I've encountered the following problem that I'm sure someone more topologically inclined can answer:

Say that a homotopy of maps $f:X\times[0,1)\to Y$ between two compact smooth manifolds $X$ and $Y$ is good (I don't know the exact term for it if there is one) if the homotopy is constant in time outside of some coordinate chart in $Y$, in which case we'd say the homotopy is supported in the chart.

Suppose you have two continuous maps $f_0$ and $f_1$ between compact smooth manifolds $X$ and $Y$ which are homotopic. Cover $Y$ with finitely many coordinate charts $U_\alpha$. My question is the following:

Are $f_0$ and $f_1$ homotopic with respect to a finite sequence of good homotopies supported in the elements $U_\alpha$ of the covering? Can the number of good homotopies required be uniformly bounded by topological data like the homotopy class of $f_0$ or the number of elements in the covering?

I am interested in this question because I have a functional which I can get a good estimate for if the above is true.


1 Answer 1


Yes, you can always connect $f_0$ and $f_1$ by a finite sequence of good homotopies supported in any given open cover (and in fact, this sequence of good homotopies can be chosen to be homotopic to $f$ relative to $X\times \{0,1\}$). By pulling back the $U_\alpha$ to an open cover of $X\times [0,1]$ via $f$, it suffices to show this in the universal case where $Y=X\times [0,1]$ and $f$ is the identity. By refining the open cover, we may assume we have an open cover consisting of all the sets of the form $V_i\times (n\epsilon,(n+3)\epsilon)$ for some finite open cover $\{V_1,\dots,V_m\}$ of $X$ and some $\epsilon>0$. Let $\{\varphi_i\}$ be a partition of unity subordinate to $\{V_i\}$. Then you can connect $f_0$ to $f_1$ through good homotopies as follows. First, homotope $f_0:x\mapsto (x,0)$ to the map $x\mapsto (x,\epsilon \varphi_1)$ through the homotopy $(x,t)\mapsto (x,t\epsilon\varphi_1)$. Then homotope to $x\mapsto(x,\epsilon(\varphi_1+\varphi_2))$ similarly, and so on; after $m$ steps you will have reached $x\mapsto (x,\epsilon\sum \varphi_i)=(x,\epsilon)$. Now repeat to get to $x\mapsto (x,\epsilon+2)$, and so on until you reach $x\mapsto (x,1)$.

On the other hand, you can't get any uniform bound on the number of steps needed using only $f_0$ or the open cover. For instance, consider the case where $X=Y=S^1$, $f_0$ is a constant map, and $f_1$ winds around $n$ times in one direction and then $n$ times in the opposite direction. By lifting to the universal cover, it is not hard to show that it takes at least $n$ good homotopies to get from $f_0$ to $f_1$ (for any covering of $Y$ by charts).

  • $\begingroup$ I had expected that, thank you! It still seems like you can uniformly bound the number of good homotopies needed by, say, the Dirichlet energy of the individual maps. $\endgroup$ Oct 16, 2015 at 23:38

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.

Not the answer you're looking for? Browse other questions tagged or ask your own question.