[Pw_forum] Backward compatibility issue on elph matrix elements between QE 4.0.3 and 5.1.2

[Samuel Poncé]

Dear developers of QE,

I noticed a backward compatibility issue on the electron-phonon matrix
element between QE 4.0.3 and 5.1.2.

I was wondering if it was a bug and if not, what was the reason.

I added the following line in the two versions of the code in
PHonon/PH/elphon.f90

if (ik==215) then
write(*,*)'elphmat(:,:,:)**2',SUM((REAL(REAL(elphmat(:,:,:))))**2)+SUM((REAL(AIMAG(elphmat(:,:,:))))**2)

This should be a physically relevant quantity that should not change
between version (the order of the mode changed but it is not an issue).

Test:
6x6x6 q-grid same pseudo, tol and inputs parameters

There is 16 irr q-points:

     (  16q-points):
       N         xq(1)         xq(2)         xq(3)
       1   0.000000000   0.000000000   0.000000000
       2  -0.166666667   0.166666667  -0.166666667
       3  -0.333333333   0.333333333  -0.333333333
       4   0.500000000  -0.500000000   0.500000000
       5   0.000000000   0.333333333   0.000000000
       6  -0.166666667   0.500000000  -0.166666667
       7   0.666666667  -0.333333333   0.666666667
       8   0.500000000  -0.166666667   0.500000000
       9   0.333333333   0.000000000   0.333333333
      10   0.000000000   0.666666667   0.000000000
      11   0.833333333  -0.166666667   0.833333333
      12   0.666666667  -0.000000000   0.666666667
      13   0.000000000  -1.000000000   0.000000000
      14   0.666666667  -0.333333333   1.000000000
      15   0.500000000  -0.166666667   0.833333333
      16  -0.333333333  -1.000000000   0.000000000

All the elphmat**2 are the same (up to 4 digits) for 13 of the 16 q-points.
For example

QE 5.1.2  q = (   -0.166666667   0.166666667  -0.166666667 )
elphmat(:,:,:)**2  0.5446267
elphmat(:,:,:)**2  0.5446253
elphmat(:,:,:)**2  0.9506719
elphmat(:,:,:)**2  0.9506719

     freq (    1) =       9.137536 [THz] =     304.795384 [cm-1]
     freq (    2) =       9.137536 [THz] =     304.795384 [cm-1]
     freq (    3) =      15.052865 [THz] =     502.109513 [cm-1]
     freq (    4) =      39.653912 [THz] =    1322.712114 [cm-1]
     freq (    5) =      39.653912 [THz] =    1322.712114 [cm-1]
     freq (    6) =      40.741956 [THz] =    1359.005377 [cm-1]

QE 4.0.3 q = (   -0.166666667   0.166666667  -0.166666667 )
elphmat(:,:,:)**2  0.5446267
elphmat(:,:,:)**2  0.9506719
elphmat(:,:,:)**2  0.5446266
elphmat(:,:,:)**2  0.9506719

     omega( 1) =       9.137611 [THz] =     304.799915 [cm-1]
     omega( 2) =       9.137611 [THz] =     304.799915 [cm-1]
     omega( 3) =      15.052641 [THz] =     502.105399 [cm-1]
     omega( 4) =      39.653770 [THz] =    1322.716173 [cm-1]
     omega( 5) =      39.653770 [THz] =    1322.716173 [cm-1]
     omega( 6) =      40.741800 [THz] =    1359.009209 [cm-1]


But the three irr-q* 5, 10* and *16* are NOT the same:
QE 5.1.2 q = (    0.000000000   0.333333333   0.000000000 )
elphmat(:,:,:)**2  0.1192376
elphmat(:,:,:)**2  0.9861287
elphmat(:,:,:)**2   1.348592
elphmat(:,:,:)**2  0.3134981

     freq (    1) =      11.633039 [THz] =     388.036416 [cm-1]
     freq (    2) =      11.633039 [THz] =     388.036416 [cm-1]
     freq (    3) =      16.188917 [THz] =     540.004156 [cm-1]
     freq (    4) =      39.110566 [THz] =    1304.588071 [cm-1]
     freq (    5) =      39.110566 [THz] =    1304.588071 [cm-1]
     freq (    6) =      40.947507 [THz] =    1365.861806 [cm-1]

QE 4.0.3 q = (    0.000000000   0.333333333   0.000000000
elphmat(:,:,:)**2  0.3075966
elphmat(:,:,:)**2   1.354492
elphmat(:,:,:)**2  0.1192375
elphmat(:,:,:)**2  0.9861288

     omega( 1) =      11.633152 [THz] =     388.042751 [cm-1]
     omega( 2) =      11.633152 [THz] =     388.042751 [cm-1]
     omega( 3) =      16.188922 [THz] =     540.007888 [cm-1]
     omega( 4) =      39.110425 [THz] =    1304.592025 [cm-1]
     omega( 5) =      39.110425 [THz] =    1304.592025 [cm-1]
     omega( 6) =      40.947359 [THz] =    1365.865939 [cm-1]

As you can see the difference for the 0,1/3,0 is not huge but significant.
As a matter of fact when
I'm using the dvscf produced by QE 4 or 5 for that irr-q it can lead to
physically very different results. Therefore I would like to know which
dvscf file is correct (QE 4 or 5)?

Thank you,

Best Regards,

Samuel Ponce
Department of Materials, University of Oxford
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20150626/f123affe/attachment.html>