The SuperCLASS Weak Lensing

1

• The SuperCLASS Weak Lensing
• Deep Field Survey

• Ian Harrison on behalf of the SuperCLASS
  collaboration
• AASTCS 2 Exascale Radio Astronomy
• 4 April 2014

2

• SuperCLuster Assisted Shear Survey
• Overview/Contents

Pathfinder for weak lensing cosmologywith the
SKA using UK e-Merlin

• Introduction to Weak Lensing
• Radio Weak Lensing
• Promises and challenges
• Shape measurement with radio data
• SuperCLASS Survey
• Description and status

3

• Weak Lensing as a Cosmological Probe

• Coherent distortion of background sources
• by baryonic and dark matter
• Measure integrated mass on line of sight between
  us and source
• Traces evolution of dark matter structures

4

• Weak Lensing as a Cosmological Probe

• Track Dark Energy equation of state and how it
  evolves with time
• Learn about DE physical nature
• Cosmological constant?
• Scalar field?
• Modifications to GR?
• Weak Lensing can be the best probe of Dark Energy

Dark Energy Task Force FoM
WL
5

• Weak Lensing as a Cosmological Probe
• Requirements

• Large numbers of resolved background galaxies
• Beat down random shape noise
• Exquisitely precise/accurate measurement of
  ellipticities
• 1 level for detection
• 0.01 level for 1 constraint on DE equation of
  state
• Systematics are key!

6

• Weak Lensing as a Cosmological Probe
• Optical Systematics

• Point-Spread-Function errors
• Uncertainty in telescope, seeing
• even in space
• Intrinsic alignments
• Galaxy ellipticities/orientations not random due
  to sharing of LSS environment
• Redshift uncertainties
• Photo-zs can put sources in wrong tomographic bin

7

• Weak Lensing as a Cosmological Probe
• Systematics How bad? Bad

8

• The Promise of Radio Weak Lensing
• Control of Systematics

• PSF Errors
• Radio interferometer beams are (in principle)
• Precisely known
• Highly deterministic
• Intrinsic alignments (Brown Battye 2011)
• Radio polarisation information tells about
  intrinsic alignment
• Polarisation angle unchanged by gravitational
  lensing
• Redshift uncertainties
• Large 21cm line surveys give spec-z for sources
• Cross Correlations
• Euclid comparable, similar timescale to SKA

9

• The Promise of Radio Weak Lensing
• Current Status

• Chang, Refregier, Helfand (2004)
• VLA FIRST data
• 5 arcsec resolution
• 1 mJy depth
• 104 deg2
• 20 sources deg-2
• 20,000 source
• 3s detection of cosmic shear
• Measure shapes in UV plane

• Patel et al (2010)
• MerlinVLA data
• 0.4 arcsec resolution
• 50 µJy depth
• Only 70 arcmin2
• 1-4 sources arcmin-2
• 50-300 sources
• No detection of cosmic shear
• Measure shapes in images

10

• The Promise of Radio Weak Lensing
• Measuring Ellipticities

• One methodshapelets
• Model image using truncated basis
• or visibilities
• FT is just a phase factor
• Gives linear problem
• Easy to solve ?2 for best-fitting coefficients
• Can estimate shear from combination of
  coefficients

11

• The Promise of Radio Weak Lensing
• Current Status

• Chang, Refregier, Helfand (2004)
• Take source positions from images
• Use Fourier-plane shapelets to model visibilities
  directly
• Model systematics with simulations of
  delta-function sources
• 3s detection

12

• The Promise of Radio Weak Lensing
• Current Status

• Patel et al (2010)
• Use real-space shapelet basis functions
• Model sources in reconstructed images
• No shear signal recovered
• Also cross-correlate with optical data
  (HDF-North)
• Find no correlation

13

• The Promise of Radio Weak Lensing
• Current Status

• Patel et al (2013)
• Simulate e-Merlin and LOFAR observations
• Known input ellipticities
• Noise free
• Measure shear using image plane shapelets
• Quantify accuracy of fit
• eobs etrue metrue c

Amara Refregier (2008) gives m lt 0.05 c lt
0.0075 For simulated survey to be dominated by
statistics, not systematics m lt 0.001 c lt
0.0002 for SKA
14

• The Promise of Radio Weak Lensing
• Challenges of Radio Shape Measurement

• Understanding of shape measurement algorithms for
  radio data currently not good
• Only 1.5 methods have been tried
• On different datasets
• Are N potential shape measurement methods
• Which galaxy model?
• Physically motivated (e.g. Sersic)
• Image decomposition (e.g. Shapelets)
• Which data?
• UV
• Image
• Method space needs exploring

15

• The Promise of Radio Weak Lensing
• Challenges of Radio Shape Measurement

• Image Plane
• Only fit one object at a time
• Optical algorithms can be easily leveraged
• Correlated noise
• Need to create image with no spurious shear from
  deconvolution!
• Is a big challenge in itself

• UV Plane
• Does not require deconvolution
• Need to fit sources simultaneously!
• 5 parameters per source
• 100 sources per FoV
• 10n data points
• (Probably) still need to image to source find
• Probably wont have visibilities any more

16

• A Radio GREAT Challenge
• (Gravitational lEnsing Accuracy Test)

• Understanding of shape measurement algorithms for
  radio data currently not good
• Optical weak lensing community has gained much
  from shape measurement challenges
• STEP, STEP2, GREAT08, GREAT10, GREAT3
• Simulate weak lensing data set
• Different algorithms compete to measure (blinded)
  shear in the data with greatest fidelity
• Winners have come from non-astronomy backgrounds

• gt A GREAT Challenge for radio data

17

• A Radio GREAT Challenge
• Plans

• (Very simple) overview
• Create sky model
• Simulate observation with a single pointing of a
  known antenna configuration
• Provide entrants with
• Visibilities
• Fiducial image with quantified systematics due to
  deconvolution
• Help and ideas welcome
• Sign up for updates!
• jb.man.ac.uk/harrison/

18

• SuperCLASS

e-Merlin legacy survey Pathfinder for radio weak
lensing with the SKA
19

• SuperCLASS
• Goals

• Develop techniques for radio shear measurement
• Prove effectiveness of polarisation for
  mitigation of intrinsic alignments
• Learn about source populations at µJy radio
  fluxes which will be probed by SKA surveys
• Number densities
• Polarisation fraction and position angle scatter
• few and rms 10-20 deg for local spirals (Stil
  et al 2009)

20

• SuperCLASS
• The Survey

• Specifications/performance goals
• 1.75 deg2
• 4µJy/beam flux rms
• L-band (1.4 GHz), 512MHz bandwidth
• 0.2 arcsecond resolution
• 1-2 arcmin-2 source density
• Dense supercluster target field
• Observing strategy
• 800 hours total
• 430 mosaic pointings
• 20TB visibilities on disk

21
SuperCLASS Collaboration
David Bacon Bob Nichol
Richard Battye (PI) Michael Brown Neal
Jackson Ian Browne Simon Garrington Paddy
Leahy Peter Wilkinson Anita Richards Scott
Kay Rob Beswick Tom Muxlowe Sarah Bridle Lee
Whittaker Constantinos Demetroullas Ian
Harrison Rafal Szepietowski
Torsten Ensslin Mike Bell
Steve Myers Chris Hales
Anna Scaife Chris Riseley
Ian Smail
Caitlin Casey
Mark Birkinshaw
Hung Chao-Ling
30 People 11 Institutions 3 Countries
Meghan Gray
Filipe Abdalla
22

• SuperCLASS
• e-Merlin Pipeline

• Currently uses standard e-Merlin data reduction
  pipeline(Argo et al, in prep)
• Requires ParselTongue, AIPS, Obit

• What it does
• Loading sorting
• Averaging
• Concatenating
• Flagging
• Diagnostic plotting
• Calibration (with caveats)

• What it doesnt (yet) do
• Perfect calibration
• Spectral line mode
• Multiple source/phcal pairs
• Wide-field imaging
• Publication-quality images

(from Megan Argo)
23

• SuperCLASS
• e-Merlin Pipeline

• Merlin data
• manual reduction

• e-Merlin data
• one button reduction

(from Megan Argo)
24

• SuperCLASS
• RFI Mitigation

• SERPent automated flagging of RFI (Peck Fenech
  2013)
• Fully parallelised
• Flags 7GbCPU-1day-1
• Uses SumThreshold algorithm (Offringa et al 2010)
• Subset of visibilites thresholded
• If above, flag to threshold level

25

• SuperCLASS
• Current Status

• Characterisation of polarisation leakage across
  field of view
• Appears to be stable in time, position
• Calibratable
• Have observed initial 7 point mosaic
• 12 hours total
• mJy sources visible in total intensity

(from Neal Jackson)
26

• SuperCLASS
• Projected Performance

(Brown Battye 2011)

• Expect up to 10s detection of shear from each
  cluster
• Lower limit should be 6.6s
• Expected across a whole randomly chosen field

27

• SuperCLASS
• Additional Data and Science

• Data

• Science

• LOFAR
• 120 180 MHz
• GMRT
• 325MHz
• JVLA (proposed)
• Short baselines
• Optical data from Subaru SuprimeCam
• Photometric redshifts

• Source populations at µJy fluxes
• Magnetic fields in super-clusters
• Dynamic state of ICM
• Strong lenses

28

• SuperCLASS
• Summary

• Radio weak lensing can do good cosmology
• Mitigates many systematics from optical surveys
• Deterministic beam
• Polarisation for intrinsic alignments (Brown
  Battye 2011)
• Cross-correlations (Euclid comparable, on same
  timescale to SKA)
• but will be difficult
• What are properties of sources?
• How will we do the shape measurement?
• radioGREAT challenge for shape measurement from
  simulations jb.man.ac.uk/harrison
• SuperCLASS providing real data to form a test bed