Re: [ESPResSo] Restarting a Simulation

[Axel Arnold]

On Thursday 08 April 2010 22:05:43 Salvador Herrera Velarde wrote:
> Hi All,
>
> First of all, thanks to ALL users and developers for your time and help
> with all my questions.
>
> I using mbtools to run  a simulation which contains a large chain of
> monomers among other molecules.
>
> I want to restart the simulation with different conditions; to save the
> actual configuration I was using the command: polyBlockWrite
> "$outputdir/$ident.[format %04d $j].gz" {time box_l npt_p_diff } {id pos
> type v f molecule}
>
> Then, I used the generated file to restart the simulation. However,  when I
> visualized the snapshots in VMD, there are many bonds spanning along the 
> box simulation. I think this is an effect of the periodic boundary
> conditions. To fix the problem I tried to fold the positions:
> polyBlockWrite "$outputdir/$ident.[format %04d $j].gz" {time box_l
> npt_p_diff } {id folded_position type v f molecule}
>
> But still the same problem.
>
> Does anybody has an idea of how to fix the problem?

First of all, for almost all purposes it is better to save the unfolded 
position in your blockfiles, since for example mean square displacements are 
otherwise incorrectly sampled after restarting. So, the first version is the 
way you want to save your data.

Regarding the spanning bonds - that has nothing to do how Espresso internally 
stores the coordinates, therefore you get exactly the same output whether you 
use folded or positions in your blockfiles. This is indeed just a problem of 
visualization. Due to the periodic boundary conditions, you always will have 
bonds spanning the whole box when drawn nonperiodic. So, what you would need 
is a way to dynamically hide bonds which are longer than a certain length, but 
that VMD can't do as far as I know. Alternatively, if there is a concept of 
molecules in your setup, you can try to fold each of the molecules 
consistently according to its center of mass; look at Olaf's pbctools how to 
do that (http://www.espresso-pp.de/projects/pbctools). 

Cheers,
Axel



-- 
Dr. Axel Arnold Tel: +49 711 685 67609
ICP, Universität Stuttgart      Email: address@hidden
Pfaffenwaldring 27
70569 Stuttgart, Germany