From CMB-S4 wiki
Back to SLAC-2017 main page
Foregrounds and simulations
Post talks here:
- r -- Lloyd Knox slides
- TT/TE/EE -- Erminia Calabrese slides
- Lensing -- Blake Sherwin slides
- Clusters -- Colin Hill slides
Notes from session
r: Lloyd Knox
- From Science Book, expect two distinct surveys for r: one at low resolution with many frequencies (for tensors), one with high resolution but fewer frequencies (for delensing)
- Defined a staged series of data challenges. Simulated maps exist at NERSC, analysis of first data challenge is underway. Start simple, update survey and signals in later rounds.
- Science Book used Fisher forecast to allocate sensitivity between eight frequency bands for low-res survey plus high-res survey.
- Important question: Do we need additional frequencies below 30 GHz?
- Survey definitions:
- X=1: Science Book case with fsky=3% (check codes to confirm that we can reproduce Science Book results)
- X=2: Updated set of bandpasses, multiple values of fsky.
- X=3: Include both low-res and high-res surveys
- Signal definitions:
- Y=0: Lensed and partially-delensed CMB, Gaussian dust and synchrotron, no extragalactic foregrounds
- Y=1: PySM 'standard' model = Galactic foregrounds (non-Gaussian, spatially varying indices for sync/dust), no extragalactic foregrounds
- Y=2: PySM 'alternate' model
- Y>2: Ben Thorne and Brandon Hensley working to add Hensley / Drain dust model, add extragalactic foregrounds consistent with CMB lensing (Alvarez/Battaglia/Bond)
- Currently working on data challenge 1.0 -- recover the answer from Science Book. Check for biases and statistical uncertainty.
- Need to connect with instrument working groups, start to include systematic effects in maps.
TT/TE/EE: Erminia Calabrese
- Study impact of CMB secondaries, Galactic and extragalactic foregrounds on science from small-scale T/E. Build a covariance matrix including foreground uncertainty.
- Use templates and nuisance parameters for tSZ, kSZ, tSZxCIB, CIB-P, CIB-C, Radio-P, Cirrus in TT, Radio-P in TE/EE.
- Make simulations and multi-frequency covariance matrix. Repeat for different experiment configurations. Then analyze to solve for nuisance parameters, CMB, and eventually cosmological parameters.
- Some constraints on secondary / Galactic / extragalactic nuisance parameters are orders of magnitude better than current limits. Will these models hold up at that precision?
- Five frequencies vs three improves secondary/foreground constraints, but doesn't impact damping tail science (Neff). TE is most constraining spectrum.
- No systematics currently included.
Lensing: Blake Sherwin
- CMB-S4 is a new regime for lensing science: 0.2% precision, dominated by EB estimator. Stringent requirements for biases.
- Biases from polarized Galactic foregrounds -- don't know what small-scale dust (or sync) polarization looks like.
- Try scaling Planck dust intensity but assume constant angle.
- Use orientation of HI filaments to trace dust polarization angles (in progress).
- Simulations with magnetic turbulence -- yields large spread of bias in lensing estimator
- Looks like foreground cleaning will be necessary for lensing science -- need multiple frequency channels at high resolution.
- Try cleaning E-modes using multiple frequencies on a 5-meter telescope. Seems to work well with little degradation of lensing measurement.
- Still need to try analogous process for delensing.
- Extragalactic foregrounds only important for temperature, but can still be important.
- Alvarez et al have developed simulations that contain tSZ, kSZ, CIB, and lensing, with appropriate correlations.
- Temperature biases from extragalactic foregrounds can cause bias for cross-correlation estimators as well.
Clusters: Colin Hill
- Use simulations to understand foreground biases for tSZ, kSZ. End to end test with set of simulations (that are also useful for high-ell science, lensing, delensing).
- Galactic foregrounds at low ell. At high ell, mixture of CMB, tSZ, CIB, radio sources -- need multifrequency data.
- For simulations, important that extragalactic foregrounds have the right correlations. Non-gaussian structure is important.
- Need more information about non-white atmospheric noise dependence on frequency.
- Extragalactic foreground sims from Alvarez et al (CITA)
- Products are available at NERSC.
- Working on component separation using ILC methods.
Action items/Next steps
Summarize action items here