The Path Lifting Theorm for Local Homeomorphisms

808 Views Asked by At

We know that the path lifting theorem is true for covering maps. The map $p: \mathbb{R}_+ \to \mathbb{S}^1$ given by $p(t)=(\cos(2\pi t), \sin(2 \pi t))$ gives an example of a local homeomorphism which is not a covering map.

I want to know what really goes wrong in this case, due to which the path lifting theorem fails to hold good here. Thanks for any help.

1

There are 1 best solutions below

7
On BEST ANSWER

Your example is the "standard" covering of $S^1$ by $\mathbb R$ restricted to $\mathbb R_+$. Since you state that this is a local homeomorphism, I assume that $0 \not\in\mathbb R_+$. From the parametrization, we see that $p$ wraps $\mathbb R_+$ around $S^1$ in counterclockwise orientation, starting at $1$. Now consider the path $$ \gamma:[0,1]\rightarrow S^1,\ t \mapsto ie^{-2\pi it}. $$ This path goes one time around $S^1$ in clockwise orientation, starting at $i$. This path can't be lifted through $p$ to the starting point $\frac{1}{4} \in \mathbb R_+$. Such a lift would move from $\frac{1}{4}$ downwards towards $0$ and "fall off" $\mathbb R_+$ at $0$ after a quarter of the way. So the path lifting theorem fails.

Here's a more detailed explanation of why a purported lift of $\gamma$ with starting point $\frac{1}{4}$ must move from $\frac{1}{4}$ straight down to $0$.

So let's assume that $\widetilde\gamma:[0,1]\rightarrow \mathbb R_+$ is a lift of $\gamma$ with starting pont $\frac{1}{4}$. This means we have $\widetilde\gamma(0) = \frac{1}{4}$, $p \circ \widetilde\gamma = \gamma$, and of course $\widetilde\gamma$ is continuous.

Denote by $T^1 = \{z \in S^1\ |\ Im\ z > 0 \}$ the open upper hemisphere and put $U := p^{-1}(T^1) \subseteq \mathbb R_+$. If we also put $J_k := (k,k+\frac{1}{2}) \subseteq \mathbb R_+$ for $k \in \mathbb N_0$, then we have $$ U = \bigcup_{k \in \mathbb N_0} J_k $$ and each of the intervals $J_k$ is a connected component of $U$. Moreover $p|_{J_k}:J_k \rightarrow T^1$ is a homeomorphism. Now, $\gamma([0,\frac{1}{4})) \subseteq T^1$, so we must have $\widetilde\gamma([0,\frac{1}{4})) \subseteq U$. Since $\widetilde\gamma([0,\frac{1}{4}))$ is connected and $\widetilde\gamma(0) = \frac{1}{4}$, we must have $\widetilde\gamma([0,\frac{1}{4})) \subseteq J_0$. So we have $$ \gamma|_{[0,\frac{1}{4})} = (p|_{J_0})\circ(\widetilde\gamma|_{[0,\frac{1}{4})}). $$ But $p|_{J_0}$ is a homeomorphism, so we have $$ \widetilde\gamma|_{[0,\frac{1}{4})} = (p|_{J_0})^{-1}\circ\gamma|_{[0,\frac{1}{4})}. $$ Now, as $t \in [0,\frac{1}{4})$ goes from $0$ to $\frac{1}{4}$, $\gamma|_{[0,\frac{1}{4})}(t)$ traces out the top right quarter of $S^1$ from $i$ to $1$. By looking at the construction of $p$, this implies that $(p|_{J_0})^{-1}\circ\gamma|_{[0,\frac{1}{4})}(t)$ traces out the line segment from $\frac{1}{4}$ down to $0$. So $\widetilde\gamma|_{[0,\frac{1}{4})}(t)$ approaches $0$ as $t$ tends to $\frac{1}{4}$ from below. So $\widetilde\gamma$ can't be continuously extended to $\frac{1}{4}$. Thus we have a contradiction and $\widetilde\gamma$ doesn't exist.