CMB-S4 Collaboration Meeting Fireslides
Wednesday August 2, 2023
Fireslide instructions
3
Zhumabek, Denissenya, Linder 2306.03154
Connecting Primordial Gravitational Waves and Dark Energy
grad Y3
postdoc
N=50-60
Quintessential inflation to unify early and late time acceleration.
Protect with symmetries and pole structure – 𝛂-attractors!
Universality: ns = 1-2/N, r = 12𝛂/N2
Dark energy attractor: w∞ = -1+2/(9𝛂)
Note r ~ 1/(1+ w∞). Thawing dynamics relates w∞ to w0, wa.
Kallosh & Linde 1306.5220
If don’t see GW, will see DE dynamics; if don’t see DE dynamics, will see GW!
Starobinsky inflation
Eric Linder (online)
Improving Constraints on Models Addressing the Hubble Tension with CMB Delensing
Joshua Ange (2nd Year Undergrad at SMU), Joel Meyers (Online); Based on arXiv:2307.01662
CMB lensing smooths peaks; delensing sharpens peaks
H0 and related parameters affect peak locations
See also: arXiv:1609.08143, arXiv:2111.15036, https://github.com/ctrendafilova/FisherLens, https://github.com/selimhotinli/class_delens
Multi-wavelength AGN Variability
Plots from: https://arxiv.org/abs/2302.14749
John C. Hood II
New Frontiers and Challenges with Future CMB Experiments
Beyond Fisher Forecasting for Cosmology
arXiv:2211.06534 - Joseph Ryan, Brandon Stevenson, CT, Joel Meyers
Cynthia Trendafilova (they/them, she/her)
Maps to Power Spectra
CLASS_delens�https://github.com/selimhotinli/class_delens/
FisherLens�https://github.com/ctrendafilova/FisherLens
Ola Kusiak (remote)
Columbia Y4 PhD
Measuring SZ x galaxy with ACT DR6
Projected-fields kSZ2 x unWISE
tSZ x unWISE
CIB contamination → “deCIBing” (2303.08121) with Kristen Surrao and Colin Hill
Idea: Use the external LSS data that is correlated with both CIB and tSZ to remove those contaminants to enhance CMB+kSZ measurements using ILC methods
Goal: Constrain pressure (tSZ) and density (kSZ) to infer thermodynamic profile of unWISE
The 3 new methods vary in terms of noise penalty and how well they clean out the CIB
akk2175@columbia.edu
Capse.jl: efficient and auto-diff CMB Cℓ emulation
New Julia-based CMB spectra emulator (w/ python wrapper!)
🔋Cheap to train (~1 hr on a laptop with an 8-core CPU)
⚡Blazingly fast (40 µs, about 100-1000x speed up wrt state of the art)
🎯Accurate (res err < 0.08𝛔 at all S4-relevant scales)
📈Auto-differentiable → efficient gradient-based samplers/minimizers!
Check out CosmoPower (Spurio-Mancini+22,Bolliet+23) and ClassNet (Günther+22) too!
Planck analysis
Jaime �Ruiz-Zapatero
(Oxford)
Marco
Bonici
INAF �Milan
Federico
Bianchini
Stanford�SLAC
Optimization techniques for Low-ℓ noise performance
Two main mitigation strategies:
Subtraction of Polarized Atmosphere: A scaled version of each observation’s coadd-removed 220 GHz map is low-pass filtered, then subtracted from each observation’s 150 GHz map. This subtracts off each day’s atmosphere, observation by observation.
Re-weighting based on PSD of Maps: The mean of the weight map of each observation (2 hours of data) is scaled by that map’s PSD between ℓ = 50-250. This takes into account correlated noise between detectors.
2.3 x reduction in noise at ℓ = 70 since initial CMB-S4 LAT + SAT r forecasts!
=
Scaling coefficient
x
U
Q
220 GHz 2-hour map - Q
150 GHz 2-hour map - Q
Cleaned 150 GHz 2-hour map - Q
150 GHz 2-hour map - U
What is RadioForegroundsPlus? The project has been approved as part of the Space Program within Horizon 2020 in the EU (HORIZON-CL4-2023-SPACE-01) and it is a continuation of the radioforegrounds.eu program (G.A. 687312) which has been operating between 2016 and 2018.
What are the Main Goals? Exploiting the Data from Radio and Microwave Surveys, C-BASS, QUIJOTE, S-PASS for improving our knowledge of Low Frequency Foregrounds for CMB B-Modes
For How Long will it be Operating? The approval was communicated on July 24th, the program becomes operational on January 1st, 2024, for 3 Years.
Which Opportunities? About 1.5 Millions to Support Analysis and 2 and 3 years Post-Doc Positions in the participating institutions.
CMB-Stage IV Collaboration Meeting, Stanford, August 2nd, 2023, Thank You CMB-S4 for the Support!
Nodes: IFCA (Barreiro, Lead), IAC (Rubino-Martin), CNRS (Banday), Manchester (Chluba), Oxford (Taylor), SISSA (Baccigalupi)
Work Packages: 1-Project coordination, management and dissemination, 2 - Low-frequency data (QUIJOTE, C-BASS, S-PASS) and Planck maps, 3 - Advanced tools for component separation and foregrounds maps production 4 - Modelling the diffuse Galactic emission, 5 - Forecasted impact of radio foregrounds in future CMB experiments and preparation of products
Carlo Baccigalupi, SISSA, On Behalf of the RF+ Consortium
Shamik Ghosh, M. Alvarez, J. Delabrouille, M. Remazeilles, E. Russier, J. Borrill, Z. Lukić
Work funded by the LBNL LDRD program.
Results using from 7.7 Gpc box at 61443 resolution using LPT calculated on the lightcone. Halo component now in development.
Avg. Nl 50 < l < 100:
WMAP K: 4.01x10-3
LFI 30GHz: 3.92x10-3
Combined: 1.79x10-3
Improved input maps for making templates of galactic emission.
Lensing potential power validation
𝛋
𝛟
Sky simulations for Stage-4/5 experiments
An update on ongoing developments
Synchrotron Polarized Intensity
CO10
CO10 maps at (150, -45)
Type 1
Type 2
GNILC
Noise level
Exploring Reionization Astrophysics Using kSZ Beyond ℓ = 3000
Divesh Jain (remote)
NCRA-TIFR
1. Self-consistent framework to evaluate shape and amplitude of kSZ (Jain et al. 2023)
Collaboration : Tirthankar Roy Choudhury (NCRA-TIFR Pune), Suvodip Mukherjee (TIFR Mumbai), Srinivasan Raghunathan (University of Illinois)
2. kSZ power variation with reionization models allowed by R21 measurement. Also shown are Cross-ILC error bars (Raghunathan & Omori 2023)
3. Model forecasts tight error bars on both homogeneous and patchy properties.
Conclusion:
Highlights the need to capture kSZ power on a broad range of multipoles to gain insights into the inhomogeneous reionization era.
Forecasts: Error bars on 𝑧50 is ~ 0.50 and Patchy B-mode at ℓ=200 is at ~1.3 nK2
Has important implications for detecting patchy reionization signal.
Jain et. al. in prep
R21:
Reichardt 2021
R21:
Reichardt 2021
Lindsay Ng Lowry llowry1@berkeley.edu
Postdoc at UC Berkeley since Sept 2021, working with Adrian Lee
Deployment and Commissioning of POLARBEAR-2b (PB-2b)
Other Work
Example PB-2b Jupiter Observation
Map courtesy of Megan Russell
Photo courtesy of Yuyang Zhou
Jason (Jaemyoung) Lee (4th year PhD student, SN Ia cosmology + LSS)
Email: astjason@sas.upenn.edu
Non-Gaussianity of the CIB & Its Gravitational Lensing
Poster Outside!
Lensing increases power spectra by 1 ~ 2% but (equilateral) bispectra by 10 ~ 20%!
Methods
Results
Inference of dust emissivity with Planck & HI data
Shabbir Shaikh (sshaik14@asu.edu)
Planck data at 353 GHz
Data - Best-fit model
[IPlanck - CMBPlanck ]HFI = 𝝴* NHI + Offset + CIB + [HI residual] + NoiseInst
Signal (S)
Gaussian likelihood, Hamiltonian Monte Carlo sampling
Results with simulations
Mock Data = S + N
Best fit signal
residual
sims
Noise (N)
at 545 GHz
Difference between
inferred parameters & input parameters.
model
GASS survey
Emissivity [Nside 32]
data
Colin Hill (Columbia, remote)
work with Kristen Surrao (Columbia)
Cosmological Parameter Inference via Needlet�Internal Linear Combination (NILC)-Based Likelihood
- Standard CMB parameter inference methods (Planck, ACT, SPT, BICEP/Keck) only use information in the auto- and cross-frequency power spectra. For non-Gaussian fields, this is sub-optimal. Can we do better, e.g., with a more optimal weighting scheme to mitigate foregrounds and upweight CMB-dominated modes?
- Consider the auto- and cross-component power spectra of NILC maps for all fields {p,q} in the sky model. These are automatically weighted optimally to mitigate foregrounds. We have derived an analytic expression for the NILC maps’ power spectra, via a MASTER-like technique:
- Applying this method to a toy model sky containing CMB, (highly amplified) tSZ, and CMB-S4-like noise, we find ~10x improvement in parameter constraints:
- Potentially powerful application: primordial B-mode constraints
Astrophysical foreground cleanup using non-local means
Guillermo F. Quispe Peña
3rd year PhD student
Simon Fraser University
gfq@sfu.ca
arXiv:2306.00211 - Guillermo F. Quispe Peña and Andrei V. Frolov
Generalized non-local means:
Feature space:
Development of DAQ Software & Critical Site Capabilities for the Simons Observatory
Develop critical site software: cooling loop flow & temp, thermometry, PWV, site power distribution, star cameras
Developed housekeeping SQL db & HK analysis tools for providing simultaneous + immediate detector + HK analysis on-site & in analysis pipeline
Pipeline Tools
DAQ (OCS) Dev
Thermometry
Constructed/calibrated ~280 thermometers to map SO’s detector fluctuations
Next: Go Back to Chile + SO Early Data Analysis
July 29, 2023
LAT at the Site
July 21, 2023
SAT1 Fully Populated
2018
Thermometry
Pipeline Tool
March 2023
Chile Site Network Setup
Reconstructing the first Planck submm polarization maps
Secondary spillover
Baffle spillover
Primary spillover
Best-fit estimate of the far sidelobes beam
857GHz polarization maps before/after FSL processing
Baseline
FSL corrected
Collaboration with: Reijo Keskitalo, Brandon Hensley, Aurelien Fraisse, Julian Borrill
Main beam
Zoomed-in Chamaeleon-Musca region 12x12 deg
galactic coordinates: (lon, lat) = (300, -13)
353GHz
857GHz
Paul Williams (Lawrence Berkeley National Lab)