PISCO Progress Carnegie A. Szentgyorgyi for A. Stark 20 March 2008
Dec 30, 2015
PISCO Progress
CarnegieA. Szentgyorgyi for A. Stark
20 March 2008
Parallel Imager for Southern Cosmology Observations (PISCO):
A Multiband Imager for Magellan
Antony Stark Antony Stark Smithsonian PISmithsonian PI
Christopher StubbsChristopher Stubbs Harvard PIHarvard PI
Matt HolmanMatt Holman Smithsonian Smithsonian — Planets, exoplanets— Planets, exoplanets
John GearyJohn Geary CCD electronicsCCD electronics
Andy SzentgyorgyiAndy Szentgyorgyi Design consultantDesign consultant
Steve AmatoSteve Amato CCD electronicsCCD electronics
Michael Wood-VaseyMichael Wood-Vasey AstronomerAstronomer — Observing — Observing
Will HighWill High Thesis project, Harvard PhysicsThesis project, Harvard Physics
Andrea LoehrAndrea Loehr Astronomer Astronomer — — Observing algorithmObserving algorithm
Brian StalderBrian Stalder PostDoc, Harvard PhysicsPostDoc, Harvard Physics
James BattatJames Battat grad student, SAOgrad student, SAO
Armin RestArmin Rest PostDoc PostDoc — — Photo-z SoftwarePhoto-z Software
Steve SansoneSteve Sansone LPPC machine shopLPPC machine shop
Antony Stark Antony Stark Smithsonian PISmithsonian PI
Christopher StubbsChristopher Stubbs Harvard PIHarvard PI
Matt HolmanMatt Holman Smithsonian Smithsonian — Planets, exoplanets— Planets, exoplanets
John GearyJohn Geary CCD electronicsCCD electronics
Andy SzentgyorgyiAndy Szentgyorgyi Design consultantDesign consultant
Steve AmatoSteve Amato CCD electronicsCCD electronics
Michael Wood-VaseyMichael Wood-Vasey AstronomerAstronomer — Observing — Observing
Will HighWill High Thesis project, Harvard PhysicsThesis project, Harvard Physics
Andrea LoehrAndrea Loehr Astronomer Astronomer — — Observing algorithmObserving algorithm
Brian StalderBrian Stalder PostDoc, Harvard PhysicsPostDoc, Harvard Physics
James BattatJames Battat grad student, SAOgrad student, SAO
Armin RestArmin Rest PostDoc PostDoc — — Photo-z SoftwarePhoto-z Software
Steve SansoneSteve Sansone LPPC machine shopLPPC machine shop
PISCO System Characteristics
Optical Passbands g, r, i, z simultaneous imaging
Plate Scale 0.16 arcsec per 10 micron pixel
Field of View 9 arcmin across diagonal (corresponds to 2.4 Mpc at z=0.3)
Detectors One MIT Lincoln Labs 3K x 6K CCD per focal planeRead noise < 4 e rms (cf. Stubbs talk)Optimal AR coating for each passbandDeep depletion CCDs for i, z bands
Readout time < 25 seconds
80% encircled energy radius < 0.2 arcseconds (seeing limited)
Optical surfaces All spherical, 8 inch diameter or smaller lenses.
Magellan Telescope
Folded f/11 port
Dichroic in Cube Optical Layout
Revised Optical Design of PISCO. Steve Schectman contributed to this design. The dichroics are embedded into cubes of fused silica, so that there is no difference in dielectric constant on either side of the dichroic. This allows the dichroics to be used at 45º. The dichroics are placed in the telecentric beam from the focal reducer, so all field positions have identical ranges of angle of incidence at the dichroics. The overall length of the instrument is reduced to 1.6 meters, and all CCDs are in a single, medium-sized dewar.
Current best desgin
Design is 1.42 meters (56 inches) from focus to focusDesign uses S-FPL51 glass
ADC Operation
• Can use PISCO on Clay telescope
• Consists of two rotating cylindrical prisms, – 1 cm thick– airspaced, multi-coated
• Initial scientific mission can be achieved without ADC
• ADC can be removed with re-focus
Operation of ADC
0.17
arc
sec
PISCO Design Concept
Shutter Dewar Wall
Small Guider Housing
ADC
Lenses
Dichroics & CCDs
Cable Wrap
Electronics mounted here
Some Optics Have Been Ordered
• Contract in place with Barr Associates for fabrication of Dichroic Cubes
• This has a long lead time (8 months) and will drive the project timetable
Electronics are done…
• We have already taken images in the lab with full control-to-image software.
• Readout noise is OK (3 electrons).
• Readout speed is OK ( < 8 seconds).
Initial detector tests look favorable
Tested 2 3K x 6K 10 micron high-rho devices in Univ of Hawaii test system.
Read noise
Dark current vs. temperature
CTE via Fe55 xrays
Gain via Fe55 xrays
Detectors work well
Analysis Software: We’ll build upon SuperMacho/ESSENCE image analysis
pipeline
Battle tested over past 6 years at CTIO for SM and ESSENCE surveys.– Flatten with dome flats, fringe flat and sky flats
– Astrometric WCS registration, warp to fixed plate scale
– Photometry to 1%
– CVS code management, easy to add new modules
– Parallel implementation, Condor on Linux boxes
– Robust and self-tracking
– Honed on crowded fields
– Need to add (1) cluster photo-z module, and (2) SQL database
– Armin Rest, pipemeister, coming to CfA in Spring 2007.
Flow diagram for real-time cluster redshift analysis pipeline
We expect that within 30 seconds of acquiring the first image, we will have produce an appraisal of whether the second 30 image will add enough integration time to obtain a cluster photometric redshift at the requisite SNR. We have in hand the middleware and pipeline structure for this, from the ESSENCE and SuperMacho surveys. We are missing only the final segment, namely the redshift estimator, which we will develop in parallel with the construction of the hardware.
Tightly coupled software/observing
Take Image 1Take Image 130 sec 30 sec
Analyze Image: Analyze Image: flatten, WCS, sextractor flatten, WCS, sextractor Galactic reddening corr.Galactic reddening corr.
Produce z, Produce z, zz
OK? OK?
OffsetOffsetTake Image 2Take Image 2
30 sec 30 sec
Slew to next targetSlew to next target
Offset if appropriateOffset if appropriate
More imagesMore images
Photometric Redshift for Clusters
• Photo-z’s for individual galaxies tend to have scatter of z/(1+z)~0.03, but with a few “catastrophic” outliers.
• Combination of morphology, magnitude, color and location can be used to establish cluster’s redshift.
• Robust statistics can be used to eliminate “outliers”.
Uniform exposure times for clustersRedshif
t zg t_g r t_r i t_i z t_z
0.1 18.5 5 18.0 5 17.5 5 17.0 5
0.2 20.5 5 19.5 5 19.0 5 18.5 5
0.4 23.0 35 21.2 5 20.5 5 20.0 5
0.6 24.5 550 23.0 90 22.0 60 21.2 30
0.8 26.2 12000 24.2 780 23.0 320 22.1 160
1.0 - 25.0 3420 24.0 2000 23 870
Magnitudes in the four filter bands (shaded) for L*/2 early type galaxies, and exposure times (in seconds, unshaded) to achieve SNR=10, as a function of redshift. The table assumes galaxy flux integrated in a 2.2 arcsec diameter aperture, in seeing of 0.8 arcsec at an airmass of 1.2 in dark time. The numbers assume deep depletion detectors in the z and i bands, like those for the SMI. The exposure time needed to achieve SNR=10 is reasonably well matched across the bands. A minimum exposure time is 5 sec.
One night to obtain 115 cluster redshifts at z < 1.5 z range N clusters N * exposure time (seconds)
0.0 – 0.2 15 15 * 60 = 900
0.2 – 0.4 30 30 * 60 = 1800
0.4 – 0.6 30 30 * 60 = 1800
0.6 – 0.8 15 15 * 200 = 3000
0.8 – 1.0 10 10 * 800 = 8000
1.0 – 1.2 9
1.2 – 1.4 4 15 * 1000 = 15000
1.4 – 1.6 2
Totals 4 hours for 100 clusters to z=14.2 add’l hrs for 15 at z > 1
The time needed to obtain 115 cluster redshifts, in good conditions, is 8.2 hours. It will not be possible to obtain redshifts for the ~10% of clusters with redshift z > 1.5; these will be flagged to obtain redshifts using other instruments.
South Pole Telescope2007 First-Look Data
•SPT data, Feb-April 2007
•4 square degrees shown
•red circles are known quasars
•green regions are significant negative regions in CMB: possible clusters
•Current, upgraded SPT detector system shows two order of magnitude improvement in observing speed.
Magellan Observations of SPT Cluster Candidates
LDSS3 multi-color photometry of SPT-selected region nr01
Abell 267, extrapolated to various redshiftsand observed with PISCO
Order of detection by PISCO
8 bright red galaxies detected first (green circles)
Black-circled detected next
Blue dots are cluster galaxies
Black dots are foreground
Galaxies in Color-Magnitude Diagram
Histogram of photo-z of the first 18 galaxies and photo-z of the color-magnitude selected galaxies.
We are building the capability to efficiently chase SZ detections in optical
1. Blanco Cluster Survey (with Mohr et al)
2. Imaging with existing Magellan instruments
3. Spectroscopy with existing magellan instruments
4. Custom simultaneous multiband imager, PISCO
Ask a restricted set of questions• At a known position on the sky, is there a cluster of galaxies?• What is the redshift of the cluster?
– We initially assume all galaxies are LRG’s– We make a redshift estimate based on this assumption– We use magnitude consistency to select the LRGs.– We then use a clustering algorithm to search for clustering in redshift space.
• This is not the general problem of finding a photo-z for some random galaxy.
• We are focusing on “luminous red galaxies”, LRG’s.
Why LRG’s?
• These elliptical galaxies are preferentially found in custers, so they exhibit “clustering” more than, say, spirals.
• They suffer minimal extinction/reddening due to dust in the galaxy, which can distort colors and therefore photo-z’s
• They’re bright, and are crude standard candles, which helps in photo-z determination.
Photometric Redshift PrincipleThe plots show how the observer-frame spectrum of a Luminous Red Galaxy (LRG) depends upon its redshift. The redshifts are indicated in the upper left corner of each panel. The flux ratios between the g, r, i, and z bands is a good indicator of galaxy redshift, as the 4000 Å break moves across the spectrum. We will develop real-time analysis code that will produce an initial cluster redshift result within 30 seconds of the acquisition of an image.
From M. Blanton’s web page
Status of Observations• Reduction of BCS data under way
• Flatfielding to better than 1%• Astrometric registration• Source Extractor photometry• Photo-z determination under development
• Deep multiband images of initial SZ 2 degree region at (RA,DEC), plus similar region at arbitrary location for statistics
• Long slit and muli-slit spectroscopy of selected galaxies in NR1 region
• Additional nights both allocated & requested
Source Extractor Photometry• Used mag_auto
fluxes from SE• Determine
galaxy colors and uncertainties
A cluster photo-z estimator1. Use Blanton’s K-correct code to predict SDSS colors for LRG
vs. redshift.
2. Assume all galaxies are LRG’s
3. For each galaxy, for each trial redshift, compute
error-weighted distance to prediction, for each color
4. Using distances for all 3 colors, calculate composite color distance vs. z
5. Pick z with minimum normalized color distance
6. Estimate redshift uncertainty by finding dz that produces color distance = 2
Forward modeling of LRG spectra
redshift
i-zi-z
r-ir-i
g-rg-r
3-d color evolution with redshift
Example of color distances vs. redshift
Overall distanceOverall distance
Photo-z estimate
What about r band magnitude?• We can use the apparent magnitude to select out
likely LRG’s. • They’re bright, r ~ 17th at redshift = 0.1• At other redshifts the r band magnitude has two
contributions, m(z)=m(0.1) + DM + K_corr(z)
cosmology filter/SED
Change in apparent magnitude due to passband redshift and luminosity distance
m=0.27=0.73h=0.7
K correction(LRG’s)
both
cosmology
Note: this Note: this ignores potential ignores potential age effects in age effects in stellar populationstellar population
Luminosity Luminosity function work function work suggests we suggests we normalize to normalize to r = 17 at redshift r = 17 at redshift of 0.1 of 0.1
Compare this with observations
Fudged LRG cut:r_cut = (predicted r(z)) - 2.5*redshift - offset
SDSS LRG’s
SDSS reg. galaxiesIntroduce an Introduce an empirical empirical correction vs. correction vs. redshift to redshift to correct for correct for evolutionary evolutionary effects effects
• Use colors and assumption of LRG spectrum to estimate the redshift
• Use lookup table to find typical LRG magnitude at this redshift
• Compute magnitude difference.• Allow for galaxies to be up to Mcut magnitudes
fainter than the LRG line.
Demand LRG consistency
Compare photoz and spectroz’s
Cut to require LRG magnitude
LRGs only
LRG catalog is produced
• RA, DEC, photoz, photoz error, magnitudes and colors with uncertainties, color distance vectors and statistics.
• Next task is to ask if there is a statistical overdensity in redshift within SPT angular footprint
Cluster finding• Visual inspection of BCS and Magellan followup
images suggest a cluster of galaxies that coincides with NR1 region.
• Cluster detection is multiparameter search• Position
• Size
• LRG cutoff magnitude
• Redshift histogram binning width
Cluster finding
• We have multiple cluster detection algorithms under development.
• One example:• Map out redshift distribution in an area A• Determine background redshift distribution in 8
surrounding regions. Use this as background estimate.• Compute histogram of excess or deficit relative to this
local background redshift distribution.
A Cluster at z = 0.3 in nr01
Recommendations
• We are close to being able to write a paper on SZ detection of clusters (author list?)
• It would help a lot to have a radio color discriminant for clusters (i.e. non-detection at 220 GHz, stronger detection at 90 GHz)
Australia Telescope Compact Array (ATCA) observations of SPT clusters
Antony Stark Antony Stark Smithsonian Astro ObsSmithsonian Astro Obs
Wilfred WalshWilfred Walsh U. New South Wales Asia U. New South Wales Asia
Joe MohrJoe Mohr U. IllinoisU. Illinois
Tom CrawfordTom Crawford U. ChicagoU. Chicago
Antony Stark Antony Stark Smithsonian Astro ObsSmithsonian Astro Obs
Wilfred WalshWilfred Walsh U. New South Wales Asia U. New South Wales Asia
Joe MohrJoe Mohr U. IllinoisU. Illinois
Tom CrawfordTom Crawford U. ChicagoU. Chicago
Relevance to SPT Cluster Survey
• SPT system is around 100 Jy/K• SPT-SZ survey will be 10 μK rms per beam• Point continuum sources that are 1 mJy or brighter
will make a significant contribution to the data, and possibly affect the detection of clusters and their derived parameters.
• The number of such sources in SPT bands is poorly known.
• With the ATCA, we can actually search for and detect such sources in SPT clusters.
Pilot Study Completed
• We were actually awarded a significant amount of observing time on ATCA
• “Extragalactic” time slot is undersubscribed, and not too hard to get observing time
• We observed 24 X-ray selected moderate redshift clusters (Mullis et al. 2003) in redshift range 0.05 < z < 0.65
• Observe at 18 GHz, because of ATCA sensitivity and map area; possible follow-up at 90 GHz
• Detected one source at ~ 2 mJy at 18 GHz• Our sensitivity was primarily limited by phase stability—we
will need good weather
Current Status
• Funded through Smithsonian Institution for expenses related to these observations.
THE ENDhttp://www.tonystark.org
Science Opportunities• Supernova followup observations
• Type Ia and type II Sne as cosmological probes• Requires multiband images, multiple epochs
• Photometric redshifts of clusters• 4 band imaging over 5 arcmin field
• Transient followup• Evolution of SED for GRBs• Microlensing light curves
• Planetary occultations• Multiband data useful for discrimination
• Followup camera for PanSTARRS/LSST
Masses and radii of transiting extrasolar planets
The dashed lines correspond to loci of constant mean density. The symbols indicate the nine known transiting planets, along with Jupiter and Saturn. Two symbols are shown for OGLE-TR-10b. In green is the result based on a fit to the OGLE photometry and available radial velocities (Konacki et al. 2005). In blue is the Holman et al. (2005) result, based on a simultaneous fit to Magellan photometry and the same radial velocity measurements.