2
$\begingroup$

Let's say I have a complex function like $$f(z)=z+e^{\frac{1}{z-1}}+\sin(\frac{1}{z-2})$$ which is holomorphic in $\mathbb{C}\setminus \left \{ 1,2 \right \}$

Then there exist entire functions $$f_{0}(z)=z$$ $$f_{1}(z)=e^{z}$$ $$f_{2}(z)=\sin(z)$$

Such that, the equality $$f(z)=f_{0}(z)+f_{1}(\frac{1}{z-1})+f_{2}(\frac{1}{z-2})$$ holds for all points except for $z=1$ and $z=2$ which are essential singularities.

My question is, can this be extended in general?

Precisely put, let's say I have a holomorphic function $f(z)$ except for a finite number of essential singularity points $a_{1},...,a_{n}$ in the complex plane. Is it true in general, that there exist $n+1$ entire functions $f_{0}(z),...,f_{n}(z)$ such that: $$f_{i}(z) \qquad is\quad holomorphic \quad in \quad \mathbb{C}\setminus \left \{ a_{i} \right \} \quad for \quad i=1,...,n $$

$$f(z)=f_{0}(z)+f_{1}(\frac{1}{z-a_{1}})+...+f_{n}(\frac{1}{z-a_{n}})$$

$\endgroup$
  • $\begingroup$ You can take a look at weierstrass factorization theorem and Mittag-Lefler's theorem $\endgroup$ – Beslikas Thanos Dec 23 '16 at 8:45
  • $\begingroup$ Mittag-Leffler's theorem is a kind of decomposition of a function in terms of its singularities. Its relation with Weierstrass factorization theorem is obvious. Many call these theorems "cousins". This is a generalization of OP's idea. $\endgroup$ – Beslikas Thanos Dec 23 '16 at 9:07
  • $\begingroup$ @BeslikasThanos I don't see. OP's idea is about functions with finitely many isolated singularties, and is the answer I wrote, while Weierstrass's factorization theorem and Mittag-Lefler's theorem are about functions with infinitely many zeros and poles (otherwise they are trivial). $\endgroup$ – reuns Dec 23 '16 at 9:09
  • $\begingroup$ Yes but he cqn see these techniques in order to obtain something similar. I didnt post as an answer but to give motivation $\endgroup$ – Beslikas Thanos Dec 23 '16 at 9:12
2
$\begingroup$
  • If $f(z)$ is holomorphic on $0 < |z| < r$ then the Cauchy integral formula in an annulus gives $f(z) =\frac{1}{2i\pi} \int_{|s|= R_1} \frac{f(s)}{s-z}ds-\frac{1}{2i\pi}\int_{|s|= R_2} \frac{f(s)}{s-z}ds$ for $ 0 < R_2 < |z| < R_1 < r$.

  • Expanding $\frac{1}{s-z}$ in a geometric series we get the Laurent series expansion $f(z) = \sum_{n=-\infty}^\infty a_n z^n$ where $a_n =\frac{1}{2i\pi} \int_{|s|= R} f(s) s^{-n-1}ds$ and $0 < R < r$.

  • Let $g(z) = \sum_{n=1}^\infty a_{-n} z^n$. It converges for $|z| > 1/r$ and hence for every $z$ : it is entire.

  • $f(z)-g(1/z)$ is clearly holomorphic at $z=0$

$\implies$ Every holomorphic function on $\mathbb{C} \setminus \{z_1,\ldots,z_K\}$ can be written as $f_0(z)+\sum_{k=1}^K f_k(\frac{1}{z-z_k})$ where $f_k$ are entire

$\endgroup$
  • $\begingroup$ Perfect :) Exactly in the lines of what I thought about and I realised what I missed $\endgroup$ – zokomoko Dec 23 '16 at 9:36

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.