It is well-known that continuous space-filling curves exist. But can they be Lipschitz?
Specifically, is there a Lipschitz map from [0,1] onto [0,1]x[0,1]?
It is well-known that continuous space-filling curves exist. But can they be Lipschitz?
Specifically, is there a Lipschitz map from [0,1] onto [0,1]x[0,1]?
On
Thanks for your answers. The answer I came up with doesn't mention Hausdorff dimension, although you might say it's there in disguise:
Suppose there is such a map. Given a positive integer $N$, subdividing $[0,1]$ into $N$ equal intervals and considering the images of these will give a cover of $[0,1]^2$ having area $\leq NC/N^2$ where $C$ is a constant independent of $N$ (this is since the image of a sub-interval lies in a ball of diameter proportional to $1/N$). Letting $N \to \infty$ gives $\big| [0,1]^2 \big|=0$, which cannot be.
It's a general fact that if $f$ is a Lipschitz mapping, then the Hausdorff dimension satisfies
$$\dim_{H} f(F) \le \dim_H F$$
for all $F$. If $F$ is the unit interval, then $\dim_H F = 1$, while the Hausdorff dimension of the square is $2$; this means no such Lipschitz map can exist.
The above inequality can be found, e.g. as Corollary $2.4$ in Falconer's Fractal Geometry.