atomic C does not converge with nolevelshifting


Clicked A Few Times
Hello Nwchem users

I am trying to converge the SCF of atomic carbon with nolevelshifting as I want to compare these energies finally with smeared metallic cluster calculations. Both the gas phase and solvent phase did not converge for specified # iterations upto 3000. I tried different strategies (damping, cgmin, input vectors from scf/b3lyp for my pbe96 calculation) for obtaining convergence but it fails always. However, with default levelshifting it converges fine. Below is my input file. Thank you for any help.

echo

start cluster
geometry autosym
C 0 0 0
end


basis "ao basis" cartesian
C library 6-311G** file /usr/local/NWChem/data/libraries/
end

dft
iterations 1000
direct
mult 1
XC xpbe96 cpbe96
convergence ncyds 1000 nolevelshifting
end

DRIVER
XYZ a.xyz
MAXITER 500
END

cosmo
dielec 80.4
solvent water
do_cosmo_smd true
end


set cosmo:map cosmo-bondi.par
task dft

Forum Vet
You are missing two essential ingredients in your input file to get the proper DFT atomic solution
1) You need to try to mimic the spherical symmetry (e.g. using the Oh point group as in the input below)
2) You need fractional occupation in place (e.g. using the smear keyword as in the input below)

echo

start cluster
geometry 
C 0 0 0
symmetry oh
end


basis "ao basis" cartesian
C library 6-311G** file  /usr/local/NWChem/data/libraries/
end

dft
 iterations 1000
 direct
 mult 1
 XC xpbe96 cpbe96
 smear 1d-7
end

cosmo
dielec 80.4
solvent water
do_cosmo_smd true
end

set cosmo:map cosmo-bondi.par

task dft

Clicked A Few Times
Thank you, it works. So as expected Oh point group does not work for atomic energy calculation for H atom (with the same input block as you specified). So I obtain H atom's energy using C1 symmetry and smear=1d-6. I tried using FON, but for 1 electron, it didn't make sense to me. So does the used setting give a proper DFT energy?

Forum Vet
Why do you need smearing for the hydrogen atom?
The only electron should occupy the alpha 1s orbital.


Forum >> NWChem's corner >> General Topics