COSMO and UHF?


Clicked A Few Times
When I run the following job I get an error which states:
Quote:
COSMO cannot be used with UHF !! cannot use COSMO and UHF


echo
memory total 3000 mb
scratch_dir /scratch/eitan1_flux/beckmw/nwchem/CuH2O4_cosmo/
PERMANENT_DIR /scratch/eitan1_flux/beckmw/nwchem/CuH2O4_cosmo/perm/
ecce_print ecce.out
start CuH2O4_cosmo

title "CuH2O4_cosmo"
charge 2

geometry units angstroms print xyz autosym
  Cu       -0.04066        0.14517        0.07708
   O       -0.08809        1.80297       -0.84268
   O       -1.78459       -0.29055        0.68287
   O        0.57948       -1.21055       -1.09548
   O        1.13071        0.27952        1.56249
   H        0.77510        1.84207       -1.32626
   H       -0.76499        1.67180       -1.55337
   H       -1.70147       -1.21247        1.03450
   H       -2.31677       -0.38835       -0.14640
   H       -0.07664       -1.21459       -1.83709
   H        0.40922       -2.05585       -0.60872
   H        0.84539        1.10531        2.02846
   H        0.85165       -0.45479        2.16534
end

dft
 iterations 20000  
 xc b3lyp
 mult 2
end

set driver:linopt 0
driver
 default
 maxiter 20000
 xyz CuH2O4_cosmo_Td
end
title "quick geometry opt"
basis
 C library "STO-3G"
 H library "STO-3G"
 O library "STO-3G"
 N library "STO-3G"
 Cu library "LANL2DZ ECP"
 Zn library "LANL2DZ ECP"
end

ecp
 Cu library "LANL2DZ ECP"
 Zn library "LANL2DZ ECP"
end

task dft optimize


title "geometry opt"
basis
 C library "6-31++G**"
 H library "6-31++G**"
 O library "6-31++G**"
 N library "6-31++G**"
 Cu library "LANL2DZ ECP"
 Zn library "LANL2DZ ECP"	
end

ecp
 Cu library "LANL2DZ ECP" 
 Zn library "LANL2DZ ECP"
end

task dft optimize

title "gas sp energy"
task dft energy

title "gas frequency"
task dft freq

cosmo
 dielec 78.0
 rsolv  1.3
end

title "sp energy in water"
task dft energy

title "freq in water"
task dft freq



The online documentation states
Quote:
. Note that COSMO for unrestricted Hartree-Fock (UHF) method can also be performed by invoking the DFT module with appropriate keywords.

but doesn't actually give what keywords to use. Anyone have any suggestions on how to get this to work?

Forum Regular
The following dft block should do the trick:

   dft
mult 2
xc hfexch
end
task dft

I.e. you should run the regular DFT implementation (not cgmin), select a non-singlet spin so that the UKS formalism is used, but select pure Hartree-Fock exchange as the functional.

Clicked A Few Times
Theory-wise how does hfexch compair to B3LYP? Are there any other levels of theory that I can use for my exchange functional?

Forum Regular
Hi, This depends on the system. B3LYP has 20% HF in it. If you look at our DFT manual pages you can find information about the other exchange-correlation functionals we currently have.

What are specifically looking at with your calculation ?

Best regards,
-Niri

niri.govind@pnl.gov

Clicked A Few Times
I am trying to find the binding constant for a ligand and metal ion in water. The papers which I have seen /am basing my calculations off of use B3LYP as the lowest level of theory. My questions is really is there a higher level of theory than HF which can be used with COSMO when I have unpaired electrons (UHF)? If so what are my options.
Thank you.

Forum Regular
In principle COSMO can be used with many different levels of theory. However, in practice I would think the highest level of theory is CASSCF. The reason is that the COSMO charges are dependent on the electron density. So your level of theory must allow that quantity to be generated readily. The COSMO charges then feed into the one-electron potential. Hence they will change the orbitals. If one were not to update the orbitals then some inconsistency would remain. However, at the moment in NWChem COSMO is implemented up-to the DFT level of theory only.

Clicked A Few Times
So why can't B3LYP be used for COSMO and other theories can?

Forum Vet
This is a quick that is being resolved as we speak. We have implemented COSMO analytical gradients, but not the Hessians so far. And, we are not properly catching this in the code. If a frequency calculation with COSMO is called the code should use analytical gradients to form the numerical hessian.

Bert

Clicked A Few Times
I read through this thread as I was having a similar problem. Bert points out:

Quote:username
We have implemented COSMO analytical gradients, but not the Hessians so far.


I took a look at my output file and realized that the error is only in the frequency portion of the calculation, so I changed:

 task dft freq


to:

 task dft freq numerical


It solved my problem (B3LYP radical opt/freq in COSMO).

regards,



Forum >> NWChem's corner >> General Topics