# [CP2K-user] [CP2K:13088] Forces in BOMD calculations

Dmitrii Nikolaev dmitrii.... at gmail.com
Thu Apr 9 12:39:33 UTC 2020

Thank you very much!
My energy is not shifting, and the forces are fluctuating around 20
kcal/(mol*A) -- 0.016 Ht/Bohr. Is it ok?

On Thursday, April 9, 2020 at 3:06:17 AM UTC+3, Lucas Lodeiro wrote:
>
> Hi,
> In the case of BO framework, the forces calculated with Hellmann-Feynman
> theorem have some instabilities, due to the theorem requires that
> <d\phi|\phi> = 0 = <\phi|d\phi>. In most cases this terms are very small,
> but requires a very accurate convergence of energy and the wfc to ensure a
> completely negligible value.
> Then the real forces are equal <dH> + E<d\phi|\phi>.
> if <d\phi|\phi> is not negligible, you are avoiding a part of the real
> forces and the summation  of them must not be 0.
>
> This in the theory. The implementation of this, uses no infinite numbers
> and some noise are in the numerical evaluation, another thing to no
> cancel completely.
>
> The important thing with this issue is if you are using and NVE system,
> the total energy (kinetic + potential) must be the conserved quantity, not
> in a perfect numeral way, but the oscillations must be small and without a
> biased direction, if the average (along the fluctuations) total energy is
> incrementing or decreasing you have problems in the forces application,
> either in the force calculation (more accurate scf cycles) or a too big
> time step integrator.
>
> El mié., 8 abr. 2020 a las 13:59, Dmitrii Nikolaev (<dmi... at gmail.com
> <javascript:>>) escribió:
>
>> Dear collegues,
>>
>> I am running the Born-Oppenheimer molecular dynamics, and I notice that
>> the sum of the forces on all atoms of my molecule is non-zero. However, as
>> I understand for the NVE ensemble the total force should be zero? What is
>> the reason for this force?
>>
>> Dmitrii
>>
>> --
>> You received this message because you are subscribed to the Google Groups
>> "cp2k" group.
>> To unsubscribe from this group and stop receiving emails from it, send an
>> email to c... at googlegroups.com <javascript:>.
>> To view this discussion on the web visit