Take the 2-minute tour ×
Mathematics Stack Exchange is a question and answer site for people studying math at any level and professionals in related fields. It's 100% free, no registration required.

I'm having some difficulty with this homework problem:

Let $A, B$ be reduced finitely generated $\mathbb{C}$-algebras, and $\psi : A \longrightarrow B$ a $\mathbb{C}$-algebra homomorphism. Let $\mathcal{M} \subset B$ be a maximal ideal. Show that $\psi^{-1}(\mathcal{M})$ is also a maximal ideal.

Here's what I have so far. The preimage of a prime ideal under a homomorphism is a prime ideal, so we know $\psi^{-1}(\mathcal{M})$ is a prime ideal. Also, if the homomorphism is surjective, then the image of an ideal is an ideal; but unfortunately we don't know $\psi$ is surjective.

So suppose that there is an ideal $\mathcal{J} \subset A$ such that $\mathcal{J} \supsetneq \psi^{-1}(\mathcal{M})$. Then $\psi(\mathcal{J}) \supsetneq \mathcal{M}$. But $\psi(\mathcal{J})$ is not necessarily an ideal, so consider the ideal it generates (which I'll denote $\langle\psi(\mathcal{J})\rangle$ - not sure if this notation is standard). This ideal contains $\mathcal{M}$, and since $\mathcal{M}$ is maximal we have $\langle\psi(\mathcal{J})\rangle = B$. Therefore any $f \in B$ can be written as a linear combination of elements of $B$ with coefficients in $\psi(\mathcal{J})$, and in particular, all of the generators of $B$ can be written this way.

I want to say that $\mathcal{J} = A$, but I'm not sure how to prove it.

share|improve this question
1  
Following your idea, maybe you can prove that, if $S=Im\psi$ then $\mathcal{M}\cap S$ is a maximal ideal of $S$. If this is the case, then $\psi^{-1}(\mathcal{M})=\psi^{-1}(\mathcal{M}\cap S)$ and you may consider the co-restriction of $\psi$ to its image, which is surjective. I think this can't be done with general ring homomorphisms, but in this case you may use that $A$ and $B$ are $\mathbb{C}$-algebras. –  Loronegro Feb 9 '12 at 3:08

1 Answer 1

up vote 8 down vote accepted

First of all note that the assumption that $A$ and $B$ be algebras over a field $k$ and $\psi$ is a $k$-algebra homomorphism is crucial, since the injection $\mathbb Z \hookrightarrow \mathbb Q$ shows that the pullback of $(0)$, which is maximal in $\mathbb Q$, is $(0)$ which is not maximal in $\mathbb Z$. Or you can take $A=k[X] \hookrightarrow B=k(X)$, the natural injection of a polynomial ring into its function field and again pull back the zero ideal.

To prove this, since it's a homework problem, I'm going to assume the following lemma for a $k$-algebra $A$: (a) If $A$ is a domain such that every element $a \in A$ is the root of a non-zero polynomial in $k[X]$ (i.e. $A$ is algebraic over $k$), then $A$ is a field; (b) if $A$ is a field and is contained in an affine $k$-algebra domain, then $A$ is algebraic over $k$.

These are basic results which you should have covered, but I'll sketch the proof of this lemma before proceeding. For part (a), let $a \in A$, it suffices to show $k[a]$ is a field, and you may assume $A=k[a]$. Consider the map $k[X] \rightarrow A$ mapping $f$ to $f(a)$, let the kernel be $I$, show $A \cong k[X]/I$, and since $a$ is algebraic over $k$, $I \neq 0$. But $k[X]$ is a PID, so $I=(f)$ for $f$ irreducible, so $I$ is maximal and $A$ is a field. Part (b) takes more work, I'm afraid, but in the case $k=\mathbb C$ is algebraically closed, it's easier to see.

Given this lemma, define a map $A \rightarrow B/ \mathfrak m$ by sending $a$ to $\psi(a)+ \mathfrak m$, which has kernel $\psi^{-1}(\mathfrak m)=: \mathfrak n$. Hence, $A/ \mathfrak n$ is isomorphic to a subalgebra of $B/ \mathfrak m$. By part (b) of the lemma above, $B/ \mathfrak m$ is algebraic over $k$, hence $A/ \mathfrak n$ is also algebraic over $k$, and a domain, so by part (a) of the lemma, it's a field and so $\mathfrak n$ is maximal.

share|improve this answer
    
Where exactly can the proofs of the two lemmas be found e.g. in Lang or Atiyah-MacDonald? –  Manos Oct 10 '12 at 19:00

Your Answer

 
discard

By posting your answer, you agree to the privacy policy and terms of service.

Not the answer you're looking for? Browse other questions tagged or ask your own question.