Austin Zadoks

Ah true, that could be pretty good. In the end, I just implemented a cleaner `simpson` function that seems to have fair performance (a bit better than QE's implementation above...

  Where "Simpson Julia" is what is used for the convergence (Ecut 140 Ha is the reference): 

> Do you have the new and old version of the Ecut convergence that you can put on the same plot easily? Looks like it's a bit better I don't,...

Here's a plot using the dot / trapezoid method, it looks very similar. I'll double-check that this is correct, but I think it's right.