About the definition of vertex

curvaturedifferential-geometry

Following is the definition of a vertex from $\textit{Elementary Differential Geometry}$ by Pressley

Definition.
A vertex of a curve $\gamma(t)$ in $\mathbb{R^2}$ is a point where its signed curvature $\kappa_s$ has a stationary point, i.e., where $d\kappa_s/dt=0$.

And here is an exercise and solution from the book about how to find number of vertices of a parameterised curve.

Show that the limacon $\gamma(t)=((1+2\cos t)\cos t,(1+2\cos t)\sin t)$ has only two vertices.

Solution.
$\dot\gamma=(\color{red}{-\sin t-2\sin2t},\color{blue}{\cos t+2\cos 2t})$ and $\Vert\dot\gamma\Vert=\color{green}{\sqrt{5+4\cos t}}$, so the angle $\varphi$ between $\dot\gamma$ and the $x$-axis is given by $$\cos\varphi=\frac{\color{red}{-\sin t-2\sin 2t}}{\color{green}{\sqrt{5+4\cos t}}},\sin\varphi=\frac{\color{blue}{\cos t+2\cos 2t}}{\color{green}{\sqrt{5+4\cos t}}}$$
Differentiating the second equation gives
$$\dot\varphi\cos\varphi=\frac{-\sin t(24\cos^2 t+42\cos t+9)}{(5+4\cos t)^{3/2}}\text{ so}$$
$$\dot\varphi=\frac{\sin t(24\cos^2 t+42\cos t+9)}{(5+4\cos t)(\sin t+2 \sin 2t)}=\frac{9+6\cos t}{5+4\cos t}$$
Hence, if $s$ is the arc-length of $\gamma$, $\kappa_s=\frac{d\varphi/dt}{ds/dt}=\frac{9+6\cos t}{(5+4\cos t)^{3/2}}$ , so $\dot\kappa_s=\frac{12\sin t(2+\cos t)}{(5+4\cos t)^{5/2}}$. This vanishes at only two points of the curve, where $t=0$ and $t=\pi$.

Since for signed curvature $\kappa_s$, have $\kappa_s=\vert\kappa\vert$, that $d\kappa_s/dt=0$ iff $d\kappa/dt=0$, is it really necessary to find $\kappa_s$ every time; In general, can we just use $\kappa$, as it's much easy to calculate? Like suggested in this post (defined vertex as where $d\kappa/dt=0$).

For example in above exercise we can just compute $\kappa=\frac{9+6\cos t}{(5+4\cos t)^{3/2}}=\kappa_s$, and find where $d\kappa/dt$ vanishes that gives the same result, which seems to be a much shorter approach than compute the sign curvature. (However, all the examples from the book are calculating $\kappa_s$ instead.) So I'm not sure, based on definition of this book, is this approach correct?

Best Answer

The reason for the signed in the definition is that you want to rule out a point like the origin on $y=x^3$. The signed curvature of this curve is $\kappa_s(x) = \dfrac{6x}{(1+9x^4)^{3/2}}$, and it has no critical point at $x=0$. However, the unsigned curvature $\kappa = |\kappa_s|$ has a minimum at $x=0$ (although it is not differentiable there).

Vertices should correspond to extreme points of curvature, not just any old critical points.

If you do not like the lack of differentiability of $\kappa$ in the example, change to $y=x^5$ instead. The same phenomenon persists, but now $\kappa$ will be differentiable at $x=0$.

Related Question