In Serre's "A Course in Arithmetic"(page $84$) he mentions that the Laurent series of the Weierstrass function $$\wp_{\Gamma}(u)= \frac{1}{u^2} + \sum_{\gamma \in \Gamma}(\frac{1}{(u-\gamma)^2} - \frac{1}{\gamma^2})$$ is given by $$(1)\qquad \frac{1}{u^2} + \sum_{k=2}^{\infty}(2k-1)G_k(\Gamma)u^{2k-2}$$ where $G_k(\Gamma)$ is the Eisenstein series of weight $2k$, and $\Gamma$ a lattice in $\mathbf{C}$.
I am attempting to prove that this is in fact the case, and have considered expanding $\frac{1}{(u-\gamma)^2} - \frac{1}{\gamma^2}$ as $\frac{1}{\gamma^2}\frac{1}{(1-\frac{u}{\gamma})^2} - \frac{1}{\gamma^2} = \frac{1}{\gamma^2}(\sum_{k=1}^{\infty}(k+1)(\frac{u}{\gamma})^k) = \sum_{k=1}^{\infty}(k+1)(\frac{u^k}{\gamma^{k+2}})$. Hence I am left with:
$$\frac{1}{u^2} + \sum_{\gamma \in \Gamma}(\sum_{k=1}^{\infty}(k+1)(\frac{u^k}{\gamma^{k+2}}))$$ Now in attempting to make this look like (1), I feel like this involves some tricky manipulation of the series to get $G_k(\Gamma)$, but I am not sure how to proceed.
Any hints would be greatly appreciated.
What you can do is reverse the order of the sums since the series is absolutely convergent to obtain that:
$$\frac{1}{u^2} + \sum_{\gamma \in \Gamma}(\sum_{k=1}^{\infty}(k+1)(\frac{u^k}{\gamma^{k+2}})) = \frac{1}{u^2} + \sum_{k=1}^{\infty}(k+1)(\sum_{\gamma \in \Gamma}(\frac{u^k}{\gamma^{k+2}}))$$ then factor out the $u^k$ to get:
$$\frac{1}{u^2} + \sum_{k=1}^{\infty}(k+1)(\sum_{\gamma \in \Gamma}(\frac{1}{\gamma^{k+2}}))u^k = \frac{1}{u^2} + \sum_{k=1}^{\infty}(k+1)G_k(\Gamma)u^k$$ We know that modular forms of odd weight are zero, and hence our sum becomes:
$$\frac{1}{u^2} + \sum_{k=2}^{\infty}(2k-1)G_k(\Gamma)u^{2k-2}$$ as desired.