Why EPW code gets stuck at wannierization-step with USPPS

[roxana]

Hi,

I have used the input files you provided for epw and I had obtain the same Wannier functions that you listed with the Wannier90 code used as standalone.

This is the result with EPW after 2000 iterations (below are snapshots from epw.out and zrs2.wout of the same epw run):
===================================
Running Wannier90

Wannier Function centers (cartesian, alat) and spreads (ang):

( 0.00000 0.00000 1.95280) : 2.08645
( 0.00000 0.00000 1.95280) : 2.08646
( -0.34821 0.20105 1.65224) : 1.45807
( 0.34816 -0.20103 2.25332) : 1.45857
( -0.34821 0.20105 1.65224) : 1.45807
( 0.65184 -0.20103 2.25332) : 1.45858
( 0.00000 -0.00000 1.95279) : 2.35948
( 0.00000 -0.00000 1.95279) : 2.35949
( -0.00000 0.00000 1.95279) : 2.36240
( -0.00000 0.00000 1.95279) : 2.36239
( -0.00000 0.57735 2.47458) : 1.26732
( -0.00000 0.57735 2.47458) : 1.26732
( -0.15184 0.66500 2.25332) : 1.45858
( 0.15184 0.66500 2.25332) : 1.45858
( -0.00000 0.40205 2.25330) : 1.45868
( -0.00000 0.40205 2.25330) : 1.45868
( 0.50000 0.28868 1.43098) : 1.26767
( 0.50000 0.28868 1.43098) : 1.26767
( 0.34821 0.20105 1.65224) : 1.45807
( 0.34821 0.20105 1.65224) : 1.45807
( 0.50000 0.46393 1.65225) : 1.45818
( 0.50000 0.46393 1.65225) : 1.45818
====================================

Final State
WF centre and spread 1 ( 0.000000, 0.000001, 7.200026 ) 2.08645354
WF centre and spread 2 ( 0.000000, 0.000001, 7.200026 ) 2.08645635
WF centre and spread 3 ( -1.283846, 0.741294, 6.091838 ) 1.45806988
WF centre and spread 4 ( 1.283689, -0.741203, 8.308060 ) 1.45857499
WF centre and spread 5 ( -1.283846, 0.741294, 6.091838 ) 1.45806964
WF centre and spread 6 ( 2.403334, -0.741202, 8.308063 ) 1.45857755
WF centre and spread 7 ( 0.000000, -0.000009, 7.200006 ) 2.35948076
WF centre and spread 8 ( 0.000000, -0.000009, 7.200006 ) 2.35949446
WF centre and spread 9 ( -0.000000, 0.000008, 7.200006 ) 2.36240130
WF centre and spread 10 ( -0.000000, 0.000008, 7.200006 ) 2.36238852
WF centre and spread 11 ( -0.000004, 2.128704, 9.123857 ) 1.26731731
WF centre and spread 12 ( -0.000004, 2.128704, 9.123857 ) 1.26731738
WF centre and spread 13 ( -0.559825, 2.451855, 8.308060 ) 1.45857523
WF centre and spread 14 ( 0.559820, 2.451856, 8.308063 ) 1.45857744
WF centre and spread 15 ( -0.000003, 1.482357, 8.307992 ) 1.45868002
WF centre and spread 16 ( -0.000003, 1.482357, 8.307992 ) 1.45868006
WF centre and spread 17 ( 1.843517, 1.064355, 5.276077 ) 1.26766934
WF centre and spread 18 ( 1.843517, 1.064355, 5.276077 ) 1.26766925
WF centre and spread 19 ( 1.283850, 0.741293, 6.091836 ) 1.45807176
WF centre and spread 20 ( 1.283850, 0.741293, 6.091836 ) 1.45807211
WF centre and spread 21 ( 1.843516, 1.710519, 6.091906 ) 1.45817538
WF centre and spread 22 ( 1.843516, 1.710519, 6.091906 ) 1.45817540
Sum of centres and spreads ( 11.061080, 19.158352,158.399334 ) 36.18694767
==========================================================

So it appears that the code version that you are using has some issues. As suggested in my previous reply, you need to download the latest version of EPW for GitLab following the steps below:

git clone https://gitlab.com/QEF/q-e.git

git clone git@gitlab.com:QEF/q-e.git (if https doesn't work)

cd q-e
git checkout develop
./configure
make epw

Best wishes,
Roxana

[Gautam Sharma Monty]

Dear Sir,

Thank you for your help. I am able to obtain MLWFs from EPW using q-e/develop version.
However, there is new issue popping up. After completing the wannierization step. Codes throws
a message saying the following:

%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
Error in routine pw2wan90epw (1):
dipole matrix calculation not implemented with USP - set vme=.TRUE.
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%

However, as suggested by code, after setting vme=.true. , it runs smoothly.
I did not understand the use of "vme" flag (velocity matrix elements beyond local approximation). In my earlier calculations for electron relaxation time using NCPPs, I did not use this flag. What is my actual question is where do we need the information of velocity matrix elements in electron_self_energy calculations?

[roxana]

Hi,

Indeed, velocity matrix elements are not used in electron_self_energy calculations, but for USPP the velocity matrix elements should always be calculated beyond the local approximation. This why vme needs to be true. We will change this as default in the code.

Best,
Roxana

[Gautam Sharma Monty]

I am thankful for the clarification.