Finer DOS plot of HSE calculation

Message

laalitha · #1 Post by **laalitha** » Fri Nov 09, 2012 7:15 pm

I have done a spin polarized calculation of Sr-Hexaferrite (64 atom hexagonal crystal structure) with HSE functional. Unfortunately the NEDOS was not set (Default NEDOS used) and the DOS plot is coarse. Is there any way to get a finer DOS plot say NEDOS = 1204
without redoing the HSE calculation?

Thanks

dhfphysics · #2 Post by **dhfphysics** » Wed Nov 21, 2012 8:13 am

I have written my own program to calculate a DOS (or PDOS) from the OUTCAR (or POSCAR) using Gaussian or Fermi smearing in energy. This is not too difficult and you can make sure that the DOS has the correct integral by making sure that each occupied state distributes 100% of it's weight among the nearby energy bins through a renormalization process.

dhfphysics · #3 Post by **dhfphysics** » Wed Nov 21, 2012 8:24 am

One caveat. Increasing NEDOS does not give a finer DOS for a calculation that was done with too few k-points. This can be a problem for HSE calculations.

Of course, having lots of atoms helps by increasing the number of bands and therefore the number of eigenvalues in a single energy bin of fixed size. Your 64 atoms might be enough to be ok with say, 2x2x2 kpoints, or perhaps even 1 kpoint.