I am a little bit confused by Serre and Hurewicz fibrations in the context of pointed spaces, i.e. in $Top_*$.

Serre and Hurewicz fibrations are defined in $Top$, i.e. for non-pointed spaces, as having the RLP for inclusions $i_0 \colon I^n \hookrightarrow I^n \times I$ and $i_0 \colon X \hookrightarrow X \times I$ respectively.

But if I consider now a pointed map $p \colon (E,e_0) \to (B,b_0)$ and a pointed homotopie $H \colon (I^n,\partial I^n) \times I \to (B,b_0)$ for Serre or $H \colon (X,x_0)\times I \to (B,b_0)$ for Hurewicz, and if $p$ is a Serre/Hurewicz fibration in Top, there is no garantee that the lift $\tilde{H}$ is also a pointed homotopy. All we can say is that $\tilde{H}(* \times I) \subset F$ where $F = p^{-1}(b_0)$ is the fiber of $p$. So a priori $p$ is not a Serre/Hurewicz fibration in $Top_*$

Thus it seems that a specific more restrictive notion of Serre/Hurewicz fibration should be used in the context of $Top_*$, as having the RLP in $Top_*$ for the mentionned inclusions, i.e. the lift is should also be a pointed homotopy.

But I have never seen explicitely such a definition in classic books or pdf, although in most theorems involving homotopy groups we are in fact in the $Top_*$ context. Why ?

Now it is true that Serre fibration also have the RLP for $i \colon J^n \hookrightarrow I^n \times I$ where $J^n = \partial I^n \times I \cup I^n \times 0$. If there is a commutative square as usual where the initialization is $J^n \mapsto e_0$, then the lift is such that $\tilde{H}(\partial I^n \times I) = e_0$ and $\tilde{H}$ is a pointed homotopy. In this way, a Serre fibration is also a "pointed" Serre fibration. But this does not work for Hurewicz.


This all works as long as the basepoint $x_0$ of $X$ is nondegenerate. The general context for this is due to Arne Strom who showed that the category of topological spaces, with the classic (i.e. Hurewicz) definitions of cofibration and fibration, is a model category. The key theorem goes as follows: Suppose given a commutative square $$(i: A \rightarrow Y) \rightarrow (p: E \rightarrow B),$$ with $i$ a Hurewicz cofibration and $p$ a Hurewicz fibration. If either one is also a homotopy equivalence, there exists a map $Y \rightarrow E$ making the two evident triangles commute (i.e. extending $A\rightarrow E$ and lifting $Y \rightarrow B$).

Now apply this to the case when $Y=X \times I$ and $A = X \times 0 \cup x_0 \times I$. You can check that $A \hookrightarrow Y$ will be a cofibration and a homotopy equivalence if $x_0$ is nondegenerate.

Besides Strom's original papers from 1966 and 1968, a nice very short proof of this general theorem appears at the beginning of section 17 of the book by May and Ponto.

| cite | improve this answer | |

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.