TMS Meeting Minutes, August 20, 2020

Attendees: Andrew, Breann, Heejoo, Matthieu, Trenton, Yang (Zoom meeting)

Meeting discussion summary:

  • Stephanie passed her Master's thesis defense successfully! Congratulations to her!!!
    • Stephanie might be an interested acquisition at GMT, as she already lives in LA
  • Andrew/Breann/Heejoo/Trenton discussed some optics related topics (and possible topics for Trenton's dissertation work)
    • Suggest conducting SVD sensitivity analysis of perturbations to truss lengths, determining eigenmodes of kinematic system
    • On optical model, consider looking at cost of pose changes is and work out an equal cost eigen-pose data set to then decompose
    • Pseudoinverse may be resolved into S, U, and V matrices, which may give useful information for truss design
    • Each column is a sensitivity to pose in that channel
    • Tying optical model and sensitivity to aberration contributions from different pose changes could result in a nice link between the physical telescope and the optical model
    • Able to connect image quality metrics directly to pose metrics
  • Andrew and Breann also shared some discussions and papers on image analysis (VST and Keck papers etc)
    • Andrew supporting Osaka Uni. on development of microlensing telescope
    • SExtractor used to extract ellipticities and FWHM, then least squares fit to optical model with perturbables as variables
    • VST image analysis system uses intra and extra focal images to decompose optical aberrations into basis functions by coupling with optical model basis set
    • One of the primary problems with FPIA (a "horrible clooge" as Andrew so lovingly refers to it as) is that, with only one side of defocus, the seeing and optical aberrations are blurred together, inseparable
    • Adding intra and extra focal images allows for the subtraction of seeing, isolating the optical aberrations and making the problem tractable
    • Focusing on image analysis because we have a beautiful TMS and crappy active optics; want to combine the capabilities of the TMS with excellent active optics (improved FPIA) to bring LBT up to speed
    • See VST Image Analysis paper for details on the process
    • In essence, make a good fit for the model, extract data (ellipticity, angle, FWHM) from a model, and do a least squares best fit, now you've got a tool
    • Andrew's work focused on the problem of linear astigmatism and its effect on microlensing observations, where ellipticity is the primary science observable
    • On LBT, Andrew suggests a "weaving" strategy of oscillating from intra to extra focal rather than having to take both for each measurement
    • Similar to on-sky phase diversity; When Breann did this on Keck, she was going from -2 mm to +6 mm defocus (WAY out of caustic)
  • The majority of the discussion has been focused on TMS test problems we have experienced:
    • Pose and change of pose calculation with missing channel information is wrong in our current software and we misunderstood it previously. So this has been hotly discussed after the first TMS test:
      • First of all, reference pose is tied to particular channel combination and needs to be explicitly stored/reconstructed for matching channel combination. This is the main source of our problem and failure.
      • We all agree we will store the reference pose implicitly and reconstruct one on-the-fly when needed (like upon receiving a channel combination with missing information).
      • What we have been debating on mostly is how to handle missing channel information when we needed to record a new reference. In the end, we have settled on reconstruct the missing channel information during a reference-setting step with Andrew's method (details will be given in a separate document later).
      • There are some discussions on the general TMS software approach (e.g., what amounts to be robust overall, different levels/views of robustness, levels of efforts, general engineering consideration and point of views, etc.).
    • The other issue in the first TMS test (where we experienced an incoming data where one channel is slightly off, it is however not enough to trigger our current sanity check failure, but enough to upset the subsequent calculation/corrections) may prompt another look at how we handle sanity checks in the TMS system and reconsider what amounts to be physically possible measurements etc.
    • Yang sent some post processed data collected from first TMS test for others to look at.

Action Items:

  • Yang will plan to prototype the new pose/change of pose calculation with missing channel information (to be tested in second TMS test).
  • Yang also plans to write a formal document defining the TMS operations/calculations in an unambiguous procedural manner.
  • Heejoo has a list of optics related tasks to do.
  • Heejoo and Trenton will work on extracting relevant LBT telemetry matching the TMS test data.
  • Andrew will take a look at the post processed TMS test data.
Topic revision: r2 - 16 Sep 2020, TrentonBrendel
This site is powered by FoswikiCopyright © by the contributing authors. All material on this collaboration platform is the property of the contributing authors.
Ideas, requests, problems regarding Foswiki? Send feedback