6
$\begingroup$

Let $X$ be the underlying space of a scheme.

  • If $X$ is irreducible of finite Krull dimension, is it necessarily quasi-compact?
  • Is it necessarily Noetherian?
  • What if we assume not only that Krull dimension is finite but also that it is 1?
$\endgroup$
3
  • 1
    $\begingroup$ a separated 1d example: Enumerate the primes: $p_1=2,p_2,...$. Let $R_n$ be the ring of rational numbers whose denominator is not divisible by the first $n$ primes $p_1,...,p_n$. This has spectrum with 1 generic point and $n$ closed points with local rings $\mathbb Z_{(p_i)}$. Each ring is a localization of the next: $R_n=R_{n+1}[p_{n+1}^{-1}]$, so its spectrum is open in the next. Then $X=\bigcup_n \mathrm{Spec}R_n$ is a noncompact scheme with global sections $\mathbb Z$. Clearly not $\mathrm{Spec}\mathbb Z$! $\endgroup$ Apr 18, 2019 at 0:11
  • 1
    $\begingroup$ A 2d example closer to practical geometry: Let $X_0=\mathbb P^2$ and $p_0$ some point on it. Let $X_{n+1}$ be the blow-up of $X_n$ at $p_n$ and let $p_{n+1}\in D\subset X_{n+1}$ be some point of the exceptional divisor. Let $U_n=X_n-\{p_n\}$. Each is open in the next. So $U=\bigcup U_n$ is a noncompact scheme... The inverse limit of the $X_n$ is probably a compact non-noetherian scheme with open set $U$... The sequence of $p_n$ specifies a 2d valuation ring, necessarily non-noetherian. It has compact spectrum, but the complement of the unique closed point is related to $U$, maybe even 1d. $\endgroup$ Apr 18, 2019 at 0:16
  • $\begingroup$ @BenWieland I think that should be an answer, it's a better example than mine and it's a pity to leave it in the comments. $\endgroup$ Apr 18, 2019 at 8:59

1 Answer 1

10
$\begingroup$

The answer is no for all these questions. Take the line with infinite origins: the scheme obtained by gluing an infinite amount of copies of $\mathbb{A}^1$ along the open subsets $\mathbb{G}_m$. This has Krull dimension 1 (there are only closed points and the unique generic point) and it is irreducible (the only proper nonempty closed subsets are the closed points) but it is not quasi-compact (it contains an infinite discrete set), and so in particular not Noetherian.

$\endgroup$
2
  • 1
    $\begingroup$ Denis Nardin, I am not sure about the second paragraph. I thought nilpotent thickening never affects the underlying topological space. So while your nilpotent thickening gives a non-locally Noetherian scheme, does it give a non-locally Noetherian space (about which is the question)? $\endgroup$
    – user137767
    Apr 17, 2019 at 17:44
  • 1
    $\begingroup$ @StepanBanach Ah sorry, I was thinking of a non-Noetherian scheme. Let me correct it. $\endgroup$ Apr 17, 2019 at 18:13

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.