I do not manage to exhib a non commutative ring of division with a discrete valuation. Can anyone show me one? Examples with quaternions would be a plus !!

Thanks in advance.


I’m sure this quaternion algebra is well known, and I apologize for presenting it in a probably nonstandard way.

Consider $\Bbb Q_p$, the $p$-adic integers, and its (unique) unramified quadratic extension $U$, definable as the field of $(p^2-1)$-th roots of unity over $\Bbb Q_p$. Consider the two-dimensional $U$-vector space $E$ with basis $\{1,\pi\}$, where $\pi^2=p$. Now consider the non-commuting relation $\pi u=u^\sigma\pi$ for any $u\in U$, where $\sigma$ is the Frobenius automorphism of $U$, sending the $(p^2-1)$-th root of unity $\zeta$ to $\zeta^p$. Now you show that $E$ is a four-dimensional vector space over $\Bbb Q_p$, it’s the central division algebra of rank four and of “invariant” $\frac12$ over $\Bbb Q_2$. I do leave it to you to show that it’s a division algebra, and that the expected valuation $\upsilon_p$ is discrete.

The “expected valuation” is, for $z\in U$, $\upsilon_E(z)=\frac12\upsilon_p\mathbf N(z)$, where $\mathbf N$ is the reduced norm from $E$ to $\Bbb Q_p$. If you don’t know about the reduced norm, use the “unreduced norm” of $z$, $\mathbf N'(z)$, which is the determinant of $z$ in the regular representation of $E$ as a four-dimensional $\Bbb Q_p$-space, and then we have $\upsilon_E(z)=\frac14\upsilon_p\mathbf N'(z)$.

By the way, I’m pretty sure you can extend the $2$-adic valuation to the familiar rational quaternion algebra $\Bbb Q(i,j,k)$ in just this way as well.

  • $\begingroup$ Thank you. I have to think to your comment. I will try to prove all your assertions. $\endgroup$ – joaopa Dec 25 '18 at 9:47

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.