[CP2K-user] [CP2K:13757] Re: H2 single molecule AIMD

Lucas Lodeiro eluni... at gmail.com
Sat Aug 15 06:05:33 UTC 2020

Hi Marcella an Travis,

I try both advice. To improve the forces, I increase the cutoff from 350 to
700 Ry and rel_cutoff from 40 to 60 Ry, and decrease the eps_scf (inner and
outer scf) from 1.0E-6 to 1.0E-8. This does not solve the rotational
problem, but the pump is lower than the no-improve input. The system shows
equal the two fake peaks but the "distance" between them is smaller,
and approaches
the corrected trajectory with the true one peak.
For the second, I set angular velocity tolerance equal to 1.0E-11. In this
case the rotation is off along all the trajectory, just a little
temperature decreasing along the trajectory (17 K in 40 ps).

Firstly I want to thank you for your disposition and advice. Secondly, I
have some questions.
In the case of improving the forces, as I am using a time step of 0.25 fs,
I think the cutoff and eps values are the main parameters to improve the
forces, isn't it? Do you expect a better improvement?
In the case of angular velocity tolerance, a value of 1.0E-11, is a too
little one? There is a typical value to balance the angular movement and to
subtract the lesser amount of kinetic energy?


El mar., 11 ago. 2020 a las 3:12, Marcella Iannuzzi (<marci... at gmail.com>)

> Hi Lucas,
> I would rather try to improve the accuracy of the integration of the
> equations of motion.
> The error in the forces is probably too large to guarantee the right
> behaviour of the conserved quantities.
> Regards
> Marcella
> On Monday, August 10, 2020 at 6:59:57 PM UTC+2 Lucas Lodeiro wrote:
>> Hi Marcella,
>> I try the MD with periodic and non periodic conditions. When ANGVEL_ZERO
>> was applied for a non periodic cell, the simulation starts with zero
>> angular velocity, but the numerical inaccuracies pumps some angular
>> velocity at the first steps... and the molecule starts rotating.
>> As you mention, for a diatomic molecule it is not possible to apply any
>> constraint... in all cases it will be not physical.
>> Is there a way to fix the rotational movement?
>> El lun., 10 ago. 2020 a las 6:03, Marcella Iannuzzi (<ma... at gmail.com>)
>> escribió:
>>> The translational and rotational degrees of freedom are subtracted,
>>> hence fixing 4 more degrees of freedom  by constraints you end up with a
>>> negative number of DoF, which is obviously not physical.
>>> Are you using PBC? The ANGVEL_ZERO keyword is ignored with PBC
>>> Regards
>>> Marcella
>>> On Sunday, August 9, 2020 at 10:26:49 PM UTC+2 Lucas Lodeiro wrote:
>>>> Hi all!
>>>> I am exploring single molecule AIMD (H2 in this case) for different
>>>> purposes. The main problem that I face is temperature/rotational modes. I
>>>> try to run a AIMD using NVE and NVT ensembles, but there are some problems.
>>>> For an NVE ensemble, the system starts rotating with a conserved
>>>> angular moment, even if I set equal to zero the total angular
>>>> moment... This disrupt the linear movement of H2 bond oscillation. The
>>>> rotation couples with the bond and generates a fake two frequency
>>>> oscillator. If, after the AIMD, I correct the trajectory to a zero total
>>>> angular moment for all frames (simple for diatomic homonuclear molecule,
>>>> but not so easy for polyatomic molecules), the oscillator recovers the
>>>> single frequency motion.
>>>> For an NVT  ensemble, the thermalization is so hard for few atoms...
>>>> and the system is very bad behaved.
>>>> Now I try an NVE ensemble with constrain MD, fixing the Y and Z
>>>> position of both atoms, and put the bond along the X direction.
>>>>     &FIXED_ATOMS
>>>>       COMPONENTS_TO_FIX      YZ
>>>>       LIST                   1 2
>>>> I think with it, I force the MD to evolve only over the X axis... but
>>>> the initial step gives me a NaN temperature and kinetic energy, no
>>>> matter what initial temperature I use. I attach both files.
>>>> The output file prints:
>>>>  Calculation of degrees of freedom
>>>>                                                       Number of atoms:
>>>>         2
>>>>                                  Number of Intramolecular constraints:
>>>>         4
>>>>                                  Number of Intermolecular constraints:
>>>>         0
>>>>                                   Invariants(translation + rotations):
>>>>         5
>>>>                                                    Degrees of freedom:
>>>>        -3
>>>>  Restraints Information
>>>>                                   Number of Intramolecular restraints:
>>>>         0
>>>>                                   Number of Intermolecular restraints:
>>>>         0
>>>>  ************************** Velocities initialization
>>>> **************************
>>>>  Initial Temperature
>>>>     NaN K
>>>>  COM velocity:                       NaN                 NaN
>>>>       NaN
>>>>  COM position:            5.669178398657      5.669178398657
>>>>  5.669178398657
>>>>  Angular velocity:                   NaN                 NaN
>>>>       NaN
>>>>  *******************************************************************************
>>>> It seems, the degrees of freedom is: 3N - constrains - invariants...
>>>> Is there a way to force the MD only along one axis to avoid rotation?
>>>> Regards!
>>> --
>>> 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 cp... at googlegroups.com.
>>> To view this discussion on the web visit
>>> https://groups.google.com/d/msgid/cp2k/aed754ac-07de-40c2-9114-4370f7af7b16n%40googlegroups.com
>>> <https://groups.google.com/d/msgid/cp2k/aed754ac-07de-40c2-9114-4370f7af7b16n%40googlegroups.com?utm_medium=email&utm_source=footer>
>>> .
>> --
> 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 cp... at googlegroups.com.
> To view this discussion on the web visit
> https://groups.google.com/d/msgid/cp2k/df7eeb5f-ef2d-424c-a9c1-d8545a3012a7n%40googlegroups.com
> <https://groups.google.com/d/msgid/cp2k/df7eeb5f-ef2d-424c-a9c1-d8545a3012a7n%40googlegroups.com?utm_medium=email&utm_source=footer>
> .
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20200815/df3af555/attachment.htm>

More information about the CP2K-user mailing list