MathOverflow is a question and answer site for professional mathematicians. Join them; it only takes a minute:

Sign up
Here's how it works:
  1. Anybody can ask a question
  2. Anybody can answer
  3. The best answers are voted up and rise to the top

This is elementary, but I have no idea how to solve it for all $s$. Actually I'm not sure if it's true. Let $\mathbb Z_{2^s}$ be the ring of integers modulo $2^s$. Considere the hamilton quaternions over $\mathbb Z_{2^s}$, that is, the ring

$\mathbb H_{2^s}=\mathbb Z_{2^s}[i, j, k]$={$a_0+a_1i+a_2j+a_3k\;\;:\;\;a_l\in\mathbb Z_{2^s}$},

where each element of $\mathbb Z_{2^s}$ commute with $i, j$ and $k$, and $i^2=j^2=k^2=ijk=-1$.

Is $\mathbb H_{2^s}$ a semicommutative ring? that is, for every $z\in\mathbb H_{2^s}$ the right annihilator $ann_r(z)$ is an ideal of $\mathbb H_{2^s}$.

Note that reversible rings (rings where $xy=0$ implies that $yx=0$) are semicommutative. My intuition is that $\mathbb H_{2^s}$ is a reversible ring.

It can help to observe that $\mathbb H_{2^s}$ is a local ring and $\mathbb H_{2^s}/J(\mathbb H_{2^s})\cong\mathbb Z_2 $, where $J$ denotes de jacobson radical. I also have simple proofs that $J(\mathbb H_{2^s})$ is generated by $1+i$, $ 1+j$ and $1+k$ and every element $z$ of $J(\mathbb H_{2^s})$ can be written as $z=2a_0+a_1(1+i)+a_2(1+j)+a_3(1+k)$, where $a_0, a_1, a_2, a_3\in\mathbb Z_{2^s}$.

share|cite|improve this question
up vote 4 down vote accepted

Assume $xy=0$ but $yx\neq 0$. We can take $x$ and $y$ to be $2$-adic quaternions, and then we have $xy=0$ modulo $2^s$ but $yx\neq 0$ modulo $2^s$. We can factor out the highest possible power of $2$ from $x$ and $y$, and then divide $2^s$ by that power, so without loss of generality $x$ and $y$ are nontrivial modulo $2$.

$xy\bar{y}=x|y|^2$. $|y|^2$ is a regular $2$-adic number and is the sum of four squares, at least one odd, and so is nonzero mod $8$. $x$ has a coefficient that is nonzero mod $2$ so that same coefficient in $x|y|^2$ is nonzero mod $8$, so $x|y|^2$ is nonzero mod $8$, so $xy$ is nonzero mod $8$.

Therefore $s$ must be $1$ or $2$. Since a-fortiori computed that reversibility holds for $s=1,2$, it holds for every $s$.

share|cite|improve this answer
The OP is not talking about $\mathbb{Z}_2$ but about $\mathbb{Z}/2^s \mathbb{Z}$. – Qiaochu Yuan Jun 25 '12 at 6:25
thank you Will, but The problem (as stated by Qiaochu) is not only for $\mathbb Z_2$ but for $\mathbb Z_{2^s}$, for a positive integer $s$. For $s=1$ the ring $\mathbb H_2$ is commutative so this case is trivial. – zacarias Jun 25 '12 at 11:34
I am fairly sure that Will is using $\mathbb{Z}_2$ to mean the $2$-adics, not the integers modulo $2$. But, as Qiaochu says, the actual problem is about the integers modulo $2^s$. – David Speyer Jun 25 '12 at 12:48
hello David Speyer, probably you're right, since over the $2$-adics the Hamiltonian quaternions have no zero divisors and thus reversibility is trivial. Probably this can help since $\mathbb H_{2^s}$, for every $s$, is an homomorphic image of the hamiltonian queternions over the $2$-adics. Also the quaternions over the $2$-adics is the inverse limite of $\mathbb H_{2^s}$, $s=1, 2, 3, \ldots$. – zacarias Jun 25 '12 at 13:20
I get that product to be $4j$. In noncommutative rings, it's not true that $a^2-b^2 = (a-b)(a+b)$. – user91132 Jun 25 '12 at 14:48

Your Answer


By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.