Hi,
i'm trying to calculate O2 atomization energy (2*E(O)-E(O2)) with PBE, hgh pseudopotentials
(i assume ATK uses
http://www.abinit.org/downloads/psp-links/psp-links/hgh).
I'm having troubles getting ~5.71 eV from atk-12.2.r1.622604e-linux64.bin,
the result I expect based on the calculations with ASE-3.6.0/abinit-5.4.4p::
python `which ase` abinit molecule O2 O --unit-cell='14.00,14.01,14.02' \
-l -p 'xc=PBE,ecut=1500,width=0.001,toldfe=1.0e-7,nstep=250,pps=hgh'
python `which ase` abinit molecule O2 O -s
My ATK script is attached. Could you help me to find out what i'm doing wrong?
I have also additional questions:
1. the SCF with current script settings does not converge, and it seems I have to increase
the electron_temperature to ~5K in order to make SCF converge. In principle i need to run at 0K,
so i tried to reduce the electron_temperature in order to get 0 entropy::
ncdump *.nc | grep "entropy ="
0K seems not accepted::
NL.ComputerScienceUtilities.Exceptions.NLValueError: The parameter, electron_temperature,
must be a positive temperature quantity - e.g. 300*Units.Kelvin
2. how can i determine what electronic state i get? I have singlet and triplet methylene (CH2) molecules.
LCAOCalculator does not seem to accept initial_spin parameter, so i'm thinking about BulkConfiguration
in order to perform such calculations. What other things need to be adjusted in order to describe
a molecule by BulkConfiguration?
