This is an exercise from Goertz/Wedhorn: Let $R$ be a PID and $f\in R$ with prime-decomposition $f=\epsilon f_1^{n_1}\ldots f_k^{n_k}$. Analyse the affine scheme $X=\operatorname{Spec} R/fR$, so the topological space, stalks and $\mathcal{O}_X(U)$ for $U$ open.
What I have so far:$X=\operatorname{Spec} R/fR$ is homeomorphic to $V(f)\subseteq \operatorname{Spec} R$, and the prime ideals in the letter set are precisely $\{(f_1)=\mathfrak{p}_1,\ldots,(f_k)=\mathfrak{p}_k\}$. Thus, as a set we have $\operatorname{Spec} R/fR=\{\overline{\mathfrak{p}_1},\ldots,\overline{\mathfrak{p}_k}\}.$ For the stalk we have $$\mathcal{O}_{X,\overline{\mathfrak{p}_s}}=(R/fR)_{\overline{\mathfrak{p}_s}}\cong (R/(f_1^{n_1}))_{\overline{\mathfrak{p}_s}}\times\ldots\times (R/(f_k^{n_k}))_{\overline{\mathfrak{p}_s}}=(R/(f_s)^{n_s})_{\overline{\mathfrak{p}_s}},$$ because for $l\neq s$ we have $(f_l^{n_l})\cap(R\setminus\mathfrak{p}_s)\neq\emptyset$, as otherwise we would have $(f_l)^{n_l}\subseteq \mathfrak{p}_s$ and then also $(f_l)\subseteq\mathfrak{p}_s$ because the latter is prime. But this would force $l=s$ since both ideals are maximal. Now the topology on $\operatorname{Spec} R/fR$ is the discrete topology, because it's finite and the closed sets are precisely the finite subsets. Given an open - so arbitrary - $U=\{\overline{\mathfrak{p}_{m_1}},\ldots,\overline{\mathfrak{p}_{m_r}}\}\subseteq X,$ we consider $g=\prod _{i\neq m_j}f_i\in R$ and then have $D(\overline{g})=U$. This gives $$\mathcal{O}_X(U)=(R/fR)_{\overline{g}}\cong (R/f_1^{n_1})_{\overline{g}}\times\ldots\times (R/f_k^{n_k})_{\overline{g}}=\prod_j (R/f_{m_j}^{n_{m_j}})_{\overline{g}}.$$
My questions:
- Is this correct?
- Am I done with this exercise? I have kind of a bad feeling regarding if I'm done with such vaguely formulated exercises. I don't know if I could simplify a bit more, for example. Maybe a bit of life advice concerning this would also be appreciated. I'm grateful for any sort of help.
The scheme $X$ is the coproduct (=disjoint union) of the schemes $X_i=\operatorname{Spec}(R/\langle f_i^{n_i} \rangle)$, namely $ X=\coprod X_i$.
Each underlying topological space $\vert X_i\vert=\{p_i\}$ is a singleton set and $\vert X\vert=\coprod\vert X_i\vert=\{p_1,\cdots,p_k\}$ has the discrete topology.
The local rings of these schemes are $\mathcal O_{X,p_i}=\mathcal O_{X_i,p_i}=R/\langle f_i^{n_i} \rangle$.
Finally, for every subset $S\subset \{1,\cdots ,k\}$ and corresponding open subset $U=\{p_i\vert i\in S\}\subset X$ we have $$\mathcal O_X(U)=\prod_{i\in S}O_{X_i,p_i}=\prod_{i\in S}R/\langle f_i^{n_i} \rangle$$
Note carefully:
a) In general an affine scheme with finitely many points need not have discrete topology.
The simplest counterexample is $S=\operatorname {Spec A}=\{\mathfrak m, (0)\}$, the affine scheme associated to a discrete valuation ring $A$ with maximal ideal $\mathfrak m$.
The open subsets of the topological space $\vert S\vert$ are $\{(0)\}$,$S$ and $\emptyset$.
Thus $\vert S\vert$ is a non discrete space even though it has only two points: the closed point $\mathfrak m$ and the open but not closed point $(0)$.
b) However our $X$ is discrete because $X$ has dimension zero since $R/\langle f\rangle$ is an Artin ring.