Galaxy Clutering and Photometric Redhift Jennifer Helby (KICP/UChicago), Johua Frieman (KICP/UChicago, Fermilab), Huan Lin (Fermilab) DESpec Workhop - May 30, 2012
overview uing cro-correlation for DES reducing ample-variance requirement cro-correlation with cluter poition of nearby pectrocopic galaxie a prior application for DESpec Helby - DESpec Workhop - 1/10
teting with the DES mock ued DES mock v3.04 (~220 q. deg.) plit into 70 field roughly ~2x2deg ~2 ~2 in each field we meaure: w p (r p ) w pp (θ) w p (θ) of pectrocopic ample (meaured in bin of pec) of photometric ample cro-correlation between photometric and pectrocopic ample (meaured in bin of pec) from thee correlation function we determine the redhift ditribution Helby - DESpec Workhop - 2/10
tet with realitic pectrocopic ditribution each over 14 field (44 q. deg.) N = 2000 per field N p 200000 per field redhift ditribution of the pectrocopic ample 2.0 10 4 dn d 1.5 10 4 1.0 10 4 5.0 10 3 complete to i = 22.5 0 0.0 0.5 1.0 1.5 Helby - DESpec Workhop - 3/10
tet with realitic pectrocopic ditribution in each field: photometric ditribution pectrocopic field/telecope pointing on the ky cae 1 mimic 1 AAT/AA pointing in one DECam pointing t 40 hour Ω i = 22.5 Helby - DESpec Workhop - 4/10 cae 2 mimic 4 Magellan/IMACS pointing in one DECam pointing t 4 hour cae 3 mimic 16 VLT/VIMOS pointing in one DECam pointing t 16 hour i = 22.5 i = 22.5
tet with realitic pectrocopic ditribution 2 cae 1 0.0>>1.4 2 cae 2 2 cae 3 1 φ p 1 φ p 1 φ p 0 0 0-1 1.5 true newman -1 1.5 true -1 1.5 true newman fractional error 1.0 0.5 fractional error 1.0 0.5 fractional error 1.0 0.5 0.0 0.0 0.5 1.0 1.5 Helby - DESpec Workhop - 5/10 0.0 0.0 0.5 1.0 1.5 0.0 0.0 0.5 1.0 1.5
cluter and cro-correlation cluter photo: no outlier δ 0.01 0.02 can ue thee object a a puedo-pectrocopic ample ditributed over entire 5000 q. deg. Helby - DESpec Workhop - 6/10 from Roo, Redmapper method
cluter and cro-correlation cluter photo: no outlier δ 0.01 0.02 can ue thee object a a puedo-pectrocopic ample ditributed over entire 5000 q. deg. 1.5 Newman Anat Helby Anat Photometric Spectrocopic recontruction 1.5 φ p 1.0 0.5 0.0 fractional difference 1.0 0.5-0.5 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Helby - DESpec Workhop - 6/10 0.0 0.0 0.2 0.4 0.6 0.8 1.0 1.2
uing nearby pectrocopic galaxie a redhift prior DEC at =? pectrocopic galaxy photometric galaxy at 2 at 1 RA P() old new output from photometric code P ZADE () = ZADE (ued in COSMOS) ee: Kovac et al. 2009 N gal (<R ZADE,)P photometry () 0 N gal (<R ZADE, )P photometry ( )d Helby - DESpec Workhop - 7/10
uing nearby pectrocopic galaxie a redhift prior DEC at =? pectrocopic galaxy photometric galaxy DEC at 2 at =? at 2 at 1 at 1 P() old new RA output from photometric code P() 1 old new RA P i ( ) w(θ)dω each probability relate to 2pt function P ZADE () = ZADE (ued in COSMOS) ee: Kovac et al. 2009 Helby - DESpec Workhop - 7/10 N gal (<R ZADE,)P photometry () 0 N gal (<R ZADE, )P photometry ( )d 2 modified method: ue clutering to compute a P() eparately for each pectrocopic galaxy N nearby gal P combined () =P photometry () P i ( ) i=1
uing nearby pectrocopic galaxie a redhift prior: example true = prior 25 20 15 _true: 0.205485 _ade: 0.43000000 _jen: 0.21000000 ann: 0.384260 carlo: 0.550000 ANN ZADE Carlo ZADE P() Jen P() Carlo Carlo P() True Nearby Gal P() 10 5 0 0.0 0.5 1.0 1.5 Helby - DESpec Workhop - 8/10
tet in DESpec 2 0.3 0.2 Jen Zade Carlo ANN tet uing DES v3.04 mock (220 q. deg.): 2.3% galaxie: pectrocopic 2.3% = 7 million 0.1 pectrocopic DES galaxie / (300 million photometric DES galaxie ) ab( phot - true )/(1+ true ) 0.30 0.25 0.20 0.15 0.10 0.05 Jen Zade Carlo ANN 0.0 0.0 0.2 0.4 0.6 0.8 1.0 1.2 0.00 0.0 0.2 0.4 0.6 0.8 1.0 1.2 ab( phot - true ) 0.4 0.3 0.2 0.1 Jen Zade Carlo ANN 0.0 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Helby - DESpec Workhop - 9/10
ummary galaxy clutering can provide additional information, allowing more accurate photometric redhift and redhift ditribution applied to DES, the ample variance iue due to LSS in pectrocopic training et can be alleviated by uing the cro-correlation method, uing available pectrocopic data and poibly the DES cluter population DESpec allow u to ue nearby galaxie a a prior - particularly ueful at low redhift (can reduce outlier rate for thee object) where photo are le accurate due to DES lack of a u-band thank for litening! Helby - DESpec Workhop - 10/10
Extra
comparion with ZADE 2 0.3 0.2 0.1 Jen Zade Carlo ANN ab( phot - true )/(1+ true ) 0.30 0.25 0.20 0.15 0.10 0.05 Jen Zade Carlo ANN 0.0 0.0 0.2 0.4 0.6 0.8 1.0 1.2 0.00 0.0 0.2 0.4 0.6 0.8 1.0 1.2 ab( phot - true ) 0.4 0.3 0.2 0.1 Jen Zade Carlo ANN 0.0 0.0 0.2 0.4 0.6 0.8 1.0 1.2
an ideal tet photometric ample: all object within the DES mock catalog pectrocopic ample: random ubample of 10% of the object in the DES mock catalog (mimic an idealied DESpec galaxy redhift urvey) Helby - DES Munich - 6/12
an ideal tet truth recontruction 0.6 newman φ p fractional error 0.4 0.2 all 70 field 0.0 0.0 0.5 1.0 1.5 Helby - DES Munich - 7/12 no photometric information i ued here
teting the photo- elect galaxie in bin of photometric redhift and recontruct the true redhift ditribution in that bin recontructed from 220 q. deg. uing repreentative pectrocopic ample 0.3 < phot < 0.4 true phot 0.7 < phot < 0.8 true phot φ p φ p Helby - DES Munich - 11/12
photometric redhift calibration for DES many training et exit already within the DES footprint Stripe 82 imaging urvey pectrocopic training et deep urvey intermediate urvey infrared urvey SZ urvey
iue 1: incompletene training et not complete down to DES photometric limit i=24 partially mitigate thi problem by getting more redhift at faint end: propoed a program of multilit pectrocopy to improve the pectrocopic training et on Magellan/IMACS (Frieman, Lin, Keler & Helby)
iue 2: ample variance ample variance effect due to fluctuation in LSS in pectrocopic training et Cunha et al. (2011)
iue 2: ample variance knowledge of true photometric ditribution can reduce followup requirement: currently do not know thi, want to find thi P ( i j p)=p ( j p i ) N i N j p Eqn 21 Cunha et al. (2011) known P ( i j p) impact of LSS greater on than P ( j p i )
iue 2: ample variance knowledge of true photometric ditribution can reduce followup requirement: currently do not know thi, want to find thi P ( i j p)=p ( j p i ) N i N j p Eqn 21 Cunha et al. (2011) Contamination Object in a photometric bin that are from the wrong pectrocopic bin known P ( i j p) impact of LSS greater on than P ( j p i )
iue 2: ample variance knowledge of true photometric ditribution can reduce followup requirement: Leakage Object in a pectrocopic bin placed into the wrong photometric bin currently do not know thi, want to find thi P ( i j p)=p ( j p i ) N i N j p Eqn 21 Cunha et al. (2011) Contamination Object in a photometric bin that are from the wrong pectrocopic bin known P ( i j p) impact of LSS greater on than P ( j p i )
pectrocopic dataet from DES cience verification document (docdb-6255):
an ideal tet truth recontruction φ p φ p all 70 field ingle field no photometric information i ued to recover thee ditribution
the newman method method to recontruct the true redhift ditribution of a ample by uing it clutering propertie require two ample: pectrocopic ample and photometric ample ample mut overlap in redhift and on the ky le enitive to incompletene than redhift etimation baed on photometry method; thu i particularly ueful at faint end, where pectrocopy i difficult/expenive can only recontruct the ditribution, not individual galaxy redhift Newman (2008)
the newman method: why doe thi work? increaing redhift object with pectrocopic redhift photometric object
the newman method: why doe thi work? meaure w p (θ, pec ) (in bin of pectrocopic redhift) increaing redhift object with pectrocopic redhift photometric object
the newman method: what going on here meaure w p (θ, pec ) (in bin of pectrocopic redhift) w p (θ, pec )= ξ p φ p ()d olve for φ p () φ p () 0 dn p ddω dn p d dω d
the newman method: what going on here meaure w p (θ, pec ) (in bin of pectrocopic redhift) w p (θ, pec )= we meaure thi ξ p φ p ()d we want thi olve for φ p () φ p () 0 dn p ddω dn p d dω d
the newman method: what going on here meaure w p (θ, pec ) (in bin of pectrocopic redhift) w p (θ, pec )= we meaure thi ξ p φ p ()d we want thi aume ξ p ξ which we can meaure in the pectrocopic ample olve for φ p () φ p () 0 dn p ddω dn p d dω d