Is this characterization of a curve on a surface using lines of curvature correct?

100 Views Asked by At

Let $\alpha$ be a curve on a surface with Gaussian curvature $K=-1$ and let $N$ be the surface normal along $\alpha$. Assume that \begin{equation} \langle\alpha'(t),N(t)\rangle=0\quad\text{and}\quad\langle\alpha'(t),N'(t)\rangle\neq 0 \end{equation} for all $t$. I came across a document where the author says that if this holds then

$\alpha'(t)$ and $N'(t)$ are everywhere or nowhere parallel, i.e., $\alpha$ is everywhere or nowhere tangent to a principal direction.

I don't seem to know how to use the two given assumptions to prove that $\alpha'(t)$ and $N'(t)$ are everywhere or nowhere parallel, nor that $\alpha$ is everywhere or nowhere tangent to a principal direction. Actually, it is not clear to me why these two conclusions are equivalent: if for each $t$ we have that $\alpha$ is tangent to a principal direction, then $\alpha$ is a line of curvature. This means, using Rodrigues' curvature formula, that $N'(t)+k_i(t)\,\alpha'(t)=0$, where $k_i(t)$ is the principal curvature associated with the direction to which $\alpha$ is tangent. Then, N'(t) is everywhere parallel to $\alpha'(t)$ provided the $k_i(t)$ are constant, which is not the case in general.

Any idea regarding this proof of why the staments are not correct is really appreciated.

1

There are 1 best solutions below

1
On BEST ANSWER

This seems to be false. What is correct is that $\alpha$ is never tangent to an asymptotic direction. The curve can certainly vary from a line of curvature without ever being tangent to an asymptotic direction. (The first condition is totally redundant.)