Dear developer
In EPW, when both eig_read = .true. and lwfpt = .true. are enabled, an error occurs:
task # 16
from setup_rotate_wavefunction : error # 1
symmetry equivalent k points have different energy. Use unfold_ksym = .false.
However, there is no unfold_ksym keyword in EPW. How can this issue be resolved?
best wishes
Donglin Guo
both eig_read = .true. and lwfpt = .true. can not be used at the same time
Moderator: stiwari
-
guodonglin
- Posts: 14
- Joined: Mon Mar 29, 2021 5:56 am
- Affiliation: CQU
Re: both eig_read = .true. and lwfpt = .true. can not be used at the same time
Dear Donglin Guo,
I hope Jae-Mo Lihm has already addressed your question.
Regarding the error message, it suggests that symmetry-equivalent k-points have inconsistent eigenvalues. Could you please check the eigenvalue file you provided and verify whether the eigenvalues for symmetry-equivalent k-points match (to at least ~6 significant digits)? If they do not, you may need to manually average the eigenvalues across these points to restore symmetry.
If you are using eig_read to input GW eigenvalues in place of DFT ones, please note that WFPT is formally not compatible with GW eigenvalues. The reason is that the WFPT splits the self-energy into the active space and rest space terms. The rest space term is computed in ph.x, and stored in the ahc_upfan_iq## file. This is done at the DFT level. If you use GW eigenvalues in EPW, you compute the active space term in the GW level, inconsistent with how the rest space term is calculated. Whether combining these two is justified formally is not very clear.
Best regards,
Shashi (on behalf of Jae-Mo Lihm)
I hope Jae-Mo Lihm has already addressed your question.
Regarding the error message, it suggests that symmetry-equivalent k-points have inconsistent eigenvalues. Could you please check the eigenvalue file you provided and verify whether the eigenvalues for symmetry-equivalent k-points match (to at least ~6 significant digits)? If they do not, you may need to manually average the eigenvalues across these points to restore symmetry.
If you are using eig_read to input GW eigenvalues in place of DFT ones, please note that WFPT is formally not compatible with GW eigenvalues. The reason is that the WFPT splits the self-energy into the active space and rest space terms. The rest space term is computed in ph.x, and stored in the ahc_upfan_iq## file. This is done at the DFT level. If you use GW eigenvalues in EPW, you compute the active space term in the GW level, inconsistent with how the rest space term is calculated. Whether combining these two is justified formally is not very clear.
Best regards,
Shashi (on behalf of Jae-Mo Lihm)