How do I prove that every meromorphic function on the extended plane is a rational function?
-
Compare http://math.stackexchange.com/questions/123259/why-is-every-meromorphic-function-on-hat-mathbbc-a-rational-function. – Martin R Nov 30 '15 at 08:37
2 Answers
Because $\infty$ is a removable singularity or pole of a meromorphic function $f(z)$, $\exists$ $R>0$, s.t. $f\in H(\{z:R<|z|<\infty\})$.
We note that $f$ has at most finite poles in $\{z:|z|\le R\}$. If not, then the finite poles $\{z_j\}$ must have a convergent sub sequence. Let $z_j\to a(j\to \infty)$. Then $|a|<R$ and $a$ is not a isolated pole, which is impossible.
Let $z_1,z_2,\cdots,z_n$ be all poles of $f$ in $\{z:|z|\le R\}$, with orders of $m_1,m_2,\cdots,m_n$. $f$ has Laurent expansion besides $z_j$, which main part is \begin{equation} h_j(z)=\sum_{k=1}^{m_j}\frac{c_{-k}^{(j)}}{(z-z_j)^k}. \end{equation}
Let $g$ be the main part of $f$'s Laurent expansion besides $\infty$.
If $\infty$ is a pole of $f$, $g$ is a polynomial; if $\infty$ is a removable singularity of $f$, $g\equiv 0$. Note \begin{equation} F(z)=f(z)-\sum_{j=1}^n h_j(z)-g(z), \end{equation} We know that $F\in H(\bf C$ $_\infty$ \ $\{z_1,\cdots,z_n,\infty\}$. The main parts of $f$'s Laurent expansion besides $z_1,\cdots,z_n,\infty$ have been eliminated, thus $f$ is holomorphic. That implies $F\in H(\bf C$ $_\infty$). From Liouville's theorem, we know $F\equiv c$. Then \begin{equation} f(z)=c+g(z)+\sum_{j=1}^n h_j(z) \end{equation} is a rational function.
- 91
- 3
Let $f$ be meromorphic on $S^2$. Let $A\subset S^2$ be the set of points where $f$ has poles.
Let suppose that $A$ is infinite. Then, by compactness of $S^2$, $A$ must have a limit point in $S^2$ which contradicts the definition of meromorphic functions. Thus $A$ is finite. Let $A=\{p_k\}_{k=1}^n$ and let $d_k$ be the degree of the pole $p_k$ for all $k\in\{1,2,\cdots,n\}$.
Then there exists $g\in H(\mathbb{C})$ s.t. $$g(z)=f(z)-\sum_{k=1}^n\left({\sum_{i=1}^{d_k}}\frac{c_i^{(k)}}{(z-p_k)^i}\right) \text{i.e.} f(z)=g(z)+\sum_{k=1}^n\left({\sum_{i=1}^{d_k}}\frac{c_i^{(k)}}{(z-p_k)^i}\right)\cdots(1)$$
Next, we define $$h(z)=f(z)\cdot\prod_{k=1}^k{(z-p_k)^{d_k}}.$$ Then, using $(1)$ we see that $h$ does not have poles in $\mathbb{C}$ and it may have a pole only at $\infty.$ This means that $h$ is a polynomial, hence, by $(1),$ $f(z)=\frac{h(z)}{\prod_{k=1}^k{(z-p_k)^{d_k}}}$ is a rational function.
- 3,719
-
Hi, how do you conclude that there exists a $g$? Are you doing a Laurent series expanded about each pole of $f$? – Nicholas Roberts Mar 01 '19 at 15:44
-
@NicholasRoberts if by Laurent series expanded about each pole of $f$ you mean the principal parts of $f$ at the poles (see Rudin's Real and Complex Analysis, page 211), then yes. – Emo Mar 01 '19 at 16:14
-
Thank you, one more question: To deduce that $h$ is a polynomial, you use the fact that there may be a pole at infinity. I'm guessing you're citing the theorem that states: "If $f$ is entire with a non-essential singularity at $\infty$ then $f$ is a polynomial. However, in the case where there is no pole at $\infty$ how do you deduce that $h$ is a polynomial? – Nicholas Roberts Mar 02 '19 at 18:05
-
By the construction of $h$ it has only a (non-essential) pole at $\infty$. – Emo Mar 02 '19 at 19:08