EPW Forum

Dear all,

I tried to calculate the electron linewidth of hexagonal 2D structure but I have got the following issue when i use a Grid size = 30 x 30 x 1
"kmesh_shell_automatic: Singular Value Decomposition has found a very small singular value".

But when I try the Grid size = 16 x 16 x 1, I get an other issue: Non-symmetric k-point neighbours!

In the diamond example, I tried to change the grid size and but the code does not find any error. I am wondering, if my problem is because of the small size of my unit cell
(only two atoms).
Can anyone give me any suggestion?
Thank you in advance

Junior kanga
Mohammed V university Morocco

Dear all,

I finally tried to generate my k-points of the nscf file with the script /espresso/wannier90-2.0.1/utility/kmesh.pl and it worked successfully for me.

Best regard

Dear JuniorKanga,

Thanks for letting us know.

The grid should be between 0 and 1 in very direction. This should work with any script in principle.

Best,

Samuel

Dear Samuel Ponc

Thank you for your reply. I am following the example of diamond and I would like to plot the carrier scattering rate as you did in your article:

( Ponc, S., Margine, E. R., Verdi, C., & Giustino, F. (2016). EPW: Electronphonon coupling, transport and superconducting properties using maximally localized Wannier functions. Computer Physics Communications, 209, 116-133.)

But my problem is that a can obtain the linewidth.elself but I have no idea how to use this file to obtain the scattering rate. And I have the same problem on how to plot eliashberg function. I have all the output files but I dont know how to use then.

Sorry for asking so many questions.

Best

Dear JuniorKanga,

That's ok, the forum is there fore that reason !

The file linewidth.elself contains:

# ik ibnd E(ibnd) Im(Sgima)(meV)


So this is the k-point and band index of the imaginary part of the electron-phonon self-energy.

The scattering rate is linked with the Im(Sgima) through Eq. 61 of the paper.

You just have to multiply by the correct factor to go from meV to picosecond and you can obtain the Fig. 8.

The factor is 2/0.6582119514 (using 2pi/hbar).

Best,

Samuel

Dear Samuel Ponc

Thank you very much for your reply.

Best

Dear Dr. Samuel

I am also getting the same error for k-mesh.
kmesh_shell_automatic: Singular Value Decomposition has found a very small singular value


But I didn't get how to follow your advice "The grid should be between 0 and 1 in very direction. "

For my case, I have used

K_POINTS automatic 8x8x8 in scf.in

nscf.in......
K_POINTS crystal
216
0.000000 0.000000 0.000000 0.004630
0.000000 0.000000 0.001157 0.004630
0.000000 0.000000 0.001852 0.004630
0.000000 0.000000 0.002778 0.004630
0.000000 0.000000 0.003704 0.004630
0.000000 0.000000 0.004630 0.004630
0.000000 0.001157 0.000000 0.004630
0.000000 0.001157 0.001157 0.004630
0.000000 0.001157 0.001852 0.004630
0.000000 0.001157 0.002778 0.004630
0.000000 0.001157 0.003704 0.004630
0.000000 0.001157 0.004630 0.004630
0.000000 0.001852 0.000000 0.004630
0.000000 0.001852 0.001157 0.004630
0.000000 0.001852 0.001852 0.004630
etc

and
epw.in
nkf1 = 60
nkf2 = 60
nkf3 = 60

nk1 = 6
nk2 = 6
nk3 = 6

nq1 = 6
nq2 = 6
nq3 = 6


So where and how should I change my values.

Thank you in advance.
Inoka

Hello,

If its a 6x6x6 grid (from your epw input, so not a 8x8x8), then your nscf grid seems wrong.
Should be something like:
0.000000 0.000000 0.000000 0.00462962962
0.000000 0.000000 0.16666666666 0.00462962962
etc ..

Best,
Samuel

Dear Dr. Samuel

Thank you

Yes , it worked.

Have a good day!

Fantastic, glad to hear.

Sam