5

I've heard that in a manifold of dimension four or higher, if one takes a two-dimensional ribbon and gives it two half-twists (ie. two $180$-degree twists), it is the same as if the ribbon has no twists. Thus in $\mathbb{R}^3$, a ribbon can have $z$ half-twists where $z \in \mathbb{Z}$ because it can have any number of half-twists in either direction, whereas in $\mathbb{R}^4$, it can only have $z$ half-twists where $z \in \mathbb{Z} / 2\mathbb{Z}$.

I don't know the theory behind this, and it doesn't seem obvious. Can anyone offer a proof or some intuition for this fact?

NoName
  • 2,975

1 Answers1

3

$\newcommand{\Reals}{\mathbf{R}}$Consider the parallel unit circles in $\Reals^{3} \times \{0\} \subset \Reals^{4}$ parametrized by $$ c_{0}(u) = (\cos u, \sin u, 0, 0),\qquad c_{1}(u) = (\cos u, \sin u, 1, 0),\quad 0 \leq u \leq 2\pi, $$ and think of these as the boundary of an untwisted ribbon. "Tip" the second so that only two points lies in $\Reals^{3} \times \{0\}$. Slide the center appropriately, then tip the circle back into $\Reals^{3} \times \{0\}$ so that the two circles are now linked, e.g., via \begin{align*} c(t, u) &= \bigl(t + \cos u, (1 - t)\sin u, 1 - t + t\sin u, \sqrt{2t(1 - t)}\sin u\bigr) \\ &= \bigl(t + \cos u, 0, 1 - t, 0\bigr) + \sin u\bigl(0, 1 - t, t, \sqrt{2t(1 - t)}\bigr),\quad 0 \leq t \leq 1, \end{align*} whose image for each $t$ is a unit circle centered at $(t, 0, 1 - t, 0)$, and for which $c(0, u) = c_{1}(u)$ and $c(1, u) = (1 + \cos u, 0, \sin u, 0)$.

The mapping $H:[0, 1] \times [0, 2\pi] \times [0, 1] \to \Reals^{4}$ defined by $$ H_{t}(u, v) = (1 - v)c_{0}(u) + vc(t, u),\quad 0 \leq t \leq 1,\ 0 \leq u \leq 2\pi,\ 0 \leq v \leq 1 $$ is a smooth homotopy through embeddings from the cylinder \begin{align*} H_{0}(u, v) &= (1 - v)c(0, u) + vc(0, u) \\ &= (\cos u, \sin u, v, 0) \end{align*} to the twice-twisted ribbon \begin{align*} H_{1}(u, v) &= (1 - v)c_{0}(u) + vc(1, u) \\ &= (v + \cos u, (1 - v)\sin u, v\sin u, 0). \end{align*} The animation shows its projection to $\Reals^{3}$.

Deforming a cylinder to a twice-twisted ribbon in four-space

  • 1
    So it's the same principle as linking unlinked circles that lie in a hyperplane of $\mathbb{R}^4$. Thanks for the animation. – NoName May 12 '17 at 01:33
  • Yes, you can think of (full-)twisting a ribbon as adding one to the linking number of two loops in a hyperplane. :) – Andrew D. Hwang May 12 '17 at 02:35