Hi,
When performing 3D classification for symmetry expanded datasets in the presence of high symmetry (high order helical, icosahedral etc), during 3D classification CS treats each symmetry expanded copy independently for the purposes of optimal scale calculation, which takes quite a long time.
Would it make sense instead to assume that all the sym expanded copies of a given particle have the same scale? This would be a lot faster to calculate, and I think it’s a reasonable assumption?
Cheers
Oli
Hi @Olibclarke,
Thank you for the request. This makes sense, although perhaps if you have used symmetry expanded particles, would it be desirable to first perform scale estimation prior to the particles have been expanded (e.g. in an upstream refinement) and then hold scales constant afterwards (by setting input scale to input)? Or do you notice that re-estimation of scales afterwards is significantly beneficial?
Best,
Michael
Ah that’s a good idea, thanks @mmclean! Yes, just setting scales to input should do the trick (assuming that symmetry expanded copies inherit scale factors from the original particle)