Let $X$ be a topological space and $V$ and $N$ are subspaces of $X$ such that $N$ deformation retracts onto $V$. I want to show that $X-V$ deformation retracts onto $X-N$. So i need to construct a retraction $r':X-V\longrightarrow X-N$ and show that there exists a homotopy between $i'\circ r'$ and $id'$ where $i':X-N\rightarrow X-V$ is the inclusion and $id'$ is the identity of $X-V$. So the main problem is to find $r'$, i know that we should use the retraction $r:N\longrightarrow V$ at some step but i don't see how to do it. thank you for your help!

  • $\begingroup$ You should also use the homotopy that witnesses the fact that $r$ is a deformation retract. $\endgroup$ – tomasz Oct 14 '12 at 18:08
  • $\begingroup$ It would be very kind if you give me a hint on how to do that!! $\endgroup$ – palio Oct 14 '12 at 18:11
  • 1
    $\begingroup$ Are you sure your statement is correct? Take $X$ a disk, $N$ its (closed) upper half and $V$ the equator. Then $N$ deformation retracts onto $V$, but $X-V$ isn't even connected. $\endgroup$ – Miha Habič Oct 14 '12 at 18:14
  • $\begingroup$ Actually we need $V$ to be closed in $X$ and $N$ to be an open neighborhood of $V$ that deformation retracts on $V$ . $\endgroup$ – palio Oct 14 '12 at 18:18
  • $\begingroup$ Still no good. If you fatten $N$ slightly in my previous example it still fits your conditions. $\endgroup$ – Miha Habič Oct 14 '12 at 18:24

The proposition stated is false. But I had the same idea and came up with this:

The deformation retract can be viewed as a homotopy $R(t,x)$ such that $R(0,x)=id$ and $R(1,x)=r$. Note that fixing $x$ defines a collection of paths

$$\gamma_x =R(t,x)$$

from $x$ to $r(x)$. Suppose that the retract has the additional property, that there exists a set $\beta \in N$ (imagine the boundary) together with a continuous function $\phi : N -V \to \beta$, such that $\gamma_{\phi(x)}$ crosses $x$ exactly once.

Then $X-V$ deformation retracts onto $(X-N)\cup \beta$ by

$$R^*(t,x)=\gamma_{\phi(x)} ((1-t_0)(1-t)) $$

where $t_0$ is the number in $I$, such that $\gamma_{\phi(x)}(t_0)=x$. (Define $R^*(t,x)=id$ outside $N$).


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.