Let $X$ be a topological space, and $PX$ the space of all paths on $X$. Then let $G\subset X$ be a path-connected subset and $p\in G$ a point. Let $\sigma:G\rightarrow PX$ be a continuous function such that

$\sigma(x)(0)=x\ \forall x\in G,$

$\sigma(x)(1)=p\ \forall x\in G.$

Then show that there exists a function $\tau:G\rightarrow PX$ such that

$\tau(x)(0)=x\ \forall x\in G$

$\tau(x)(1)=p\ \forall x\in G$

$\tau(p)(t)=p\ \forall t\in[0,1].$

Clearly $G$ is contractible, and then the question relates to showing that it is SDR-contractible. I know that this is not always the case (see comb space) but I'm hoping that the addition of the path-connected property makes this true.

I need to show this as part of a lemma in my masters project and would really appreciate any help. Thanks


I think the assumption you need is that $p$ is a non-degenerate basepoint of $G$, which is to say that the inclusion $\lbrace p\rbrace \hookrightarrow G$ is a cofibration, which is to say the pair $(G,\lbrace p\rbrace)$ has the homotopy extension property. It follows from this that the pair $$ (G,\lbrace p\rbrace)\times (I,\partial I) = (G\times I, \lbrace p\rbrace \times I \cup G\times \partial I) $$ also has the HEP. Now you can take your contracting homotopy $\sigma\colon\thinspace G\to PX$, or rather its adjoint $\tilde\sigma \colon\thinspace G\times I\to X$, and show that it is homotopic to a homotopy which fixes $p$. (You will be extending a homotopy $$ (\lbrace p \rbrace \times I \cup G\times \partial I)\times I \to X $$ given by $(x,t,s) \mapsto \tilde\sigma (x,(1-s)t)$.)

A good online reference for this might be the notes of Jesper Møller here, in particular Section 5.

  • $\begingroup$ Thank you, exactly what I was after. My project is a study of topological complexity, so this question aside, I still have much to thank you for. $\endgroup$ – Sam Bullen Mar 24 '13 at 12:10
  • $\begingroup$ Hi Sam, you're welcome! This kind of question crops up a lot when studying LS-category-like invariants. Good luck with the masters project. $\endgroup$ – Mark Grant Mar 25 '13 at 9:59

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.