How to pick the right value of degaussw

General discussion around the EPW software

Moderator: hlee

Post Reply
YSW
Posts: 19
Joined: Fri Jun 18, 2021 5:24 am
Affiliation: Shanghai University of Electric Pow

How to pick the right value of degaussw

Post by YSW »

Dear all,

I have a trouble.
In my case, I calculate superconductivity with 10 10 10 k/q grids, and I get a same lambda with 20 20 20 k/q grids.(degaussw = 0.5,lambda = 1.799)
but, when I change degaussw,the value of lambda have a big change with 10 10 10 k/q grids.(degaussw = 0.4,lambda = 1.899).
Unfortunately, my computer have not enough mem to compute with finer k/q grids.
How should I choose this value of degaussw to get a credible result ?

I see someone say "In theory, those two parameters can be 0 if you have dense enough k/q-points grids."
But, the value of lambda alaways change with different degaussw.I think it is not a convergent results when two parameters to be 0,even though he have dense enough k/q-points grids. right?

I am looking forward to your reply.

Sincerely,

YSW

hlee
Posts: 293
Joined: Thu Aug 03, 2017 12:24 pm
Affiliation: The University of Texas at Austin

Re: How to pick the right value of degaussw

Post by hlee »

Dear YSW:

(from my previous message at viewtopic.php?p=3991#p3991)
Regarding the convergence, you need to converge quantities w.r.t both the broadening and the # of grids.
As I already said, you need to converge lambda with respect to both degaussw and # of grids.
Please check Fig. 5 or 6 of the paper at https://journals.aps.org/prb/pdf/10.110 ... .89.094515 .

Sincerely,

H. Lee

YSW
Posts: 19
Joined: Fri Jun 18, 2021 5:24 am
Affiliation: Shanghai University of Electric Pow

Re: How to pick the right value of degaussw

Post by YSW »

Dear H. Lee:


Thank you for your patient answer.
I will try and see what happens.


Sincerely,

YSW

Post Reply