2

I was trying to show that if $\gamma : [a,b] \to \mathbb R^3$ is a curve and $$ p(t) = \int_{t_0}^t |\gamma'(\tau)|d\tau$$ where $t_0 \in [a,b]$ then $p^{-1}: [c,d] \to [a,b]$ is a reparametrisation of $\gamma$. This is exercise 2 here.

But I am not even sure what I need to show:

  1. $p^{-1}$ is a bijection

  2. and $(p^{-1})'(s) \neq 0$ for all $s$

  3. and $\gamma ( p^{-1}(s)) = \gamma (s)$ for all $s$?

I tried to show 3. but the problem is that it does not even have same domain on both side. How to show that $p^{-1}$ is a reparametrisation of $\gamma$? Thank you for help.

dolan
  • 537

2 Answers2

2

You must show that $p^{-1}$ is continuous and injective. To achieve this it is enough to show that $p$ is strictly increasing and continuous.

some details

You have a $C^1$ function $\gamma\colon [a,b]\to \mathbb R^3$ with $|\gamma'(t)|\neq 0$ for all $t$ and define $$ p(t) = \int_a^t |\gamma'(\tau)|\, d\tau. $$ Let $L=p(b)$. L is the length of the curve $\gamma$. The function $$ p\colon [a,b] \to [0,L] $$ is differentiable and increasing because by the fundamental theorem of calculus: $$ p'(t) = |\gamma'(t)| > 0. $$ Hence $p$ is a bijection. It is injective because it is strictly increasing. It is surjective in view of the intermediate values theorem since $p(0)=0$, $p(b)=L$. So $p$ is invertible. Since $p'(t)\neq 0$ the inverse is also differentiable and $$ (p^{-1})'(s) = \frac{1}{p'(p^{-1}(s))}> 0. $$

  • is nonsense as you have already noticed in your post
  • – Emanuele Paolini Feb 19 '13 at 09:37
  • In $2$ I meant to write "$(p^{-1})' \neq 0$". Is it nonsense still? – dolan Feb 19 '13 at 16:19
  • The derivative different from 0 makes sense. But only if you are interested in differentiable reparameterizations, in general I would not require it. – Emanuele Paolini Feb 19 '13 at 16:22
  • How did you find out that $p^{-1}$ must be continuous and injective? What does it not have to be surjective also? (It seems to me that if I want to go from one map to the other an vice versa then the parametrisation $p$ has to be bijective?) I am sorry but I am new to this. – dolan Feb 19 '13 at 16:29
  • I added some details. – Emanuele Paolini Feb 19 '13 at 16:47
  • Many thank you. I understand now. We show it is injective because that shows that it is invertible. – dolan Feb 20 '13 at 07:53
  • I now understand it better! Thank you! The 3 should be: $p^{-1}(s) = \gamma (s)$ (using information on reparametrization on Wikipedia). – dolan Feb 20 '13 at 08:31
  • @user58975: 3 should be $\gamma(p^{-1}(p(t))) = \gamma(t)$ which is somewhat trivial. – Emanuele Paolini Feb 20 '13 at 08:34
  • I posted a summary of my understanding XD – dolan Feb 20 '13 at 08:44
  • Maybe we can say $p$ is surjective by definition since $[0,L]=p([a,b])$? Or would it be wrong and we have to use intermediate value theorem? – dolan Feb 20 '13 at 09:06
  • By definition $L=p(b)$ and $0=p(a)$, so the extremes are reached. To prove the intermediate values are reached you must use the corresponding theorem. – Emanuele Paolini Feb 20 '13 at 09:20