7
$\begingroup$

In field theory, the following fact is used in the construction of splitting fields: Given a field $F$ and an irreducible polynomial $f \in F[x]$, the quotient $F[\alpha]/(f(\alpha))$ is a field extension of $F$ which contains a root of $f$ (namely the congruence class of $\alpha$).

Let $n$ be a positive integer and let $\lambda_{1} + \dotsb + \lambda_{k} = n$ be a partition of $n$. Does there exist a field $F$ and an irreducible polynomial $f \in F[x]$ of degree $\deg f = n+1$ such that, if we define $K := F[\alpha]/(f(\alpha))$, the factorization of $f$ in $K[x]$ is of the form $f = (x-\alpha) \cdot f_{1} \dotsb f_{k}$ where each $f_{i} \in K[x]$ is irreducible and $\deg f_{i} = \lambda_{i}$?

$\endgroup$

1 Answer 1

15
$\begingroup$

Let us show that for the partition $2+1+1=4$, there is no such $f$.

If $f$ were inseparable, then over $K$ it would factor as a constant times $(x-\alpha)^5$. If $f$ were separable, its Galois group $G$ would be a transitive subgroup of $S_5$ containing a transposition, so $G=S_5$, but then the stabilizer of a point would act transitively on the other four points, so $f$ would factor over $K$ into $x-\alpha$ and an irreducible polynomial of degree $4$.

$\endgroup$
2
  • 5
    $\begingroup$ You could replace $S_5$ by $S_p$ for each prime $p\geq 5$ and the partition having one $2$ and all other terms equal to $1$. $\endgroup$
    – KConrad
    Commented Apr 9, 2022 at 4:46
  • 1
    $\begingroup$ For $n \geq 7$ there is different partition of $n$ that can't occur for separable $f$. By Bertrand's postulate, there is a prime $p$ s.t. $(n+1)/2 < p \leq n-2$. Use the partition $p + 2 + 1 + \cdots + 1 = n$; it has $k := n-p$ terms. The Galois group $G$ of $f$ over $F$ has a permutation $\sigma$ of the $n+1$ roots of type $(p,2,1,\ldots,1)$. Then $\sigma^2$ is a $p$-cycle on the roots and $\sigma^p$ is a $2$-cycle on the roots. A transitive subgroup of $S_{n+1}$ having a $2$-cycle and $p$-cycle is $S_{n+1}$, so Bjorn's argument shows $f/(x-\alpha)$ is irred. of degree $n$: contradiction. $\endgroup$
    – KConrad
    Commented Apr 9, 2022 at 20:18

You must log in to answer this question.

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