~ 1 eV difference in HSE@TDDFT excitation energies between WEST6.2.0 and WEST6.0.0 #19
Description
Dear developers,
I have tested ZnO pristine supercell using hybrid HSE (alpha = 0.378) using WEST 6.2.0 and WEST 6.0.0 versions and obtain large difference for excitation energies using same. For WEST 6.2.0 I compiled QE-7.4.1 and for WEST 6.0.0 I compiled QE-7.3. I checked the QE KS eigenvalues and total energies and they do not differ. Nevertheless for the "wbse_calculation: D" I get different solutions. I attach screenshots of the wbse.out files for each calculation
WEST 6.0.0:
*----------* *----------* *----------*
# Iteration = | 31 | WBSE_dim = | 5 | Diago_dim = | 56 |
*----------* *----------* *----------*
s-DIAGOX done in < 00.1s with cuSOLVER
--------------------------------------------------------------------------------------------
Max diff = 0.183E-05
0.23158 0.23164 0.23408 0.23412 0.23772 0.24021 0.24697 0.24698 0.24703
0.24704 0.24972 0.24973 0.24976 0.24977 0.26520
WEST 6.2.0:
*----------* *----------* *----------*
# Iteration = | 31 | WBSE_dim = | 5 | Diago_dim = | 56 |
*----------* *----------* *----------*
s-DIAGOX done in < 00.1s with cuSOLVER
--------------------------------------------------------------------------------------------
Max diff = 0.183E-05
0.16353 0.16358 0.16603 0.16607 0.16967 0.17215 0.17892 0.17893 0.17898
0.17899 0.18167 0.18167 0.18171 0.18172 0.19715
--------------------------------------------------------------------------------------------
Both calculations achieve convergence and have the same input file.
Thanks,
Erik