0

If we have commutative rings $R$ and $T$ and a homomorphism between them $\phi : R \to T$, then we define $Spec(\phi): Spec(T) \to Spec(R)$ by $p \mapsto \phi^{-1}(p)$.

There is then a theorem in the notes that I am reading that states that if $\phi$ is surjective, then $Spec(\phi)$ is injective.

My question is how $Spec(\phi)$ could ever not be injective, since that would mean that $\phi^{-1}(p)=\phi^{-1}(q)=s$, but $p \not= q$, so $\phi(s)=p \not= q = \phi(s)$.

Moreover I am also unsure as to how $Spec(\phi)$ is even well-defined in the case where $\phi$ is not surjective, or at least if not all prime ideals in $T$ are in $Im(\phi)$.

My other question is about a theorem that says that if $T$ is finite over $R$ then $Spec(\phi)$ has finite fibres, i.e. the set $Spec(\phi)^{-1}(\{p\})$ is finite, and again I fail to see how this set could have size greater than 1.

Any help would be appreciated.

Orlly
  • 175
  • 2
    If $\mathfrak s = \phi^{-1}(\mathfrak p)$, then not necessarily $\phi(\mathfrak s) = \mathfrak p$, just $\phi(\mathfrak s) \subseteq \mathfrak p$. – azif00 Apr 21 '21 at 19:09
  • Ok, thank you, that does make sense for the first bit of my question. I'm still struggling to understand the intuition behind what $Spec(\phi)^{-1}({p})$ actually means – Orlly Apr 21 '21 at 19:36

1 Answers1

3

We can look at this simple example: let $R=\mathbb{C}[x]$ and $S=\mathbb{C}[x,y]$. Consider the inclusion map $\phi: R \to S$ and the induced map $\operatorname{Spec}(\phi): \operatorname{Spec}(S) \to \operatorname{Spec}(R)$.

Take $p=(x-a)$ (where $a$ is a number) a prime ideal in $\operatorname{Spec}(R)$. What is $\operatorname{Spec}(\phi)^{-1}(\{p\})$? It just consists of all prime ideals $Q$ of $S$ such that $\phi^{-1}(Q)=p$. Can we find more than one of such $Q$? Yes! In fact, we can find infinitely many of them, for example, all prime ideals of the form $(x-a,y-b)$ where $b$ is any number.

Geometrically speaking, you can think of $\operatorname{Spec}(S)$ as a plane and $\operatorname{Spec}(R)$ as a line, say, the $x-$axis of the plane. The map $\operatorname{Spec}(\phi)$ is just the projection map from the plan to the $x-$axis. Prime ideal $p$ corresponds to the point $(a,0)$ and $\operatorname{Spec}(\phi)^{-1}(\{p\})$ is what called the fiber over the point $(a,0)$ (everything that project down to $(a,0)$. Obviously, all the points $(a,b)$ projects to $(a,0)$ and they correspond to prime (maximal) ideals $(x-a,y-b)$.

For an example that $T$ is finite over $R$ and there is a fiber that has more than one points, we can take the following simple one:

Take $R=\mathbb{C}[x]$ and $S=\mathbb{C}[x,y]/(x-y^2)$. Then at each prime ideals $(x-a)$, $a\not = 0$, of $\operatorname{Spec}(R)$, we can see that there are two prime ideals $(x-a,y-a)$ and $(x-a,y+a)$ in $\operatorname{Spec}(\phi)^{-1}((x-a))$. The picture is just the projection of horizontal parabola to the $x-$axis, so above each point $a\not =0$ we have two points of the parabola that maps down to it.

mathmathmath
  • 1,433