EPW Forum

Dear all,

I would like to thank EPW developer group for this excellent effort to integrate it with QE and also like to congratulate for this latest version.

I have few questions related to output files. For all of these following questions I am referring to folder ~/qe-6.0/EPW/tests/Inputs/t03/ which is basically FCC-Pb test. For comparison I am referring to the file ~/qe-6.0/EPW/tests/Refs/t03/seq/pb.a2f.01

Q1. I tried to reproduce a2f results and compared with reference (pb.a2f.01) and found good numerical match. However, my a2f plot (and also the results given in /refs/ folder) looks like mirror reflection with respect to the results shown in website documentation (http://epw.org.uk/Documentation/FCCLead) figure 3 (a) and (c). To be more specific I get the peak near 1meV instead of near 9meV. Beside my reproduced calculation, I also get similar plot for the data in given in reference folder (pb.a2f.01). So I am confident that compilation and calculations ended correctly as I checked output files of each individual calculations.


https://www.dropbox.com/s/6fwk4t3afm43c6w/a2f.png?dl=0

Here is the plot I reproduced, I reproduced it for phonon smearing 0.25meV. No matter what phonon smearing I take it gives almost same location for peaks. I wish anybody correct me in case I am overlooking some other crucial steps or making any silly mistake.

Q2. I am confused about linewidth and imaginary part of self-energies in several places. I use 2*Im(\Sigma) for linewidth for electron and for phonon 2*Im(\Gamma). In https://arxiv.org/abs/1604.03525, page 2, document says line width as just Im(\sigma). Its okay to use them interchangeably. What matters to me is about output files linewidth.phself and linewidth.elself. In first one, in linewidth.phself file, right most column definition is phonon linewidth which I assumed as 2*Im(\Gamma) but in epw.out these numbers are consistent with \Gamma. Do I need to multiply with 2 with this last column to get phonon linewidth?

Q3. Again in linewidth.elself right most column is explicitly saying Im(\Sigma), so to obtain linewidth I should multiply by 2, is that correct?

Many thanks for your time. I appreciate your help and thought.

---------------------------------------------------------------------------------------------------------
Mohsin
Graduate student,
Nanoscale Circuit Integration Lab,
Electrical Engineering and Computer Science Division
Louisiana State University,
Baton Rouge, LA-70802.
U.S.A.

Dear Mohsin,

Thank you for your interest in the EPW software.

Q1) The test folder of EPW is intended for numerical test (i.e. to see if the code runs correctly) but will not give meaningful physical results. All the convergence parameters are severely undercoverged. You can refer to the EPW paper http://www.sciencedirect.com/science/ar ... 5516302260 to see the convergences parameters that was required (very dense k and q grids and also sufficiently dense k/q coarse grids). If your calculations are converged but you still get this behaviour, please check carefully your interpolated Wannier phonon bandstructure.

Q2) The definition of linewdith that I use (both in the paper and in the code is Im(\sigma)). So if you want 2*Im(\Sigma) you do indeed need to multiply by 2.
2*Im(\Sigma) should be the scattering rate, not the linewidths. See for example Eq. (61) of the EPW paper.

Q3) According to me you shouldn't, except if you want scattering rates. In that case, yes.

Best,

Samuel

Hi Samuel,

Thank you very much for all your answers. Actually, I am clear now after reading all your replies.

Thanks again.


---------------------------------------------------------------------------------------------------------
Mohsin
Graduate student,
Nanoscale Circuit Integration Lab,
Electrical Engineering and Computer Science Division
Louisiana State University,
Baton Rouge, LA-70802.
U.S.A.