8
$\begingroup$

I am stuck in a question about finite fields and would like to ask you for some help.

Given an integer $m\geq 2$ and $p$ a prime number, show that $(\mathbb F_{p}[x]/(x^m))^{\times}$ (the group of all invertible elements of $\mathbb F_{p}[x]/(x^m)$) is not a cyclic group.

I don't really know how to start. I know a lemma which tells me that for every finite field $\mathbb F_{q}$ with $q$ elements, the multiplicative group of the nonzero elements $\mathbb F_{q}^{\times}$ is cyclic. But I don't think that I should use it here...

Does anyone have an idea? Thank you in advance!

$\endgroup$
  • $\begingroup$ Related to this and this. $\endgroup$ – user26857 Oct 5 '13 at 18:47
  • $\begingroup$ Sorry, but i am still new on MSE. I will correct my mistake. Thank for you editing my question! $\endgroup$ – Lullaby Oct 6 '13 at 15:08
2
$\begingroup$

Similar arguments to those I have already used here give the following group isomorphism: $$(\mathbb F_{p}[x]/(x^m))^{\times}\simeq\mathbb F_p^{\times}\times\mathbb F_p^{m-1}.$$ (This shows that you should assume $m\ge 3$ in order to get a non-cyclic group.) For $m\ge 3$ if you assume that the group of units is cyclic, then $\mathbb F_p^{m-1}$ is also cyclic (or $\mathbb F_p\times\mathbb F_p$ if you like), a contradiction.

$\endgroup$
4
$\begingroup$

Let $R = \mathbb{F}_p[X]/(X^m)$, then for any $f(X) = a_0 + a_1X + \ldots + a_nX^n \in \mathbb{F}_p[X]$, let $\overline{f}$ denote its image in $R$. Then $$ \overline{f} \in R^{\ast} \Leftrightarrow (f,X^m) = 1 \Leftrightarrow X\nmid f(X) \Leftrightarrow a_0 \neq 0 $$ Hence, if $\overline{f} \in R^{\ast}$, we may write $$ g(x) = 1 + a_0^{-1}a_1X + a_0^{-1}a_2X^2 + \ldots + a_0^{-1}a_nX^n $$ Define $$ G = \{1 + b_1\overline{X} + b_2\overline{X}^2 + \ldots + b_{m-1}\overline{X}^{m-1}\} $$ Then, $G$ is a multiplicative subgroup of $R^{\ast}$, and $$ \overline{f} \mapsto (a_0, \overline{g}) $$ defines an isomorphism $$ R^{\ast} \cong \mathbb{F}_p^{\ast} \times G $$ In particular, $|R^{\ast}| = (p-1)p^{m-1}$

  1. Note that for any $a \in \mathbb{F}_p^{\ast}$, $a^{p-1} = 1$
  2. For any $\overline{g} \in G$, the Frobenius map gives $$ \overline{g}^p = 1 + a_1\overline{X}^p + a_2\overline{X}^{2p} + \ldots + a_{m-1}\overline{X}^{p(m-1)} $$ But since $p > m$, $X^m \mid X^p$, and hence $$ \overline{g}^p = 1 $$ Hence,

    The order of any element of $R^{\ast}$ is atmost $(p-1)p$.

Since $|R^{\ast}| = (p-1)p^{m-1}$, $R^{\ast}$ cannot be cyclic, provided $m>2$.

Added : If $m=2$, then $|G| = p$, and $G$ is cyclic. We also know that $\mathbb{F}_p^{\ast}$ is cyclic. Since $(p,p-1) =1$, it follows that

$R^{\ast}$ is cyclic if $m=2$

$\endgroup$
  • $\begingroup$ Thank you very much, dear Prahlad! Now i am going to try to understand every step of your proof :) Thanks again! $\endgroup$ – Lullaby Oct 5 '13 at 15:28
  • $\begingroup$ Yes, of course. I was so keen on finding the orders of elements that I didn't think of that! $\endgroup$ – Prahlad Vaidyanathan Oct 5 '13 at 15:58
  • 1
    $\begingroup$ I don't think that $1+a_1 X + \dotsc + a_{m-1} X^{m-1} \mapsto (a_1,\dotsc,a_{m-1})$ (you certainly didn't mean $X^p$, which vanishes) is a group homomorphism (unless $m=2$). This already fails for products $(1+a_1 X)(1+ b_1 X)$. $\endgroup$ – Martin Brandenburg Oct 5 '13 at 16:05
  • $\begingroup$ Hmm. Yes. My brain is refusing to cooperate now. Will just delete that bit, and think about it later. $\endgroup$ – Prahlad Vaidyanathan Oct 5 '13 at 16:08
2
$\begingroup$

A cyclic group has at most one subgroup of each order, to it is enough to show that there exists two invertible elements in $F_p[x]/(x^m)$ which have the same order and generate different subgroups. In particular, in a cyclic group there is at most one element of order two.

If $p$ is odd, there is a nonzero element $\alpha$ in the field which is different to $1$, and the elements $a=1+x^{n-1}$ and $b=1+\alpha x^{n-1}$ are two different units of order two.

If $p$ is even and $m\geq4$, then $a=1+x^{m-2}+x^{m-1}$ and $b=1+x^{m-1}$ are two different elements of order two.

We are left with $p=2$ and $m=3$, which you can do by hand :-)

$\endgroup$
  • 1
    $\begingroup$ (Notice that in the case I did not do we do have a cyclic group) $\endgroup$ – Mariano Suárez-Álvarez Oct 5 '13 at 20:38

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.