Meaning of headers in output .csv from homogeneous refinment?

Hi all,

Is there documentation about what the different columns in the refinement output .csv file mean? The ones starting with ctf_params and data_input make sense to me. I’m less sure about the data related to alignments. Can anyone help me fill in the gaps or suggest a resource? I read the paper and the supplement which answered some of my questions.

  • alignments.model.E - image alignment error E(r,t) ?
  • alignments.model.E_min_alpha - ?
  • alignments.model.alpha - ?
  • alignments.model.cross_cor - ?
  • alignments.model.dr - ?
  • alignments.model.dt - ?
  • alignments.model.image_pow - ?
  • alignments.model.phiC - related to pose variable?
  • alignments.model.r.0 - angle phi?
  • alignments.model.r.1 - angle theta?
  • alignments.model.r.2 - angle psi?
  • alignments.model.resid_pow - ?
  • alignments.model.slice_pow - ?
  • alignments.model.t.0 - translational parameter x?
  • alignments.model.t.1 - translational parameter y?


If you are curious, I suggest you read the code for The most important model fields (particle poses and class membership probability) are documented (or are self-evident) in the code.

Your guesses are mostly right, except cryoSPARC uses the axis-angle representation and r.0-2 are the elements of the axis-angle vector.

Euler angles are so 18th century, the 1840s are really where it’s at when it comes to rotations! :wink:

Oh right! Great point, @DanielAsarnow . I will check it out. :slight_smile: Thanks as always for your work.

PS: I’m disappointed that Euler angles are so 18th century, because I finally started pronouncing “Euler” correctly. UGH what a waste of time.

Haha, fortunately, William Rowan Hamilton is pretty straightforward to say. Unfortunately, Olinde Rodrigues less so.