2

I believe it is, because I did a similar problem for $\mathbb{Z}+\mathbb{Z}\omega$ where $\omega=e^{2\pi i/3}$ is 3rd root of unity and proved the element $1-\omega$ is prime.

I did this by showing $\mathbb{Z}+\mathbb{Z}\omega$ is Euclidean with respect to $N(a+b\omega)=(a+b\omega)(a+b\omega^2)=a^2+b^2-ab$, then showing $1-\omega$ is irreducible using properties of the norm.

I don't see how I would be able to do this for the 23rd root of unity since I don't know what the norm should be. It was already pretty tedious for then 3rd root of unity so I imagine this line of thought would be even more tedious for the 23rd root of unity.

Any help is greatly appreciated!

Shean
  • 877
  • 3
    Nope. $\mathbb Q(\omega)$ contains $\mathbb Q(\sqrt{-23})$ as a subfield. And $2$ is not prime in $\mathbb Z[\frac12(1+\sqrt{-23})]$. The norm of $2$ will be $2^{22}$. – Mathmo123 Feb 18 '22 at 13:37
  • 3
    $23$ divides $2^{11}-1$ which implies that $x^{23}-1$ splits completely in $\Bbb{F}_{2^{11}}$. If $2$ was prime then $\Bbb{Z}[\omega]/(2)$ would be the field with $2^{22}$ elements. – reuns Feb 18 '22 at 13:38
  • Ok I see, so to prove it's not prime I guess I should use a counter example. Obviously in $\mathbb{Q}(\frac{1+\sqrt{-23}}{2})$ I would use $2|(1-\sqrt{-23})(1+\sqrt{-23})=24$ but what would those elements be in $\mathbb{Q}(\omega)$? – Shean Feb 18 '22 at 13:50
  • @EwanDelanoy $\omega$ is a root of the polynomial $f_{23}(x)=x^{22}+x^{21}+...+x+1$ so $\omega^{22} = -(\omega^{21}+...+\omega+1)$ – Shean Feb 21 '22 at 11:43
  • @MinecraftPlayer69 Got it, thanks. – Ewan Delanoy Feb 21 '22 at 13:23
  • 1
    This field is important historically. – Jyrki Lahtonen Feb 21 '22 at 17:54

2 Answers2

2

All the important ideas have already been expressed in the comments, but let me flesh it out in a full answer.

Let $p=23$, let $\omega=e^{\frac{2\pi i}{p}}$ and let $R$ be the ring $R={\mathbb Z}[\omega]$ (this is simpler to write than your sum!). The square residues modulo $p$ are $S=\lbrace 1, 2, 3, 4, 6, 8, 9, 12, 13, 16, 18 \rbrace$.

By the theory of quadratic Gauss sums, we know that

$$ \sqrt{-p}= \sum_{j=1}^{p-1} \big(\frac{j}{p}\big)\omega ^ j = 1+2\sum_{s\in S} \omega^s $$

and hence, if we put $x=\frac{1-\sqrt{-p}}{2}, y = \frac{1+\sqrt{-p}}{2}$, we have

$$ x=-\sum_{s\in S} \omega^s \in R, y=1+\sum_{s\in S} \omega^s \in R. $$

Now, $xy=\frac{1+p}{4}=6$ is divisible by $2$ in $R$, but neither $x$ nor $y$ is divisible by $2$ in $R$. So $2$ is not a prime in $R$.

Ewan Delanoy
  • 61,600
  • Nice proof, I haven't learned quadratic Gauss sums yet but they seem very useful. I think I have come up with an elementary proof, could you check it? – Shean Feb 21 '22 at 19:14
0

Let $R=\mathbb{Z}+\mathbb{Z}\omega+\mathbb{Z}\omega^2+...+\mathbb{Z}\omega^{21}$

Let $$\alpha=1+\omega^2+\omega^4+\omega^5+\omega^6+\omega^{10}+\omega^{11}\in R$$ $$\beta=1+\omega+\omega^5+\omega^6+\omega^7+\omega^9+\omega^{11}\in R$$ then, $$\alpha\beta=1+\omega+\omega^5+\omega^6+\omega^7+\omega^9+\omega^{11}+\\+\omega^2+\omega^3+\omega^7+\omega^8+\omega^9+\omega^{11}+\omega^{13}+\\+\omega^4+\omega^5+\omega^9+\omega^{10}+\omega^{11}+\omega^{13}+\omega^{15}+\\+\omega^5+\omega^6+\omega^{10}+\omega^{11}+\omega^{12}+\omega^{14}+\omega^{16}+\\+\omega^6+\omega^7+\omega^{11}+\omega^{12}+\omega^{13}+\omega^{15}+\omega^{17}+\\+\omega^{10}+\omega^{11}+\omega^{15}+\omega^{16}+\omega^{17}+\omega^{19}+\omega^{21}+\\+\omega^{11}+\omega^{12}+\omega^{16}+\omega^{17}+\omega^{18}+\omega^{20}+\omega^{22}\\$$ Using the fact that $\omega^{22}+\omega^{21}+\omega^{20}+...+\omega^2+\omega+1=0$ we get, $$\alpha\beta=2(\omega^5+\omega^6+\omega^7+\omega^9+\omega^{10}+3\omega^{11}+\omega^{12}+\omega^{13}+\omega^{15}+\omega^{16}+\omega^{17})$$ Thus $2\mid(\alpha\beta)$ but $2\nmid\alpha$ because if it did, then, $$\alpha=1+\omega^2+\omega^4+\omega^5+\omega^6+\omega^{10}+\omega^{11}=2z$$ for some $z\in R\implies\exists n_0,n_1,n_2,...,n_{21}\in\mathbb{Z}$ such that, $$1+\omega^2+\omega^4+\omega^5+\omega^6+\omega^{10}+\omega^{11}=2(n_0+n_1\omega + n_2\omega^2+...+n_{21}\omega^{21})$$ Thus $n_0=1/2$ which is a contradiction. Thus, $2\nmid\alpha$. Similarly, $2\nmid\beta$.

Therefore, $2$ is not prime in $R$

Shean
  • 877