[gmx-developers] remove periodicity

[Gaurav Goel]

Dear Anton,

Thanks for your help with this. I printed out the coordinates from
'do_update_md' subroutine and all the values for particle coordinates
are in the range [0,L], where L is my box size. Does that mean those
coordinates are already in the 'atom in the box' state?

Thanks,
Gaurav

On Mon, Jul 5, 2010 at 8:25 AM, Anton Feenstra <feenstra at few.vu.nl> wrote:
> On 02/07/10 23:24, Gaurav Goel wrote:
>>
>> On Fri, Jul 2, 2010 at 3:20 AM, Anton Feenstra<feenstra at few.vu.nl>  wrote:
>>>
>>> On 01/07/10 22:22, Gaurav Goel wrote:
>>>>
>>>> If I know the coordinates at time 't-del_t' and 't', I can remove the
>>>> periodicity as follows:
>>>>   /* Remove periodicity */
>>>> ...
>>>> ...
>>>>
>>>>       while(xcur[ind][m]-xprev[ind][m]<= -hbox[m])
>>>>        rvec_inc(xcur[ind],box[m]);
>>>>       while(xcur[ind][m]-xprev[ind][m]>     hbox[m])
>>>>        rvec_dec(xcur[ind],box[m]);
>>>> ...
>>>> ...
>>>
>>> What do you mean with 'remove periodicity'? Make molecules whole? Put all
>>> atoms in the box? Put all molecules' center of mass in the box? Give all
>>> atoms (or molecules) a continuous trajectory (i.e., they will fly 'out of
>>> the box')?
>>>
>>> There are a number of PBC related options in trjconv that might include
>>> what
>>> you want. If you want to use that in a program of you own, I suggest
>>> copying
>>> over the routine(s) from trjconv.
>>>
>>
>> Thanks for your reply Anton. I need to modify the md
>> algorithm...basically scale the velocities of atoms up or down in two
>> slabs at a fixed distance from the face of the box. By "remove
>> periodicity" I meant to put all atoms in the box.
>
> Just a not of caution, if you are going to modify the md code, do *not* use
> code from trjconv. The latter is meant for analysis, and uses some
> constructions that should not be used during simulation (md or otherwise).
>
>> I will take a further look at the md algorithm and try to locate where
>> in the code coordinates are generated corresponding to 'atoms in the
>> box' state. I am not able to understand why atoms need to be put in
>> the box to calculate non-bonded interactions? Isn't it right that for
>> non-bonded interactions we need the closest periodic image.
>
> This is not my main expertise, but AFAIK the non-bonded interaction code and
> the related neigbour-search code optimizes best with all atoms in the box.
>
>
> --
> Groetjes,
>
> Anton
>  _____________ _______________________________________________________
> |             |                                                       |
> |  _   _  ___,| K. Anton Feenstra                                     |
> | / \ / \'| | | IBIVU/Bioinformatics - Free University  Amsterdam     |
> |(   |   )| | | De Boelelaan 1083A - 1081 HV Amsterdam - Netherlands  |
> | \_/ \_/ | | | Tel +31 20 59 87783 - Fax +31 20 59 87653 - Room P136 |
> |             | Feenstra at few.vu.nl - www.few.vu.nl/~feenstra/         |
> |             | "Just a Minute While I Reinvent Myself" (Red Hot      |
> |             | Chili Peppers)                                        |
> |_____________|_______________________________________________________|
>