5
$\begingroup$

Given a nowhere vanishing continuous map $f:\mathbb{C}^*\to \mathbb{C}^*$, how do I show that there exist an integer $m\in \mathbb{Z}$ and a continuous map $g:\mathbb{C}^*\to \mathbb{C}$ such that $f$ is of the form $$f(z)=z^me^{g(z)}, ~~~~~ \forall z\in \mathbb{C}^*.$$ Where $\mathbb{C}^*= \mathbb{C}\setminus\{0\}.$

I come across this problem while trying to show that the cokernel of the exponential map in the following sequence $$0\to \mathbb{Z}\stackrel{2\pi i}{\rightarrow} \mathcal{O}(\mathbb{C}^*)\stackrel{exp} {\rightarrow}\mathcal{O}^*(\mathbb{C^*})\to 1$$ is isomorphic to $\mathbb{Z}=H^1(\mathbb{C}^*, \mathbb{Z}).$ Where $\mathcal{O}(\mathbb{C^*})$ denotes the group of all continuous maps from $\mathbb{C}^*$ to $\mathbb{C}$ and $\mathcal{O}^*(\mathbb{C^*})$ denotes the group of all continuous maps from $\mathbb{C}^*$ to $\mathbb{C}^*$.

I have proved that for all integers $m\in \mathbb{Z}$, the function sending $z\mapsto z^m$ belongs to distinct classes of the quotient group $\frac{\mathcal{O}^*(\mathbb{C^*})}{exp\left(\mathcal{O}(\mathbb{C^*})\right)}$ and I want to show that these are the only classes in the quotient. For that I have to solve the above problem.

Is there an elementary proof for this problem where I dont have to use cohomology of the punctured complex plane. Thanks in advance.

$\endgroup$
9
  • 2
    $\begingroup$ $\mathcal{O}$ usually denotes holomorphic functions, not continuous functions. Regardless, you can reduce to the case where $f(x)=1$ where $x > 0$ (because maps $(0,\infty) \rightarrow \mathbb{C}^{\times}$ are exponentials). We can then define continuous functions $\theta_r: [0,2\pi] \rightarrow \mathbb{R}$ such that $f(re^{it})=|f(re^{it})|e^{2\pi i\theta_r(t)}$ and $\theta_r(0)=0$. Now, analysis shows that $r \longmapsto \theta_r$ must be continuous and $\theta_r(2\pi)$ is an integer, which is your $m$ – if $\theta_r(2\pi)=0$, then $\theta_r(t)=g(re^{it})$ for some continuous $g$, QED. $\endgroup$
    – Aphelli
    Commented Sep 22, 2022 at 12:48
  • $\begingroup$ @Aphelli Can you please expand as answer.. I could not understand.. $\endgroup$
    – Uncool
    Commented Sep 23, 2022 at 5:26
  • $\begingroup$ What part was unclear? The reduction to the case $f((0,\infty))=\{1\}$? The construction and continuity of the $\theta_r$ (a continuous argument of $f$ on the circle of radius $r$)? The fact that $f$ is a global exponential if $m=0$? $\endgroup$
    – Aphelli
    Commented Sep 23, 2022 at 14:37
  • $\begingroup$ All the parts was unclear to me. $\endgroup$
    – Uncool
    Commented Sep 23, 2022 at 15:08
  • 1
    $\begingroup$ Indeed, that’s the main idea of the construction. It just takes a little work to make it explicit without homotopy or cohomology. I’m willing to elaborate on the details but I’d rather the author tried to work on it for a bit (even if it’s just making their questions more precise). $\endgroup$
    – Aphelli
    Commented Sep 23, 2022 at 22:37

1 Answer 1

3
+50
$\begingroup$

We hide everything fancy in

Lemma. Let $U\subseteq \Bbb C$ be simply connected and $h\colon U\to\Bbb C^\times$ continuous. Then there exists continuous $\ell_h\colon U\to \Bbb C$ such that $h=\exp\circ \ell_h$. Moreover, such $\ell_h$ is determined up to an additive constant $\in2\pi i\Bbb Z$.

Proof. Easy for sufficiently small open sets. These can be glued together. $\square$

With $U_1:=\Bbb C\setminus (-\infty,0]$, $U_2:=-U_1=\Bbb C\setminus [0,+\infty)$, $f_i:=f|_{U_i}$ we thus find continuous functions $\ell_{f_j}\colon U_j\to \Bbb C$ such that $f(z)=e^{\ell_{f_j}(z)}$ for $z\in U_j$. Moreover, $\ell_{f_1}-\ell_{f_2}$ is a locally constant multiple of $2\pi i$, i.e., there exist $m_+,m_-\in\Bbb Z$ such that such that $\ell_{f_1}(z)-\ell_{f_2}(z)=2\pi i m_+$ for positive and $\ell_{f_1}(z)-\ell_{f_2}(z)=2\pi im_-$ for negative imaginary part of $z$. In fact, we may assume wlog that $\ell_{f_1}(z)=\ell_{f_2}(z)$ on the upper half plane and $\ell_{f_2}(z)=\ell_{f_1}(z)+2\pi i m$ on the lower half plane (with $m=m_+-m_-$).

Using the identity instead of $f$ on $U_1$, we likewise find $\ell_{\operatorname{id}_j}\colon U_j\to\Bbb C$ (in other words, branches of $\ln$) such that $z=e^{\ell_{\operatorname{id}_j}(z)}$ for $z\in U_j$. Again, we may assume $\ell_{\operatorname{id}_1}(z)=\ell_{\operatorname{id}_2}(z)$ on the upper half plane and then can verify that $\ell_{\operatorname{id}_2}(z)=\ell_{\operatorname{id}_1}(z)+2\pi i$ on the lower half plane.

It follows that $$ g(z):=\begin{cases}\ell_{f_1}(z)-m\ell_{\operatorname{id}_1}(z)&z\in U_1\\\ell_{f_2}(z)-m\ell_{\operatorname{id}_2}(z)&z\in U_2\end{cases}$$is a well-defined continuos map $\Bbb C^\times\to\Bbb C$ with $$ e^{g(z)}=\frac{f(z)}{z^m}$$ for all $z\in\Bbb C^\times$.

$\endgroup$
0

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .