Fundamental cosmology

Phil BullUniversity of Oslo

with the SKA

1) Overview of the SKA

2) Cosmology with radio surveys

3) Testing fundamental physics

4) Challenges...

Outline

Overview of the SKA

Square Kilometre Array

ASKAP36 x 12m dishes36-element multi-pixel feedsW. Australia (completed 2012)

MeerKAT64 x 13.5m dishesChoice of two bands for HIS. African site (due 2017)

KAT 77 x 12m dishesS. African siteAlready producing results (2014)

ASKAP

KAT 7

Dish arrays~96-250 steerable dishesInterferometer with dense coreBaselines from 20m – 100km+SKA1-MID, SKA1-SUR (2023)

Low-freq. aperture arrayVery cheap, large field of viewElectronic beam formingSKA1-LOW (2023)

Mid-freq. aperture arrayCheap, large field of viewElectronic beam formingSKA Phase 2 (~2030)

SKA2

SKA1

SKA1

K. Vanderlinde

CHIMEFive 20 x 80m cylindersGPU correlation (1280x recv.)Site in British ColumbiaPathfinder up and running (2013)

BINGO~40m dual reflector system50x autocorrelation receiversUK-based, site in Uruguay/Brazil

Others

FAST (China)HIRAX (S. Africa)LOFAR (Netherlands)TIANLAI (China)

BINGO

CHIME

Cosmology withradio surveys

The future of radio cosmology

Big surveys...● Large instantaneous field of view (e.g. multiple beams)● Large collecting area (many dishes/aperture array tiles)● Many, many low-noise receivers● Wide instantaneous bandwidth (100's of MHz)

The future of radio cosmology

Big surveys...● Large instantaneous field of view (e.g. multiple beams)● Large collecting area (many dishes/aperture array tiles)● Many, many low-noise receivers● Wide instantaneous bandwidth (100's of MHz)

...and big problems● Need very (read:impossibly) high performance computing:

● High sample rates, narrow channels → very high data rates● Massive cross-correlation● On-the-fly calibration (can't store all the data)

● Frequency-dependent beams, gigantic foregrounds, polarisation leakage...

Cosmology from LSS

SDSS

Shape of the power spectrum

K. Heitmann et al. (2014)

“Ultra-large”

Baryon acoustic oscillations

Non-linear

HI Galaxy Redshiftsurvey

Continuum survey

HI Intensity Mappingsurvey

HighLow

Angular resolution

Low

Hig

h

Red

shif

t re

solu

tio

n SKA LSS surveys

Galaxy redshift survey

BOSS (2014)

SKA 1 (~2025)

~ 1.5 million galaxies

10,000 sq. deg. (z < 0.7)

~ 5 million galaxies

5,000 sq. deg. (z < 0.4)

Good redshift resolution → small bandwidth → higher noise

Galaxy redshift survey

BOSS (2014)

SKA 1 (~2025)

Euclid (~2025)

Full SKA (~2030)

~ 1.5 million galaxies

10,000 sq. deg. (z < 0.7)

~ 5 million galaxies

5,000 sq. deg. (z < 0.4)

~ 60 million galaxies

15,000 sq. deg. (0.7 < z < 2.0)

~ 1 billion galaxies

30,000 sq. deg. (0.2 < z < 1.7)

Good redshift resolution → small bandwidth → higher noise

Intensity mapping

Why detect individual galaxies?● Only care about larger scales...● High SNR detection of galaxy 'wastes' photons● Spectroscopic redshifts take a long time

→ Map out emission integrated over many galaxies

Intensity mapping

Why detect individual galaxies?● Only care about larger scales...● High SNR detection of galaxy 'wastes' photons● Spectroscopic redshifts take a long time

→ Map out emission integrated over many galaxies

21cm intensity maps● Low-resolution still preserves large-scales (c.f. CMB)● Integrated emission is easier to detect / no thresholding● Detecting an emission line → get redshifts for free

Continuum survey

Galaxy spectra have a continuum component● Synchrotron from star-forming galaxies / AGN● Spectra are smooth; no features => no redshift info

(but can distinguish different populations)● Integrate emission over frequency → high SNR

M. Brown et al. (2015)

Testing fundamental physics

Important observables

BAO feature is a “standard ruler”

● Measure combination of H(z) and DA(z) from moments of corr. fn.

● Constrains geometry and energy content of the Universe

● Key quantity: Dark energy equation of state, w(z)

Important observables

BAO feature is a “standard ruler”

● Measure combination of H(z) and DA(z) from moments of corr. fn.

● Constrains geometry and energy content of the Universe

● Key quantity: Dark energy equation of state, w(z)

Redshift-space distortions

● Peculiar velocities are sensitive probe of growth history

● See deviations from GR growth in modified gravity theories

● Key quantity: Linear growth rate, f(z)

Important observables

BAO feature is a “standard ruler”

● Measure combination of H(z) and DA(z) from moments of corr. fn.

● Constrains geometry and energy content of the Universe

● Key quantity: Dark energy equation of state, w(z)

Redshift-space distortions

● Peculiar velocities are sensitive probe of growth history

● See deviations from GR growth in modified gravity theories

● Key quantity: Linear growth rate, f(z)

Ultra-large scales (H ~ k)

● Key quantities: See the rest of this talk!

Why is Λ so fine-tuned? Why is the cosmic expansion accelerating?

Maybe we have the wrong theory of gravity?

Huge zoo of MG theories

Modified gravity

Why is Λ so fine-tuned? Why is the cosmic expansion accelerating?

Maybe we have the wrong theory of gravity?

Huge zoo of MG theories; they do make some generic predictions though:

Modified gravity

Poisson eqn.

Slip relation

Why is Λ so fine-tuned? Why is the cosmic expansion accelerating?

Maybe we have the wrong theory of gravity?

Huge zoo of MG theories; they do make some generic predictions though:

Modified gravity

Gauge-invariantdensity contrast

Poisson eqn.

Slip relation

=1 in GR

Most studies have looked at effects on:● Background (different expansion history)● Linear perturbations (different growth history)● Non-linear scales (screening effects)

How can we detect MG?

Most studies have looked at effects on:● Background (different expansion history)● Linear perturbations (different growth history)● Non-linear scales (screening effects)

But if MG is responsible for acceleration, the Hubble scale is a natural scale to look at

How can we detect MG?

Most studies have looked at effects on:● Background (different expansion history)● Linear perturbations (different growth history)● Non-linear scales (screening effects)

But if MG is responsible for acceleration, the Hubble scale is a natural scale to look at:

How can we detect MG?

Scale-indep.(fn. of time)

Scale-dep.correction

New MGtimescale

New MGmass scale

Most studies have looked at effects on:● Background (different expansion history)● Linear perturbations (different growth history)● Non-linear scales (screening effects)

But if MG is responsible for acceleration, the Hubble scale is a natural scale to look at:

How can we detect MG?

Details in Baker et al.(arXiv:1409.8284)

Scale-indep.(fn. of time)

Scale-dep.correction

New MGtimescale

New MGmass scale

Scales that we can probe

There are corrections to LSS observables on ultra-large scales even in GR.

Physics on ultra-large scales

There are corrections to LSS observables on ultra-large scales even in GR.

Perturbation to galaxy number counts:

Physics on ultra-large scales

See work by (e.g.) Yoo et al., Bonvin & Durrer, Challinor & Lewis, Di Dio et al.

There are corrections to LSS observables on ultra-large scales even in GR.

Perturbation to galaxy number counts:

Physics on ultra-large scales

See work by (e.g.) Yoo et al., Bonvin & Durrer, Challinor & Lewis, Di Dio et al.

Density

RSDs

LensingLensing

ISW

T. Baker & PB (arXiv:1506.00641)

T. Baker & PB (arXiv:1506.00641)

Simple MG ansatz:

λ = 10, fμ = 0.1, ,

T. Baker & PB (arXiv:1506.00641)

~1%

0.1%

λ = 10, fμ = 0.1, ,

T. Baker & PB (arXiv:1506.00641)

Non-Gaussianity

● Primordial non-Gaussianity is a powerful test of inflation

● NG => Scale-dependent bias at small k (Dalal et al. 2008)

Camera et al. (2013)

fNL = 10

Only important at ultra-large scales

Non-Gaussianity

● Primordial non-Gaussianity is a powerful test of inflation

● NG => Scale-dependent bias at small k (Dalal et al. 2008)

(p=1)

Size of NG signal depends on relative bias (larger = better)

D. Alonso, PB et al. (arXiv:1505.07596)

σ(fNL)~3

Multi-tracer effect

Cosmic variance: fundamental limit to large-scale obs.

● Only see a finite number of Fourier modes

● Biased populations probe the same DM field → deterministic

Multi-tracer effect

Cosmic variance: fundamental limit to large-scale obs.

● Only see a finite number of Fourier modes

● Biased populations probe the same DM field → deterministic

● Tracer-dependent quantities not CV-limited (Seljak 2008)

Alonso & Ferreira (2015)

Multi-tracer effect

Cosmic variance: fundamental limit to large-scale obs.

● Only see a finite number of Fourier modes

● Biased populations probe the same DM field → deterministic

● Tracer-dependent quantities not CV-limited (Seljak 2008)

Alonso & Ferreira (2015)

Challenges

Bias / HI evolution

● Tracer evolution not well understood – important consequences for detection of the signal!

Padmanabhan et al. (arXiv:1407.6366)

e.g. Obreschkow et al. (2009); Villaescusa-Navarro et al., Yahya, PB et al. (2014)

Galaxy survey systematics

● Bright point sources complicate angular selection function on large scales (c.f. stellar weights in BOSS, Ross et al. 2013)

● Multi-tracer analysis only possible if populations can be reliably distinguished (continuum: using morphological info only)

● Need lots of very big simulations (including GR effects) for calibration etc. (extremely large volumes out to high redshift)

● Efficient radio frequency interference (RFI) flagging needed

Foreground contamination

Total foreground is ~105 times larger than cosmo signal

But: spectrally smooth, so subtraction is possible

Subtracting smooth functions also takes away large-scale modes

Alonso, PB, Ferreira, Santos (2014)

Polarisation leakage

● HI signal is unpolarised, but polarised foregrounds leak into total intensity channel of receivers

● Polarised foregrounds not smooth due to Faraday rotation

Cosmological signal

Polarisation leakage

Alonso et al. (2014)

Autocorrelation

● Autocorrelation suffers from correlated noise → striping

● Need to scan quickly or have very stable receivers, or lose large-scale modes

E. Keihanen et al. (2010)

(Well-known problem in CMB experiments)

Summary

Summary

The SKA will produce very wide + deep surveys, enabling exciting new tests of fundamental physics in the radio

● Dark energy / linear growth with Phase 1 intensity mapping and Phase 2 spectro-z

● Ultra-large scale observations allow novel tests of inflation (NG) and gravity (Hubble-scale GR corrections)

● Difficult (but tractable) systematics abound

Image credits: Flickr: Daniel Foster / Berge Gazen / Mattia Mc / luxomedia (CC:BY-NC-SA); Alex Cherney

[Extra Slides]

Dark Energy

IM / Galaxy z's

Modified gravity: Growth

See Bull et al. (arXiv:1405.1452)IM / Galaxy z's

Intensity mapping: Resolution

● Single-dish resolution limited by dish size

● Interferometer resolution set by max. baseline

But can't see angular scales bigger than the minimum baseline!

Interferometer vs. dish survey

15m dishes, 25,000 sq. deg survey

Curvature

● Curvature is a strong prediction of inflation

● Eternal inflation would be ruled-out if curvature was detected above the |10-4| level

(Kleban & Schillo, Guth & Nomura)

● Fundamental limit of detectability at ~|10-4| too

(Vardanyan et al. 2009, PB & Kamionkowski 2013)

Kleban & Schillo (2012)

Curvature

See PB et al. (arXiv:1405.1452)

Future LSS + CMB curvature constraints will approach this limit:

IM / Galaxy z's

(incl. Planck)

Other observables

● ISW cross-correlation (all)

● Weak lensing (continuum)

● Redshift drift – measure dz/dt (galaxy z's)

● Tests of isotropy (continuum / IM)