I'm trying to prove that a local ring $(R,\mathfrak{m})$ contains a field if and only if $\mathrm{char}(R)$ and $\mathrm{char} (R/\mathfrak{m})$ are equal. To this we must relate the characteristic of $R$ to the characteristic of it's residue field $K=R/\mathfrak{m}$. If $\mathrm{char}(K)=0$, then, $\mathrm{char}(R)=0$. So $R$ contains a copy of $\mathbb{Z}$. Now, since $R/\mathfrak{m}$ has characteristic $0$, none of the images of non-zero integers in $R$ can be in $\mathfrak{m}$, else their images in $R/\mathfrak{m}$ will be zero, contradicting $\mathrm{char}(K)=0$. Then by the universal property of localization, $R$ contains $\mathbb{Q}$.

Now for the converse, I know that if $\mathrm{char}(K)=p$, then $\mathrm{char}(R)$ is $0$ a power of $p$, but I don't know how to completely prove this part. Why can't the characteristic be some non-prime integer? And furthermore, why can't $R$ contain a field in any case besides when $\mathrm{char}(R)= p$?

  • $\begingroup$ Am I missing something here? The statement you want to prove seems to be false... What about $\Bbb F_p\subseteq \Bbb F_p[\![x]\!]$? $\endgroup$ – Stahl Sep 22 '18 at 21:00
  • $\begingroup$ @Stahl It might be false. The original thing that the OP wanted to prove some statement behind a link, but the link was dead. See the original post. I tried to "reverse engineer" what the actual question was based on OPs writing and the answer below. Do you have any idea what the actual claim might have been? $\endgroup$ – Mike Pierce Sep 22 '18 at 21:23
  • $\begingroup$ @MikePierce I didn't realize how old this question was, I'm not sure why it popped up in my feed just now! In any case, I think the correct statement is "$R$ contains a field iff the characteristic of $R$ is the same as the characteristic of the residue field," and I've posted a proof of this below. $\endgroup$ – Stahl Sep 22 '18 at 21:50
  • $\begingroup$ @Stahl It popped up because I edited the post. Editing a questions will (sometimes unfortunately) bump the post to the front page. In this case its a good thing that it got bumped and you saw it since I got it wrong, and Arturo's answer wasn't quite right. Thank you for posting an answer! :) $\endgroup$ – Mike Pierce Sep 22 '18 at 22:01

The reason the characteristic of the local ring $R$ will be a prime power if it is not equal to $0$ is that the maximal ideal must contain all zero divisors (consider the image of a zero divisor in $R/\mathfrak{m}$ since the maximal ideal in a local ring contains all nonunits). If the characteristic is $d=ab$, with $\gcd(a,b)=1$, then both $a$ and $b$ must lie in $\mathfrak{m}$, hence $1\in\mathfrak{m}$, which is impossible. Thus, the characteristic must be either $0$ or a prime power.

If $\mathrm{char}(R) = 0$ and $\mathrm{char}(K)=p\gt 0$, then $R$ cannot contain a field: if it contained one, the field would contain $1$, hence $\mathbb{Z}$, hence $\mathbb{Q}$, but then $p$ would be a unit lying in $\mathfrak{m}$, which is impossible.

If $\mathrm{char}(R) = p^m$, $m\gt 0$, and $\mathrm{char}(K)=p$, then $R$ cannot contain a field: the field would necessarily contain $1$ and be of characteristic $p$ (since the characteristic of a subring must divide the characteristic of the ring), but then the subring generated by $1$ would be $\mathbb{Z}/p\mathbb{Z}$; however, the condition that $\mathrm{char}(R)=p^m$ means that the additive order of $1$ in $R$ is $p^m$, which shows $R$ cannot contain a field.

Added. As pointed out in comments, in fact the field wold contain $1$, and hence be of the same characteristic as $R$ (since the characteristic equals the additive order of $1$).

  • $\begingroup$ Thank you, Arturo. I understand all of your answer except the paranthetical comment in the last paragraph - since the characteristic of a subring must divide the characteristic of the ring. Shouldn't the characteristic of any subring of a ring be the same as that of the ring. This would still give us the contradiction we are seeking, since if $R$ contained a field, it will have a characteristic which is $p^m$ with $m>1$ but fields can only have a characteristic which is $0$ or a prime. $\endgroup$ – Brittany Murphy Jun 30 '11 at 11:59
  • $\begingroup$ @Brittany: There is actually a specious argument before (the reason $\mathfrak{m}$ contains all zero divisors is that it contains all non-units). As to your comment: if your subrings must include $1$, then you are correct, as the characteristic of the ring is the additive order of $1$. But if subrings need not contain $1$, the characteristic is the smallest $n$ such that $na=0$ for all $a\in R$, so if you quantify over a subring, the set of possible $n$ is potentially larger, so the characteristic could be smaller (but must divide it anyway); think $Z_3$ in $Z_3\times Z_2\cong Z_6$. $\endgroup$ – Arturo Magidin Jun 30 '11 at 14:37
  • 2
    $\begingroup$ @Arturo Magidin In your second case, we should say that if $m>1$, then $R$ cannot contain a field; but for $m=1$ it's possible, for instance, $R$ could be $\mathbb{Z}/p\mathbb{Z}$. $\endgroup$ – Lao-tzu Jan 29 '17 at 10:48
  • 1
    $\begingroup$ @Lao-tzu Or even worse, if you think fields don't count as a counterexample: $R$ could be something like $\Bbb F_p[\![x]\!].$ $\endgroup$ – Stahl Sep 22 '18 at 21:02

The result you want is that $R$ contains a field $K$ iff $\operatorname{char}(R) = \operatorname{char}(R/\mathfrak{m}).$

$\implies$ If $K\subseteq R,$ then clearly $K$ and $R$ must have the same characteristic, as $1 + 1 + \dots + 1 = 0$ in $K$ if and only if $1 + 1 + \dots + 1 = 0$ in $R.$ Moreover, the composition $K\to R\to R/\mathfrak{m}$ gives a homomorphism of fields $K\to R/\mathfrak{m},$ so that $\operatorname{char}(R/\mathfrak{m}) = \operatorname{char}(K) = \operatorname{char}(R).$

$\impliedby$ Conversely, suppose $\operatorname{char}(R) = \operatorname{char}(R/\mathfrak{m}).$ There are two cases to consider.

  1. $\operatorname{char}(R) = \operatorname{char}(R/\mathfrak{m}) = 0:$ If the characteristic is $0,$ then every integer $n$ is invertible in $R$. Indeed, $n$ is invertible if and only if $n\not\in\mathfrak{m}.$ If $n\in\mathfrak{m},$ then $n = 0$ in $R/\mathfrak{m}.$ But this implies that $\operatorname{char}(R/\mathfrak{m})\mid n.$ Thus, there exists a map $\phi : \Bbb Q = (\mathbb{Z}\setminus\{0\})^{-1}\Bbb Z\to R$ by the universal property of localization, which is necessarily an injection.
  2. $\operatorname{char}(R) = \operatorname{char}(R/\mathfrak{m}) = p:$ In this case, $R$ contains a copy of $\Bbb F_p.$ (Just map $\Bbb F_p\to R$ by $1\mapsto 1,$ and verify that this is a well-defined ring homomorphism and an injection.)

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.