electron DOS smearing...

[odelaire]

Dear VASP users,

I have a question regarding the different electronic smearing options in VASP:
when using Fermi smearing (ISMEAR = -1), to what quantity is the smearing (SIGMA) applied?
To be more specific: the Methfessel Paxton prescribes a smearing corresponding to the approximation of delta functions in energy delta(E-E_k) by functions with finite smearing widths. This induces a smearing of features in the electronic DOS.

What I wonder, regarding the Fermi smearing option in VASP, is whether the energy levels are similarly broadened in energy (as in a MP scheme of order zero), or wether it is just the _occupation numbers_ that follow the Fermi distribution?
The section of the manual on ISMEAR did not answer my question precisely, hence the post.

Since I get smoother electronic DOS curves with VASP upon increasing SIGMA (when using Fermi smearing, ISMEAR=-1) for a fixed CHGCAR, it seems to me that the energy levels are indeed broadened (by the derivative of the Fermi function?). Is this correct?

Thanks for your help.

Olivier

[admin]

you are right, the smearing (SIGMA) is applied to broaden the energy levels (KS-eigenvalues) themselves.

[odelaire]

To clarify this further: what is the functional form of the Broadening for the KS-eigenvalues in this case (ISMEAR = -1) ?
Is it the energy derivative of the Fermi function?

Thanks in advance.

[odelaire]

Also, in the case of ISMEAR=-1, is SIGMA the value of k*Tel,
or is it defining some other measure of the width of the derivative of the Fermi-Dirac function?