RIM problem

[yanghuang]

Dear Developers and Users,

I have got the file ndb.E_RIM (BZ_RIM_Nk=2000), but GW calculation (-g n -p p) is still very slow, did I miss some steps?

Sorry to disturb you.

Best

Yang

[yanghuang]

What is the meaning of "using the E-RIM to average the denominators"?

[Davide Sangalli]

Dear Yang,
I'm not sure I understood what you are doing.

Can you describe more in details what are you using yambo for and which is the problem you are facing ?

Best,
D.

[yanghuang]

Dear Davide,

Thank you so much for your kind help.

There is a word on http://www.yambo-code.org/input_file/ypp/ypp_rim.php
"Moreover other properties that depend from the integral of epsilon, as G0W0 corrections, will converge even faster."
Does this mean that E-RIM method can be used to calculate GW corrections? What is the meaning of "using the E-RIM to average the denominators"?

Best,
Yang

[Davide Sangalli]

Dear Yang,

the E-RIM method is used to speed up the convergence of the response function and thus the screening epsilon against the number of k-points.
Since the screening is used to compute the W in the GW self-energy the same apply also to GW calculations

For more informations see:
http://www.yambo-code.org/theory/docs/doc_Xd.php

and the references at the end of the page.

Best,
D.

[yanghuang]

Dear Davide,

Does this mean that if I get the file ndb.E_RIM, the GW calculations will be faster?


Best,
Yang

[Davide Sangalli]

No, they will be slower given the same SAVE

[yanghuang]

Dear Davide,

Thanks for your detailed reply.

Best,
Yang

[yanghuang]

Dear Davide,

You mentioned in your article PRB 93, 195205 (2016)
"Equation (1) is solved on a 16X16X16 k grid in the Brillouin zone, with the KS wave functions and energies computed from a non-self-consistent DFT calculation starting from a 4X4X4 k grid."

Does this mean you set BZ_RIM_Nk=16X16X16?

Best,
Yang

[Davide Sangalli]

Dear Yang,
yes, this is what we do.

Best,
D.