Symmetry fudging
Clicked A Few Times
Hi,

I'm pretty sure this must have been discussed before but I cannot find it so quickly. I'm confused about the " symmetry fudging" that occurs during the SCF in NWchem 6.0.

I'm performing DFT calculations one a range of organic molecules and when I use a relatively small basis-set (e.g. 6-31G*) the SCF converges smoothly and without any warnings. When I perform a calculation on exactly the same structure with the same input-file but now only with a def2-TZVPP basis-set I suddenly get the following behavior:

 convergence    iter        energy       DeltaE   RMS-Dens  Diis-err    time

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

d= 0,ls=0.0,diis     1  -1229.6972526956 -4.00D+03  1.76D-03  5.83D+00  3411.3

WARNING: error on integrated density =  0.10D-05

greater than required accuracy of  0.10D-05



!! scf_movecs_sym_adapt:  881 vectors were symmetry contaminated



 Symmetry fudging



!! scf_movecs_sym_adapt:   20 vectors were symmetry contaminated



d= 0,ls=0.0,diis     2  -1230.0606455141 -3.63D-01  4.06D-04  1.08D-01  5692.6

WARNING: error on integrated density =  0.10D-05

greater than required accuracy of  0.10D-05



!! scf_movecs_sym_adapt: 1157 vectors were symmetry contaminated



 Symmetry fudging



!! scf_movecs_sym_adapt:   50 vectors were symmetry contaminated



d= 0,ls=0.0,diis     3  -1230.0607072429 -6.17D-05  2.35D-04  1.05D-01  7969.4

WARNING: error on integrated density =  0.10D-05

greater than required accuracy of  0.10D-05



!! scf_movecs_sym_adapt:  986 vectors were symmetry contaminated



 Symmetry fudging



!! scf_movecs_sym_adapt:   30 vectors were symmetry contaminated



d= 0,ls=0.0,diis     4  -1230.0733545208 -1.26D-02  7.70D-05  1.56D-02 10242.5

WARNING: error on integrated density =  0.10D-05

greater than required accuracy of  0.10D-05



!! scf_movecs_sym_adapt:  543 vectors were symmetry contaminated



 Symmetry fudging



!! scf_movecs_sym_adapt:    6 vectors were symmetry contaminated



d= 0,ls=0.0,diis     5  -1230.0754934434 -2.14D-03  1.65D-05  7.02D-04 12520.6

WARNING: error on integrated density =  0.10D-05

greater than required accuracy of  0.10D-05



!! scf_movecs_sym_adapt:  182 vectors were symmetry contaminated



 Symmetry fudging



!! scf_movecs_sym_adapt:    4 vectors were symmetry contaminated



d= 0,ls=0.0,diis     6  -1230.0755749648 -8.15D-05  5.86D-06  1.29D-04 14804.5


What is going on? If this is really a symmetry issue why does this not occur for the small basis-set calculation. I use autosym, I guess without it these issues might go away, but the SCF calculation is only the starting point for a subsequent TD-DFT calculation in which I would like to have symmetry so I can label the irreps.

Regards,

Martijn

Forum Vet
11:02:29 AM PDT - Wed, Apr 25th 2012
It simply means the orbitals in the initial atomic guess were not well symmetry adapted, which can happen with more diffuse basis sets. As part of the SCF process the orbitals and density get symmetry adapted (symmetry fudging is just a term used in the code, means that orbitals get symmetry adapted as much as possible). As you can see in your output, the remaining number of symmetry contaminated vectors is getting smaller and smaller, and should disappear. So, you shouldn't worry about it.

Bert



I'm pretty sure this must have been discussed before but I cannot find it so quickly. I'm confused about the " symmetry fudging" that occurs during the SCF in NWchem 6.0.

I'm performing DFT calculations one a range of organic molecules and when I use a relatively small basis-set (e.g. 6-31G*) the SCF converges smoothly and without any warnings. When I perform a calculation on exactly the same structure with the same input-file but now only with a def2-TZVPP basis-set I suddenly get the following behavior:

 convergence    iter        energy       DeltaE   RMS-Dens  Diis-err    time

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

d= 0,ls=0.0,diis     1  -1229.6972526956 -4.00D+03  1.76D-03  5.83D+00  3411.3

WARNING: error on integrated density =  0.10D-05

greater than required accuracy of  0.10D-05



!! scf_movecs_sym_adapt:  881 vectors were symmetry contaminated



 Symmetry fudging



!! scf_movecs_sym_adapt:   20 vectors were symmetry contaminated



d= 0,ls=0.0,diis     2  -1230.0606455141 -3.63D-01  4.06D-04  1.08D-01  5692.6

WARNING: error on integrated density =  0.10D-05

greater than required accuracy of  0.10D-05



!! scf_movecs_sym_adapt: 1157 vectors were symmetry contaminated



 Symmetry fudging



!! scf_movecs_sym_adapt:   50 vectors were symmetry contaminated



d= 0,ls=0.0,diis     3  -1230.0607072429 -6.17D-05  2.35D-04  1.05D-01  7969.4

WARNING: error on integrated density =  0.10D-05

greater than required accuracy of  0.10D-05



!! scf_movecs_sym_adapt:  986 vectors were symmetry contaminated



 Symmetry fudging



!! scf_movecs_sym_adapt:   30 vectors were symmetry contaminated



d= 0,ls=0.0,diis     4  -1230.0733545208 -1.26D-02  7.70D-05  1.56D-02 10242.5

WARNING: error on integrated density =  0.10D-05

greater than required accuracy of  0.10D-05



!! scf_movecs_sym_adapt:  543 vectors were symmetry contaminated



 Symmetry fudging



!! scf_movecs_sym_adapt:    6 vectors were symmetry contaminated



d= 0,ls=0.0,diis     5  -1230.0754934434 -2.14D-03  1.65D-05  7.02D-04 12520.6

WARNING: error on integrated density =  0.10D-05

greater than required accuracy of  0.10D-05



!! scf_movecs_sym_adapt:  182 vectors were symmetry contaminated



 Symmetry fudging



!! scf_movecs_sym_adapt:    4 vectors were symmetry contaminated



d= 0,ls=0.0,diis     6  -1230.0755749648 -8.15D-05  5.86D-06  1.29D-04 14804.5


What is going on? If this is really a symmetry issue why does this not occur for the small basis-set calculation. I use autosym, I guess without it these issues might go away, but the SCF calculation is only the starting point for a subsequent TD-DFT calculation in which I would like to have symmetry so I can label the irreps.

Regards,

Martijn


Forum >> NWChem's corner >> Running NWChem