Calculating the q resolved Eliashberg function

Running Abinit Response calculations
1

Hi,

I’m trying to calculate the q resolved Eliashberg function on a grid of q-points in the BZ, which is a thing I’ve seen done in various papers with Wannier interpolation in QE+W90, but I was wondering if it is possible to do in Abinit with the new eph code.

I tried to follow the tutorial, from what I understand the q dependent Eliashberg function is stored in the A2F.nc file, but I’m having trouble extracting the data from this file. Usually I read NetCDF files with Matlab, but in this case I’m struggling with understanding the meaning of the quantities contained in the A2F.nc file. I would have expected to just find a vector containing the reduced coordinates of the q-points and another vector of the same size containing the values of alpha^2F in those points, but it doesn’t look like that to me. Can anybody give me some insight?

Thanks,
Luca

2

Hi Luca,

the file contains several different versions of a2F, but not q resolved. They are calculated on different q grids, with and without interpolation (qintp and qcoarse resp.)
Should be easy to add as an output option, but it’s not done afaik.

In the log/ouput files there are mode and q presentations of the lambda and line width values, for the original input q points. Might be a start for your analysis.

best

Matthieu

3

Hi Matthieu,

thanks for your reply. So, for example, if I take one version of a2F, it looks like the netcdf file contains these 3 arrays of data

    a2ftr_mesh_qintp              
           Size:       473x1
           Dimensions: a2ftr_nomega_qintp
           Datatype:   double
    a2ftr_values_qintp            
           Size:       473x3x3x4x1
           Dimensions: a2ftr_nomega_qintp,three,three,natom3p1,number_of_spins
           Datatype:   double
    a2ftr_lambdaw_qintp           
           Size:       473x3x3x4x1
           Dimensions: a2ftr_nomega_qintp,three,three,natom3p1,number_of_spins
           Datatype:   double

could you help me understand what they are? I wasn’t able to find any documentation on the Abinit website about this.

Luca

4

Note that a2ftr is for transport calculations (see the corresponding tutorials and papers by PB Allen). The 3x3 matrix versions are tensors wrt electron group velocities. I think the 3natom+1 dimension is a phonon band decomposition, plus the total. Lambdaw is related to values the way the epc superconducting lambda is related to the linewidths gamma