Let $L$ be a positive real, $(X,d)$ a complete metric space, and $\gamma:[0,L[\rightarrow X$ be a local distance preserving map(where $[0,L[$ is equipped with the Euclidean metric). Must $\gamma$ be extendable to a local distance preserving map from $[0,L]$ to $X$ ? Can at least one extend gamma continuously to $[0,L]$ ?
Thank you,
Here is a sketch of the construction of a counter-exmaple. Consider the map $\gamma : [0, 1) \to \Bbb{C}$ defined by
$$ \gamma(t) = \frac{\exp(2\pi i \cdot 2^{k+1} t) - 1}{2\pi \cdot 2^{k+1}} \qquad \text{for } t \in [1 - 2^{-k}, 1 - 2^{-k-1}). $$
Its image $X = \gamma([0, 1))$ looks like the following figure:
$\hspace{12em}$
Equip $X$ with the intrinsic metric $d$. Then it is possible to check that $\gamma : I \to X$ is a local distance-preserving map. Intuitively, this is because $\gamma$ is distance-preserving on each circular arc.
Now, since $\gamma(1 - 2^{-k}) = 0$ for each $k$, the path $\gamma$ extends to a continuous function $\tilde{\gamma} : [0, 1] \to X$ with $\tilde{\gamma}(1) = 0$. However, this also shows that $\tilde{\gamma}$ cannot be distance-preserving near $1$.