[Math] If $\alpha$ is a unit speed curve of constant curvature lying in a sphere, then $\alpha$ is a circle.

differential-geometryvector analysis

I'm trying to solve the following problem but got stuck along the way. I would like some help on getting this through.

Prove that if $\alpha$ is a unit speed curve of constant curvature lying in a sphere, then $\alpha$ is a circle.

Solution: My goal is to show that the torsion is zero.

We have $\alpha \cdot \alpha =r^2$, so taking the derivatives, $T \cdot \alpha =0$. So we can let $\alpha=xN+yB$ for some functions $x$ and $y$. Now differentiating the previous equation again, we get $T' \cdot \alpha + T\cdot T=0$, so using Frenet Formula, we get $\kappa N \cdot \alpha =-1$, and another differentiation and Frenet Formula yield $\tau B\cdot \alpha =0$.

So by the assumption on $\alpha$, we get $\tau y=0$.

However, here is where I have a problem. If I know that $y\neq0$, then I'm done. But I cannot guarantee that $y$ is nonzero, so I can't show that $\tau$ must be zero. How can I solve this?

I would greatly appreciate any solutions or suggestions.

Best Answer

Suppose $ \tau $ is non-zero. Then $ B \cdot \alpha = 0 $, so that $ \alpha = -\frac{1}{\kappa} N $. Differentiating this once gives: $$ T = -\frac{1}{\kappa} \frac{dN}{dt} = -\frac{1}{\kappa} ( -\kappa T + \tau B ) = T -\frac{\tau}{\kappa} B $$ Which means $ \tau = 0 $!

Related Question