(x^{27})+\sigma(x^9)+\sigma(x^3)+\sigma(x) \\ = x^{81}+x^{27}+x^9+x^3 = x+x^{27}+x^9+x^3 = q(x), \end{multline*} where we have used the fact that every $x\in GF(81)$ is a root of $x^{81}-x$. In particular, $q(x)\in GF(3)$ (the fixed field of $\sigma$) for all $x\in GF(81)$. But each of $q(x)=0$ and $q(x)=1$ and $q(x)=2$ can have at most $27$ solutions due to the degree of $q$; it follows that each of them has exactly $27$ solutions. In particular, $q(x) = xp(x)$ has $27$ roots in $GF(81)$ and thus splits completely, which implies that $p(x)$ itself splits completely in $GF(81)$.

$\endgroup$
1
  • 1
    $\begingroup$ This is surely the way to go. Your $q(x)$ is the trace map $tr: GF(81)\to GF(3)$ (the range verified by the same calculation). As a linear combination of iterates of the Frobenius it is linear over the prime field. The image is 1-dimensional, so by rank-nullity the kernel is 3-dimensional, hence it has $27$ zeros. All of the above is just commentary, nothing new needs to be added to the excellent answer $\endgroup$ Commented May 12 at 7:57

You must log in to answer this question.

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