Let $l$ be an odd prime number and $\zeta$ be a primitive $l$-th root of unity in $\mathbb{C}$. Let $K = \mathbb{Q}(\zeta)$. Let $A$ be the ring of algebraic integers in $K$. Let $k$ be a rational integer not divisible by $l$. How would you prove that $(1 - \zeta^k)/(1 - \zeta)$ is a unit of $A$?

This is a related question.

  • $\begingroup$ I tried to compute the norm of it. $\endgroup$ – Makoto Kato Jul 25 '12 at 1:40
  • $\begingroup$ That is doable but unnecessary. It will be easier to try to write the inverse as an element of $A$. $\endgroup$ – Qiaochu Yuan Jul 25 '12 at 1:45
  • $\begingroup$ @MakotoKato Can you show that $ (1-\zeta^k)/(1-\zeta) \in A $ ? $\endgroup$ – Ragib Zaman Jul 25 '12 at 1:47
  • $\begingroup$ @QiaochuYuan I think N($1 - \zeta^k$) = N($1 - \zeta$) is clear. $\endgroup$ – Makoto Kato Jul 25 '12 at 1:56

Just to verify that this is indeed an element of $A$, and also because it hints at a way of showing that your element is a unit in that ring: a very old identity tells you that \[ \frac{1 - \zeta^k}{1 - \zeta} = 1 + \zeta + \dots + \zeta^{k - 1}. \] Since $k$ is prime to $l$, you can find an integer $k'$ such that $kk' \equiv 1 \bmod l$. Then writing \[ \frac{1 - \zeta}{1 - \zeta^k} = \frac{1 - \zeta^{kk'}}{1 - \zeta^k} \] should help quite a bit. These are called, somewhat confusingly, the cyclotomic units. A book on Iwasawa theory should talk about them quite a bit.

| cite | improve this answer | |
  • $\begingroup$ I think I learned this from a book [Washington, maybe?] — I definitely am not clever enough to have discovered it independently. $\endgroup$ – Dylan Moreland Jul 25 '12 at 5:14

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.