The double-delta approximation

Running Abinit Response calculations
1

Hi Abifriends,

I am calculating phonon line widths (and eventually the electron spectral function) in a metal near \Gamma, where I have good reason to believe they are broad. However, the implementation in Abinit uses the so called double-delta approximation where, among other things, the phonon energy is neglected relative to the electronic transitions. In a metal, where there are \Delta E \rightarrow 0 transitions, this causes the linewidths to blow up as we get near \Gamma. I am wondering if it is reasonably straightforward to implement the full sum over states that doesn’t use this approximation. I looked at the source code, but I am not familiar enough with Abinit to implement this myself without a lot of guidance. I’d be happy to work on it if anyone is interested in helping? Of course, maybe there are good reasons that this isn’t done in first place and I’d be happy to learn that too.

Alternatively, I could just calculate the phonon spectral function. However, it doesn’t seem to implemented for q \neq \Gamma. Am I missing something?

Thanks!
Ty

2

Hi Ty,

For metals the ephtask 1 formalism should work, it has been used for aluminum and many others in the past. You might nevertheless need a non adiabatic formalism. The full self energy expressions are in eph task 4 (full SE) and -4 (Im part only)

I don’t think you need to implement anything, though if you are interested there may be further things to develop. Have a look at the full eph transport tutorial and get back to us.

Best

Matthieu

3

PS: for acoustic Gamma point modes with omega → 0 the line width should go to 0 as well physically. Numerically it might be delicate to converge though

4

Thanks for the quick reply. I’ll check out the tutorial and report back!