I currently am looking for a DFT+GW program (preferably all-electron) with which I would like to implement and test a way to obtain the inverse epsilon/W.
The idea is pretty simple. The main idea is to calculate the poles of the inverse epsilon by diagonalizing a certain matrix (see for example eq. 9 and 10 in this paper: /pdf/1402.5433v1.pdf on arxiv.org) and then to perform the GW integration analytically.
There are a few things I would like to test with this approach, mostly regarding diagonalization of diagonal+rank 1 matrices.
Anyways I would love to get some insight into the code. Maybe an opinion if you guys think this is feasible with the current exciting code, which files to look at and understand and why you have the current implementation (Pade+analytic continuation if I'm not mistaken).