scf: no. of closed-shell electrons is not even!


  • Guest -
8:04:46 AM PST - Sat, Nov 19th 2011
I have been running calculations on fluorinated vanadylnapthalocyanate, and was having some issues with convergence and bad molecular orbitals. I talked to a theoretical chemistry professor who suggested I add "task scf optimize" before dft in my input file. By running a hartree-fock first, it should make the transition to dft easier and have fewer problems. However, my job came back with the following error pertaining to scf:

scf: no. of closed-shell electrons is not even!

I googled this error message, and found that the only solution is to "define multiplicity." What does this mean, and how do I do it?

Thank you!

Forum Regular
8:50:58 AM PST - Mon, Nov 21st 2011
Hi,

Can you post or send us your input file ?

Thanks.

Best,
-Niri

Niri Govind
niri.govind@pnnl.gov

  • Guest -
3:17:22 PM PST - Tue, Nov 22nd 2011
start voncfjob11

ECHO
ECCE_PRINT /dtemp/bock788/VONcF/VONcFjob11/eccejob11.out

charge 0
geometry units angstroms
[coordinates would be here]
symmetry c4v print
end

basis spherical
  • library 6-311g except V
V library 6-31g*
end

driver
maxiter 300
end

task scf optimize

dft
vectors input hcore
mult 2
iterations 300
xc pbe0
end


task dft optimize
task dft property
property
dipole
end


Also, if you have any information about "restarting" a job, that would be helpful. I was told to use this format for a restart, but the job was unable to run:

restart voncfjob11

permanent_dir /dtemp/bock788/VONcF/VONcFjob11

ECHO
ECCE_PRINT /dtemp/bock788/VONcF/VONcFjob12/eccejob12.out


When I tried to do this, I got an error with line:0

Thank you!

Forum Regular
12:20:39 AM PST - Wed, Nov 23rd 2011
Hi,

With regard to your first question:

charge 0
geometry units angstroms
[coordinates would be here]
symmetry c4v print
end

Have you checked to see if more atoms are being created as a result of the symmetry specified ?
Perhaps you want to do the following:

geometry units angstroms
[specify the coordinates of all the atoms]
end

Let the code determine the symmetry.

Best,
-Niri
niri.govind@pnnl.gov

Gets Around
7:57:38 AM PST - Thu, Nov 24th 2011
The reason seems to be that you calculate singlet with Hartree-Fock for (assumed in dft part) doublet.
Count the electrons in the system and set the correct multiplicity.
For Hartree-Fock:
http://nwchemgit.github.io/index.php/Hartree-Fock_Theory_for_Molecules#Wavefunction_type
for dft with:
http://nwchemgit.github.io/index.php/Density_Functional_Theory_for_Molecules#ODFT_and_MULT_-...
In my impression (i may be wrong - don't know your system)
doing initially Hartree Fock geometry optimization and/or hybrid DFT is an overkill.
Maybe try just with PBE, and refine your results with PBE0 later.
This will offer you also a possibility to check dependence of your results on functional.
Note also that sometimes non-physical systems show convergence problems.
Or problems could be related to the basis set you use, maybe try ECP's (http://nwchemgit.github.io/index.php/ECP)?
Check the example in the Nwchem source:
QA/tests/cu2_m06-l/cu2_m06-l.nw
I don't know what properties you are interested in, but the tiny 6-311G/6-31Gs basis will give you error of tenths of kcal/mol
in atomization energy of a medium/large (~50 atoms) Vanadium complex.


Forum >> NWChem's corner >> Running NWChem