CSHIFT: Difference between revisions
No edit summary |
No edit summary |
||
| Line 8: | Line 8: | ||
as the grid spacing at low frequencies. If {{TAG|CSHIFT}} is smaller than the grid spacing, | as the grid spacing at low frequencies. If {{TAG|CSHIFT}} is smaller than the grid spacing, | ||
the QP energies might show erratic behavior (for instance large re-normalization factors Z). | the QP energies might show erratic behavior (for instance large re-normalization factors Z). | ||
== Related tags and articles == | == Related tags and articles == | ||
{{TAG|OMEGAMIN}}, | {{TAG|OMEGAMIN}}, | ||
Revision as of 10:36, 26 September 2022
CSHIFT = [real]
| Default: CSHIFT | = OMEGAMAX*1.3 / max(NOMEGA,40) | for GW calculations |
| = 0.1 | for linear response calculations |
Description: CSHIFT sets the (small) complex shift η in the Kramers-Kronig transformation and Hilbert transformation in linear response and GW calculations, respectively.
The default CSHIFT=0.1 is perfectly acceptable for most calculations and causes a slight smoothing of the real part of the dielectric function. If the gap is very small (i.e. approaching two times CSHIFT), slight inaccuracies in the static dielectric constant are possible, which can be remedied by decreasing CSHIFT. If CSHIFT is further decreased, it is strongly recommended to increase the parameter NEDOS to values around 2000.
For GW type calculations, one should manually check that CSHIFT is at least as large as the grid spacing at low frequencies. If CSHIFT is smaller than the grid spacing, the QP energies might show erratic behavior (for instance large re-normalization factors Z).
Related tags and articles
OMEGAMIN, OMEGAMAX, LOPTICS, LNABLA