Re: [h5md-user] box data as part of trajectory/position

[Felix Höfling]

Am 18.09.2012, 11:14 Uhr, schrieb Pierre de Buyl  
<address@hidden>:

> Hi Felix,
>
> On Mon, Sep 17, 2012 at 04:26:45PM +0200, Felix Höfling wrote:
> > Hi Pierre,
> >
> > Am 17.09.2012, 15:18 Uhr, schrieb Pierre de Buyl
> > <address@hidden>:
> >
> > >A conclusion would be that "box" is mandatory within a trajectory
> > >group with
> > >identical "step" and "time" datasets (that may or not be linked,
> > >this is an
> > >implementation detail). Then, as you write, wether or not it is
> > >also found in
> > >"observables" depends on the simulation.
> > >
> >
> > Mostly yes. I just thought that datasets position/box/{time,step} are  
> > not
> > needed as they coincide with position/time and position/step by
> > definition. The reader would refer to the latter (and ignore box/time if
> > present). Thereby consistency
> > would be automatically ensured without demanding that position/box/time  
> > ≡
> > position/time. But I have no strong preference here, my main point was  
> > to
> > provide the box information along with the position data in trajectory.
> >
> > So you may link the current box group from observables, but the reader  
> > is
> > adviced to access the time and step information from the datasets in
> > position rather than box.
> >
> > In my diagram above I noted a slight mistake. I meant it to look like
> > this, box is at the same level as value:
> >
> > trajectory
> >      \-- group1
> >      |  \-- position
> >      |    \-- value
> >      |    \-- step
> >      |    \-- time
> >      |    \-- box
> >      |         +-- type
> >      |         \-- edges [D][D]
> >      |         \-- offset [D]
>
> Ok, this is quite different. I see a small problem with that: "box" does  
> not
> appear with a normal H5MD group structure. Also, one may one to also use  
> the box
> information (that give the box volume) along with the velocity and  
> species data.
> This is not very natural with this organization. I propose to make "box"  
> at the
> same level as "position", "velocity" and "species" but make it mandatory  
> to have
> equal time and step than "position".
>
> Thoughts on that?
>
> Pierre

Hi Pierre,

Why would one need the box information to interpret velocities or even the  
species? If the total box volume is needed, one should resort to  
observables/box. For position data, on the other hand, the box has the  
very special meaning to describe the periodic continuation of the  
positions. That's why I wanted to put the box information _inside_  
position.

Second, the structure of the box data set is already different from the  
other data sets because it holds edges and the offset, not just a "value"  
field. (The type attribute may be ignored for our discussion.) Hence, I  
suggested to drop the redundant time and step fields for the position/box  
group (of course not for observables/box). The box information is really  
different.

Peter and Olaf: are you still following the list? Do you have an opinion  
on that?

Regards,

Felix