French Group Toronto Group Reynald Pain, Pierre Astier, Ray - PowerPoint PPT Presentation

French Group Toronto Group Reynald Pain, Pierre Astier, Ray Carlberg, Mark Julien Guy, Nicolas Regnault, Victoria Group Sullivan[->Oxford], Andy Jim Rich, Stephane Basa, Chris Pritchet, Dave Howell, Kathy Perrett, Dominique Fouchez Balam Alex Conley UK Gemini PI: Mark Sullivan, Isobel Hook Richard McMahon Richard McMahon USA LBL: Saul Perlmutter CIT: Richard Ellis, Don Neill ��������������������������������������������������������������

Dark Energy � Einstein � GR � E Eqns metric constant � 6 � 6 DE ~ (1+z) 3(1+w) , w=-1 ~ (1+z) 3(1+w) , w=-1 � Scaler Field Theory � Inflation inspiration � Quintessence � Interesting values near w=-0.8

Huterer & Turner 2000 From LSS � = � � � ρ � Slightly modified version of figure in Huterer & Turner 2000 (astro-ph/0012510)

4

SNLS1w=-1.02±0.09 (stat)

SNLS1 (Astier 06) weaknesses � Sample of only 72 at z>0.1 � Patchy 4 filter data � Calibration somewhat uncertain � Sample of ~40 at z<0.1 � Sample of ~40 at z<0.1 � Unknown selection function � Landolt Vega calibration (1950’s) � Systematic errors not fully accounted

WMAP Beam recalibration

SDSS/2dF, BAO

SNLS Third Year

244 spectro’ed Ia to Aug 07

<w>=- <w>= -0.982 0.982 ± ± 0.044, 0.044, Ω Ω M M =0.267 =0.267 ± ± 0.014 0.014 <w>= <w>= - - 0.982 0.982 ± ± 0.044, 0.044, Ω Ω =0.267 =0.267 ± ± 0.014 0.014 M M Factor of 3 below SNLS1 Factor of 3 below SNLS1 Ω Factor of 3 below SNLS1 Factor of 3 below SNLS1 Ω Ω Ω M M ! ! ! ! M M About factor of 2 better than early SNLS3 About factor of 2 better than early SNLS3 About factor of 2 better than early SNLS3 About factor of 2 better than early SNLS3

SALT2/SIFTO Distance Model � First, fit the Light Curves � No cosmology except redshift � Separates the LC problem from cosmology � m* = m B + α(s!1) + βc , � O=m*-M has all of the cosmology � O=m*-M has all of the cosmology � m B is the B brightness at peak � s is stretch � c is roughly a U-B color � α,β,M are fitted “nuisance parameters” � No compelling evidence so far of anything more. � Need more colors to be measured (IR, UV)

Light Curve fitters � SALT2 : many parameter fit � Trained at low-z � Color relations built in (all filter fits) � SIFTO : SCP heritage, advanced � SIFTO : SCP heritage, advanced � LC’s defined at low z � Each filter independent

SIFTO LC’s and errors

Light Curve Fitters (low redshift training, same high redshift A06 data) stretch varies with wavelength—redder slower Method RMS (mag) comment SALT SALT 0.179 0.179 A06 (SNLS1) A06 (SNLS1) SALT2 0.159 SiFTO 0.160 MLCS2k2 0.205 Hubble Bubble “on”

What is New? � More, better SNLS data (~10x) � More, better low-z data (CfA, SDSS) � Calibration to BD+17 4708 (F0 sD) � Both Megaprime and nearby data � Both Megaprime and nearby data � Removes an R and I band problem � LC stretch depends on wavelength � U-B color to predict B-V statistically � Massive MC simulations of biases

SNLS color-color relations important to determine “c”

Errors for Hubble diagam fits 6 error (diagonal) + 6 systematics matrices Usual minimization: χ 2 = ∑ [(m i – m(b i ,s i ,c i )/σ i ] 2 � i = m i – [b i +α(s i �1)+βc i ] Generalized to matrix: χ 2 = � T V �1 � , whereV error matrix Statistical : V stat = σ bb + α 2 σ ss +β 2 σ cc +aσ 2 Statistical : V = σ + α 2 σ +β 2 σ +aσ 2 bs + +βσ 2 + +βσ 2 bc +aβσ 2 +aβσ 2 s Systematic : � Each source of error traced through to distance model � Many are correlated (e.g. filter calibration effects most distances) � Creates a nearly filled matrix (NxN is number of supernova) � Need to iterate because α ,β determined in fit. Leads to 6 statistical and 6 systematic error matrices

Cosmological constraints (flat) Cosmological constraints (flat)

SNLS3 prelim w(a) analysis

Color Nightmare 2: β � Color not understood: � Temperature, � Metallicity, � Explosion Details and � Explosion Details and � Dust. � Intrinsic + Extrinsic � Dust β~4.1 � Dustless β~2 � Redshift variation (selection effects…)

Warning: Bogus evolution! (Emma makes no mistakes) (Emma makes no mistakes) Sample Sne become bluer and brighter with increasing redshift—gives many false trends that are due to selection.

Where next? � Ia continue to be surprisingly good � Astrophysical interest � Much potential for cosmological refinement � σ w ~ 2-3σ mags , σ w of 2-3% should be possible � Careful design ( JDEM + low z Hubble flow) � Multi-color (next step 3 c’s) � Multi-color (next step 3 c’s) � UV diversity? � Dust � metallicity � Calibration (low z, high z, space) � Note: same data fit other cosmologies about equally well (e.g. DGP)