CUTOFF_TYPE

CUTOFF_TYPE = erfc | gaussian | fermi

Default: CUTOFF_TYPE

= erfc

Description: CUTOFF_TYPE chooses the type of cutoff function to be used before performing the singular-value decomposition (SVD) of the initial projections matrix.

The selected columns of the density matrix LSCDM method works by fitting a unitary matrix $U_{mn\mathbf {k} }$ that transforms from the Bloch states obtained by VASP to a Wannier basis. This is done using a one-shot method through a singular-value decomposition (SVD).

This tag governs how much weight should be given in the SVD decomposition to a certain orbital with energy $\epsilon _{n\mathbf {k} }$ . If the weight is zero the orbital is not included in the fitting while if it is one it is included with the maximum importance. The behavior is similar to the wannier90 disentanglement window.

In order to obtain a good Wannierization, a certain level of freedom should be given to the localized orbitals to adequately accommodate the Bloch states. This is particularly important for systems with entangled states.

CUTOFF_TYPE	Function
erfc	${\frac {1}{2}}{\text{erfc}}\left[(\epsilon _{n\mathbf {k} }-\mu )/\sigma \right]$
gaussian	$e^{-(\epsilon _{n\mathbf {k} }-\mu )^{2}/\sigma }$
fermi	${\frac {1}{e^{(\epsilon _{n\mathbf {k} }-\mu )/\sigma }+1}}$

with $\sigma$ specified by the CUTOFF_SIGMA tag and $\mu$ by CUTOFF_MU.

Related tags and articles