Electron self-energy in polar material

General
1

Dear EPW users,

I have a question about computing electron self-energy in polar materials.

As described in the code, computing the long-range and short-range terms separately may improve convergence. To get the self-energy for each k-point, can I simply sum that from long-range and short-range terms, using the shortrange or longrange tag? Can you talk about how the self-energy from each term is computed in EPW? Thank you!

Best regards.
Tianshi Wang
Graduate student, University of Delaware

2

Hello Tianshi,

This is something I have coded to test. It should work but there is no test/example available.

In turns out, at least for the case I tested, that it did not really speed up much the convergence with respect to Cauchy Gamma centred random points.

If I remember correctly, you can just sum them.
However, I would advise you to make a small test

  1. compute shortrange
  2. compute longrange
  3. compute all

Verify that long+short gives the same results as all.

Best,

Samuel

3

Hi Dr. Ponce,

Thank you for your prompt reply. I appreciate it.

You remark in the code that the short-range term still contains a linear long-range part. Does it mean that the short-range term in polar materials has the form as |g[size=70:1prjm9s8]S[/size:1prjm9s8]^2+2g[size=70:1prjm9s8]L[/size:1prjm9s8]g[size=70:1prjm9s8]S[/size:1prjm9s8]| where g[size=70:1prjm9s8]S[/size:1prjm9s8] is the matrix elements in non-polar material? Thank you!

Tianshi

4

Hello,

Yes the short-range part is defined as
g_s = sqrt(gg - g_lg_l)

I would need go back to the code to be sure but I think the idea is that you do (g_s+g_l)^2 because you interpolate
the matrix element and not the matrix element square.

5

Dear Dr. Ponce,

Got you. Thanks a lot!

Tianshi