Recent results in the Supernova Legacy Survey SNLS
Patrick El-Hage
Rencontres de Blois, June 2015 c Z z dz0 d (z) = L 0 H0 0 H(z )/H0
2 2 X −3(1+wi ) H = H0 Ωi a i
µ = 5 × log10(dL) − 5 × log10(H0/c)
Luminosity distance evolves with redshift depending on the expansion rate. The expansion rate depends on cosmological parameters through the Friedmann equations. In practice use distance modulus µ.
Supernova Cosmology Recent Improvements Preliminary Results
The Hubble Diagram In a flat universe with constant equations of state, and a constant luminosity L for all objects considered :
L F = 2 4πdL
By definition, flux is related to luminosity distance.
2/15 2 2 X −3(1+wi ) H = H0 Ωi a i
µ = 5 × log10(dL) − 5 × log10(H0/c)
The expansion rate depends on cosmological parameters through the Friedmann equations. In practice use distance modulus µ.
Supernova Cosmology Recent Improvements Preliminary Results
The Hubble Diagram In a flat universe with constant equations of state, and a constant luminosity L for all objects considered :
L F = 2 4πdL
c Z z dz0 d (z) = L 0 H0 0 H(z )/H0
By definition, flux is related to luminosity distance. Luminosity distance evolves with redshift depending on the expansion rate.
2/15 µ = 5 × log10(dL) − 5 × log10(H0/c)
In practice use distance modulus µ.
Supernova Cosmology Recent Improvements Preliminary Results
The Hubble Diagram In a flat universe with constant equations of state, and a constant luminosity L for all objects considered :
L F = 2 4πdL
c Z z dz0 d (z) = L 0 H0 0 H(z )/H0
2 2 X −3(1+wi ) H = H0 Ωi a i
By definition, flux is related to luminosity distance. Luminosity distance evolves with redshift depending on the expansion rate. The expansion rate depends on cosmological parameters through the Friedmann equations.
2/15 Supernova Cosmology Recent Improvements Preliminary Results
The Hubble Diagram In a flat universe with constant equations of state, and a constant luminosity L for all objects considered :
L F = 2 4πdL
c Z z dz0 d (z) = L 0 H0 0 H(z )/H0
2 2 X −3(1+wi ) H = H0 Ωi a i
µ = 5 × log10(dL) − 5 × log10(H0/c)
By definition, flux is related to luminosity distance. Luminosity distance evolves with redshift depending on the expansion rate. The expansion rate depends on cosmological parameters through the Friedmann equations. In practice use distance modulus µ.
2/15 ? µB = mB - MB + α × stretch
- β × color Such a model allows for standardization corrections (i.e. implementing magnitude corrections based on color and stretch).
Compare the SN at same phase and wavelength. Fit model to data spanning both phase and wavelength.
Supernova Cosmology Recent Improvements Preliminary Results
How to compare standard candles ? L must be constant for all µ.
3/15 ? µB = mB - MB + α × stretch
- β × color Such a model allows for standardization corrections (i.e. implementing magnitude corrections based on color and stretch).
Fit model to data spanning both phase and wavelength.
Supernova Cosmology Recent Improvements Preliminary Results
How to compare standard candles ? L must be constant for all µ. Compare the SN at same phase and wavelength.
3/15 Such a model allows for standardization corrections (i.e. implementing magnitude corrections based on color and stretch).
Supernova Cosmology Recent Improvements Preliminary Results
How to compare standard candles ? L must be constant for all µ. Compare the SN at same phase and wavelength. Fit model to data spanning both phase and wavelength.
? µB = mB - MB + α × stretch
- β × color
3/15 Supernova Cosmology Recent Improvements Preliminary Results
How to compare standard candles ? L must be constant for all µ. Compare the SN at same phase and wavelength. Fit model to data spanning both phase and wavelength.
? µB = mB - MB + α × stretch
- β × color Such a model allows for standardization corrections (i.e. implementing magnitude corrections based on color and stretch).
3/15 + α × stretch - β × color
Add correction based on observed correlation of stretch and magnitude. Similarly for color.
Supernova Cosmology Recent Improvements Preliminary Results
−2.5 log10(F /F0)
? µB = mB - MB
Begin with naive estimator.
4/15 - β × color
Similarly for color.
Supernova Cosmology Recent Improvements Preliminary Results
−2.5 log10(F /F0)
? µB = mB - MB + α × stretch
Begin with naive estimator. Add correction based on observed correlation of stretch and magnitude.
4/15 Supernova Cosmology Recent Improvements Preliminary Results
−2.5 log10(F /F0)
? µB = mB - MB + α × stretch - β × color
Begin with naive estimator. Add correction based on observed correlation of stretch and magnitude. Similarly for color.
4/15 Supernova Cosmology Recent Improvements Preliminary Results
−2.5 log10(F /F0)
? µB = mB - MB + α × stretch - β × color
F = x0 × [ +x1× ]× ec×CL(λ)
Photometric data available in multiple bands. Properties extracted by fitting to a model: SALT2. Model parameters include apparent magnitude, color,& stretch.
Guy et al. 2007, “SALT2: using distant supernovae to improve the use of Type Ia supernovae as distance indicators”
4/15 2 Get a spectrum for identification and redshift. 3 Compute lightcurve. 4 Fit lightcurve model. 5 Construct Hubble diagram.
Supernova Cosmology Recent Improvements Preliminary Results
A review of the steps 1 Look for the Supernova.
5/15 3 Compute lightcurve. 4 Fit lightcurve model. 5 Construct Hubble diagram.
Supernova Cosmology Recent Improvements Preliminary Results
A review of the steps 1 Look for the Supernova. 2 Get a spectrum for identification and redshift.
5/15 4 Fit lightcurve model. 5 Construct Hubble diagram.
? µB = mB - MB + α × stretch - β × color
Supernova Cosmology Recent Improvements Preliminary Results
A review of the steps 1 Look for the Supernova. 2 Get a spectrum for identification and redshift. 3 Compute lightcurve.
5/15 5 Construct Hubble diagram.
Supernova Cosmology Recent Improvements Preliminary Results
A review of the steps 1 Look for the Supernova. 2 Get a spectrum for identification and redshift. 3 Compute lightcurve. 4 Fit lightcurve model.
? µB = mB - MB + α × stretch - β × color
5/15 Supernova Cosmology Recent Improvements Preliminary Results
A review of the steps 1 Look for the Supernova. 2 Get a spectrum for identification and redshift. 3 Compute lightcurve. 4 Fit lightcurve model. 5 Construct Hubble diagram.
5/15 123 118 ∼ 210 101 374 ∼ 380 231 239 ∼ 400 14 9 ∼ 10
Supernova Cosmology Recent Improvements Preliminary Results
Table : Evolution of the cosmology sample.
SNLS 1 SNLS 3 JLA SNLS5 Low z 44 SDSS 0 SNLS 73 HST 0
6/15 123 118 ∼ 210 101 374 ∼ 380 231 239 ∼ 400 14 9 ∼ 10
Supernova Cosmology Recent Improvements Preliminary Results
Table : Evolution of the cosmology sample.
SNLS 1 SNLS 3 JLA SNLS5 Low z 44 SDSS 0 SNLS 73 HST 0
SNLS 1 Motley crew of low redshift SN. Analysis of first year SNLS data. With BAO prior we get w = −1.023 ± 0.087
Astier et al. 2008, The Supernova Legacy Survey: Measurement of ΩM ,ΩΛ and w from the First Year Data Set
6/15 118 ∼ 210 374 ∼ 380 239 ∼ 400 9 ∼ 10
Supernova Cosmology Recent Improvements Preliminary Results
Table : Evolution of the cosmology sample.
SNLS 1 SNLS 3 JLA SNLS5 Low z 44 123 SDSS 0 101 SNLS 73 231 HST 0 14
6/15 118 ∼ 210 374 ∼ 380 239 ∼ 400 9 ∼ 10
Supernova Cosmology Recent Improvements Preliminary Results
Table : Evolution of the cosmology sample.
SNLS 1 SNLS 3 JLA SNLS5 Low z 44 123 SDSS 0 101 SNLS 73 231 HST 0 14
SNLS 3 Bigger SNLS sample. Add low redshift surveys. Add first SDSS data release. With WMAP and BAO prior we get w = −1.069 ± 0.07
Guy et al. 2010, The Supernova Legacy Survey 3-year sample: Type Ia Supernovae photometric distances and cosmological constraints Sullivan et al. 2011, SNLS3: Constraints on Dark Energy Combining the Supernova Legacy Survey Three Year Data with Other Probes Conley et al. 2011, Supernova Constraints and Systematic Uncertainties from the First Three Years of the Supernova Legacy Survey
6/15 ∼ 210 ∼ 380 ∼ 400 ∼ 10
Supernova Cosmology Recent Improvements Preliminary Results
Table : Evolution of the cosmology sample.
SNLS 1 SNLS 3 JLA SNLS5 Low z 44 123 118 SDSS 0 101 374 SNLS 73 231 239 HST 0 14 9
46 HST
C 44 β −
1 42 X
α SNLS
+ 40 ) G
( SDSS 38 − M B ? 36 m Low-z = 34 µ
0.4 .
CDM 0 2 Λ 0.0 µ .
− 0 2 −
µ 0.4 − 2 1 0 10− 10− 10 z
6/15 ∼ 210 ∼ 380 ∼ 400 ∼ 10
Supernova Cosmology Recent Improvements Preliminary Results
Table : Evolution of the cosmology sample.
SNLS 1 SNLS 3 JLA SNLS5 Low z 44 123 118 SDSS 0 101 374 SNLS 73 231 239 HST 0 14 9
JLA New calibration. Better understanding of systematics. With Planck and BAO prior we get w = −1.027 ± 0.055
Betoule et al. 2014, Improved cosmological constraints from a joint analysis of the SDSS-II and SNLS supernova samples.
6/15 Supernova Cosmology Recent Improvements Preliminary Results
Table : Evolution of the cosmology sample.
SNLS 1 SNLS 3 JLA SNLS5 Low z 44 123 118 ∼ 210 SDSS 0 101 374 ∼ 380 SNLS 73 231 239 ∼ 400 HST 0 14 9 ∼ 10
46 44 C β 42 − 40 1)
− 38 s (
α 36 + B ? 34 m 32 =
µ 30
28 2 1 0 10− 10− 10
) 1.5 0 H c L 1.0 d ( 0.5 10 0.0
5 log 0.5
− −
µ 1.0 2 1 0 − 10− 10− 10 z 6/15 Supernova Cosmology Recent Improvements Preliminary Results
Table : Evolution of the cosmology sample.
SNLS 1 SNLS 3 JLA SNLS5 Low z 44 123 118 ∼ 210 SDSS 0 101 374 ∼ 380 SNLS 73 231 239 ∼ 400 HST 0 14 9 ∼ 10
SNLS 5 In progress. New data alone shaves off ∼ 0.5%. New training and other improvements could remove another ∼ 0.5%. No preliminary value on w because analysis is blind. As an indicator, it will take half a billion dollars to lower it a further ∼ 2% (Euclid mission).
6/15 R T (λ)λdλ R φS (λ)T (λ)λdλ
Supernova Cosmology Recent Improvements Preliminary Results
φ(λ)
Flux 80
60
40
20
Supernova spectrum
0 λ 0 0.2 0.4 0.6 0.8 1
7/15 R φS (λ)T (λ)λdλ
Supernova Cosmology Recent Improvements Preliminary Results
R φ(λ)T (λ)λdλ
Flux 80
60
40
20
Supernova spectrum
0 λ 0 0.2 0.4 0.6 0.8 1
7/15 Supernova Cosmology Recent Improvements Preliminary Results
R φ(λ)T (λ)λdλ R φS (λ)T (λ)λdλ
Flux 80
60
40
Field star spectrum
20
0 λ 0 0.2 0.4 0.6 0.8 1
7/15 Supernova Cosmology Recent Improvements Preliminary Results
R φ(λ)T (λ)λdλ R φS (λ)T (λ)λdλ
Flux 80
60 Primary star spectrum
40
20
0 λ 0 0.2 0.4 0.6 0.8 1
7/15 Supernova Cosmology Recent Improvements Preliminary Results
R φ(λ)T (λ)λdλ R φS (λ)T (λ)λdλ
Flux 80
60 Primary star spectrum
40
20
0 λ 0 0.2 0.4 0.6 0.8 1
7/15 Supernova Cosmology Recent Improvements Preliminary Results
Regnault et al. 2006, Photometric Calibration of the Supernova Legacy Survey Fields
SNLS SDSS
MegaCam SDSS 2.5 m
LANDOLT PT patches
KPNO SDSS PT
HST
No direct comparison to primary standards. Reach ∼ 2% accuracy in absolute calibration.
8/15 Supernova Cosmology Recent Improvements Preliminary Results
Betoule et al. 2013, Improved photometric calibration of the SNLS and the SDSS supernova surveys
MegaCam SNLS SDSS
MegaCam MegaCam SDSS 2.5 m
LANDOLT PT patches
KPNO SDSS PT
HST
Direct observation of HST stars. Direct SNLS/SDSS cross-calibration. Reaches ∼ 0.5% accuracy in absolute calibration. Has become the default calibration scheme for CFHT.
8/15 Supernova Cosmology Recent Improvements Preliminary Results
In progress
MegaCam SNLS SDSS
MegaCam MegaCam SDSS 2.5 m
LANDOLT PT patches
KPNO SDSS PT
HST
∼ 10 times more observations comparing SNLS fields to HST stars. PanStarrs calibration allows for more consistency checks. Using DICE (a LED based instrument) to better measure our filter throughput. Changed the model used for interpreting primary star observations (from Hubeny to Rauch models).
8/15 Simulations guarantee linearity to (1.34 ± 2.72) × 10−4 Largest systematic impacting calibration is1 .3 × 10−4
Supernova Cosmology Recent Improvements Preliminary Results
Photometry Improvements New PSF statistically optimal photometry algorithm without image resampling.
Mi,p = {[fi × φref (~xp − ~xSN ) + galref ] ⊗ Ki }p + si
9/15
Astier et al. 2013, Photometry of supernovae in an image series: methods and application to the SuperNova Legacy Survey (SNLS) Largest systematic impacting calibration is1 .3 × 10−4
Supernova Cosmology Recent Improvements Preliminary Results
Photometry Improvements New PSF statistically optimal photometry algorithm without image resampling. Simulations guarantee linearity to (1.34 ± 2.72) × 10−4
Flux 80
60
40
20
0 λ 0 0.2 0.4 0.6 0.8 1 9/15
Astier et al. 2013, Photometry of supernovae in an image series: methods and application to the SuperNova Legacy Survey (SNLS) Supernova Cosmology Recent Improvements Preliminary Results
Photometry Improvements New PSF statistically optimal photometry algorithm without image resampling. Simulations guarantee linearity to (1.34 ± 2.72) × 10−4 Largest systematic impacting calibration is1 .3 × 10−4 Flux 80
60
40
20
0 λ 0 0.2 0.4 0.6 0.8 1 9/15
Astier et al. 2013, Photometry of supernovae in an image series: methods and application to the SuperNova Legacy Survey (SNLS) Changes to SALT2 model well within model uncertainty. Model uncertainty reduced ∼ 2 over most of the model range. Impact on cosmology well within SALT2 contribution to σw . Has largest impact on projected improvement on σw .
Reduces σw by ∼ 0.5%
Supernova Cosmology Recent Improvements Preliminary Results
Table : Expanding the training sample.
JLA SNLS 5 Lightcurves 424 ∼ 800 Spectra 482 ∼ 3000
10/15 Model uncertainty reduced ∼ 2 over most of the model range. Impact on cosmology well within SALT2 contribution to σw . Has largest impact on projected improvement on σw .
Reduces σw by ∼ 0.5%
Supernova Cosmology Recent Improvements Preliminary Results
Changes to SALT2 model well within model uncertainty.
Table : Expanding the training sample.
JLA SNLS 5 Lightcurves 424 ∼ 800 Spectra 482 ∼ 3000
10/15 Impact on cosmology well within SALT2 contribution to σw . Has largest impact on projected improvement on σw .
Reduces σw by ∼ 0.5%
Supernova Cosmology Recent Improvements Preliminary Results
Changes to SALT2 model well within model uncertainty. Model uncertainty reduced ∼ 2 over most of the model range.
Table : Expanding the training sample.
JLA SNLS 5 Lightcurves 424 ∼ 800 Spectra 482 ∼ 3000
10/15 Supernova Cosmology Recent Improvements Preliminary Results
Changes to SALT2 model well within model uncertainty. Model uncertainty reduced ∼ 2 over most of the model range. Impact on cosmology well within SALT2 contribution to σw . Has largest impact on projected improvement on σw .
Table : Expanding the training sample.
JLA SNLS 5 Lightcurves 424 ∼ 800 Spectra 482 ∼ 3000
Reduces σw by ∼ 0.5%
10/15 Supernova Cosmology Recent Improvements Preliminary Results
Scenario 1 Scenario 2
End to end Monte Carlo comparing recovered and input cosmologies. Scenario 3 Scenario 4 Test for impact of various modeling choices. No significant bias found.
Mosher et al. 2013, Testing Models of Intrinsic Brightness Variations in Type Ia Supernovae and Their Impact on Measuring Cosmological Parameters
11/15 When splitting SNIa by host galaxy mass, a ∼ 10−1 magnitude step appears.
Supernova Cosmology Recent Improvements Preliminary Results
Multiple studies hint at 2 separate SNIa populations, seemingly related to its environment. Kelly 2010, Sullivan 2010, Lampeitl 2010, Gupta 2011, D’Andrea 2011, Childress 2013, Rigault 2013
12/15 Supernova Cosmology Recent Improvements Preliminary Results
Multiple studies hint at 2 separate SNIa populations, seemingly related to its environment. Kelly 2010, Sullivan 2010, Lampeitl 2010, Gupta 2011, D’Andrea 2011, Childress 2013, Rigault 2013 When splitting SNIa by host galaxy mass, a ∼ 10−1 magnitude step appears.
12/15 For SNLS5 we’ve cross calibrated the mass estimates from different surveys.
Supernova Cosmology Recent Improvements Preliminary Results
K magnitudes used as proxy for galaxy mass when little reliable information is available.
Roman, Hardin et al. 2015 in prep.
13/15 Supernova Cosmology Recent Improvements Preliminary Results
K magnitudes used as proxy for galaxy mass when little reliable information is available. For SNLS5 we’ve cross calibrated the mass estimates from different surveys.
Roman, Hardin et al. 2015 in prep.
13/15 Supernova Cosmology Recent Improvements Preliminary Results
46 44 C β 42 − 40 1)
− 38 s (
α 36
+ For JLA analysis we had σw = 0.055 B ? 34 m 32 New SALT2 model improves σw by =
µ 30 ∼ 0.5%
28 2 1 0 10− 10− 10 New data improves σw by ∼ 0.2%
) 1.5 0 Still unclear what to expect from H c L 1.0 d ( 0.5 latest calibration improvements. 10 0.0
5 log 0.5
− −
µ 1.0 2 1 0 − 10− 10− 10 z
14/15 Outlook New probes are maturing and giving ever more precise constraints on w (BOSS, DES, LSST, Euclid, DESY etc. etc.) For SN to remain competitive in 2 generations we need to go to higher redshifts. Need space based observations to observe high redshift SN in IR.
Supernova Cosmology Recent Improvements Preliminary Results
Current state Final SNLS cosmology analysis underway. New training of SALT 2 model. Still expect to provide the best measurement of w to date.
15/15 Supernova Cosmology Recent Improvements Preliminary Results
Current state Final SNLS cosmology analysis underway. New training of SALT 2 model. Still expect to provide the best measurement of w to date.
Outlook New probes are maturing and giving ever more precise constraints on w (BOSS, DES, LSST, Euclid, DESY etc. etc.) For SN to remain competitive in 2 generations we need to go to higher redshifts. Need space based observations to observe high redshift SN in IR.
15/15