A set in $\mathbb{C}^n$ is called Zariski-closed if it can be written as the set of zeroes of some set of polynomial equations $$ V(f_1,...,f_m) = \left\{ z \in \mathbb{C}^n \mid f_1(z)=...=f_m(z)=0 \right\} $$ and in general as $V(I)$ where $I \subset \mathbb{C}[x_1,...,x_n]$ is an ideal. The Zariski-closed sets form a topology called "Zariski topology". A Zariski-closed set embedded in the affine space $\mathbb{C}^n$ is called affine variety. An open set of an affine variety is called a quasi-affine variety.

It is easy to see that in the case of 1-dimensional affine space $\mathbb{C}$ the Zariski-closed sets are $\emptyset, \mathbb{C}$ and all the finite sets $\{ z_1,...,z_k \}$ (since a polynomial has only a finite number of roots). Then $$ [0,1] = \{ z \in \mathbb{R} \subset \mathbb{C} \mid 0 \le z \le 1 \}$$ is clearly not Zariski-closed, but it isn't evident it is Zariski open. So I ask: is it Zariski open? I think it is niether, since its complement is also infinite and hence not Zariski-closed.

Second question, suppose $W$ is an affine variety (and hence Zariski closed), is $W \times [0,1]$ (embedded in higher dimensional affine space) is Zariski closed or Zariski open? Or quasi-affine variety? A typical example is a band or a strip, let $\mathbb{C}^2$ and $W = \{ (x,y) \mid x = 0 \} \in \mathbb{C}^2$ and then $$ W \times [0,1] \cong \{ (x,y,z) \mid x = 0 \mbox{ and } 0 \le z \le 1 \}$$ but there are more complicated examples.

The next generalization is as follows. Suppose we have two affine varieties in $\mathbb{C}^n$, call them $X$ and $Y$. The join $J(X,Y)$ of $X$ and $Y$ is defined as follows: $$J(X,Y) = \bigcup_{p \in X, q \in Y} \overline{pq} $$ where $p \in X$ and $q \in Y$ are points in $\mathbb{C}^n$ and $\overline{pq}$ is the line segment joining them, $$ \overline{pq} = \left\{ \lambda p + (1-\lambda)q \mid \lambda \in [0,1] \right\} \ . $$ I want to determine if the join is Zariski closed, Zariski open, a quasi-affine variety or niether of these?

Again, a (relatively) simple example is the join of $y=0$ and $y=1$ in $\mathbb{C}^2$.

Remark: it is hard to visualize the varieties involved since we work over the algebraically closed field $\mathbb{C}$ and not over $\mathbb{R}$ so the Euclidean geometry may give wrong intuition.

References for definitions from Wikipedia: https://en.wikipedia.org/wiki/Zariski_topology and https://en.wikipedia.org/wiki/Join_(topology)


If a polynomial $f\in\mathbb{C}[x]$ satisfies $f(z)=0$ for all $z\in[0,1]$, then we must have that $f=0$ (this is true of any analytic function).

Let's say that $W\subseteq\mathbb{C}^n$. Applying the above observation, if $f\in \mathbb{C}[x_1,\ldots,x_n,y]$ vanishes on $W\times[0,1]\subseteq\mathbb{C}^{n+1}$, then for any $(z_1,\ldots,z_n)\in W$, the polynomial $f(z_1,\ldots,z_n,y)\in\mathbb{C}[y]$ must be zero everywhere, so that $f$ vanishes on $W\times\mathbb{C}$. Thus, the closure of $W\times[0,1]$ in the Zariski topology on $\mathbb{C}^{n+1}$ is $W\times\mathbb{C}$.

The same reasoning applies to any line segment, and hence in particular to any $\overline{pq}$; thus, $J(X,Y)$ will never be Zariski-closed except in degenerate cases (e.g. $X$ or $Y$ being $\varnothing$ or $\mathbb{C}$, or $X=Y$). The same reasoning applies to the complement of any line segment, so that $J(X,Y)$ will never be Zariski-open.

| cite | improve this answer | |
  • $\begingroup$ Thanks, I thought so. When you talk about the degenerate cases you mentioned $X = \mathbb{C}$. What about the case when $X = \mathbb{C}^3$ and $Y = \mathbb{C}$ and they are algebraic affine varieties embedded in $\mathbb{C}^4$? I thik that $J(X,Y)$ won't be closed or open by the above arguments (this is a sort of the strip example I mentioned in my question. $\endgroup$ – LinAlgMan Mar 10 '13 at 16:26

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.