Working with the cyclic group of order n, $C_n = \{ e,g,g^2,...,g^{n-1} \} $ and the group algebra of $\mathbb{C}[C_n]$ with multiplication $g^k \cdot z=\eta ^k z$ for $\eta = e^{2\pi i/n}$.
We are given that $$\hat{a}(m)= \Sigma^{n-1}_{k=0} a_k \eta^{-mk}$$ for $m=0,...,n-1$
$$a'_l = \frac{1}{n} \Sigma^{n-1}_{m=0} \hat{a}(m) \eta^{ml}$$ for $l=0,...,n-1$
We want this equation in the form $$\Sigma^{n-1}_{k=0} a_k c_{k,l}$$ for some $c_{k,l}, k,l=0,...n-1$
So far I have got this equation to be $$a'_l= \frac{1}{n} \Sigma^{n-1}_{k=0} a_k \Sigma^{n-1}_{m=0} \eta^{m(l-k)}$$
I'm not too sure how to get this summation to some $c_{k,l}$ but from this I need to look at the values of k & l for which the stabilizer is trivial/non-trivial, I am not too sure what this means?
The answer is $c_{k,l}=$$\delta_{k,l}$, i.e. $a'_l=a_l.$ This is the inversion formula for discrete Fourier transform. You can easily (re-)prove it, using that for any integer $r>1,$ the sum of the $r$ complex $r$-th roots of unity is $0.$