A ring $R$ (henceforth assumed to be commutative and with unity $1$) has a zero divisor $a$ iff $a$ is nonzero and there exists a nonzero element $b$ such that $a*b=0$. Also, by definition, a nilpotent element $n$ is a nonzero element of $R$ such that some positive integer power of $n$ is $0$. My question is, is there a ring $R$ such that $R$ has zero divisors but no nilpotent elements?
-
$\begingroup$ What about $\Bbb R[x,y]/(xy)$? $\endgroup$– ajotatxeCommented Aug 5 at 17:14
-
3$\begingroup$ What about $\mathbb Z/6\mathbb Z$? $\endgroup$– J. W. TannerCommented Aug 5 at 17:33
-
$\begingroup$ @J.W.Tanner Yes, a simpler example $\endgroup$– ajotatxeCommented Aug 5 at 17:37
-
3$\begingroup$ To give a wide class of examples, any product of (at least two) reduced rings will satisfy your criteria. J.W. Tanner's nice example is $\mathbb{Z}/2\mathbb{Z} \times \mathbb{Z}/3\mathbb{Z}$ via the Chinese remainder theorem. $\endgroup$– Alex WertheimCommented Aug 5 at 17:44
-
2$\begingroup$ DaRT query $\endgroup$– rschwiebCommented Aug 5 at 20:25
6 Answers
A simple example is the ring $R=\mathbb Z/6\mathbb Z$.
This ring has elements represented by $0$, as well as $1$ and $5$; the latter two are self-inverses.
The rest of the elements are zero divisors, because $2\times3\equiv3\times4\equiv0\pmod6$
but they are not nilpotent because $2^n\equiv\pm2, 3^n\equiv3$, and $4^n\equiv4\pmod6$,
for $n>0$, where the sign in front of $2$ depends on the parity of $n$.
-
2$\begingroup$ Generally only $\,0\,$ is nilpotent in $\,\Bbb Z/q,\,$ i.e. $\,[\,q\mid n^k\Rightarrow q\mid n\,]\iff q\,$ is squarefree. See here for this and many other characterization of "squarefree". $\ \ $ $\endgroup$ Commented Aug 5 at 19:11
Let be $A=\Bbb R[x,y]/(xy)$.
Crearly, $A$ has zero divisors because $xy=0$ in $A$. Let's see that it has no nilpotent elements.
Assume that $P\in\Bbb R[x,y]$ and $P^n\in (xy)$ for some $n\ge 1$. Let
$$P(x,y)=\sum_{j,k} a_{j,k}x^jy^k$$ $$P(x,y)^n=\sum_{j,k}b_{j,k}x^jy^k$$ but $b_{j,k}$ is $0$ whenever $j=0$ or $k=0$.
If $a_{j,k}\neq 0$ and $j=0$, then $b_{0,nk}=a_{0,k}^n\neq 0$, so $P^n\notin(xy)$. Also, if $a_{j,k}\neq 0$ and $k=0$, $P^n\notin(xy)$.
Thus, if $P\notin(xy)$, then $P^n\notin(xy)$.
-
$\begingroup$ Is it not easier to witness $(x)$ and $(y)$ are distinct primes? Hence, their intersection = their product ideal $(xy)$ is radical? $\endgroup$– FShrikeCommented Aug 5 at 18:15
An example coming from analysis would be $R=C(X)$, where $X$ is a compact Hausdorff space with more than one connected component.
For instance $R=C([0,1]\cup[2,3])$. Any function supported on one of the components is a divisor of zero. Explicitly, $f=1_{[0,1]}$ and $g=1_{[2,3]}$ are nonzero elements of $R$ with $fg=0$.
Many good examples already, but one very simple example not yet mentioned: the direct product ring $\mathbb{Z} \times \mathbb{Z}$, or more generally $R \times S$ for any nontrivial nilpotent-free rings $R$, $S$. There are non-trivial zero divisors in the product ring: for instance, $(1,0)\cdot(0,1) = (0,0)$. But there are no non-trivial nilpotents: if $(a,b)^n = 0$, that means $(a^n,b^n) = (0,0)$, so (since $\mathbb{Z}$ is nilpotent-free) $a = b = 0$.
-
1$\begingroup$ +1, a nice class of examples indeed. (But I did in fact actually mention this in the comments. :) ) $\endgroup$ Commented Aug 7 at 3:43
More generally, the ring of sequences such as $\mathbb R^{\mathbb N}$ has zero divisors but no nilpotents. If one "quotients out by all the zero divisors" (in a suitable sense), one can get a proper ordered field extension of $\mathbb R$.
As an example, split-complex numbers. In this ring there are no nilpotents, but there are zero divisors, for instance, $1+j$ and $1-j$: $$(1+j)(1-j)=1^2-j^2=0$$