[Pw_forum] How to improve the DOS plot?

[Rolly Ng]

Dear QE experts,

I have managed to do spin polarized calculation on gold-doped graphene 
supercell with total of 72 atoms. There are 71 carbon atoms and 1 gold 
atom at the center of the supercell.

I did scf with 6x6x1 (automatic) k-points, then nscf with 6x6x1 and 
12x12x1 (automatic) k-points respectively. I believe the denser the 
k-points, the better the DOS plot. Then I did dos.x and it produced the 
plot.

However, I found there are lots of noise (oscillations) in the DOS in 
comparison to the 6x6x1 k-points we did previously with DMol3. The DMol3 
is much smoother than that of QE v5.3.0. I am attaching both output 
plots, how can I improve the smoothness of the DOS plot?

I know increasing the k-points for nscf calculation does help but it 
take 120GB of RAM to run 12x12x1 and it is impossible to increase the 
k-point density indefinitely.

Thanks for your advice,
Rolly

-- 
PhD. Research Fellow,
Dept. of Physics & Materials Science,
City University of Hong Kong
Tel: +852 3442 4000
Fax: +852 3442 0538

-------------- next part --------------
A non-text attachment was scrubbed...
Name: DMol3.png
Type: image/png
Size: 58537 bytes
Desc: not available
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20160722/e2edbd28/attachment.png>
-------------- next part --------------
A non-text attachment was scrubbed...
Name: QE.png
Type: image/png
Size: 26254 bytes
Desc: not available
URL: <http://lists.quantum-espresso.org/pipermail/users/attachments/20160722/e2edbd28/attachment-0001.png>