My textbook says: Suppose that $E$ is a convex region in the plane bounded by a curve $C$. [Where a convex region is defined as for all $x, y \in E, sx + ty \in E$ where $0 \le s, t \le 1$ and $s + t = 1$.] Show that $C$ has a tangent line except at a countable number of points.

So my thinking is roughly that points without tangent lines look like sharp corners with some angle $\theta < 180^\circ$ and so if $\theta_m$ is the largest $\theta$ in $C$ then the most corners you can pack into $C$ is the regular $n$-gon with $n=\frac{2}{180-\theta_m}$. Except that seems to suggest that $C$ must have a tangent line except at finitely many points whereas the question clearly asks about countably many. So what's the pathological convex curve with infinitely many discontinuities that thwarts my proof?

  • 5
    $\begingroup$ Start horizontally. For $n$ from $1$ to $\infty$, move forward a distance of $2^{-n}$, and turn counterclockwise by an angle of $2^{-n}\cdot 90°$. Close the curve. $\endgroup$ – Daniel Fischer Jun 11 '15 at 7:35
  • $\begingroup$ Bahaha, so literally nothing I said above is remotely valid. Time to burn all my notes and start again, thanks! $\endgroup$ – Tobias Jun 11 '15 at 7:42
  • $\begingroup$ @Tobias: You can, in fact, arrange that the boundary of $E$ contains no interval, i.e., you can make the corners of $C$ dense in $C$. $\endgroup$ – Andrew D. Hwang Jun 11 '15 at 13:04

Line segments from $(n,n^2)$ to $((n+1),(n+1)^2)$ for each integer $n$ you should draw. A nice and simple convex region you will get.

  • $\begingroup$ Liking the Yoda grammar! :D $\endgroup$ – 5xum Jun 11 '15 at 7:49
  • $\begingroup$ @5xum: Hmm.. Like it I did not think anyone would. Better it is now. $\endgroup$ – user21820 Jun 11 '15 at 7:53

Here's a compact example. Start with a circle of radius $1$. Cut off a minor segment with a chord of length $1$, and close the resulting major segment. From this major segment, cut off a minor segment with an adjacent chord of length $\frac12$, and close the rest. Continue this process, cutting off minor segments with adjacent chords successively of length $\frac14,\frac18,...,1/2^n,...$. The resulting figure is convex with no defined tangent at the infinitely many corners between the successive chords.


Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.