I'm exploring rings that have the following property:


For all $x \in R$, and for all $n \in \mathbb{N}$, there exists $y \in R$ such that $\sum_{i=1}^{n} x = y \cdot x$. $\rule{10cm}{0.4pt}$

Any ring with unity satisfies this, $y = \sum_{i=1}^{n} 1$. Any Boolean ring also satisfies this, $y = \sum_{i=1}^{n} x$.

My general questions are:

(i) Are there non-Boolean rings without unity that have this property? Or better: are there any rings with elements of infinite order without unity that have this property? (ii) Is this property interesting? (iii) What if we require that $y$ be unique?

Thank you


1) yes, any ring with local right identities has this property. That is, if for every $x$, there exists $e$ such that $xe=x$. Such an example is $\prod _{i=1}^\infty \mathbb Z$.

Actually if you think about it, this is equivalent to your condition, since $\sum_{i=1}^n=y$ does the trick.

2) rings with local identities are interesting.

3) Let me share a bit about what I puzzled out about what happens if you additionally required right identities to be unique.

The first thing you notice is that all the right identities are idempotent:

$x=xe=xe^2\implies e=e^2$

So each idempotent is its own unique right identity.

But $0\cdot e=0$ for every idempotent, so $e=0$, if $0$ has a unique right identity. Unfortunately this means that $R$ is the ring $\{0\}$. So with the uniqueness assumption on all right identities, everything collapses.

Perhaps it can be saved by supposing that just nonzero elements have a unique right identity?

Supposing that, let $e$ be a nonzero idempotent and $f$ be an idempotent such that $ef=e$, it must be that $f=e$, because $e$ is its own unique right identity.

For any element $r$ and idempotent $e$, $e+re-ere$ is an idempotent, and in particular if $r=x$ and $xe=x$, you have $f=e+x-ex$ is an idempotent. Furthermore, $ef=e$. If $e$ is nonzero, it must be that $e=e+x-ex$ and then we see that $x=ex$ also, so the right identity is actually a two-sided identity.

Another thing, suppose that $e$ and $f$ are two idempotents that commute. Then $ef$ is an idempotent such that $(ef)e=ef$ and $(ef)f=ef$, therefore either $ef=0$ or else $e=ef=f$. The upshot is that if two different idempotents commute, they have to be orthogonal. In a commutative ring with identity, it would mean there are no nontrivial idempotents.

  • $\begingroup$ Very clear, thank you. $\endgroup$ – Jonathan Hebert Jan 11 '18 at 3:11
  • $\begingroup$ @JonathanHebert Added a few of the thoughts that I had which I couldn't enter with my phone. $\endgroup$ – rschwieb Jan 11 '18 at 15:01

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.