Closed dagewa closed 6 years ago
I can access it from here:
I think it's only used in ADT3D but I'd have to read it first to be sure
cheers
TODO mention the flexibility and refinement of the geometry in software like XDS and DIALS in the response letter.
How about "DIALS refines the unit cell parameters, the cell orientation, and the geometric parameters, i.e. detector distance, beam position, the axis of the rotation axis. The refined parameters are left at the users' control. For details, see (add the appropriate DIALS reference). This feature is of great importance in electron diffraction studies, as often the initial values are only poorly known. Their refinement has reasonable radius of convergence." The ED community is going to read this paper with quite some interest, as I believe that these features, in addition to data scaling, are rather unknown. It is worth being a little more detailed than only adding a reference.
@tgruene I like it, thanks. I will make minor modifications (the default behaviour in DIALS is to refine the detector orientation, leaving the rotation axis canonical in the laboratory frame) and work it into the right place in the document
Reply: We have added a reference to Dorset (1976) and Kolb et al. (2009). These procedures are not so well known in the X-ray community. We also added text here and in Section 2.6 to highlight the flexibility of the geometrical description and strength of the refinement procedures in software like XDS and DIALS, which are features not so well known by the ED community. These procedures are able to correct for rather gross errors in the experimental model, providing the input reflections are correctly indexed. This can be seen, for example, by editing the input geometry for dials.refine so as to perturb the geometry in a severe way, such as changing the rotation axis, shifting the beam position or detector by much more than the spot separation, or changing the unit cell. Experience shows that refinement from the perturbed state almost always converges back at the correct solution.
Does anybody have access to the Kolb paper (@tgruene @cmaxtb)? I am happy to cite, if it does describe an algorithm widely used in dedicated ED software, but I would need to read it first.