|
From: | Felix Höfling |
Subject: | Re: [h5md-user] Case of 1D systems |
Date: | Tue, 16 Aug 2011 09:15:34 +0200 |
User-agent: | Opera Mail/11.50 (Linux) |
Hi Pierre,I would avoid special handling of limiting cases as much as possible. It appears to introduce more headache in the end as one needs special routines on both sites: when writing the file and when analysing the results. One important aspect of the H5MD format is the possibility of sharing analysis tools. I would prefer a tool that can handle generic dimensions (and which does so in a generic way). Of course, in a specialised tool the additional index appears weird, for the benefit of the generic tool, however, I would like to keep it.
Best regards, FelixAm 15.08.2011, 21:46 Uhr, schrieb Pierre de Buyl <address@hidden>:
In 1D systems, some parts of the specification may need modification. For instance, requiring a [variable][N][D] shape when D=1 is absurd. Should we allow to throw away the last index for 1D systems ?This is already the case for the species trajectory, which does not possess dimensions anyway.Pierre
[Prev in Thread] | Current Thread | [Next in Thread] |