1

that is my trouble:

Let $\alpha : I \rightarrow R^3 $ a regular curve parameterized by its arc lenght, that is contained in the unit sphere.

If $k(s)>0$ and $\tau(s)\neq 0$ , $\forall s \in I$ then

$$ \frac{1}{k^2(s)} + \frac{(k'(s))^2}{k^4(s) \tau^2(s)} = 1 $$

my efforts were useless

dedekind1
  • 231

1 Answers1

2

We may easily generalize to the case in which $\alpha(s)$ lies in the sphere of radius $R$ centered at some point $C \in \Bbb R^3$, that is, to

$(\alpha(s) - C) \cdot (\alpha(s) - C) = R^2. \tag 1$

The key idea here is to express $\alpha(s) - C$ itself in terms of $T(s)$, $N(s)$, and $B(s)$, the Frenet frame of the curve $\alpha(s)$. We obtain, upon differentiating (1) with respect to the arc-length $s$,

$(\alpha(s) - C)' \cdot (\alpha(s) - C) = 0, \tag 2$

or

$\dot \alpha(s) \cdot (\alpha(s) - C) = 0; \tag 3$

since

$\dot \alpha(s) = T(s), \tag 4$

(3) becomes

$T(s) \cdot (\alpha(s) - C) = 0, \tag 5$

which shows that the $T(s)$ component of $\alpha(s) - C$ is zero. We next differentiate (5):

$\dot T(s) \cdot (\alpha(s) - C) + T(s) \cdot T(s) = 0, \tag 6$

and using the first Frenet-Serret equation

$\dot T(s) = \kappa(s) N(s) \tag 7$

and the fact that $T(s)$ is a unit vector,

$T(s) \cdot T(s) = 1, \tag 8$

we transform (6) into

$\kappa(s) N(s) \cdot (\alpha(s) - C) + 1 = 0, \tag 9$

and readily express the $N(s)$ component of $\alpha(s) - C$ as

$N(s) \cdot (\alpha(s) - C) = -\dfrac{1}{\kappa(s)}; \tag{10}$

taking the $s$ derivative of this equation yields

$\dot N(s) \cdot (\alpha(s) - C) + N(s) \cdot T(s) = \dfrac{\dot \kappa(s)}{\kappa^2(s)}; \tag{11}$

since

$N(s) \cdot T(s) = 0, \tag{12}$

and the second Frenet-Serret equation affirms that

$\dot N(s) = -\kappa(s) T(s) + \tau(s) B(s), \tag{13}$

we may now write (11) in the form

$(-\kappa(s) T(s) + \tau(s) B(s)) \cdot (\alpha(s) - C) = \dfrac{\dot \kappa(s)}{\kappa^2(s)}, \tag{14}$

and so by virtue of (5),

$ \tau(s) B(s) \cdot (\alpha(s) - C) = \dfrac{\dot \kappa(s)}{\kappa^2(s)}, \tag{15}$

immediately leading to

$B(s) \cdot (\alpha(s) - C) = \dfrac{\dot \kappa(s)}{\kappa^2(s) \tau(s)}, \tag{16}$

the $B(s)$ component of $\alpha(s) - C$; we combine (5), (10) and (16) and arrive at

$\alpha(s) - C = -\dfrac{1}{\kappa(s)} N(s) + \dfrac{\dot \kappa(s)}{\kappa^2(s) \tau(s)} B(s), \tag{17}$

now using

$N(s) \cdot B(s) = 0, \tag{18}$

it follows from (1) and (17) that

$R^2 = (\alpha(s) - C) \cdot (\alpha(s) - C) = \dfrac{1}{\kappa^2(s)} + \dfrac{(\dot \kappa(s))^2}{\kappa^4(s) \tau^2(s)}, \tag{19}$

the requested formula; we observe that when $R = 1$ we obtain the exact relationship given in the text of the question itself. $OE\Delta$.

See also this question.

Robert Lewis
  • 72,871
  • 1
    You've posted this same solution at least a dozen times :) I just link to my previous solution rather than continuing to retype. Admittedly, this one isn't quite the usual post. – Ted Shifrin Dec 03 '20 at 22:23
  • @TedShifrin: OK Ted, I know you are speaking hyperbolically (to borrow a term from geometry) when you say "at least a dozen times". I didn't even remember my previous solution until I found it in a file I keep which has the text of all my answers. But thanks. *Cheers!!!* – Robert Lewis Dec 03 '20 at 22:26
  • 1
    I've personally seen well over six. :) I think we should just tell all these people that there's one and only one technique: Expand the position vector as a linear combination of the Frenet frame vectors. Let them work on it. – Ted Shifrin Dec 03 '20 at 22:28
  • @Ted Shifrin: Oh come on, six? Would you mind showing me the links? But perhaps you're right about a "generic technique". – Robert Lewis Dec 03 '20 at 22:29
  • I'm not going to do all that work. Just search your answers under user:67071 [differential-geometry] – Ted Shifrin Dec 03 '20 at 22:31
  • @TedShifrin: OK Ted, I did a quick preliminary search (by the way your seach syntax (user:67071 [differential-geometry] was new to me, so thanks for that as well) and went back about 60 posts (I'll look more later) and only found this and the one I linked. Now I'm curious to search more. – Robert Lewis Dec 03 '20 at 22:40