Let $ f_t(x) := \begin{cases} \frac{xe^{tx}}{e^x-1} & \quad x\neq0\\ 1 & \quad x=0\\ \end{cases} $ where $t\in \mathbb{R}$.
(a) Prove that there is a $\delta$ which doesn't depend on $t$ such that $f_t(x)$ has a power series expansion about 0 on $(-\delta, \delta)$.
(b) Let $f_t(x)= \sum_{n=0}^{\infty} a_n(t)\frac{x^n}{n!}$ for $x\in (-\delta, \delta)$. First, show that the identity $$ \sum_{n=0}^{\infty} a_n(t)\frac{x^n}{n!} = e^{tx} \sum_{n=0}^{\infty} a_n(0)\frac{x^n}{n!} $$ holds and using this identity prove that $a_n(t) = \sum_{m=0}^n \binom{n}{m} a_m (0) t^{n-m}$.
For part (a), I have tried find a rule to be able to express $f_t(x)$ in an equivalent power series form, but I have no idea how to deal with the extra $t$ component.
As for (b), I assume that showing that the identity holds involves some part of using the fact that $e^{tx} = \sum_{k=0}^{\infty} \frac{(tx)^k}{k!}$, but I have not yet figured out how...
What would be the method to tackle both problems? Any tips or help would be most appreciated!
EDIT: I have tried and done (I think?) part (b)! The identity follows from the fact that $f_t(x) = e^{tx} \cdot f_0(x)$, and the proof follows from finding the Cauchy product of $e^{tx}$ and $f_0(x)$. Still no luck with (a), though...