NWChem vs Gaussian energies


Just Got Here
Dear all,
I'm a NWChem newbie and I'm performing single point calculations on a water geometry with both NWChem and Gaussian. I'm working at the DFT level, using the BP86 exchange-correlation functional and the 6-311G* basis set. It may seem really trivial, but I can't get comparable energies (i.e. equal to the 7-8th digit) from the two programs. I also tried following the suggestions posted here: http://comp.chem.umn.edu/info/NWChem.html with little change. Here are my input files:

NWChem
title "w sp"

geometry
O             0.00000           0.00000           0.00000
H             0.92264           0.65241           0.00000
H            -0.92264           0.65241           0.00000
end

charge 0

scf
 singlet
 rhf
end

dft
 xc slater 1.0 becke88 nonlocal 1.0 perdew86 1.0 vwn_1_rpa 1.0
 iterations 5000
 grid lebedev 90 14 ssf euler
end

basis "ao basis" spherical
 o library 6-311g*
 h library 6-311g*
end

task dft

ecce_print w_dft


Gaussian
%nproc=4

# bp86/6-311g* scf=(tight) nosym integral(grid=ultrafine)

w sp

 0 1
O             0.00000           0.00000           0.00000
H             0.92264           0.65241           0.00000
H            -0.92264           0.65241           0.00000


The energies I get are:

NWChem  :  -76.600926730992
Gaussian:  -76.4024349388


which are indeed very different. The same happens if I change the functional to B3LYP, even if the difference is much less dramatic.
I am probably doing something wrong but I don't understand what. Do you have any suggestions?

Thank you

Forum Vet
use xc becke88 perdew86
If you use xc becke88 perdew86, the energy computed by NWChem is -76.402352654964

dft
# xc slater 1.0 becke88 nonlocal 1.0 perdew86 1.0 vwn_1_rpa 1.0
xc becke88 perdew86
 iterations 5000
 grid lebedev 90 14 ssf euler
end

Just Got Here
Thank you! I had used that explicit expression for BP86 because of an e-mail I had found (here). Is that wrong?

Another thing, I see that nwchem supports the resolution of identity approximation for MP2, but I'm not sure about dft... is it possible?

Thank you again

Forum Vet
Clef,
I am not commenting on definitions that don't see to work ...

The DFT equivalent of RI-MP2 requires the definition of a second basis set, i.e. charge density fitting basis set

More details at

http://nwchemgit.github.io/index.php/Release61:Density_Functional_Theory_for_Molecules#Speci...

Just Got Here
I see; so to use the ri approximation I simply have to define the auxiliary basis set. Very nice! I will try it as soon as possible

Thank you again


Forum >> NWChem's corner >> NWChem functionality