I assume each algebra here is commutative with a unit.

Let $\nu : A \hookrightarrow B$ be a finite embedding of algebras (that is $B$ is finitely-generated as an $A$-module). I want to show the following:

  1. That the map $\nu^{-1} : \mathrm{Spec}(B) \to \mathrm{Spec}(A)$ is onto and has finite fibers. That is, every prime ideal in $A$ has a preimage which is a prime ideal in $B$, and there are only finitely many such ideals in $B$.

  2. An ideal $P \lhd B$ is maximal if and only if $\nu^{-1}(P) \lhd A$ is maximal.

My attempts:

My thoughts on 1) where to extend the ideal $P \lhd A$ to $B$ via $P \mapsto B\nu(P)$ and then take $\nu^{-1}(B \nu(P))$ and show it is equal to $P$, however, there is no $\nu^{-1}(B)$ since $\nu$ is not surjective.

We note that $B\nu(P)$ is not necessarily a prime ideal. Therefore we must take the maximal ideal $I$ in $B$ containing $B\nu(P)$ and such that $I \cap \nu(A) = \nu(P)$. We need to prove that $I$ is prime ideal and that $\nu^{-1}(I) = P$.

My thoughts on 2) where to use the characteriziation that $P \lhd B$ is maximal if and only if $B/P$ is a field. However, lack of surjectivity still causing me troubles. By using a Proposition that say that if $\beta$ is algebraic over a field $F$ then $F[\beta]$ is a field, using the fact that $B$ is finite over $A$ I can prove that $$ A/\nu^{-1}(P) \mbox{ is a field } \Rightarrow B/P \mbox{ is a field } $$ since the quotient $B/P$ is finite over $A/\nu^{-1}(P)$ and has a finite number of generators, all are algebraic. Now I need to prove the other direction. Any thoughts?

  • $\begingroup$ Does it obvious that $I \cap \nu(A) = \nu(P)$ and in particular $B\nu(P) \cap \nu(A) = \nu(P)$ or it needs a proof? If so, how do I prove it? $\endgroup$ – LinAlgMan Nov 18 '12 at 19:03

Finite algebras are integral and the results you want (except for finiteness of fibers) are valid more generally if you replace "finite" by "integral": the proofs are in Atiyah-Macdonald's Introduction to Commutative Algebra, Corollary 5.8 and Theorem 5.10.

For the finiteness of the fiber over the prime ideal $\mathfrak p\subset A$, use that this fiber is in bijection with the primes of the ring $B\otimes_A Frac (A/\mathfrak p)$, which is also finite over $Frac (A/\mathfrak p)$, to reduce to the case where $A$ is a field.
Then use that $B$ is trivially artinian to conclude that $B$ has only finitely many primes (which are maximal) thanks to Proposition 8.3 of our reference.

  • $\begingroup$ Thank you. I guess in frac you mean the field of fractions. Also, could you elaborate on the bijection between the fibers and the tensor product? I know (in the definition level) what is an Artinian ring, but we havn't cover it in class. I do have Atiyah-McDonalds and read the above mentioned propositions. $\endgroup$ – LinAlgMan Nov 19 '12 at 18:36
  • $\begingroup$ Dear LinAlgMan, I recommend you try to solve Exercise 21 of Chapter 3 in Atiyah-Macdonald: part iv) gives the required bijection (which is even a homeomorphism). $\endgroup$ – Georges Elencwajg Nov 22 '12 at 19:18

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.