systematicswg:telecons:2018-05-03 [Probe Mission Study Wiki]

This page is read only. You can view the source, but not change it. Ask your administrator if you think this is wrong.
May 3 Face to face meeting

present: Jacques, Maurizio, Reijo, Julian, Shaul, Brendan


  * Jacques thinks we should consider other precession time scales. **Schedule a discussion with Kris about precession time scales.**

   * Review feedback from the workshop
       * No one questioned choices we made, but not a lot of time for audience to think about our list
       * Bandpass mismatch
           * IQUS solution (spurious signal projected out at the same time as I/Q/U solution) with a PICO scan strategy?  for Planck, this increased noise by factor 5, but with wide variety of scan angles, this might work well.
           * In Banerji & Delabrouille in prep,  shows how much of a noise boost you get by including a solution for this additional component in the case of the CORE scan strategy. Probably on order of 2-3 with a PICO-like parameters.  If you use a template, there is no noise penalty.
           * On-orbit measurement will always be the best way to correct.
           * A suggestion from Reijo: take map domain templates, compute posterior covariance of template fit parameters.
   * Any additional updates on ongoing work
       * Full mission sims run into issues: Madam is a memory hog and 2x 1 PICO freq x 1 year doesn't fit in RAM: either replace Madam, or some possible workarounds: thin the focal plane
       * Do we really need all 13k detectors for a sim?
       * Gain solution (da Capo): 
	   * A limiting factor is Galaxy bias… but let's say in the future we solve this.
           * Maurizio would like to do a CMB-only sim with 1/f noise solving for calibration and determining time scale over which this is feasible.  Assesses impact of noise on the calibration vs. time.  Provide some bound on detector stability requirement.
	   * In a week, get the gains.  Then.. Use these recovered gains as inputs to produce sky maps.  Hopefully done in a few weeks.
	   * In CORE paper, got 0.01% with a 96 hour time scale, 12 hours may not be good enough.
	   * Get a reference for stability of TES bolos as seen in SPIDER (from Hannes and Jeff)
	   * OK, great.  But this level of reconstructed gain at the ~0.1 to 1% level may not be good enough.
	   * Determine time scale / accuracy of calibration that is needed on the detector side.
   * Absolute angle rotation
	   * TBC: Relative angle of each detector to mean angle could be iteratively solved (?)
	   * Solve for error of individual detectors vs. the mean, and look at the distribution -> gives lower bound on final error in absolute angle
           * How well can we jointly solve for cosmic birefringence and an absolute polarimeter angle error using EB (along lines of Eric's presentation)? **Need to ask Eric about this.**
   * Sidelobes
	   * Convolution of timeline with GRASP beams.
	   * Karl's GRASP output was converted to b_lm's (w/help from Ashdown)
	   * Andrea now has 4pi beam files: small issue being fixed by Andrea: ask Andrea status
	   * Power spectrum of sidelobe signal or map of sidelobe pickup
	   * Generate convolution with the the x- and y- polarization detectors in one pixel
	   * Have to cut out the main lobe of the beam prior to convolution Either central pixel (1x1 degree), or a little bit more of the near sidelobes just to be safe.  Karl to take care of that.
	   * What sky do we convolve with? Just use pysm that is interfaced with toast: include Galaxy and dipole, maybe separately (?)
	   * Look at amplitude of map: compare to 1 nK -> maybe a BB spectrum with a sky cut.
	   * Ask Andrea for not just an entire year, but also maybe one precession in order to identify where the pickup is coming from.
		* Possibly identify design changes to baffles
		
   * An important question: will ground calibration be carried out at individual detector level at a high level of accuracy? (bandpass, time constants, linearity, etc.)
	    * What does S4 / Simons do really?
	    * What did Roger O'Brient assume about integration & test for PICO
	    * Do we need new processes? AI?
User Tools

Page Tools
