Transcript SKADS-DS2 Sky and Data Simulations Steve Rawlings & Hans-Rainer Kl (Oxford)

Qui ckTime™ and a
TIFF (U ncompr essed) decompressor
are needed to see thi s pi cture.
SKADS-DS2
Sky and Data Simulations
Steve Rawlings & Hans-Rainer Klöckner
(Oxford)
SKADS meeting Perth 2008
SKADS DS2
Qui ckTime™ and a
TIFF (U ncompr essed) decompressor
are needed to see thi s pi cture.
To establish the scientific requirements in
a quantitative manner; to establish the
capabilities of the observing system
necessary to deliver them and thus to
derive a quantitative set of design
specifications.
 Dish versus AA
SKADS interactions
Qui ckTime™ and a
TIFF (U ncompr essed) decompressor
are needed to see thi s pi cture.
SKADS
concept
Sky Simulation (DS2T1)
WIKI
Telescope Simulation (DS2T2)
Technology
Demonstrator
(DS5& DS6)
Technology
Development
(DS4)
Network
Sims + costing
(DS3)
Pathfinder (LOFAR, KAT,ASKAP, MWA)
Simulating the Radio Sky
• The need for sophisticated simulations
- demonstrate SKA can beat systematics
- drive technologies that generate
transformational SKA science
• SKADS
- what’s been done 2006-2008 and how to
use it
- what will be done before mid-2009
- what won’t be done by then, and
future plans
Qui ckTime™ and a
TIFF (U ncompr essed) decompressor
are needed to see thi s pi cture.
‘Back of Envelope’
Z = 1 – 1.5
Z = 0.5 – 1
mn= 0.25 eV
mn= 0.05 eV
mn=0.5 eV
Z = 1.5 – 2
Abdalla & Rawlings (2007)
Important to keep adding to the science case
Qui ckTime™ and a
TIFF (U ncompr essed) decompressor
are needed to see thi s pi cture.
Measurement of Neutrino masses
and hierarchies
Qui ckTime™ and a
TIFF (U ncompr essed) decompressor
are needed to see thi s pi cture.
Abdalla & Rawlings (2007)
• If Smn is ~0.05 eV, SKA+Planck sufficient and necessary to measure it, and
hierarchy must be normal. If Smn > 0.25 eV, measurement of Nn too!
• With particle physics experiments, prospect of evidence for sterile neutrinos or
time evolution of neutrino mass
• But all based on `perfect’ (10 catalogue) SKA survey: no longer adequate
Simple `e2e’ Simulations now inadequate
kk
Sivia & Rawlings
w
Qui ckTime™ and a
TIFF (U ncompr essed) decompressor
are needed to see thi s pi cture.
Also, things are moving on
Qui ckTime™ and a
TIFF (U ncompr essed) decompressor
are needed to see thi s pi cture.
QuickTime™ and a
TIFF (Uncompressed) decompressor
are needed to see this picture.
Spectroscopic All-sky Cosmic Explorer: 5 billion redshifts over 3/4 sky to
z~2.5, multi-slit near-IR
Competition/Synergies: DUNE+SPACE (EUCLID): M-class (300 MEuro),
could be as early as 2017
Best Technology Solution
Qui ckTime™ and a
TIFF (U ncompr essed) decompressor
are needed to see thi s pi cture.
Dishes
~300 MHz: HI at z~4
Aeff
~800 MHz: HI at z~0.75: ~day
with dish-SKA to get multiple
P(k): perfect because few x
365~20,000/30
~500 MHz: HI at z~2: ~month
with the AA-SKA to get multiple
P(k): perfect because
10x12~20,000/250
SPARSE AA CRITICALLYSAMPLED AA Frequency
Collecting area below 500 MHz rises as (1+z)2 for sparse arrays, but
only worth going to z~4 because of sky temperature (Braun 2006)
Using multi-beaming to ‘shape’ FOV a (1+z)4 gives ~constant mapping speed,
and a plausible FOV at the lowest frequency
Technology and Phasing
Qui ckTime™ and a
TIFF (U ncompr essed) decompressor
are needed to see thi s pi cture.
Cosmic volume ~ (time) x (FOV) x (A/T)2 for radio HI
If we relied on observing time, errors on P(k) would improve by a
factor 10 every 100 years!
Technology I: increase dish FOV by ~100 and `dedicate’ array, ~105
galaxies to z~0.2 (1% SKA pathfinders HI P(k) by ~2010)
Technology 2: mass production to increase A/T by ~10. ~107 galaxies
over 20,000 deg2 to z~0.75 (10% SKA in dishes ~2015)
Technology 3: additional technology (500-800 MHz). Further
increase in A/T and FOV. ~109 galaxies to z~2 (100% SKA in
aperture arrays ~2020).
Technology 4: Benefit from increased processing power as you
integrate (300-800 MHz). >1010 galaxies to z~4 (2020-2030)
SKADS `DS2-T1’ Short-Term
Deliverables: T0=July2006
•
Qui ckTime™ and a
TIFF (U ncompr essed) decompressor
are needed to see thi s pi cture.
Continuum Surveys – Oxford, Leiden, (Herts, UCL):
PAPER SUBMITTED:
Richard Wilman (Oxford); simulation deliverable T0+18
RICHARD WILMAN
Ilse van Bemmel (Leiden); deliverable to be finalised (MeqTree Ionosphere module?)
•
Line HI Surveys – Oxford, Groningen, Swinburne
PAPER in prep: SEE
Danail Obreschkow (Oxford); simulation deliverable T0+12
DANAIL OBRESCHKOW
Rense Boomsa (Groningen); deliverable to be finalised (High-res HI module?)
•
Magnetism – Cambridge, Bonn
Martin Krause+replacement (Cambridge); simulation deliverable T0+18
Tigran Arshakian (Bonn); deliverable to be finalised (Galactic Foreground module?)
•
Pulsar Surveys – Manchester
Roy Smits (Manchester), simulation deliverable T0+15
•
EOR – Paris, Lisbon;
Paola Di Matteo (Paris), simulation deliverable T0+21
PAPERS in prep:
ARIS
KARASTERGIOU/ROY SMITS
Two types of extragalactic simulations
Qui ckTime™ and a
TIFF (U ncompr essed) decompressor
are needed to see thi s pi cture.
•
‘Line’ semi-analytic approach
(Obreschkow).
DM haloes from Millennium Simulation ascribed HI and H2,
star formation rates and AGN properties: provides insufficient
FOV for SKADS benchmark
4x4 deg2 out to z~20 (HI/CO)
~107 objects
star-formation continuum OK
AGN continuum needs work
• ‘Continuum’ semi-empirical approach (Wilman). DM
density field evolved under linear theory, populated with objects from known
radio luminosity functions, and with other important physics (e.g. non-linear
structures, source models) `pasted on’
currently 20x20 deg2 out to z~20
~2.5x108 simulated sources
Semi-analytic simulations
Qui ckTime™ and a
TIFF (U ncompr essed) decompressor
are needed to see thi s pi cture.
Millennium Simulation
Semi-analytics
Post-processing
(Springel et al. 05)
(De Lucia et al. 06/07)
(Obreschkow et al. 08)
Dark matter
Visible matter
Neutral atomic hydrogen
Physical Model at z=0
Obreschkow et al.
Qui ckTime™ and a
TIFF (U ncompr essed) decompressor
are needed to see thi s pi cture.
Evolution of HI and H2
Qui ckTime™ and a
TIFF (U ncompr essed) decompressor
are needed to see thi s pi cture.
1.2E-03
1.0E-03
8.0E-04
6.0E-04
HI
4.0E-04
Zwaan 2005 (HIPASS)
2.0E-04
Prochaska 2005 (SDSS-DR3)
Peroux 2005
0.0E+00
H2
0
z=2.5
Note link with ALMA
1
2
3
4
5
Qui ckTime™ and a
TIFF (U ncompr essed) decompressor
are needed to see thi s pi cture.
SKA Predictions
Abdalla & Rawlings
Obreschkow et al
No evolution
• In ~100 days, phased arrays deliver >109 galaxies over
~20,000 deg2 to z~2 (and multiple P(k) to at least z~1)
Continuum and Post-Processing Options
Qui ckTime™ and a
TIFF (U ncompr essed) decompressor
are needed to see thi s pi cture.
Wilman
simulation
Lensing
model: Ben
Metcalf
Francois Levrier + OeRC (Oxford)
Qui ckTime™ and a
TIFF (U ncompr essed) decompressor
are needed to see thi s pi cture.
SKADS Sims as a Tool
Qui ckTime™ and a
TIFF (U ncompr essed) decompressor
are needed to see thi s pi cture.
Obreschkow
Simulation
WSRT data
Sky simulations will increasingly be checked against real data
how does it fit together
Qui ckTime™ and a
TIFF (U ncompr essed) decompressor
are needed to see thi s pi cture.
I
Q
U
V
EoR
galaxies, clusters
Milky Way
Ionosphere /
Troposphere
Jones Matrix
EoR
cosmological
parameter
space density
catalogues of
astronomical
sources
population of
galaxies
population of
Pulsars
& distribution
Galactic
parameter (B, ne)
Galactic
foreground
extrapolate to SKA sensitivities
the concept
Qui ckTime™ and a
TIFF (U ncompr essed) decompressor
are needed to see thi s pi cture.
image
catalogue
catalogue
image
s-cubed.physics.ox.ac.uk
Qui ckTime™ and a
TIFF (U ncompr essed) decompressor
are needed to see thi s pi cture.
Galactic magnetism
RM
Sun & Reich et al.
Qui ckTime™ and a
TIFF (U ncompr essed) decompressor
are needed to see thi s pi cture.
Average of RM |b|<2deg
Average of PI |b|<2deg
PI
PA
I
RM too large for l=-60deg (not a big problem) PI fits well
painting onto
Qui ckTime™ and a
TIFF (U ncompr essed) decompressor
are needed to see thi s pi cture.
HI
Groningen
QuickTime™ and a
TIFF (Uncompressed) decompressor
are needed to see this picture.
polarization /
depolarization
(source IQUV &
cluster) Cambridge
B - field configuration
Bonn
Synthesized Catalogue
Qui ckTime™ and a
TIFF (U ncompr essed) decompressor
are needed to see thi s pi cture.
Sky Simulation (DS2T1)
Telescope Simulation (DS2T2)
catalogue
synthesized catalogue
The MapMaker
•
•
•
•
•
•
Qui ckTime™ and a
TIFF (U ncompr essed) decompressor
are needed to see thi s pi cture.
Standalone application in Python
SQL/Python interface
Source list = query + postprocessing
FITS output with annotation file
2D images or 3D cubes
Backend routines to be
parallelized and plugged into VO
Measurement Set Generator (De
Mesmaeker)
Qui ckTime™ and a
TIFF (U ncompr essed) decompressor
are needed to see thi s pi cture.
Python interface to a glish script from T. Willis
Uses AIPS++ “newsimulator” task
Uses existing configurations or generates
random ones
Builds a skeleton MS to be used by MeqTrees
Configuration Library
Qui ckTime™ and a
TIFF (U ncompr essed) decompressor
are needed to see thi s pi cture.
analysis of the continuum sky
Mauch
reliability & completeness
test, crosscheck number density, clustering
Qui ckTime™ and a
TIFF (U ncompr essed) decompressor
are needed to see thi s pi cture.
Source HI Extraction
Qui ckTime™ and a
TIFF (U ncompr essed) decompressor
are needed to see thi s pi cture.
R. Auld (Cardiff)
MapMaker
HI cube
Duchamp
•
•
•
•
•
•
•
ATNF (M. Whiting)
Locates sets of contiguous voxels above some threshold
Uses spectral, spatial or wavelet smoothing for enhancement
Very quick (30 minutes for 512 x 512 x 1024 cube on a laptop)
VO-compliant
Memory and process intensive
Completeness issues (misses high S/N sources)
Duchamp
The MeqTrees Software
Developed @ ASTRON [Smirnov, Noordam] with outside contributions [Willis]
Open source code Linux / UNIX (virtual machine available)
Code repository for beam shapes, ionospheric models, ...
Trees are declared by python scripts and executed by a C++ kernel
Qui ckTime™ and a
TIFF (U ncompr essed) decompressor
are needed to see thi s pi cture.
MeqTrees Features
•
•
•
•
Powerful calibration : almost any parameter can be solved for
Fast and smart evaluation on time/frequency grids
Per-station effects and polarization effortlessly included
Visual assessment of different models via difference trees
symmetrical vs asymmetrical beam
error level: 2%
•
•
•
average vs per-station beam
error level: 0.02%
No parallelization yet : limited number of sources and antennae
Uses Measurement Sets but does not generate them
Input skies are source lists, not images
Qui ckTime™ and a
TIFF (U ncompr essed) decompressor
are needed to see thi s pi cture.
Sky through a simulated interferometer
Qui ckTime™ and a
TIFF (U ncompr essed) decompressor
are needed to see thi s pi cture.
Wilman
simulation
+ E-LOFAR
simulation
(Heywood)
STFC
permitting,
by~2010
Lensing
model: Ben
Metcalf
What do we need / have?
Qui ckTime™ and a
TIFF (U ncompr essed) decompressor
are needed to see thi s pi cture.
Sky Sim ( cont, HI , Pulsar , Magnetism )
Analysis (Cont , Line , Pulsar  Magnetism )
simulator ( AIPS,  AIPS ++, MeqTree )
calibration ( AIPS,  AIPS ++, MeqTree )
aperture arrays (OeRC, Amsterdam)
configuration studies (Bonn)
need concepts to evaluate technical
simulations (dish/AA): sharpen questions
Concluding Remarks
Qui ckTime™ and a
TIFF (U ncompr essed) decompressor
are needed to see thi s pi cture.
- reference skies exist http://s-cubed.physics.ox.ac.uk
- please use them (for SKA simulations and real science)
and provide feedback
- e2e data simulations now urgently needed, and
tools will appear on this same web site as they are
developed (map-maker, post-processing options,
configurations, instrument simulators)
- must be linked to functional simulators and cost model
- must keep an eye on synergies with other facilities, e.g. EUCLID
• The Future: PrepSKA + Path2SKA: FP7 Sep2
Path2SKA - being organised by SR on behalf of ESKAC needs to be strongly linked to international SKA project