disp vdw 3 with BP86

Just Got Here

8:37:20 AM PDT - Mon, Oct 22nd 2012

Hello

I'm having problems with the DFT-D3 dispersion correction when I use the BP86 functional with nwchem 6.1 on kraken. Using 'disp vdw 3' works fine for other functionals (M06-2X, B3LYP, etc.). Based on the documentation, it should also work for BP86.

Here's my input:

echo

start water

geometry units angstrom
O 0. 0. 0.
H 0. 1. 0.
H 0. 0. 1.
end

basis

 * library 6-31g*

end

dft

 xc becke88 perdew86

 disp vdw 3

end

task dft optimize

From the output, the dispersion contribution blows up:
parksjm@krakenpf4:/lustre/scratch/parksjm/water$ grep vdW water.out_bp86

             vdW contrib  =       -0.247461279735

             vdW contrib  =       -0.515877050996

             vdW contrib  =       -1.203812040935

             vdW contrib  =       -1.590906030809

             vdW contrib  =       -2.136036330957

             vdW contrib  =       -2.910663005651

             vdW contrib  =       -4.058929248674

             vdW contrib  =       -5.739391072862

             vdW contrib  =       -8.363364231227

             vdW contrib  =      -12.400051448054

             vdW contrib  =      -19.041849671375

             vdW contrib  =      -29.745922513719

             vdW contrib  =      -48.376395376068

             vdW contrib  =      -81.453683120641

             vdW contrib  =     -145.309472389670

             vdW contrib  =     -271.803385895461

             vdW contrib  =     -549.054686340327

             vdW contrib  =    -1209.068840024983

             vdW contrib  =    -3031.468199314070

             vdW contrib  =    -8816.993686658659

             vdW contrib  =   -32408.672730546739

             vdW contrib  =  -170468.441171195154

             vdW contrib  = -1736394.661521459930

             vdW contrib  =-76543982.987982153893

             vdW contrib  =**********************

             ...

Any ideas what's going on here?

Thanks,
Jerry

Forum Vet

5:31:24 PM PDT - Thu, Oct 25th 2012
Hi Jerry, There is a bug that will be fixed in the next release for DFT-D3 with BP86. Essentially, it is not picking up the dispersion coefficients (sets them to zero) and this leads to incorrect energies and gradients. If you compiled the code yourself I can send you the code changes you need. Bert Quote:Parksjm Oct 22nd 3:37 pm Hello I'm having problems with the DFT-D3 dispersion correction when I use the BP86 functional with nwchem 6.1 on kraken. Using 'disp vdw 3' works fine for other functionals (M06-2X, B3LYP, etc.). Based on the documentation, it should also work for BP86. Here's my input: echo start water geometry units angstrom O 0. 0. 0. H 0. 1. 0. H 0. 0. 1. end basis * library 6-31g* end dft xc becke88 perdew86 disp vdw 3 end task dft optimize From the output, the dispersion contribution blows up: parksjm@krakenpf4:/lustre/scratch/parksjm/water$ grep vdW water.out_bp86 vdW contrib = -0.247461279735 vdW contrib = -0.515877050996 vdW contrib = -1.203812040935 vdW contrib = -1.590906030809 vdW contrib = -2.136036330957 vdW contrib = -2.910663005651 vdW contrib = -4.058929248674 vdW contrib = -5.739391072862 vdW contrib = -8.363364231227 vdW contrib = -12.400051448054 vdW contrib = -19.041849671375 vdW contrib = -29.745922513719 vdW contrib = -48.376395376068 vdW contrib = -81.453683120641 vdW contrib = -145.309472389670 vdW contrib = -271.803385895461 vdW contrib = -549.054686340327 vdW contrib = -1209.068840024983 vdW contrib = -3031.468199314070 vdW contrib = -8816.993686658659 vdW contrib = -32408.672730546739 vdW contrib = -170468.441171195154 vdW contrib = -1736394.661521459930 vdW contrib =-76543982.987982153893 vdW contrib =********************** ... Any ideas what's going on here? Thanks, Jerry

Just Got Here

4:50:41 AM PDT - Fri, Oct 26th 2012
Bert Yes, it would be great it you could provide the code changes. Thanks, Jerry

Forum Vet

4:37:34 PM PDT - Fri, Oct 26th 2012
Jerry, Bert Quote:Parksjm Oct 26th 11:50 am Bert Yes, it would be great it you could provide the code changes. Thanks, Jerry

Forum >> NWChem's corner >> NWChem functionality