[QE-users] Wrong representation solved by nosym but implies crazy frequencies and looses phonons symetries

[Lorenzo Paulatto]

I'm not sure why you get the error you get, supercell calculations cause 
every kind of problems with symmetry. However, something easy and quick 
you can do, is to generate the dynamical matrix of the super-cell 
starting from a unit cell calculation with matdyn. All you have to do a 
unit-cell phonon calculation and that feed matdyn and input like this:

&input
   flfrc='mat2R'
   asr='crystal'
   q_in_cryst_coord = .true.
   l1=2
   l2=2
   l3=2
   fldyn='sc222_dyn'
/
  1
0 0 .5

Where l1,l2 and l3 specify the size of the supercell. You can also 
specify the supercell as at=a1x,a1y,...a3y,a3z which should allow you to 
go from FCC to SC.

The resulting dynamical matrix file does not contain the header, but 
this can easily be solved adding a line like this
   CALL write_dyn_mat_header( filout, ntyp, nat, ibrav, nspin_mag, &
             celldm, at, bg, omega, atm, amass, tau,ityp,m_loc,nqq)

just before the call to write_dyn_on_file, line 621 of matdyn.x

hth

On 3/18/20 2:06 PM, Antoine Jay wrote:
> Dear all,
> I ckecked the phonon frequencies at q=gamma of a 512-atoms cubic 
> supercell of bulk silicon. (kpoint=gamma also)
> As the phonon code output the "wrong representation" error, I added 
> nosym and noinv.
> The output frequencies seem to be right (80-507cm^-1) except for the 
> last one that is 1132cm^-1
> Hopefully, a simple ASR was enough to decrease this frequency down to 
> 507cm^-1.
> 
> This calcul implies three questions:
> 1) How is it possible to have such a decrease between before ASR and 
> after ASR, is the calcul wrong (input in PJ)?
> 2) Why is there a "wrong representation" error, whereas the atomic 
> coordinates are exactelly symetric and not quasi-symetric?
> 3) Is it possible to know which modes are degenerated (nosym implies C_1 
> space group which looses the symetry analysis)?
> For this last, I cannot use only the phonons frequencies because they 
> are not exactelly equals, and due to the BZ folding in the supercell, 
> several frequencies are very close from each other.
> 
> Thank you very much for your help.
> 
> Antoine Jay
> LAAS-CNRS
> Toulouse France
> 
> _______________________________________________
> Quantum ESPRESSO is supported by MaX (www.max-centre.eu/quantum-espresso)
> users mailing list users at lists.quantum-espresso.org
> https://lists.quantum-espresso.org/mailman/listinfo/users
> 

-- 
Lorenzo Paulatto - Paris