i would need some help for the following problem.
Consider the $M/E_r/1$ queue with arrival rate $\lambda$ and mean service time $r/\mu$. Let $P(z)$ be the generating function of the probabilities $p_n$, so
$$P(z)=\sum_{n=0}^{\infty} p_{n}z^{n},|z|\leq 1$$ with normalization equation $$\sum_{n=0}^{\infty} p_n=1$$
We are given that $P(z)=\frac{p_0\mu}{\mu-\lambda(z+\ldots+z^{r-1}+z^{r})}$ (*)
(iii) Show, by partial fraction decomposition of $P(z)$, that the probabilities $p_n$ can be written as $$p_n= \sum_{k=1}^{r} c_k (\frac{1}{z_k})^n, n=0,1,2,...$$ where $z_1,...,z_r$ are the zeroes of the denominator in (*).
This is where i have got so far but still cannot relate how i can get $p_n$ eventually:
$P(z)=\frac{p_0}{1-\frac{\lambda}{\mu}(z+\cdots+z^{r})}$, we let $c_k=\frac{p_0}{1-\frac{\lambda}{\mu}}$ and rewriting $\frac{1}{z+\cdots+z^r}=\frac{1}{(z-z_1)\cdots(z-z_k)}$
we have that \begin{align*} P(z) &=\sum_{k=1}^{r} c_k\frac{1}{(z-z_k)} &=\frac{c_1}{z-z_1}+...+\frac{c_k}{z-z_k}=\frac{1}{(z-z_1)...(z-z_k)} \end{align*}
Hence we get \begin{align*} c_1\prod\limits_{j\neq1}^k(z-z_j)+\cdots+c_k\prod\limits_{j\neq k}^k(z-z_j)=1 \end{align*}
For $z=z_1$, $c_1\prod\limits_{j\neq1}^k(z_1-z_j)=1$ so $c_1=\frac{1}{\prod\limits_{j\neq1}^k(z_1-z_j)}$, iterating n times
For $z=z_k$, $c_k\prod\limits_{j\neq1}^n(z_k-z_j)=1$ so $c_k=\frac{1}{\prod\limits_{j\neq1}^n(z_k-z_j)}$
Thanks in advance