Time reversal symmetry or not for spinor BSE

152 views
Skip to first unread message

Hong Tang

unread,
Oct 12, 2021, 12:37:43 PM10/12/21
to BerkeleyGW Help
Dear Meng Wu and other BGW masters,

I have a question about time reversal symmetry. 
In quantum espresso, it says " In the noncollinear/spin-orbit case, starting with zero
starting_magnetization on all atoms imposes time reversal
symmetry. The magnetization is never calculated and is
set to zero (the internal variable domag is set to .FALSE.). "

It seems to say if we use nonzero starting_magnetization in qe, then we may break the time reversal symmetry.  Then my question is that for a broken time reversal symmetry system, does spinor BSE work correctly?  or is there anything we should care? 

Best,
Hong
 

Meng Wu

unread,
Oct 12, 2021, 2:41:17 PM10/12/21
to Hong Tang, BerkeleyGW Help
Hi Hong,

Good question. BGW uses time-reversal symmetry in different ways. It has nothing to do spinor (nspin=4 in QE). The following statements also apply to the collinear spin case (nspin=2 in QE).

1. k/q-points are generated without time-reversal symmetry (kgrid.x).
2. RPA dielectric function (epsilon.x) is calculated with time-reversal symmetry.
3. The self-energy (full-frequency or generalized-plasmon pole) calculated in sigma.x assumes time-reversal symmetry.

If the system has inversion symmetry, E_{-k} = E_{k}, you would expect that time-reversal symmetry will only have a small effect on the dielectric function or self-energy (and therefore quasiparticle bandstructure). But you have to USE IT AT YOUR OWN RISK. We are working to add support for broken time-reversal symmetry in the code. Stay tuned! ^_^

Best!
Meng

--
You received this message because you are subscribed to the Google Groups "BerkeleyGW Help" group.
To unsubscribe from this group and stop receiving emails from it, send an email to help+uns...@berkeleygw.org.
To view this discussion on the web visit https://groups.google.com/a/berkeleygw.org/d/msgid/help/564f8fa2-5f3d-4ff2-a113-872904188fc0n%40berkeleygw.org.

Hong Tang

unread,
Oct 12, 2021, 5:00:17 PM10/12/21
to BerkeleyGW Help, wu1m...@berkeley.edu, BerkeleyGW Help, Hong Tang
Dear Meng,

Thank you so much for helping.

In the quantum espresso run, in the out file, it says "No symmetry found". So my system should has no  inversion symmetry.   If I do 01-scf to 06-wfnq_fi with nonzero starting_magnetization in qe, then the following  epsilon.x  and  sigma.x runs, is there any problems? 

Best,
Hong

Hong Tang

unread,
Oct 12, 2021, 5:17:25 PM10/12/21
to BerkeleyGW Help, Hong Tang, wu1m...@berkeley.edu, BerkeleyGW Help
Dear Meng,

"In the quantum espresso run, in the out file, it says "No symmetry found". So my system should has no  inversion symmetry.   If I do 01-scf to 06-wfnq_fi with nonzero starting_magnetization in qe, then in the following  epsilon.x  and  sigma.x runs, is there any problems? " 

I mean for calculation process, is there any problems, like crash? 
You mean this has impact on how correct the results are? right? 

Best,
Hong

Meng Wu

unread,
Oct 12, 2021, 7:02:34 PM10/12/21
to Hong Tang, BerkeleyGW Help
Hi Hong,

BGW will not crash in this case.

Yes, you are right. In an extreme case where E_{nk} and E_{n,-k} differ a lot due to broken time-reversal symmetry, the results are not reliable.

Best!
Meng

Hong Tang

unread,
Oct 12, 2021, 10:33:13 PM10/12/21
to BerkeleyGW Help, wu1m...@berkeley.edu, BerkeleyGW Help, Hong Tang
Thanks! 
HT
Reply all
Reply to author
Forward
0 new messages