Slightly differnet energies with C1V symmetry.

[asaf]

Hi,
I am using Turbomole 6.0 version.
I am trying to calculate energies of two interacting molecules with ridft option (with marij) and $disp keyword in the control file.
The molecular system poses C4V symmetry.
The problem is that when reducing the symmetry of the molecular system (via sy in define) to C1V the energy is slightly different then what should be:
-6973.526764092 instead of -6973.525937287.
The correct  result  ( -6973.525937287) is obtained at the same level of calculation with no use of symmetry  and also with  C2V symmetry.
Is there any problem with C1V ?
the conv option of the scf iteration is 9 , the func is pbe and grid is m4.
Thanks.

[uwe]

Hi,

no, that should not happen. I tried it with a small molecule, and C4v, C2v, C1v, C1 gave all the very same energy. Did you try Cs also (the difference is just that the mirror plane is taken as xy instead of xz)?

Please read in the C4v symmetric coordinates again, run define and use susy instead of sy down to Cs. Save the coordinates and run an energy calculation. Call define again and force C1v symmetry with the Cs coordinates (they might be rotated to change the mirror plane).
Then the next ridft run should give the same energy. If not, please send this as a bug report to turbomole at cosmologic.de.

Regards,

Uwe

[asaf]

Hi,
Following your instructions gives the correct and the same result, both with c1v and cs symmetries.
The different between the old c1v and the new c1v calculation is that in the old calculation the xz plan was xz` (between the atoms) and in the current it is the conventional xz plan.
Can it be the reason for the slightly energy differences ?
Regards.
Asaf.

[uwe]

Hi,

no, that should not change the energy. I have also tried both orientations and it gives the very same result. There must have happened something else in addition. But nothing obvious like wrong charge, basis set, convergence problems - the difference is too small.

Regards,

Uwe