Bug in cp2k_pdos for plotting smoothed PDOS output

106 views

Skip to first unread message

Dobromir A Kalchevski

unread,

Jan 3, 2023, 7:17:47 PM1/3/23

to cp2k

Hello,

I was using cp2k_pdos from cp2k-output-tools to plot a smooth PDOS graphic with gnuplot and I noticed there are always errors. The peaks are shifted to wrong eV values and the gaps are not the same size. I use Band Structure diagram to see where should I have peaks and where I shouldn't + I read the original .pdos output from cp2k. The Band Structure and the .pdos are in a perfect agreement. The converted file from cp2k_pdos is the problem.

I tried adjusting --sigma (smoothing) and --de (integration step size) and nothing changes no matter the range of values / combinations.

I am attaching the Band Structure, plotted .pdos and 2 different plots from cp2k_pdos to visualize the problem.

I really hope you fix this soon.

Best Regards,

Dobromir

right_pdos.png

wrong_pdos.png

wrong_pdos2.png

bs.png

Dobromir A Kalchevski

unread,

Jan 4, 2023, 3:43:47 PM1/4/23

to cp2k

Hello,

may be I was mistaken and didn't know how to use the script. It turned out it was adjusted for 0 being the Fermi level. On top of that not the HOMO energy level as Fermi level (as all graphics in publications I have seen) but a Fermi level calculated in another way.

If it is of any help to anyone, this is how I plot s,p,d and total density from one file in gnuplot:

D =
plot "pdos-k1-1.con2" u ($1+D):2 w l lw 1.5 t "", "pdos-k1-1.con2" u ($1+D):(sum[x=3:5] column(x)) w l lw 1.5 t "", "pdos-k1-1.con2" u ($1+D):(sum[y=6:10] column(y)) w l lw 1.5 t "", "pdos-k1-1.con2" u ($1+D):(sum[y=2:10] column(y)) w l lw 1.5 dt 2 t ""

Best Regards,

D is the difference between the Fermi level in .pdos and the HOMO energy level in the same .pdos