[QE-users] Role of ecutwfc and ecutrho in optimization - Reg

[singaravelan T R]

Dear all,
Qn 1: I have a question that when to start use ecutrho ?
That is I am using USPP, that means my pseudopotential depends on both
ecutwfc (majorily) and ecutrho. So, as a thumb rule ecutrho should be from
8-12 times ecutwfc. I tried both ways as follows:

Way 1: I have done optimization of ecutwfc (Ry) , with default ecutrho (not
specifying ecutrho in input file), because I felt most important is ecutwfc
since it works on varational principle. optimization of k-point, degauss,
lattice constant all these stuffs with default ecutrho(4x ecutwfc).
Completed optimization. Then with those optimised results, I also tried to
find out which is better value of ecutrho, whether 8,9,10,11,12 times
ecutwfc and found that 10 times is better. That is if my ecutwfc is 60 Ry ,
ecutrho is 600 . Now I started my scf , nscf, dos, bandstructure etc.

Way 2: From first onwards, for optimization part itself I started to use
also ecutrho. Then k-point, on going to lattice constant I found that,
lattice constant of
 10.10783(Bohr)  = -429.99249320 Ry
 10. 11783(Bohr) = - 429.99249304 Ry.
It's very close, still I have to optimise three more atleast two more
digits in lattice constant.

Also I found that, default ecutrho only has the lower energy compared to
this, while optimization.
10.10783(Bohr) (default ecutrho ) = -429.99268209
10.11783 (Bohr) (default ecutrho) = -429.99267906

If this would be the case, then it will be most difficult for me to compare
which atomic positions are better. If I use ecutrho as 8-12 times ecutwfc.

Qn: 2 , when there is any comparison or optimization , I feel using default
ecutrho is better But on calculation scf, nscf dos bandstructure etc we may
stick on to thumb rule 8-12 times ecutwfc = ecutrho. Is my opinion is
correct or otherwise which Way of doing things are better way 1 or way 2 ?
with thanks
Singaravelan T R
I also tried with other compound, that shows the same trend as above.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20210520/3ea43eaf/attachment.html>