[Math] Let $I$ be a prime ideal. Show that $f^{-1}(I)$ is a prime ideal of $R$. Is this also true for maximal ideals

abstract-algebraidealsring-theory

I'm solving some exercises to prepare for my ring theory exam.

Let $f:R→R'$ be a ring homomorphism, with $f(1)=1$, and $R,R'$
commutative rings with $1$. Let $I$ be a prime ideal. Show that
$f^{-1}(I)$ is a prime ideal of $R$. Is this also true for maximal
ideals? Prove or disprove.

I was able to prove this for prime ideals. But I don't know how to prove anything for maximal ideals. I think it is not true. But I find it hard to think of a counterexample.

Best Answer

Here is a possibly different way to prove the result for prime ideals, and it will help us determine a sufficient condition for the inverse image of a maximal ideal to be maximal.

Let $f:A\to B$ be any commutative ring homomorphism, and let $\mathfrak b$ be any ideal of $B$. Then $\mathfrak a=f^{-1}\mathfrak b$ is an ideal of $A$ such that $f$ determines an injection $$\varphi:A/\mathfrak a\to B/\mathfrak b$$ This is basically the lattice theorem from group theory. Now, an ideal $\mathfrak q\subset B$ is prime if and only if $B/\mathfrak q$ is an integral domain. Let $\mathfrak p=f^{-1}\mathfrak q$ and note that $A/\mathfrak p$ embeds $B/\mathfrak q$. Since a subring of an integral domain is in turn an integral domain, $\mathfrak p$ is necessarily prime.

Now, an ideal $\mathfrak n\subset B$ is maximal if and only if $B/\mathfrak n$ is a field. Moreover, if we set $\mathfrak m$ to be the preimage of $\mathfrak n$, $A/\mathfrak m$ embeds the field $B/\mathfrak n$. So far so good. But the problem is that subrings of fields are not necessarily fields; this is the only reason why we cannot draw the same conclusion with maximal ideals as we did with prime ideals. However, it can be seen that if $f$ is surjective, then the embedding $\varphi$ is surjective--hence, $\varphi$ is an isomorphism--and $A/\mathfrak m$ is a field.