When is the loop space of a fibration a fibration? (Possibly up to homotopy equivalence)

algebraic-topologyhomotopy-theory

On Page 409 of Hatcher's Algebraic Topology, we see a sequence of maps

$\Omega B\to F\to E\overset{p}\to B$,

where $p:E\to B$ is a fibration and $F$ is the fiber $p^{-1}(b_0)$. $\Omega B\to F$ is a map that is homotopy equivalent to a fibration $F_i\to F_p$. He then applies the functor $\Omega$ to continue this sequence to the left:

$\ldots\to\Omega^2 B\to \Omega{F}\to \Omega{E}\overset{\Omega p}\to\Omega B\to F\to E\overset{p}\to B$.

My question is about the following statement, "Here, any two consecutive maps for a fibration, up to homotopy equivalence," that the author makes. Why is this true? For example, in the case of $\Omega F\to \Omega E\overset{\Omega p}\to \Omega B$, I don't believe that this is necessarily a fibration, but how can I prove that maybe this is a fibration up to homotopy equivalence? Any hints are appreciated!

Best Answer

Suppose that $p:E\rightarrow B$ is a fibration over a well-pointed space $B$. Then for any locally compact space $K$, the map $p:Map_*(K,E)\rightarrow Map_*(K,B)$ is a fibration.

Here, $E,B$ are pointed spaces and the map $p$ preserves basepoints. By well-pointed we mean that the inclusion of the basepoint of $B$ is a closed cofibration. The pointed mapping spaces are to carry the compact-open topology.

The proof of the assertion runs by taking a lifting problem \begin{array}{ccc}X &\rightarrow & Map_*(K,E)\\ \downarrow{} & &\ \downarrow \\ X\times I &\rightarrow& Map_*(K,B) \end{array} and considering the adjointed diagram \begin{array}{ccc}K\times X &\rightarrow & E\\ \downarrow{} & &\ \downarrow \\ K\times X\times I &\rightarrow& B. \end{array} Because $p:E\rightarrow B$ is a fibration, the latter problem solves, and admits a diagonal lift $\Phi:K\times X\times I\rightarrow E$. However, we need to be a little careful about which solutions we will admit. Specifically, the adjoint of $\Phi$ will be a function $X\times I\rightarrow Map(K,E)$ which will be continuous, since $K$ is locally compact, but may not take values in the space of pointed maps.

For the adjoint of $\Phi$ to take values in the pointed mapping space we must ensure that the following condition is met $$(\ast)\qquad\Phi(\ast,x,t)=\ast,\qquad \forall\, x\in X,t\in I.$$ This is the reason that we have assumed $B$ to be well-pointed, since this is sufficient to ensure that the fibration $p$ is regular, meaning that constant paths in $B$ can be lifted to constant paths in $E$. Because the arrow $K\times X\times I\rightarrow B$ satisfies the corresponding condition, regularity is therefore enough to ensure that a diagonal $\Phi$ exists satisfying condition $(\ast)$.

With those details sketched, the only thing to do is study the fibre of $p_*:Map_*(K,E)\rightarrow Map_*(K,B)$. This is the subspace $$p_*^{-1}(\ast)=\{f\in Map_*(K,E)\mid p(f(K))=\ast\}=\{f\in Map_*(K,E)\mid f(K)\subseteq F\}\cong Map_*(K,F)$$ where $F=p^{-1}(\ast)$ is a closed subset of $E$.

Finally, aplying the above with $K=S^1$ we obtain a fibration sequence $$\Omega F\rightarrow \Omega E\xrightarrow{p} \Omega B.$$