[Top][All Lists]

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

Re: quaternions

From: Felix Höfling
Subject: Re: quaternions
Date: Wed, 18 Dec 2019 22:36:02 +0100
User-agent: Opera Mail/12.16 (Linux)

Hi Konrad,

Thanks for bringing it to the point. So "rotation" is out and "orientation" sounds attractive, but we may have different representations of it - also depending on the type of particle.

I see two options:

1) the data field is always called "orientation" and we use an HDF5 attribute to distinguish between "quaternion" or "principal_axis" or so. This leaves room for other choices too (storing rotation matrices or Euler angles?). The disadvantage is that one cannot have different representations next to each other (although one can always use private names for special tasks).

2) We leave "orientation" for the rather general quaternion and use e.g., "first_principal_axis" for the unit vector of the main symmetry axis. Of course, this can be extended too as the space of English words is pretty large.

What do you think? My tendency is for 1) which appears to be cleaner and perhaps slightly more H5MD-ish.

NB: perhaps we should think of a set of body-fixed vectors such as a body frame instead of only a specific axis. The precise meaning of the vectors depends on the application, as is the case with H5MD at various places.

Am 18.12.2019, 20:17 Uhr, schrieb Konrad Hinsen <address@hidden>:

Practical question: is it required (in H5MD) that the quaternions/axis
vectors are normalised? Should it be stated?

For quaternions, normalization is a requirement: only normalized
quaternions form a representation of the rotation group. So I'd say yes,
it should be a requirement, and therefore it should be stated.

I agree.

For axis vectors, I have no opinion - I have never used them in

For proper rotational motion, the length of such vectors should be preserved, at least. So choosing them as unit vectors makes sense.

Making the normalisation mandatory in both cases is good practice since then the file contains sound (=consistent) data. During the simulation, some people may admit small violations of the normalisation for performance reasons. Yet from my view, it is not asking too much to bring this in order before writing to the file.

Kind regards,


reply via email to

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