Abnormal Phonon Spectrum in EPW Including Quadrupole Polarization

[guodonglin]

I am currently using the EPW code to calculate electron-phonon coupling and transport properties for a 3D material. I encountered an issue regarding the phonon spectrum output when including quadrupole polarization effects.

Here are the details of the problem:

When I perform calculations without quadrupole polarization, the phonon spectrum is physically reasonable and matches expectations.

However, when I enable quadrupole polarization, the resulting phonon spectrum appears unphysical .


Could you kindly provide guidance on the following?

[stiwari]

Hi,

Can you kindly provide more information (including all input and output (.out) files) to debug the issue you are facing.

Best regards,
Sabya.

[guodonglin]

for quadrupole and none quadrupole, the input file is the same, namely, epw1.in. the output files are epw1.out and epw1.out-quad.

[Nina Girotto]

Hello,

Comparing your output files, you can notice the wannierization produced different results in the two calculations. Is it possible to provide initial projections explicitly, instead of using the 'random' option?

Best,

Nina.

[sponce]

Hello,

I confirm that this is the issue. Your Wannierization is of bad quality and different in the two outputs because you use random Wannier functions.

The use of 'random' is strongly discouraged as it often produce bad results.

Best wishes,
Samuel

[guodonglin]

@Nina Girotto @sponce @stiwari

When I modify the initial projection orbitals of Wannier and have completed the fitting, the initial k-points and q-points are 666 and 333, respectively. When considering quadrupole polarization, the phonon spectrum output by EPW is still abnormal. I am unsure whether the issue lies in the setting of a certain keyword in the input file or some other problem. I would appreciate any suggestions to help me resolve this issue. Thank you.


<<< Delta < 1.000E-12 over 4 iterations >>>
<<< Wannierisation convergence criteria satisfied >>>

Final State
WF centre and spread 1 ( 1.801377, 1.047726, 3.455151 ) 2.36771146
WF centre and spread 2 ( 1.516266, 0.441110, 4.378067 ) 2.29301712
WF centre and spread 3 ( 0.672396, 1.047710, 3.738731 ) 2.36199510
WF centre and spread 4 ( 1.516242, 1.654246, 4.378072 ) 2.29304525
WF centre and spread 5 ( 9.765575, 3.143117, 0.959982 ) 2.36767206
WF centre and spread 6 ( 10.050691, 3.749586, 0.037004 ) 2.29305458
WF centre and spread 7 ( 10.894562, 3.143071, 0.676370 ) 2.36197407
WF centre and spread 8 ( 10.050678, 2.536453, 0.037091 ) 2.29301443
WF centre and spread 9 ( 3.982151, 3.143038, 1.247566 ) 2.36769106
WF centre and spread 10 ( 4.267228, 2.536503, 2.170534 ) 2.29300986
WF centre and spread 11 ( 5.111093, 3.143085, 1.531225 ) 2.36201758
WF centre and spread 12 ( 4.267194, 3.749652, 2.170471 ) 2.29301281
WF centre and spread 13 ( 7.584796, 1.047684, 3.167565 ) 2.36765062
WF centre and spread 14 ( 7.299779, 1.654251, 2.244609 ) 2.29302174
WF centre and spread 15 ( 6.455848, 1.047717, 2.883877 ) 2.36199295
WF centre and spread 16 ( 7.299717, 0.441105, 2.244621 ) 2.29300489
WF centre and spread 17 ( 9.592615, 1.047615, 1.929666 ) 1.46722131
WF centre and spread 18 ( 9.850282, 0.604148, 2.625898 ) 1.41287574
WF centre and spread 19 ( 10.455850, 1.047696, 2.118040 ) 1.46579076
WF centre and spread 20 ( 9.850219, 1.491297, 2.625771 ) 1.41286905
WF centre and spread 21 ( 1.974335, 3.143162, 2.485449 ) 1.46722238
WF centre and spread 22 ( 1.716646, 3.586588, 1.789205 ) 1.41287068
WF centre and spread 23 ( 1.111098, 3.143025, 2.297075 ) 1.46578023
WF centre and spread 24 ( 1.716753, 2.699437, 1.789358 ) 1.41286751
WF centre and spread 25 ( 7.757809, 3.143136, -0.277896 ) 1.46722946
WF centre and spread 26 ( 7.500163, 3.586591, 0.418337 ) 1.41286901
WF centre and spread 27 ( 6.894567, 3.143069, -0.089515 ) 1.46575999
WF centre and spread 28 ( 7.500202, 2.699439, 0.418212 ) 1.41286218
WF centre and spread 29 ( 3.809141, 1.047767, 4.693011 ) 1.46723246
WF centre and spread 30 ( 4.066805, 1.491222, 3.996775 ) 1.41286627
WF centre and spread 31 ( 4.672385, 1.047671, 4.504631 ) 1.46575294
WF centre and spread 32 ( 4.066729, 0.604069, 3.996908 ) 1.41286301
Sum of centres and spreads (185.071192, 67.051987, 70.641861 ) 60.29781855

Spreads (Ang^2) Omega I = 50.871835985
================ Omega D = 0.035732075
Omega OD = 9.390178224
Final Spread (Ang^2) Omega Total = 60.297746284

[guodonglin]

this are the output and input files

[Nina Girotto]

Hello!

To me everything looks fine now. I would recommend checking the convergence of the matrix elements. The grids might be too coarse. Would it be possible to increase the k/q grids and repeat the calculations?

Best regards,

Nina.