# Why there is no continuous argument function on $\mathbb{C}\setminus\{0\}$?

An argument function $\phi$ on $\mathbb{C}\setminus\{0\} = \mathbb{R}^2\setminus\{0\}$ is a function such that for every $z\neq 0$ it holds that $$z = |z|\exp(i\phi(z)).$$

Is there an elementary and easy proof that there is no continuous argument function on $\mathbb{C}\setminus\{0\}$? I would like to see a proof which uses as less complex analysis as possible. Probably only topological arguments and no complex numbers whatoever?

-
No complex numbers? Could you please elaborate on what that would mean in the context of this problem? –  Jonas Meyer Dec 13 '11 at 15:19

Since the question itself uses complex numbers, I don't think it's possible to give a proof that doesn't use them.

Anyway, the following is a simple argument that doesn't use any complex analysis.

Suppose such a function $\phi$ exists, and consider the function $\psi(t) = \frac{1}{2 \pi}(\phi(e^{i t}) - t)$, for real $t$. $\psi$ is continuous and $\mathbb{Z}$-valued, so it must be constant, say $\psi \equiv k$.

So $\phi(e^{it}) = 2 \pi k + t$, but then

$$2 \pi k = \phi(1) = \phi\left(e^{2 \pi i}\right) = 2 \pi (k + 1),$$

The $\phi(2\pi)$ is too much, right? –  Dirk Dec 13 '11 at 17:25
If there were a continuous argument function, its restriction to the unit circle would be a homeomorphism onto its image in $\mathbb R$. (It is injective, the circle is compact, and $\mathbb R$ is Hausdorff.) But the image of the circle would be a compact and connected subset of $\mathbb R$, thus a closed and bounded interval, which is not homeomorphic to the circle. This contradiction shows that such a function doesn't exist.
Dirk, For one thing, note that it is not just any subset of the line, but an interval $[a,b]$ for some $a<b$. From $[a,b]\subset \mathbb R$ you can remove 2 points and have a connected set remaining, which is not true for the circle. Or, perhaps simpler, show that if $\varphi:S^1\to [a,b]$ is a continuous surjective map, $\varphi(z)=a$ and $\varphi(w)=b$, then $\varphi$ must map both of the arcs between $z$ and $w$ onto $(a,b)$, and therefore $\varphi$ is not injective. As in Paolo's argument, connectedness of $S^1$ is essential. (Or, the circle is not contractible.) –  Jonas Meyer Dec 13 '11 at 17:26