[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [h5md-user] [EXTERNAL] Re: topology

From: Hart, David Blaine
Subject: Re: [h5md-user] [EXTERNAL] Re: topology
Date: Fri, 9 May 2014 16:14:34 +0000

> Apart from the attributes, this is what I suggested :-)
> To be clear:
> time-dependent H5MD element:
> <bond_list>
>     \-- step [variable]
>     \-- time [variable]
>     \-- value: Integer [variable][number of items stored][X]
> where X is the number of indices that is needed (2 for pairs, etc.)
> time-independent
> <bond_list>: Integer [number of items stored][X]
> The dimension is given by the shape of the dataset. I don't know what "type"
> is supposed to contain.

The "type" attribute was a place to put the contents of the list - like "bonds" 
or "angles" - that could be a limited set of values, like the "none" and 
"periodic" that go in box/boundary. However, module-dependent group structures 
would take care of that issue, too. The other concern was that if we call the 
element a "bond_list", it might be (probably will be) used for things that 
aren't bonds, and we don't want to limit the use or encourage misuse. So 
perhaps something like <atom_list> would be clearer. Although that name isn't 
terribly clear, either.

> I hope that I understood correctly what both of you meant and I have the
> feeling that the needs (for storing the indices, at least) are similar to all 
> of us.
> Additional data would be module dependent.
> Cheers,
> Pierre

Yup, that is what I meant! :-) Have a great weekend!


reply via email to

[Prev in Thread] Current Thread [Next in Thread]