Let $f \colon X \to Y$ be a morphism of schemes, where $X$ and $Y$ are separated integral Noetherian schemes. Does there necessarily exist a nonempty open affine $U \subset Y$ such that $f^{-1}(U)$ is affine? If $X \to Y$ is not dominant, then the answer is clearly yes (take $U$ to have empty inverse image). Even for dominant morphisms, I feel like this should be related to Chevalley's theorem that the image of a constructible set is constructible, but I can't put it together into a proof.

Note: this question is silly (see comments/answer below). The specific case I had in mind was when $f$ is unramified over the generic point of $Y$; however, if I decide I want the answer to this, I'll ask in a separate question.

  • $\begingroup$ I do not think so. What if you take $X=Y \times \mathbb{P}^1$ and $f$ the projection onto the first factor? $\endgroup$ Jul 24 '10 at 17:47

I think the answer is no. Take any field $k$, and consider the natural morphism $\mathbb P^n_k \to spec(k)$, where $\mathbb P^n_k$ is the $n$-dimensional projective space. Since $spec(k)$ is, as topological space, a single point and $\mathbb P^n_k$ is not affine, your open subset cannot exist.

  • $\begingroup$ Thanks for the answer. I just realized this also, so I'm modifying the question to make it (hopefully) more subtle. $\endgroup$ Jul 24 '10 at 17:57
  • 2
    $\begingroup$ Charles, instead of trying to make it more subtle and re-posting, why not remove the question and think about the more subtle one on your own for a few days (or more) and discuss with classmates? $\endgroup$
    – BCnrd
    Jul 24 '10 at 17:59
  • 1
    $\begingroup$ I'm not sure about removing the question (I prefer not to delete a question that someone has already answered), but for the rest, if you'll notice, I came to a similar conclusion myself. $\endgroup$ Jul 24 '10 at 18:07

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.