Question about epsilon subsampling: is_q0=0 points still reading from WFNq

14 views

Skip to first unread message

Channing Hou

unread,

Jun 7, 2025, 4:39:03 AMJun 7

to BerkeleyGW Help

Subject: epsilon subsampling: is_q0=0 still reads from WFNq causing error

Hi,

I'm using epsilon with subsampling and encountering an issue:

Setup:

q-point in epsilon.inp :

begin qpoints

9.258971095898462E-03 9.258971095898462E-03 0.000000000000000E+00 1 1

3.703588438359385E-02 3.703588438359385E-02 0.000000000000000E+00 1 1

8.333073986308615E-02 8.333073986308615E-02 0.000000000000000E+00 1 1

0.000000000 0.333333333 0.000000000 1 0

0.000000000 0.666666667 0.000000000 1 0

0.333333333 0.333333333 0.000000000 1 0

0.333333333 0.666666667 0.000000000 1 0

0.666666667 0.333333333 0.000000000 1 0

end

Generated k-points for wfnq using setup_subsampling_nns.x from wfn calculation :

0.009258971 0.009258971 0.000000000 1.0
0.009258971 0.342592304 0.000000000 1.0
0.009258971 0.675925638 0.000000000 1.0
0.342592304 0.342592304 0.000000000 1.0
0.342592304 0.675925638 0.000000000 1.0
0.675925638 0.342592304 0.000000000 1.0
0.037035884 0.037035884 0.000000000 1.0
0.037035884 0.370369217 0.000000000 1.0
0.037035884 0.703702551 0.000000000 1.0
0.370369217 0.370369217 0.000000000 1.0
0.370369217 0.703702551 0.000000000 1.0
0.703702551 0.370369217 0.000000000 1.0
0.083330740 0.083330740 0.000000000 1.0
0.083330740 0.416664073 0.000000000 1.0
0.083330740 0.749997407 0.000000000 1.0
0.416664073 0.416664073 0.000000000 1.0
0.416664073 0.749997407 0.000000000 1.0
0.749997407 0.416664073 0.000000000 1.0

Problem: Despite is_q0 = 0 (which should read from WFN according to manual), epsilon still tries to read from WFNq and fails:

ERROR: genwf_mpi: No match for rkq point: 0.000 0.333 0.000 in file WFNq

The missing k-point (0.000, 0.333, 0.000) wasn't included in WFNq by the subsampling setup.

Question: Why does is_q0 = 0 still require WFNq in subsampling mode? The documentation suggests it should only use WFN.