Cosmology with SPT-3G



Cail Daley

CosmoStat, CEA Paris-Saclay

on behalf of the SPT-3G collaboration


ESLAB Euclid Symposium

March 25, 2025

SPT3G Funding Seals

Outline

  • Cosmology with the Cosmic Microwave Background (CMB)

  • The South Pole Telescope (SPT) and the SPT-3G camera


  • Recent and upcoming 2019+2020 SPT-3G results:
    • Three analyses: MUSE, TT/TE/EE, and CMB lensing (\(\phi\phi\))
  • Beyond 2019+2020: the Ext-10k survey


  • SPT-3G and Euclid

Cosmology with the CMB

most distant observable electromagnetic radiation \(\implies\) early-Universe physics

illuminates the Universe’s evolution: reionization, structure growth, astrophysics..

NAOJ

Cosmology with the CMB


Turner (2022): The Road to Precision Cosmology

\(\Lambda\)CDM Concordance

six-parameter model predicts a wide range of cosmic observables:

\(\Lambda\)CDM Phenomenology

David Spergel / AAAS / Science

but we don’t much about \(\Lambda\) or CDM!

\(\Lambda\)CDM Cracks

  • Tensions between early- and late-time measurements of \(H_0\) and \(S_8\):
  • Low neutrino masses?

The CMB Beyond Planck

Planck is noise-variance-limited beyond:

  • \(\ell > 1800\) in TT, \(\ell > 1100\) in TE;
  • everywhere in EE and lensing (\(\phi\phi\)).

CMB experiments filling this niche:

SPT      BICEP/Keck

Aman Chokshi

ACT

Debra Kellner

The South Pole
Telescope

3 bands:      95,    150,   220 GHz     
resolution:  1.6,     1.2,     1.0 arcmin

SPT-SZ (2007)              SPTpol (2012)             SPT-3G (2017)

~1000 detectors          ~1500 detectors         ~16,000 detectors

Photo Credit: Aman Chokshi

SPT-3G 2019+2020 Analyses

*successful cosmology analyses already published with SPT-3G 2018 data

Kevin Zagorski

2019+2020 Analyses


Same 9 maps used by three
independent pipelines:

  1. MUSE: Bayesian joint inference of cosmology, systematics, and pixels

  2. Traditional TT/TE/EE power spectrum

  3. Traditional quadratic estimator (QE) lensing

1. MUSE

Recently accepted! arxiv:2411.06000

Methods in the spirit of
simulation-based inference (SBI).


EE & \(\phi\phi\) bandpowers probe new scales and are in agreement with Planck \(\Lambda\)CDM:

        Fei Ge            Marius Millea

1. MUSE Results: \(H_0\)

1. MUSE Results: \(S_8\)

Joint ACT-SPT-Planck \(S_8\)
measurement in the works!

1. MUSE Insights on Structure Growth






Scales and redshifts that
contribute to MUSE constraints:
total and non-linear contributions.

1. MUSE Insights on Structure Growth

\(A_{\rm mod}\) to scale the non-linear matter power spectrum:

Planck T&E + SPT\(\phi\phi\) gives first \(3\sigma\) detection of
non-linear structure with CMB lensing!

1. MUSE vs. Planck and ACT

Comparison of polarization-only
(EE + \(\phi \phi_{\rm pol}\)) constraints from
SPT, Planck, and ACT:

SPT-3G polarization data is incredible!


but, this isn’t a complete picture of the constraining power of these instruments:


\(\implies\) upcoming SPT TT/TE/EE/\(\phi\phi\)

2. Upcoming 19+20: TT/TE/EE

Etienne Camphuis


Wei Quan

2. Upcoming: TT/TE/EE + MUSE \(\phi\phi\)

Targeting publication this summer.



As a taste, SPT-3G TT/EE/TE + MUSE \(\phi\phi\)
(mock!)
vs. ACT DR6, MUSE and Planck.



With the addition of full QE \(\phi\phi\) (see following slides), joint constraints on \(\Lambda CDM\) parameters will be
comparable to Planck!

3. Upcoming 19+20: QE Lensing

Yuuki Omori

Myself

Federico Bianchini

Yuka Nakato

forecast: 1.5% measurement \(\sigma_8 \Omega_m^{0.25}\) from SPT-3G lensing alone!
vs. 2.7% for Planck lensing

3. Upcoming 19+20: QE Lensing

3. Upcoming 19+20: Cross-correlations

SPT-3G lensing: large scales, high \(z\)
(mostly linear)


Cosmic shear: smaller scales, low \(z\)
(mostly non-linear)


Cross-correlation bridges the gap

Aaron Ouellette

SPT-3G lensing

DES Y3 bin 4

Beyond 2019+2020: the Ext-10k survey

Kevin Zagorski

Beyond 2019+2020: the Ext-10k survey

Survey Area
(deg²)
Depth
(\(\mu\)K-arcmin)
Main 1,500 2.1
Summer 2,600 9
Wide 6,000 12

Ext-10k = Main + Summer + Wide

Combined Ext-10k survey will improve on Planck constraints by as much as:
2x for \(\Lambda\)CDM parameters, 3x for single-parameter extensions!

Credit: Federica Guidi

Cosmology from CMB Secondaries

Foreground signals also contain rich cosmological information.




SPT has provided the best constraints on the Epoch of Reionization to-date from kinetic Sunyaev-Zel’dovich 4-pt:

Raghunathan et al. (2024)

Cosmology from CMB Secondaries

Foreground signals also contain rich cosmological information.



100 deg² SPT-deep cluster catalog on arXiv as of last week: arXiv:2503.17271


Looking forward to cluster cosmology with SPT-3G + Euclid!

Kornoelje et al. (2025)

SPT-3G and Euclid

Kevin Zagorski

SPT-3G and Euclid

Joint analyses of SPT-3G and Euclid data will be incredibly powerful!

  • DR1 largely overlaps SPT-3G Main and Summer fields.

SPT has conducted dedicated observations
of Euclid Deep Field South (EDFS)
and is planning a release of:

  • temperature maps & analysis products
  • emissive source catalogs
  • cluster catalogs.

SPT-3G and Euclid

Joint analyses of SPT-3G and Euclid data will be incredibly powerful!

  • DR1 largely overlaps SPT-3G Main and Summer fields.



Check out Melanie Archipley’s talk tomorrow at noon to learn more about our upcoming data release and other SPT science!

SPT-3G and Euclid

I am working on EDFS lensing maps to be cross-correlated with Euclid Q1 data:

Will provide useful pathfinding for Euclid DR1 \(\times\) CMB analyses.


Euclid-CMB \(6 \times 2\)pt
will be among Euclid’s most powerful cosmological probes.

Summary & Looking Forward

  • The CMB still has a lot to tell us about the Universe:

    • small scales, polarization, & lensing.
  • SPT-3G Main 2019+2020:

    • MUSE polarization-only results probe new scales & are consistent with \(\Lambda\)CDM

    • coming temperature+polarization constraints will be comparable to Planck!

  • SPT-3G Ext-10k will improve on Planck constraints by as much as
    2x for \(\Lambda\)CDM parameters, 3x for single-parameter extensions.
  • Joint SPT-3G + Euclid cosmology constraints will be incredible!

Thank You!

Backup

Kevin Zagorski

MUSE Algorithm



Marginalize over CMB \(f\) and lensing \(\phi\)
maps/pixels to determine parameters \(\theta\):

\[ \mathcal{P}(\theta \mid d)=\int \mathrm{d} f \mathrm{~d} \phi \, \mathcal{P}(f, \phi, \theta \mid d) \]

Algorithm similar to simulation-based inference (SBI) with semi-analytic compression statistic.

MUSE Blinding

Blinded analysis with post-unblinding beam change:

  • Evidence for different beam in polarization.
    • constrained by inter-frequency agreement.
  • Shifted cosmological parameters by up to 1\(\sigma\) (not necessarily closer to Planck).

MUSE Blinding

Similar slope seen by ACT DR6:

MUSE Insights on \(A_{\rm lens}\)

Decreasing neutrino mass enhances structure (measured by \(\phi \phi\)).


CMB prefers more lensing power than predicted by \(\Lambda\)CDM given BAO.
\(\implies\) low \(\sum m_{\nu}\) when allowed to vary.


  • \(A_{\rm recon}\): scales lensing power used to predict lensed CMB power spectra

  • \(A_{\rm 2pt}\): scales lensing power used to predict lensed CMB power spectra

  • \(A_{\rm lens}\): if \(A_{\rm recon} = A_{\rm 2pt}\)