13
$\begingroup$

I'm wondering about the theoretical placement of quasifibrations.

One nice thing about "weak fibrations" (maps homotopy equivalent in the category of maps to Hurewicz fibrations) is that a pullback square involving (one) weak fibration is a homotopy pullback square.

Is the corresponding result true for quasifibrations in the Serre-Quillen context? That is, suppose $E\to B$ is a quasifibration, and the square $$ \begin{array}{ccc} P & \to & E \cr\downarrow&pb&\downarrow \cr X& \to &B \end{array} $$ is a categorical pullback. Then is it a homotopy pullback in the Quillen-Serre model structure?

$\endgroup$
2
  • $\begingroup$ Dold and Lashof compare their construction for a monoid M to Milnor's when M is a group G. They give an explicit comparison for the first stage of the constructions. Somewhere I've seen the general n-stage comparison formula - anyone remember where? $\endgroup$ Jul 9, 2023 at 12:54
  • 1
    $\begingroup$ The desired formula can be produced by careful iteration of the case given by Dold&Lashof. It also appears in Peter May's Geometry of iterated loop spaces pp.98-99 without mention of Dold&Lashof. $\endgroup$ Jul 10, 2023 at 15:18

2 Answers 2

20
$\begingroup$

The definition of quasifibration (according to Dold & Thom, 1958) is: a map $f:E\to B$ such that for all $b$ in $B$, the canonical map from the fiber to the homotopy fiber is a weak equivalence. Pullbacks with respect to such maps are not generally homotopy pullbacks; an example was given in that 1958 paper (Bermerkung 2.3), which goes something like this:

Let $\newcommand{\R}{\mathbb{R}}B=\R\times \R$. Then $E$ will have the same underlying set as $B$, and $f$ will be the identity map. But we topologize $E$ by "tearing" along the positive $y$-axis. For instance, let $E$ have the smallest topology such that $f$ is continuous and the set $[0,\infty)\times (0,\infty)$ is open.

The space $E$ is still contractible with this topology (it deformation retracts to $\R\times -1$). Therefore, the homotopy fiber over any point of $B$ must be weakly contractible, and thus weakly equivalent to the actual one-point fiber. So $f$ is a quasi-fibration.

Let $X= \mathbb{R}\times 1\subset B$, and let $P$ be the pullback of $E$ over $X$. Then $P$ has two path components, while $X$ is contractible; this is not a homotopy pullback!

$\endgroup$
1
  • $\begingroup$ This answers my question, and makes me wish I could read German much more quickly. $\endgroup$
    – Jeff Strom
    Jan 30, 2011 at 22:50
12
$\begingroup$

It rather depends what you mean by quasi-fibration. The most useful resource I know for questions about quasi-fibrations is this message of Goodwillie, posted to the APGTOP mailing list in 2001.

$\endgroup$
4
  • $\begingroup$ This is nice to have in my back pocket, thanks. $\endgroup$
    – Jeff Strom
    Jan 30, 2011 at 22:50
  • 8
    $\begingroup$ Wow, I enjoyed writing that, but I had lost track of it. It's nice to see it again. $\endgroup$ Jan 31, 2011 at 1:58
  • 4
    $\begingroup$ But at the end where it says "Prop 1"it should say "Lemma 2". $\endgroup$ Feb 1, 2011 at 2:05
  • 1
    $\begingroup$ It seems like Goodwillie UQF's might coincide with Hopkins/Rezk 'sharp maps' (arxiv.org/pdf/math/9811038.pdf) $\endgroup$ Jul 9, 2023 at 13:36

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.