Re: [h5md-user] box and observables

[Felix Höfling]

Am 25.09.2013, 11:42 Uhr, schrieb Konrad Hinsen  
<address@hidden>:

> Peter Colberg writes:
>
>  > > Next, the reader: if there is a single box group but differently
>  > > sampled positions for different subsystems, reading the box
>  > > information is a lot of work and slow.
>  >
>  > Well, this is the essence of the step/time scheme.
>  >
>  > I added a section [1] on the use of the step/time datasets, which
>  > refers to the binary search, upon which the step/time scheme relies.
>  > This is the basic building block for any H5MD program.
>  >
>  > [1] http://h5md.nongnu.org/implementation.html
>
> So the idea is that we have only a single box information for the
> whole trajectory, right? My impression is that Felix still disagrees.
>
> In that case, the following sentence must be removed from the
> specification:
>
>   A specific requirement for box groups inside particles is that the
>   step and time datasets exactly match those of the corresponding
>   position groups, which may be accomplished by hard-linking the
>   datasets.
This sentence is the essence of a really long discussion and we all agreed  
on that. I'm convinced that this solution serves modularity and simplicity  
w.r.t. reading and writing H5MD.

> If left in, this indirectly implies that all position data must be
> sampled on the same time grid.
Each subgroup may be sampled at different time grids. Only the sampling of  
position and box within the same group must be the same (velocity or  
species might be sampled differently).

The note about hard-linking is to be understood as an implementation hint  
for the case that the sampling intervals of the box in different subgroups  
coincide. Maybe the spec is not precise enough here.

>  > For the case of all data that depends on the box for interpretation,
>  > the replicated box groups complicate the matter, while providing no
>  > benefit, as a step/time lookup is needed anyway.
>
> I agree that if there is only a single set of box information per
> trajectory, then it would be best to have it once in a single place.
>
> Konrad.

In the current version, the observables tree contains a single (mandatory)  
box group at a defined location. A source of confusion may be the phrase  
"may be stored in appropriate subgroups similarly to the particles tree".  
It was not meant to imply anything about the box group.

It is my hope that we come to a conclusion on the box soon. Let me recall  
that we have already the status of a release candidate and are about to  
finalise version 1.

Felix