Set $X = \mathbb{A}_K^2 = K^2$, $K$ an algebraically closed field. I am considering $A \, \colon= X - \{(0,0)\}$ viewed as a locally closed subspace of $X$. By definition a locally closed subset is one that is the intersection of a Zariski open and a Zariski closed subset. Since $A = V(\langle x,y \rangle)^c \cap X$ the definition is satisfied, in particular $A$ is an open subspace. View $A$ as a quasi-affine algebraic variety, $(A, \mathscr{O}_A)$ with the structure induced from $(X, \mathscr{O}_X)$. If it matters, we know this space is irreducible since $X$ is irreducible and any open subspace of irreducible is also irreducible.

I would like to check that the functions $Q_1 = x$ and $Q_2 = y$ are regular on $A$. I feel like this should be extremely trivial, or obvious, but it is not. To be honest, the concept of regular functions have always been a bit slippery to me. My approach here would be to simply check that my definition of regular function is satisfied. My definition is $Q_i$ is regular on $A$, or $Q_i \in \Gamma(A, \mathscr{O}_A)$, if

1) $Q_i \colon A \to k$ is continuous,
2) if $x \in A$ and $f \in \mathscr{O}_{x,K}$ then $f \circ Q_i \in \mathscr{O}_{x,A}.$

Are there better ways, more intuitive ways, or easier ways to be thinking of regular functions so that the justification I seek is immediate?

  • $\begingroup$ I think $Q_1$ and $Q_2$ are regular on $\mathbb{A}^2$, so the restriction to $A$ should be regular as well? $\endgroup$ – red_trumpet Mar 7 at 22:01

$\newcommand{\spec}[1]{\mathrm{Spec}(#1)}$ $\newcommand{\Ohol}{\mathcal{O}}$ $\newcommand{\ideal}[1]{\mathfrak{#1}}$

I must admit, that I do not understand the notation in point 2) of your definition of regular function, but I would propose you a definition that is both standard and easy to apply to your current example:

Let $X = \spec{A}$ be the prime spectrum of a commutative ring $A$ (with the Zariski-Topology of course). Then the first task is to define the sheaf of regular functions $\Ohol_X$ on $X$. A regular function $s$ on an open subset $U \subseteq X$ is then an element of $\Ohol_X(U)$.

So now to the definition: Take $\Ohol_X(U)$ to be the set of all functions $s:U \to \coprod_{\ideal{p} \in U} A_\ideal{p}$ with $s(\ideal{p}) \in A_\ideal{p}$. ($\ideal{p}$ is a prime of $A$, that is a point of $X$).

Furthermore require the following: For every $\ideal{p} \in U$, there is an $f \in A$ with $f \notin \ideal{p}$ and an element $a/f^d \in A_f$ such that for all $\ideal{q}$ with $f \notin \ideal{q}$ (that is $\ideal{q} \in D(f)$, to use a commonplace notation) we have $s(\ideal{q}) = a/f^d$ considered as element of $A_\ideal{q}$.

This sounds maybe like a bit scary pileup of notations, but it is actually a very natural definition. If $A=k[x_1,\ldots,x_n]$ is a polynomial ring this says that a regular function is given locally on $D(f)$ around $P = \ideal{p}$ by a quotient of polynomials $g/h$ , where the denominator $h$ does not vanish where $f$ does not vanish.

Now this definition above has two beautiful consequences: One can proof

$$(*) \quad \Ohol_X(D(f)) = A_f$$

for every $f \in A$ and

$$(**) \quad \Ohol_{X,\ideal{p}} = A_\ideal{p}$$

for every $\ideal{p} \in A.$

It should be now obvious that your function $x, y \in k[x,y]=B$ are regular in $X=\spec{B}$ and of course stay so in $U = D(x) \cup D(y) \subseteq X$.

You should really understand the proof of $(*)$ and $(**)$ (it is given in Hartshorne's "Algebraic Geometry") - in "practice" you will in most types of reasoning with regular functions use these isomorphisms.


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.