For a local field $E$, denote by $U(E)$ the units of the corresponding valuation ring $\mathcal{O}_E$, and denote by $U_n(E)$ the prinicipal $n$-units, i.e. $U_n(E)=1+M_E^n$ where $M_E$ is the maximal ideal of $\mathcal{O}_E$.

Now let $L/K$ be a finite unramified extension of local fields, of characteristic $0$, say. Then we have the norm map $N_{L/K}:U(L)\rightarrow U(K)$.

The Question: is there a simple description of $N_{L/K}^{-1}(U_n(K))$?

I know it has to contain $U_n(L)$, and by Local Class Field Theory, I know that $[U(L):N_{L/K}^{-1}(U_n(K))]=(q-1)q^{n-1}$, where $q$ is the size of the residue field of $K$. It is also not hard to find which roots of unity of $L$ are contained in $N_{L/K}^{-1}(U_n(K))$. Apart from that, I haven't been able to say much. I thought that maybe there is some simple description of this group, given that this is the unramified case, but couldn't find a reference.

Thank you!


Let $\pi$ be a uniformizer for $K$ and $L$.

For sufficiently large $m$, the $\log$ and $\exp$ maps give isomorphism between $U_m(L)$ and $\pi^m \mathcal{O}_L$. Same for $K$. This $m$ should be thought as a fixed "small" integer.

These maps commute with Galois action, i.e. for any $\sigma \in Gal(\overline{K}/K)$, viewed as a map $\sigma: L \rightarrow L^\sigma$, we have: $\log(x^\sigma) = (\log(x))^\sigma$.

Consequently, for the norm map $N=N_{L/K}$, we have: $\log(N(x)) = Tr(\log(x))$, where $Tr=Tr_{L/K}$ is the trace map.

So the question is essentially: what is the inverse image of $\pi^n \mathcal{O}_K$ in $\pi^m \mathcal{O}_L$ under the trace map?

Example: $L = K[\sqrt{d}]$ is the unramified quadratic extension. Then the thing we are looking for is the set $\{a + b\sqrt{d}: a \in \pi^n\mathcal{O}_K, b \in \pi^m\mathcal{O}_K\}$. It is an $\mathcal{O}_K$ submodule, but not an $\mathcal{O}_L$ submodule.

I'm not sure if this answers your question, since "a simple description" is hard to define. It then all depends on what kind of particular question you have in mind.

  • $\begingroup$ Interesting. And do you know of a simple/canonical description of this inverse image of the trace map? Perhaps one that does not require a choice of a basis. $\endgroup$ – Gal Porat Jul 6 '17 at 10:11
  • $\begingroup$ The general case is similar to the example above, you just replace $\mathcal{O}_K \sqrt{d}$ by the trace zero hyperplane. No choice of basis is needed on that hyperplane. $\endgroup$ – WhatsUp Jul 6 '17 at 11:51

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.