LAT pointing model and ffp updates #1216
Conversation
| if "r2" in aman.pointing: | ||
| r2 = aman.pointing.R2[msk][srt][mapping] |
| q_mir = quat.rotation_xieta(params["mir_xi_offset"], params["mir_eta_offset"]) | ||
| q_el_roll = quat.euler(2, el.copy() + params["el_offset"] - np.deg2rad(60)) | ||
| q_tel = quat.rotation_xieta(params["el_xi_offset"], params["el_eta_offset"]) | ||
| q_cr_roll = quat.euler(2, -1 * cr - params["cr_offset"]) | ||
| q_rx = quat.rotation_xieta(params["rx_xi_offset"], params["rx_eta_offset"]) |
There was a problem hiding this comment.
So with regards to q_mir, q_tel, q_rx. The coordinates that go into these are all offsets, in a sense. But not in the same way as encoder offsets are... so I might drop that word in favor of some other descriptive token.
But more importantly we should decide whether we want to apply them as rotation_xieta(xi, eta) or rotation_xieta(-xi, -eta).
Consider the combo q_cr_roll * q_rx. If q_rx = rotation_xieta(+xi, +eta), then you're saying "where (xi, eta) represents a shift applied to the focal plane, prior to rotating by the corotator angle." If you use (-xi, -eta), then you can say "(xi, eta) represents the position in the focal plane which that remains fixed under corotation."
It's equivalent, mathematically. But the signs of the parameters change. So it's really a choice of how you want to be able to describe this, in words, in code comments and supporting writeups, going forward. I think it's nice to be able to say that second thing -- "xi, eta is a position in the focal plane where some thing happens". This is the rationale for the docstring re-wording above. But there is probably a reasonable opposite view.
(In the current SAT parametrization I think these might be mixed ... fp_rot seems to the "position of" convention but fp_offset uses the "shift applied" convention. Something to consider for v2...)
| az, el, roll: naive horizon coordinates, in radians, of the | ||
| boresight. | ||
| The implemented model parameters are: |
There was a problem hiding this comment.
On naming of the parameters ... here are some things that I like about the SAT model naming:
- (xi, eta) coordinate pairs are of the form
something_{xi,eta}0. The 0 indicates it's a single number, rather than a vector. The xi, eta last means you can easily just saysomething, to mean the pair / the whole rotation. - Encoder offsets are called "enc_offset_{axis}".
I'm on board with "cr" for the corotator.
I left some comments in the model about how the various (xi, eta) pairs should be described, and whether or not that means you apply (xi, eta) or (-xi, -eta) in the model... it bears a bit of thought, in case we ever want to describe this carefully to someone. The specific problem I have with the current param descriptions is that "the offset between A and B" is ambiguous to me ... it doesn't have a specific direction attached to it (whereas "the position of A relative to B" does).
No description provided.