Dear Dr. Balabi

> So this is a bug to be fixed in 6.2, right?
Yes, Prof. Gainnozzi fixed this at the developing version yesterday.

> "The tetrahedron method is presently not implemented"
Sorry, this description should be updated.

> "tetrahedral-opt"  better in any case than "tetrahedra"
* For the DOS computation, "tetrahedra-opt" is advantageous.
* For PDOS, DFPT, and the electron-phonon coupling, "tetrahedra" is not 
implemented because the tetrahedron method with the Bl\"ochl correction is not 
applicable. "tetrahedra_opt" is available.
* For the total energy and the charge/spin density, sometimes I found 
"tetrahedra" shows a little faster k-convergence than "tetrahedra_opt".
* For the structure optimization, "tetrahedra" is not recommended because it is 
not variational. Although "tetrahedra_opt" does not have such a problem, 
sometimes charge-sloshing appears during the optimization and it makes the 
SCF-convergence worse. "smearing" is the robust method.

Best regards,
Mitsuaki Kawamura

From: [email protected] [mailto:[email protected]] On Behalf 
Of balabi
Sent: Thursday, April 13, 2017 9:51 AM
To: pw_forum <[email protected]>
Subject: Re: [Pw_forum] Why projwfc.x gives all zero pdos result while kpdos

Thank you so much, Kawamura and Paolo. So this is a bug to be fixed in 6.2, 
right?


Another thing I don't understand is that why we need "tetrahedral-opt" instead 
of "tetrahedra"? As I tested, "tetrahedra" works for no SOC pdos projection. 
“tetrahedra-opt" is just not mentioned in input_projwfc.html. And also 
according to the html file, "The tetrahedron method is presently not 
implemented" for pdos, and smearing is always implemented. Then, why does 
"occupation=tetrahedra" matters? 

And what is more, is  "tetrahedral-opt"  better in any case than "tetrahedra" 
in terms of efficiency and quality? Should we always use "tetrahedral-opt" in 
the future?


_______________________________________________
Pw_forum mailing list
[email protected]
http://pwscf.org/mailman/listinfo/pw_forum

Reply via email to