[CP2K-user] atomes go far away from box in NVT simulation

[Matt W]

No. But if there are no external fields then the centre of mass motion 
should be zero. 

If it is not (and it won't be) there are some numerical errors. You need to 
decide what level of inaccuracy is OK for your simulation.

On Monday, October 14, 2019 at 4:40:13 AM UTC+1, mh H wrote:
>
> without considering  COMVEL_TOL in MD section, calculation is wrong? 
>
> On Tuesday, February 9, 2016 at 6:06:40 PM UTC+8, Matt W wrote:
>>
>> Hi Tao,
>>
>> The keyword COMVEL_TOL in MD section will allow you to remove center of 
>> mass motion.
>>
>> However, if it is drifting very far it probably indicates that your setup 
>> is not quite good enough in conserving energy/momentum etc.
>>
>> If particles are just diffusing out of the central periodic image, this 
>> is not a problem, as you note above. You, or VMD or whatever visualizer you 
>> use, can just wrap them back.
>>
>> Remember, in CP2K a nice thing is you can run a quick classical 
>> calculation to check how it handles COM motion etc with a nicely conserving 
>> setup, rather than waiting for a DFT run to finish.
>>
>> Matt
>>
>> On Tuesday, February 9, 2016 at 9:38:42 AM UTC, tao liu wrote:
>>>
>>> Hi All,
>>>
>>> I have tested the keyword &CENTER_COORDINATES, but this keyword just 
>>> works for the first step, CP2k will center all the atoms to the default 
>>> position cell/2.. But in the following steps, the center of atoms will 
>>> change along with simulation time. Do I understand correctly?  If yes, is 
>>> there anyway to carry out this &CENTER_COORDINATES keyword in every 
>>> timestep? Thanks.
>>>
>>> Tao
>>>
>>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <https://lists.cp2k.org/archives/cp2k-user/attachments/20191014/a373271e/attachment.htm>