As the title suggests, I'm interested whether $\mathbb{Z}[\mathbb{Z}/(p)]$ a PID or not. Assume $p$ is prime.

My feeling is that it is a PID, since $\mathbb{Z}/(p)$ is cyclic an morally if an ideal is generated by elements of $\mathbb{Z}/(p)$, it's enough to consider the element whose order is the GCD of the orders. But I'm unable to find a (clever) way to turn the general case into something like the easy case above.

  • $\begingroup$ If you mod out the ideal generated by $1-\overline{1}$ then you get a ring isomorphic to $\Bbb{Z}(\zeta_p)$, $\zeta_p=e^{2\pi i/p}$ (the isomorphism maps $\overline{k}$ to $\zeta_p^k$). This is the ring of integers $\Bbb{Q}(\zeta_p)$. That is probably known to have class number $h=1$ for many a choice of $p$, but it's complicated. May even be the case that the answer to that question is not known fully? $\endgroup$ – Jyrki Lahtonen Mar 10 '16 at 17:42
  • $\begingroup$ Äsh. I meant to mod out the ideal generated by the sum $s=1+u+u^2+\cdots+u^{p-1}\in \Bbb{Z}[C_p]$ (or, if you insist on additive notation, $s=\sum_{k=0}^{p-1}1\cdot\overline{k}.$) $\endgroup$ – Jyrki Lahtonen Mar 10 '16 at 18:09
  • $\begingroup$ @JyrkiLahtonen I was about to ask you what was that notation, but I found that relation on some old note of mine and therefore I got what you meant $\endgroup$ – Luigi M Mar 10 '16 at 20:25

$\mathbb Z[C_p]$ is not even a domain.

Take for instance $p=2$. Then there is an element $u\in C_2$ such that $u^2=1$ but $u\ne \pm1$.

More generally, if $G$ has an element $u$ of finite order, then $\mathbb Z[G]$ is not a domain because $(u-1)(u^{n-1}+\cdots+u+1)=u^n-1=0$. In particular, $\mathbb Z[G]$ is never a domain if $G$ is finite.

The problem of when a group ring has non-trivial zero divisors is an open problem; see wikipedia.


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.