Page 1 of 1

One confusion regrading the subroutine "evaluate_a2f_lambda"

Posted: Fri Apr 06, 2018 2:32 am
by hellolori
Dear all,

I am reading the code of EPW (version 4.3) recently and confused by a section of the subroutine "evaluate_a2f_lambda" in the file "eliashberg_setup.f90". In the lines between 238 and 280, some loops are used to compute anisotropic Eliashberg spectrum. Part of the code reads:

Code: Select all

 DO iq = 1, nqfs(ik)
         ! iq0 - index of q-point on the full q-mesh
                 iq0 = ixqfs(ik,iq)
              DO jbnd = 1, nbndfs
                 IF ( abs( ekfs(jbnd,ixkqf(ik,iq0)) - ef0 ) .lt. fsthick ) THEN
                     weight = wkfs(ik) * wqf(iq) * w0g(ibnd,ik) * w0g(jbnd,ixkqf(ik,iq0))
                     lambda_eph = 0.d0


In my opinion, the wqf(iq), which appears in the last but one line of the above code, should be replaced by wqf(iq0), because the array "wqf" is the weight of the q point on the full q-mesh, and the index iq0 is no other than the index of q-point on the full q-mesh. So I think the index iq should be replaced by iq0.

Can anyone help me to clarify my the confusion?

Bests!

Re: One confusion regrading the subroutine "evaluate_a2f_lam

Posted: Fri Apr 06, 2018 11:13 am
by roxana
Hi,

You can indeed replace the weight by wqf(iq0) but this will be the same as wqf(iq) since we are using the full q-mesh and all q-points have equal weight (1/total nr. of q-points).

Best,
Roxana

Re: One confusion regrading the subroutine "evaluate_a2f_lam

Posted: Fri Apr 06, 2018 12:40 pm
by hellolori
roxana wrote:Hi,

You can indeed replace the weight by wqf(iq0) but this will be the same as wqf(iq) since we are using the full q-mesh and all q-points have equal weight (1/total nr. of q-points).

Best,
Roxana

Dear Roxana,

Yes, it indeed the same. So, are the weights of the q points always the same?


Bests

Re: One confusion regrading the subroutine "evaluate_a2f_lam

Posted: Fri Apr 06, 2018 5:36 pm
by roxana
Hi,

Yes, if you use a full uniform q-mesh or a random q-mesh, the weight is defined as 1/total nr. of q-points. If you want to know more details you can look at loadqmesh.f90

Best,
Roxana