4
$\begingroup$

Let $X$ be a smooth scheme over a field $k$, and let $Z\subset X$ be a closed sub-scheme of codimension $2$. Let $Bl_Z(X)$ denote the blow-up of $X$ at $Z$, and let $\pi\colon Bl_Z(X)\to X$ denote the projection. Suppose I have a section $\sigma\colon Z\to Bl_Z(X)$ of $\pi$ over $Z$ (i.e. $\pi\sigma=1_Z$).

Question: (when) is the map $Bl_Z(X)\setminus\sigma(Z)\to X$ a weak equivalence?

The references to blow-up theorems which I have found (e.g. Voevodsky's Seattle lectures) suggest that it becomes an equivalence after suspension, but I'd like to avoid suspending if possible.

I'm also happy to restrict the choice of $Z$ and $X$. The case I'm most interested has $X$ being an iterated blow-up of affine space at (proper transforms) of linear sub-spaces, and $Z$ being a (proper transform of a) linear sub-space.

$\endgroup$
2
  • $\begingroup$ I'm confused about your question: if $Z$ is already a Cartier divisor, then you seem to be asking whether $X-Z\to X$ is a weak equivalence. Clearly this is not always the case... $\endgroup$ Apr 23, 2014 at 10:18
  • $\begingroup$ Marc, good point. The examples I'm thinking about are all for $Z$ of codimension $>1$. $\endgroup$ Apr 23, 2014 at 10:35

1 Answer 1

6
$\begingroup$

I assume that $Z$ is also smooth over $k$. Then the base change of the map

$$Bl_Z(X)-\sigma(Z)\to X$$

to $Z$ is the map

$$\mathbb{P}(N_{X,Z})-\sigma(Z) \to Z$$

where $\mathbb{P}(N_{X,Z})$ is the bundle of lines in the normal bundle $N_{X,Z}$. The second map is an equivalence iff $Z$ has codimension $2$. If not, the two sides have different $\ell$-adic cohomology after pulling back to a geometric point, for example. So the first map can only be a weak equivalence when the codimension is exactly $2$.

But if the codimension is $2$, I don't know if your map an equivalence. I haven't heard of any progress towards removing the suspension in Voevodksy's blow-up theorem...

$\endgroup$
5
  • $\begingroup$ Thanks for the answer, and you're right that the question I wrote requires codimension 2. My intuition is that $Bl_Z(X)\setminus\sigma(Z)$ is obtained from X by gluing in an affine bundle over $Z$. Since affine spaces are $\mathbb{A}^1$-contractible, my intuition was that we could contract away the fibers over $Z$ and recover $X$. In higher codimension, I need to delete a hyperplane section over $Z$ to get the right picture, but this is the situation I had in mind. Do you know if there is reason to be suspicious of this intuition? $\endgroup$ Apr 23, 2014 at 23:38
  • $\begingroup$ Actually, this intuition is wrong, and this is not the question I should have asked. Thanks for helping clear this up. $\endgroup$ Apr 24, 2014 at 3:21
  • 1
    $\begingroup$ If you remove a hyperplane section from the exceptional divisor, then I agree that it looks like the map could be an $\mathbb{A}^1$-equivalence, but you need a finer topology than the Nisnevich one. Maybe this will be useful: Unstable motivic homotopy categories in Nisnevich and cdh-topologies. $\endgroup$ Apr 24, 2014 at 10:41
  • 1
    $\begingroup$ I think it's unlikely that it will ever be an $\mathbb{A}^1$-equivalence. Consider the blow up of $\mathbb{A}^2$ at a point. This is the total space of the tautological line bundle over $\mathbb{P}^1$. If we remove a point from the exceptional divisor, we get something whose topological realization looks like $S^3\vee S^2$. FWIW, the situation I was considering turns out to involve blow-ups of affine space where we remove the fiber over a hyperplane section of the exceptional divisor (I only saw the hyperplane at first). After removing the whole fiber, we get an $\mathbb{A}^1$-equivalence. $\endgroup$ Apr 25, 2014 at 21:42
  • $\begingroup$ Ah, that's a great example. In this case the blow-up square is trivially a pushout (without suspending), but there's no reason for it to remain a pushout when you remove a point... I missed this point before. $\endgroup$ Apr 28, 2014 at 10:11

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.