My Community

I have tried to get the frequency-dependent dielectric function in a HSE calculation. Specified NEDOS = 2000, as recommended in the manual, but got only 200 frequency values in OUTCAR -- more than 1 eV apart from each other.
I would appreciate any help to obtain a reasonable epsilon(energy) function.

By the way: does anyone know, how to control broadening in such (or in normal DOS) calculations?

Thanks

Peter Deak

"200 frequency values"

Here one thing is not very clear ! is it no of frequency grid or value of frequency ?

I think it is your frequency grid, which is value of NOMEGA.
And if it is the case then you should not except that NOMEGA should have same value than the NEDOS !!


If you look to the DOSCAR file you will see that the broading is like : (EMAX-EMIN)/NEDOS .......so increase the value of NEDOS and get more fine DOS.

I was talking about the number of frequencies, so probably I'd need to set NOMEGA indeed. (Thanks!) The problem is that I get eps(omega) between omega = 0 and 280 eV, while I need it only up to about 15 eV! To get a reasonable curve over a range of 280 eV I would need something like NOMEGA - 2000 which is unreasonably expensive.
As for the DOS: I meant the use of a Gaussian broadening function at each point with predefined width.

in that case of epsilon plot for certain energy range you need to set OMEGAMAX and OMEGATL (see USER GUIDE) and at the same time you need to set less frequency grid NOMEGA as energy range in only up to 15 eV not 280 or higher.