Different results between using DFT and SCF reference WF in TCE CCSD(T)
Clicked A Few Times
1:26:30 PM PDT - Thu, Aug 28th 2014
Howdy!

I have been doing some geometry optimization calculations using TCE CCSD(T) of NWChem 6.3. I am puzzled by different optimized energy and geometry between using DFT and SCF reference wave functions (RWF) for CCSD(T) calculations.

With DFT RWF

     ----------------------

     Optimization converged

     ----------------------




 Step       Energy      Delta E   Gmax     Grms     Xrms     Xmax   Walltime

 ---- ---------------- -------- -------- -------- -------- -------- --------
@ 3 -1.21194125 -1.7D-07 0.00004 0.00004 0.00007 0.00012 57.8
                                    ok       ok       ok       ok





                               Z-matrix (user)

                               --------



Units are Angstrom for bonds and degrees for angles



     Type          Name      I     J     K     L     M      Value     Gradient

     ----------- --------  ----- ----- ----- ----- ----- ---------- ----------

   1 Stretch                  2     1                       0.75972    0.00004


With SCF RWF

     ----------------------

     Optimization converged

     ----------------------




 Step       Energy      Delta E   Gmax     Grms     Xrms     Xmax   Walltime

 ---- ---------------- -------- -------- -------- -------- -------- --------
@ 3 -1.17245645 1.7D-10 0.00000 0.00000 0.00006 0.00011 17.7
                                    ok       ok       ok       ok





                               Z-matrix (user)

                               --------



Units are Angstrom for bonds and degrees for angles



     Type          Name      I     J     K     L     M      Value     Gradient

     ----------- --------  ----- ----- ----- ----- ----- ---------- ----------

   1 Stretch                  2     1                       0.74265   -0.00000



Here are my inputs.

With DFT RWF

start h2-ccsd

echo

charge 0

geometry units angstroms
symmetry C1
zmat
H
H 1 0.80
end
zcoord
 cvr_scaling 1.2
end
end

dft
  odft

  mult 1

  vectors input atomic output h2.movecs

  xc m06

  iterations 200
end

TCE
DFT

CCSD(T)

FREEZE core
END

driver
 maxiter 80
end

basis "ao basis" cartesian print
  • library "cc-pVTZ"
END

task TCE optimize


With SCF RWF

start h2-ccsd

echo

charge 0

geometry units angstroms
symmetry C1
zmat
H
H 1 0.80
end
zcoord
 cvr_scaling 1.2
end
end

SCF
SINGLET

UHF

vectors input atomic output h2.movecs
END

TCE
SCF

CCSD(T)

FREEZE core
END

driver
 maxiter 80
end

basis "ao basis" cartesian print
  • library "cc-pVTZ"
END

task TCE optimize


I am wondering where is wrong?

Thanks

Clicked A Few Times
5:33:58 PM PDT - Wed, Sep 24th 2014
Using different references gives you different energies.
Formally, CCSD(T) based on the DFT reference and CCSD(T) based on the SCF reference are different methodologies (letting alone the fact that the Fock matrix is no longer diagonal for the DFT case, which leads to some complications in defining perturbative corrections).

Best,
Karol

Clicked A Few Times
5:12:12 PM PDT - Thu, Sep 25th 2014
Uuups,
I haven't noticed that you are using CCSD(T) for two-electron system.
(T) correction is simply not defined for this case ((T) correction involves three body effects).


Forum >> NWChem's corner >> Running NWChem