Implementation of PBE0 functional in NWCHEM

From NWChem

Viewed 2248 times, With a total of 5 Posts
Jump to: navigation, search

Clicked A Few Times
Threads 11
Posts 37
Hello all,

I would like to extend a work involving the PBE0 functional started with MOLPRO and to continue it with NWCHEM. However, it seems that the energies provided by the default PBE0 methods are not the same for a calculation of the ground state of the carbon atom.

Here are the parameters of the DFT functional in MOLPRO:

Density functional group PBE0=PBEX+PBEC
Functional: PBEX(Automatically generated (new) PBEX) Gradient terms: 1
Functional: PBEC(Automatically generated (new) PBEC) Gradient terms: 1
Exact exchange will be calculated, factor: 0.2500
Functional: PBEX Factor: 0.7500
Functional: PBEC Factor: 1.0000

and the same in NWCHEM:

                        PBE0 Method XC Functional
Hartree-Fock (Exact) Exchange 0.250
PerdewBurkeErnzerhof Exchange Functional 0.750
Perdew 1991 LDA Correlation Functional 1.000 local
PerdewBurkeErnz. Correlation Functional 1.000 non-local

What are the differences between the two software packages' default implementation of PBE0, i.e. can their PBE0 definitions/implementations be adjusted?

I tried the default B3LYP functional as well and I have the same problem.

My nwchem input for carbon atom is:

C library 6-31g
thresh 1.0e-5
task scf
xc pbe0
convergence energy 1.0e-6
convergence gradient 1.0e-6
task dft

Thanks in advance,
Sincerely yours,

Forum Vet
Threads 7
Posts 1287
could you please try an atom where all shells are completely filled, e.g. a gas such as
helium or neon?

If you need to compute DFT atomic energies, the situation requires to somewhat enforce spherical symmetry
and there a few different ways to do it ... and it is possible that Molpro is doing that.

In the case of atoms with completely filled shells there is not any ambiguity of the sort mentioned above.

Cheers, Edo

Clicked A Few Times
Threads 11
Posts 37
Deers Edo

I have already done a calculation of the ground state of a neon atom and I have the same problem.

Moreover, I compute using a cartesian basis set for HF method in MOLPRO and NWCHEM. For the HF method, the value of energies are exactly the same (in the case of carbon and neon atom).

I am going to take a look at the spherical symmetry in MOLPRO.


Forum Vet
Threads 7
Posts 1287
If the Ne energy disagrees, we can exclude the atomic symmetry as a factor.
A possible source of discrepancy is the LDA component used by the Correlation PBE functionals.
NWChem makes use of the Perdew-Wang 1991 functional for the LDA bit of PBE Correlation,
can you tell what is Molpro using?

Forum Vet
Threads 7
Posts 1287
Agreement to 5 digits
I have tried the calculation myself on both codes and I am getting a 5 digits agreement on the total energy of
-128.78985 hartrees

NWChem output file (input echeod at the top)

Molpro input file

Clicked A Few Times
Threads 4
Posts 9
Which version of MOLPRO are you using? MOLPRO 2009 and older versions define a different PBE0 functional.

Forum >> NWChem's corner >> NWChem functionality

Who's here now Members 0 Guests 1 Bots/Crawler 0

AWC's: 2.5.10 MediaWiki - Stand Alone Forum Extension
Forum theme style by: AWC