Differentiability of rectifiable curves

618 Views Asked by At

I have the following question. Let $\gamma:[a,b]\rightarrow X$ be a rectifiable curve in a metric space $(X,d)$. If we consider the length function of $\gamma$, $L:[a,b]\rightarrow [0,L(\gamma)]$, we see that it is clearly a continuous, monotone increasing function. Therefore it is differentiable. In other words $L':[a,b]\rightarrow \mathbb{R}$ exists and $L(b)-L(a) = \int L'(t)dt$.

Now take $\delta>0$. Then we have that $d(\gamma(t),\gamma(t+\delta)) \leq L(t+\delta)-L(t)$. By dividing by $\delta$, we get:

$\frac{d(\gamma(t),\gamma(t+\delta))}{\delta} \leq \frac{L(t+\delta)-L(t)}{\delta}$.

Since $L$ is differentiable, the limit $\lim_{\delta\rightarrow 0}\frac{L(t+\delta)-L(t)}{\delta}$ exists. If I could bound $\frac{d(\gamma(t),\gamma(t+\delta))}{\delta}$ on the other side by the same limit I could define $|\gamma'(t)|$ as the limit

$\lim_{\delta \rightarrow 0}\frac{d(\gamma(t),\gamma(t+\delta))}{\delta}$.

I have tried to do this in many ways but haven't been able to. I know it should be possible since this construction is mentioned (without any details) in a survey about metric spaces of bounded curvature below.

Can you help me?

EDIT: I don't know if this would be of any use to anyone but I found a complete proof of this in the book The Geometry of Geodesics by Busemann.

1

There are 1 best solutions below

2
On

"Therefore it is differentiable"?

Continuous monotone increasing functions need not be differentiable at every point. Consider the function $f : [0,1] \to \mathbb{R}$ defined by $$f(t) = \begin{cases} t & \text{if $t \in [0,\frac{1}{2}]$} \\ 2t - \frac{1}{2} &\text{if $t \in [\frac{1}{2},1]$} \end{cases} $$