total drift calculation

[Danny]

Dear forum members,

As part of some work we are doing, we are looking into the total force-vector(sum over all atoms of the force-vector on each atom). In the OUTCAR file we find at the end of the set of forces (block TOTAL-FORCE) a line saying total drift: xxx yyy zzz.
In the manual it is mentioned that this is related to the forces in the system, and that you should not look at forces smaller than these values. However, it is not mentioned how this drift is obtained and what the physical meaning of it is.

Initially me and my colleagues were convinced this drift is the total force or a sum of the forces listed above, however trying to verify this showed this is not the case. Could someone shine some light on this darkness?

Danny

[admin]

The x-, y-, and z-components of the force acting on an atom summed over all atoms
should be zero. Due to the numerical noise the sums differ from zero by some
small value (drift). To keep the system of atoms not moving in each (ionic) step
the drift is subtracted from the force acting on each atom.
At lower/higher precision the numerical noise is higher/lower, therefore calculated drift
is larger/smaller.

[Danny]

That I know, but how is it calculated in VASP?
(it is not merely the sum of the forces given in the block above)

[admin]

The drift is a sum of forces in the block above.
The forces printed in the TOTAL-FORCE table
are already corrected, i.e. the sum over the block is zero.

[Danny]

The drift is a sum of forces in the block above.
The forces printed in the TOTAL-FORCE table
are already corrected, i.e. the sum over the block is zero.

oh, that explains a lot. The set of forces is already corrected in the block above with the force-vector given in drift.

thank you very much for this elucidation.

Danny