I'm reading Milne's book about algebraic geometry and I have following questions?

  1. Suppose I have an irreducible algebraic variety $V$, i.e. an algebraic prevariety with the separation condition, then why is $U \cap U'$ an open affine if $U$ and $U'$ are$?$

  2. Now, since $U$ and $U'$ are nonempty open subsets then they are irreducible and so $K[U],K[U']$ and $K[U \cap U']$ are domains, then why $K[U \cap U'] \subset K(U)?$

  3. Why $K(V) = K(U)$ for every open affine $U \subset V$?

Everything it's on page 85. Thanks for your help.


Let $k$ be the ground field.

  1. Consider the diagonal morphism $\Delta:V\to V\times_kV$. Then $$U\cap U'=\Delta^{-1}(U\times_kU').$$ Since $\Delta$ is a closed immersion (by definition of separatedness), the preimage of an affine open is affine, and our $U\times_kU'$ is certainly affine.

The answer to 2 and 3 (once $U\cap U'$ is assumed affine) does not depend on separatedness but only on irreducibility of $V$.

There are three crucial facts: (i) an irreducible variety $V$ has a unique generic point (call it $\eta$), (ii) for every $x\in V$, the stalk $\mathcal O_{x,V}$ is equal to $\mathcal O_{x,U}$ if $U$ is open in $V$ (and contains $x$!), (iii) the generic point lies in every open subset of $V$ and it is the generic point in each of them.

Let $U=\textrm{Spec }A\subset V$ be open. The function field $K(V)$ is by definition the stalk $\mathcal O_{\eta,V}$ at the generic point $\eta$ of $V$, and $$K(U)\overset{(iii)}{=}\mathcal O_{\eta,U}\overset{(ii)}{=}\mathcal O_{\eta,V}=:K(V).$$

In other words, all the domains $A=\bigcap_{P\in U}A_P\subset \mathcal O_{\eta,V}$ share the same function field! you can think it this way as well: it is clear that $K(U)\subset K(V)$ because $A\subset \mathcal O_{\eta,V}=K(V)$. The reverse inclusion is given by $K(V)=\mathcal O_{\eta,V}\to\mathcal O_{\eta,U}=K(U)$, corresponding to $U\hookrightarrow V$ (sending $\eta\mapsto \eta$).

The fact that every open affine has the same function field of the whole variety answers question 2 as well.

Edit. Avoiding generic points and stalks. Claim: if $U,U'$ are affine open subsets of an irreducible variety $V$, then $\textrm{Frac }\mathcal O_V(U)=\textrm{Frac }\mathcal O_V(U')$.

Let us take an affine open subset $P\subset U\cap U'\subset U$, principal in $U$, and an affine open subset $P'\subset P\subset U'$, principal in $U'$. Then, say, $\mathcal O_V(P)=\mathcal O_V(U)_f$ and $$ \textrm{Frac }\mathcal O_V(U)=\textrm{Frac }\mathcal O_V(U)_f=\textrm{Frac }\mathcal O_V(P). $$ (An integral domain shares its field of fractions with all of its localizations.) On the other hand, $$ \textrm{Frac }\mathcal O_V(U')\subseteq \textrm{Frac }\mathcal O_V(P)\subseteq \textrm{Frac }\mathcal O_V(P')=\textrm{Frac }\mathcal O_V(U'), $$ hence $\textrm{Frac }\mathcal O_V(U')=\textrm{Frac }\mathcal O_V(P)=\textrm{Frac }\mathcal O_V(U)$.

So one can define the function field of $V$ to be $\textrm{Frac }\mathcal O_V(U)$, where $U$ is any affine open in $V$.

  • $\begingroup$ Thanks a lot, but can you explain me the same omitting 'generic points', 'schemes' and 'stalks'? $\endgroup$ – Diego Silvera Feb 17 '13 at 23:54
  • $\begingroup$ I edited my answer. Please, let me know if it is still unclear. $\endgroup$ – Brenin Feb 18 '13 at 1:04
  • $\begingroup$ It's all right now. Another thing, can you recommend me another books or notes in the spirit of Milne, but mooore explicit with its arguments? Because I have had serious headaches with this book witch my teacher decided to follow. Thanks again. $\endgroup$ – Diego Silvera Feb 18 '13 at 14:07
  • $\begingroup$ I have to say that I never used Milne's book on AG. I feel very comfortable with Ravi Vakil's notes Foundations of Algebraic Geometry where all the concepts are neatly explained. You find the current version of these notes on his webpage. $\endgroup$ – Brenin Feb 18 '13 at 14:28

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.