For analytic $f$, how can I represent the expression $f(z)\cdot\exp\left({s\,\log(z)}\right)$, i.e. $f(z)\cdot z^s$ in the form

$$\sum_{n}^\infty\left(\sum_{k}^\infty a_k s^k\right)z^n,$$

at least as a formal power series, where the index runs over the number needed?

I got to $$f(z)\cdot\exp\left({s\,\mathrm{log}(z)}\right)$$ $$=\left(\sum_{m}^\infty\frac{1}{m!}f^{(m)}(0)\,z^m\right) \left(\sum_{j=0}^\infty \frac{1}{j!} \left(s\,\log{(z)}\right)^j\right)$$ $$=\sum_{m}^\infty\sum_{j=0}^\infty \frac{1}{m!\,j!}f^{(m)}(0)\,s^j\cdot z^m\, \log{(z)}^j,$$

and I know

$$\log(z)=\sum_{l=1}^\infty (-1)^{l+1}\frac{1}{l}(z-1)^{l}.$$

I'm motivated by wanting to understand the Mellin transform (and the Laplace transform, for that matter) and here I approach this by looking at what it does to series components of a function

$$f(z)=\sum c_n z^n\mapsto \mathcal M(f)=\sum c_n^\mathcal{M} z^n$$

  • $\begingroup$ Is $x$ a typo ? $\endgroup$ – Claude Leibovici Feb 25 '14 at 8:29
  • $\begingroup$ @ClaudeLeibovici: Yes, thx. $\endgroup$ – Nikolaj-K Feb 25 '14 at 8:46

Unless $s$ is an integer and non-negative, your function has a singularity at 0, and no such series is possible.

  • $\begingroup$ The particular case $f(z) = 1$ asks us to expand $z^s$ as a power series centered at $z=0$. $\endgroup$ – GEdgar Feb 25 '14 at 18:10
  • $\begingroup$ @GEdgar, that is the point: Unless $s$ is a non-negative integer, $z^s$ has no power series expansion around $z = 0$. $\endgroup$ – vonbrand Feb 25 '14 at 18:12

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.