Let $R$ be a commutative ring with $1$ and $e$ be an idempotent element of $R$ with the property that if $e=x+y$ (where $x, y\in R$), then there exists $r\in R$ such that either $e=rx$ or $e=ry$. Can we deduced that for such an idempotent, the ideal $\langle e\rangle$ is a minimal ideal of $R$?

  • $\begingroup$ By a minimal ideal of $R$, do you mean a minimal non-zero ideal of $R$? Note that if $R = (\mathbb{Z}/4\mathbb{Z})^2$, then $e = (1, 0)$ satisfies your requirements. $\endgroup$ – Luc Guyot Jul 2 at 20:16

If you take $e=1$, then the condition imposed on $e$ translates to: for every $x \in R,$ either $x$ or $1-x$ is a unit. This will be satisfied in any local ring $R$, while $R=(e)$ will not be minimal unless $R$ is a field. This already shows that the conclusion does not hold in general.

(If $e=1$ seems too trivial, this example should survive replacing the local ring $R$ by any ring of the form $R\times S$ and $e=1$ by $(1, 0)$).

On the other hand, if the Jacobson radical $J(R)$ of $R$ is $0$ and a nonzero idempotent $e$ has the imposed property, then $(e)$ is indeed minimal:

One needs to check that for $0 \neq x$ divisible by $e$, i.e. $ex=x$, one has $e$ divisible by $x$. In other words, if $x=ex$ is an element not dividing $e$, then $x=0$. The strategy is to show that $x$ lies in this case in the Jacobson radical, hence is $0$.

The imposed condition, together with the fact that $x$ does not divide $e$, implies that $(e-x)=(e-ex)=e(1-x)$ divides $e$. In particular, $(1-x)$ divides $e$, and $e$ divides $x$, so $(1-x)$ divides $x$. It also divides $(1-x)$, so altogether $1-x$ divides $1=(1-x)+x$. Thus, $(1-x)$ is a unit.

To finish the argument, note that $x$ in the above argument can be replaced by any multiple $x'=rx$. So we obtain that for all $r,$ $1-rx$ is a unit, i.e. $x \in J(R)=0$.


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.