DEG THRESHOLD: Difference between revisions
More detailed explanation. |
Schmiedmayer (talk | contribs) No edit summary |
||
| (6 intermediate revisions by 2 users not shown) | |||
| Line 1: | Line 1: | ||
{{DISPLAYTITLE:DEG_THRESHOLD}} | |||
{{TAGDEF|DEG_THRESHOLD|[real]}} | {{TAGDEF|DEG_THRESHOLD|[real]}} | ||
{{DEF|DEG_THRESHOLD| $2 | {{DEF|DEG_THRESHOLD| $2 \times 10^{-3}$ | }} | ||
Description: {{TAG|DEG_THRESHOLD}} specifies whether two eigenvalues should be treated as degenerate during clustering of orbitals in the linear response code. {{TAG|DEG_THRESHOLD}} is specified in units of eV. | Description: {{TAG|DEG_THRESHOLD}} specifies whether two eigenvalues should be treated as degenerate during clustering of orbitals in the linear response code. {{TAG|DEG_THRESHOLD}} is specified in units of eV. | ||
---- | ---- | ||
This parameter is used to find degenerate clusters of orbitals. It is used within VASP whenever linear-response calculations are performed (either with respect to external fields or with respect to ionic displacements). VASP internally uses an inverse-iteration solver with | This parameter is used to find degenerate clusters of orbitals. It is used within VASP whenever linear-response calculations are performed (either with respect to external fields or with respect to ionic displacements). VASP internally uses an inverse-iteration solver with imaginary shifting to solve the Sternheimer linear-response equation. If the value is too large, non-degenerate eigenvalue pairs may be detected as degenerate pairs. This can lead to very poor convergence or erroneous results, in particular, for systems with low symmetry and flat semi-core bands. Conversely, too small a value might not lead to a proper detection of all degenerate eigenvalues. For insulating systems, this usually does not lead to substantial errors, since the imaginary shift can handle degenerate eigenvalue/eigenvector pairs (so VASP implements two strategies to handle degenerate pairs and combining them yields more robust results). | ||
It is recommended to use | It is recommended to use small values for systems without symmetry (e.g., around $10^{-5}$). For highly symmetric insulators, results should be largely insensitive to the choice of {{TAG|DEG_THRESHOLD}}. For metals, degenerate states with partial occupancies must be correctly resolved to obtain highly accurate results (so {{TAG|DEG_THRESHOLD}} must be set to a value that allows safe detection of degenerate states). | ||
{{NB|tip|If you experience large drifts in the Born effective charges and adding more k-points does not help, try adjusting this parameter.}} | {{NB|tip|If you experience large drifts in the Born effective charges and adding more k-points does not help, try adjusting this parameter. The total drift in the Born effective charges (calculated with <code>{{TAG|LEPSILON}} = .TRUE.</code>) can be found in the {{FILE|OUTCAR}} file, in a block with the following form (see the final line): | ||
POSITION DIRECTION 1 BORN EFFECTIVE CHARGE (rigid.aug. ionic) | |||
----------------------------------------------------------------------------------------- | |||
*** *** *** *** *** *** ( *** *** ) | |||
----------------------------------------------------------------------------------------- | |||
total drift (improves with k-points): XXX XXX XXX | |||
This block is printed separately for each Cartesian direction. For example, <code>DIRECTION 1</code> is the $x$ direction. | |||
}} | |||
== Related tags and articles == | == Related tags and articles == | ||
{{TAG|LSUBROT}} | {{TAG|LSUBROT}}, {{TAG|LEPSILON}} | ||
[[Category:INCAR tag]][[Category: | [[Category:INCAR tag]][[Category:Linear response]] | ||
Latest revision as of 11:39, 1 June 2026
DEG_THRESHOLD = [real]
| Default: DEG_THRESHOLD | = $2 \times 10^{-3}$ |
Description: DEG_THRESHOLD specifies whether two eigenvalues should be treated as degenerate during clustering of orbitals in the linear response code. DEG_THRESHOLD is specified in units of eV.
This parameter is used to find degenerate clusters of orbitals. It is used within VASP whenever linear-response calculations are performed (either with respect to external fields or with respect to ionic displacements). VASP internally uses an inverse-iteration solver with imaginary shifting to solve the Sternheimer linear-response equation. If the value is too large, non-degenerate eigenvalue pairs may be detected as degenerate pairs. This can lead to very poor convergence or erroneous results, in particular, for systems with low symmetry and flat semi-core bands. Conversely, too small a value might not lead to a proper detection of all degenerate eigenvalues. For insulating systems, this usually does not lead to substantial errors, since the imaginary shift can handle degenerate eigenvalue/eigenvector pairs (so VASP implements two strategies to handle degenerate pairs and combining them yields more robust results).
It is recommended to use small values for systems without symmetry (e.g., around $10^{-5}$). For highly symmetric insulators, results should be largely insensitive to the choice of DEG_THRESHOLD. For metals, degenerate states with partial occupancies must be correctly resolved to obtain highly accurate results (so DEG_THRESHOLD must be set to a value that allows safe detection of degenerate states).
Tip: If you experience large drifts in the Born effective charges and adding more k-points does not help, try adjusting this parameter. The total drift in the Born effective charges (calculated with LEPSILON = .TRUE.) can be found in the OUTCAR file, in a block with the following form (see the final line):
POSITION DIRECTION 1 BORN EFFECTIVE CHARGE (rigid.aug. ionic)
-----------------------------------------------------------------------------------------
*** *** *** *** *** *** ( *** *** )
-----------------------------------------------------------------------------------------
total drift (improves with k-points): XXX XXX XXX
This block is printed separately for each Cartesian direction. For example, |