From CMB-S4 wiki
Revision as of 15:06, 7 March 2018 by Cbischoff (talk | contribs)
(diff) ← Older revision | Latest revision (diff) | Newer revision → (diff)
Jump to navigationJump to search

Back to Argonne 2018 main page

Parallel session P3-2: Forecasting - Instrumentation / systematics for the Decadal input document (chair: C. Bischoff)

  1. Summary of forecasting to date (25 min)
  2. Forecasts for Atacama and South Pole (15 min)
  3. Systematics studies and coordination with technology development working groups [Colin Bischoff] (10 min)
  4. Forecasting and simulation plan for Decadal input document (30 min)
    • Spectral vs map vs time domain forecasting and simulation
    • Updates to performance-based forecasting: large vs small aperture, Atacama vs South Pole
    • Priorities for systematics forecasting
    • Coordination between forecasting for r, light relics, and extended science

Notes from session (Tuesday, March 6, 11:15-12:45)

Notes from Denis Barkats and Ben Racine. Compiled/edited by Colin Bischoff.

Victor: Performance-based forecasting

  • Forecasting loop alternates between Fisher-based optimization and validation using map-based sims.
    • If we have a reference design, does this mean the end of optimization steps? No, we have to lock the reference design but should still think about further optimizations.
  • Current forecasts are based on BICEP/Keck achieved performance, not necessarily applicable for instruments in Atacama.
  • Delensing is currently being assumed -- need to actually demonstrate this.

Jeff: SO forecasting

  • SO forecasting tool started with white noise, evolving to include atmosphere, foregrounds, and systematics. Used for optimization and forecasts / requirements.
  • How well does forecasting tool agree with actual performance of ACT, etc? Atmospheric models reproduce noise spectra for ACT detectors. Reproduces map-level noise. Haven't tried to compare at bandpower level.
    • Discussed plan to try to use calculator to closely reproduce ACTpol results.
  • Is there a path to make an S4 version of this calculator? Seems plausible, but we didn't set a deadline or identify effort for this.

Akito: Achieved performance of ABS

  • Rapidly-rotating warm HWP greatly suppressed 1/f noise, didn't significantly degrade sensitivity.
    • Worse NET than forecast is due to detector issues (Psat, optical efficiency, etc)
  • Observing efficiency was negatively impacted due to lack of personnel and cobbled-together instrument.
  • Need to think hard about some of the issues that ABS faced before trying to use these results for forecasting.

Yuji: Achieved low-ell performance

  • Derived N_ell for many published experiments from BB error bar, assuming stated fsky.
  • Small aperture experiments (BICEP/Keck, ABS, QUIET) all have ell_knee of ~50. Large apertures (POLARBEAR, ACT, SPT) all have ell_knee ~500.
  • POLARBEAR with rapidly-rotating waveplate has knee frequency of ~20 mHz, but no published BB spectrum yet.

Colin: Systematics forecasting

  • For CDT report, we treated systematics in an abstracted, hardware-agnostic way. With reference design, we can try to get more specific.
  • Technical working groups have already produced a laundry list of potential systematics, but we need to choose which ones to model.
  • Need to bridge between low-level hardware specification of systematic and high-level description of science impact. Limited TOD sims could be helpful here.

General discussion

  • Next round of forecasting should deal with the two different sites.
    • How well can South Pole and Atacama scan strategies overlap? Is this coming from the instrument working groups, or should we figure it out?
    • How do we forecast instrument performance in Atacama? Scaling from ABS presents problems, but what else do we have?
  • We should coordinate with the people who are forecasting "extended science." This involves very different simulations, so the overlap will be limited. At least we should agree on how to forecast map depth, etc.

Notes from Plenary: Conclusions from parallel sessions

Notes from Brad Benson.

  • There was a request to know exact instrument inputs that went into CDT sims (e.g., NEP, spread in detector performance, yield, obs. efficiency) to help guide instrument TD groups. Kovac mentions that most of this is in the Logbook.
    • Action: Suggestion to go over this in the Splinter session tomorrow
    • Action: Also a request for other experiments to give their demonstrated performance, to understand range of performance.
  • Some discussion about tweaking CDT inputs if there are other plausible / separable gains to instrument performance, e.g., a dilution fridge would give increased observing efficiency. Some thought this was ok, but also that we should be careful about doing this.