total dft energy differs from partial energies


Clicked A Few Times
Dear all, i'm facing either very serious lack in my theory background or compilation/nwchem problem. I did several calculations and just recently i noticed very strange thing (i checked three different calculations) - see part of my output below:

        Total DFT energy =    -1144.144378448133
One electron energy = -5654.477100400897
Coulomb energy = 2607.095668751906
Exchange-Corr. energy = -156.929943044225
Nuclear repulsion energy = 2058.598637126281
Numeric. integr. density = 178.000323079243


when you sum all the energies you get different number (?1145.7127...) than the one reported in Total DFT energy.
i've got this in all three calculations (probably all of them). Also, the difference in reported and "summed" energy is always different, approximately ~1.2 to 1.7 Hartrees. This is the input corresponding to the upper energies:

start dimerBO4-MM1c33_GO-B3LYP-321G

MEMORY stack 2600 heap 500 global 800 mb

title "beta-O4-dimer in ccl4, initial structure from oBabel weighted conf search (MM1) - c33, geometry optimization B3LYP 3-21G"


geometry units an print xyz autosym
H          1.09605       -2.24257       -1.90815
C          1.41847       -2.08854       -0.87674
C          1.84098       -3.17525       -0.07199
O          1.81819       -4.39938       -0.71640
C          2.26350       -2.93298        1.25811
O          2.79497       -3.97266        1.98226
C          2.10800       -1.64323        1.80510
O          2.33657       -1.45588        3.14564
C          1.74719       -0.58915        0.97642
H          1.73301        0.42869        1.34122
C          1.39752       -0.79060       -0.35917
C          0.90359        0.40502       -1.15331
C         -0.54163        0.75608       -0.75684
H         -1.16737       -0.14184       -0.68909
O         -1.10710        1.65905       -1.73269
H         -0.49837        1.18660        0.24421
C         -2.06553        2.54149       -1.28458
C         -3.01765        3.05298       -2.19910
O         -2.94863        2.67755       -3.52330
C         -4.00905        3.94736       -1.72167
H         -4.75432        4.37610       -2.37384
C         -4.06928        4.30063       -0.38461
H         -4.84743        4.97387       -0.04043
C         -3.11615        3.83955        0.49294
H         -3.13149        4.13319        1.53682
C         -2.10145        2.99245        0.04157
H         -1.34790        2.73593        0.77732
C          1.89481       -0.23345        3.71973
C          3.84170       -3.67014        2.90678
C          1.63138       -5.56450        0.09157
H          0.73470       -5.46345        0.71315
H          1.48749       -6.43465       -0.55333
H          2.51469       -5.79038        0.69721
H          3.54935       -4.04304        3.89421
H          4.75448       -4.18066        2.59112
H          4.12551       -2.61952        2.99152
H          2.32886        0.64722        3.23285
H          0.80152       -0.19671        3.70476
H          2.23703       -0.20924        4.75333
C         -3.98692        3.19145       -4.36561
H         -4.96530        2.79737       -4.06074
H         -3.81279        2.87185       -5.39740
H         -3.99778        4.28856       -4.37035
O          1.75905        1.51930       -0.84503
H          1.36853        2.29218       -1.28841
H          0.99887        0.22346       -2.22515
end
basis small
 * library 3-21G
end

basis large
 * library 6-31+G*
end

set "ao basis" small

cosmo
 dielec 2.24
end

dft
 xc b3lyp
 CONVERGENCE density 1e-8
 CONVERGENCE energy 1e-8
 GRID lebedev 90 14 ssf euler
 MAXITER 60
end

driver
 EPREC 1e-8
 GRMS 0.000004
 MAXITER 900
 XYZ geomSteps_r2
end

task dft optimize




I really hope that i'm missing something. Thank you very much for help.

Clicked A Few Times
i forgot to mention, that i checked results obtained on two different clusters, version 6.6 on both of them, but compiled independently on each other, by different person, with delay of ~6 months.

Forum Regular
The confusion is due to the solute-solvent interaction energy not being written explicitly in the output, i.e. the total energy is correct, you are just missing an energy component in the output.

The patch is available at
nwchemgit.github.io/download.php?f=Cosmo_dftprint.patch.gz

Clicked A Few Times
that patch is new, right? also, i'm not sure, but in the COSMO part of the output i didn't attach originally is:
         Total DFT energy =    -1144.144378448133
      One electron energy =    -5654.477100400897
           Coulomb energy =     2607.095668751906
    Exchange-Corr. energy =     -156.929943044225
 Nuclear repulsion energy =     2058.598637126281

 Numeric. integr. density =      178.000323079243

     Total iterative time =     76.1s


                  COSMO solvation results
                  -----------------------

                 gas phase energy =     -1144.1300273697
                 sol phase energy =     -1144.1443784481
 (electrostatic) solvation energy =         0.0143510784 (    9.01 kcal/mol)

                       DFT Final Molecular Orbital Analysis


So I would assume that solvation energy is the solute-solvent interaction energy that you mentioned, isn't it? If so, than this energy is much smaller - only 0.01 Hartree, whereas the difference was bigger than 1 Hartree.
Of course, I'll apply the patch and try the calculation again.

Forum Regular
Yes, the patch is new, created this morning in response to your post.

The listed solvation energy is simply the difference in energy between the gas phase result and the solution phase result and not the solute-solvent interaction energy to which I was referring.


Forum >> NWChem's corner >> Feedback