Transcript Cosmology with Supernovae: Lecture 2 Josh Frieman I Jayme Tiomno School of Cosmology, Rio de Janeiro, Brazil July 2010

Cosmology with Supernovae:
Lecture 2
Josh Frieman
I Jayme Tiomno School of Cosmology,
Rio de Janeiro, Brazil
July 2010
1
Hoje
• V. Recent SN Surveys and Current
Constraints on Dark Energy
• VI. Fitting SN Ia Light Curves &
Cosmology
• VII. Systematic Errors in SN Ia Distances
2
Coming Attractions
•
•
•
•
•
VIII. Host-galaxy correlations
IX. SN Ia Theoretical Modeling
X. SN IIp Distances
XI. Models for Cosmic Acceleration
XII. Testing models with Future Surveys:
Photometric classification, SN Photo-z’s, &
cosmology
3
Luminosity
m15
15 days
Time
Empirical Correlation: Brighter SNe Ia decline more slowly
and are bluer
Phillips 1993
Brighter  Slower,
Bluer
Use to reduce
Peak Luminosity
Dispersion:
peak
Peak Luminosity
SN Ia Peak Luminosity
Empirically correlated
with Light-Curve
Decline Rate and
Color
M i  c i  f i (m15 )
in rest- frame passbandi
Phillips 1993
Garnavich, etal
Rate of decline
Peak brightness
correlates with
decline rate
Variety of algorithms
for modeling these
correlations:
corrected dist. modulus
After correction,
~ 0.16 mag
(~8% distance error)
Luminosity
Type Ia SN
Peak Brightness
as calibrated
Standard Candle
Time
6
Published Light Curves for Nearby Supernovae
Low-z SNe:
Anchor Hubble
diagram
Train Lightcurve fitters
Need wellsampled, wellcalibrated,
multi-band light
curves
7
Low-z Data
8
Correction for
Brightness-Decline
relation reduces scatter
in nearby SN Ia
Hubble Diagram
Distance modulus for
z<<1:
 v 
m  M  5log
 5logh  15
km/sec
Corrected distance modulus is
not a direct observable:
estimated from a model for
light-curve shape
Riess etal 1996
9
Acceleration
Discovery Data:
High-z SN Team
10 of 16 shown;
transformed to SN
rest-frame
V
B+1
Riess etal
Schmidt etal
10
Riess, etal High-z Data (1998)
11
High-z Supernova
Team data (1998)
12
Likelihood Analysis
2ln L   2  
(i  mod (zi ;m , ,H 0 ) 2
 2,i
i
ˆ  mod (H 0  70)
Since mod  5log(H 0 dL )  5log(H 0 ), let 
ˆ . If we fix H 0 , then we are minimizing
and define i  i  
ˆ 

2
i
2i
 i2
T o marginalize over log
H 0 with flat prior,we instead minimize
This assume


 C 
B2
2
ˆ 
 mar  2ln d(5logH 0 )exp  /2 
 ln ,
2 
C




where
2
B
i
i
 i2
, C
i
2
1
 i2
 2 ,i   2 , fit   2 ,int   2,vel
Goliath etal 2001
13
High-z SN Team
Supernova Cosmology Project
14
1998-2010 SN Ia Synopsis
• Substantial increases in both quantity and quality of SN
Ia data: from several tens of relatively poorly sampled
light curves to many hundreds of well-sampled, multiband light curves from rolling surveys
• Extension to previously unexplored redshift ranges:
z>1 and 0.1<z<0.3
• Extension to previously underexplored rest-frame
wavelengths (Near-infrared)
• Vast increase in spectroscopic data
• Identification of SN Ia subpopulations (host galaxies)
• Entered the systematic error-dominated regime, but
with pathways to reduce systematic errors
15
Supernova Legacy Survey (2003-2008)
Observed 2 1-sq deg regions every
4 nights
~400+ spectroscopically confirmed
SNe Ia to measure w
Used 3.6-meter CFHT/“Megacam”
36 CCDs with good blue response
4 filters griz for good K-corrections
and color measurement
Spectroscopic follow-up on 8-10m
telescopes
Megaprime Mosaic CCD camera
16
Magellan
VLT
Spectra
SN Identification
Redshifts
120 hr/yr: France/UK
FORS 1&2 for types,
redshifts
3 nights/yr: Toronto
IMACS for host
redshifts
Gemini
Keck
8 nights/yr:
LBL/Caltech
DEIMOS/LRIS
for types,
intensive study,
cosmology with
SNe II-P
120 hr/yr:
Canada/US/UK
GMOS for types, redshifts
Power of a Rolling Search
SNLS Light curves
SNLS 1st Year Results
First-Year SNLS Hubble Diagram
Astier et al.
2006
Using 72 SNe
from SNLS
+40 Low-z
19
Wood-Vasey, etal (2007), Miknaitis, etal (2007):
results from ~60 ESSENCE SNe (+Low-z)
20
60 ESSENCE SNe
72 SNLS SNe
21
22
Higher-z SNe Ia from ACS
Z=1.39
Z=0.46
Z=0.52
Z=1.23
50 SNe Ia, 25 at z>1
Z=1.03
Riess, etal
(m-M)
HST GOODS Survey (z > 1) plus
compiled ground-based SNe
Riess etal 2004
24
Supernova Cosmology Project
SN Ia Union Compilation
Data tables and updates at
http://supernova.lbl.gov/Union
Kowalski et al., ApJ, 2008
2ln Pposterior  
i
(i  mod (zi ;w,m ,DE ) 2
  ,i
2
2
2
  BAO
  CMB
where lat ter terms incorporate BAO
and CMB priors:
Likelihood
Analysis with
BAO and CMB
Priors
BAO (SDSS LRG, Eisenstein etal 05)
:
2/3
z1




m
1
dz 
1/ 2

A(z1;w,m ,DE ) 
S




k
k
1/ 3
E(z1 ) 

0 E(z) 
z1 k 

with
 BAO2  [(A(z1;w,m ,DE )  0.469) /0.017]2 for z1  0.35
CMB (W MAP 5,Komatsu et al 08):
z
m   1/ 2 CMB dz 
Sk k
R(zCMB ;w,m ,DE ) 



k 
0 E (z) 
 

with
 CMB 2  [(R(zCMB ;w,m ,DE ) 1.710) /0.019]2 for zCMB  1090
26
Recent Dark Energy
Constraints
Improved SN constraints
Inclusion of constraints from
WMAP Cosmic Microwave
Background Anisotropy
(Joana) and SDSS Largescale Structure (Baryon
Acoustic Oscillations; Bruce,
Daniel)
assuming w = −1
Only statistical errors shown
27
28
assuming flat Univ.
and constant w
Only statistical errors shown
29
SNLS Preliminary 3rd year Hubble Diagram
Conley et al, Guy etal (2010): results with ~252 SNLS SNe
Independent analyses with 2 light-curve fitters: SALT2, SiFTO
Frieman, et al (2008); Sako, et al (2008)
Results published from 2005 season
Kessler, et al 09; Lampeitl et al 09; Sollerman et al 09
SDSS II Supernova Survey Goals
• Obtain few hundred high-quality* SNe Ia light curves in the
`redshift desert’ z~0.05-0.4 for continuous Hubble diagram
• Probe Dark Energy in z regime complementary to other
surveys
• Well-observed sample to anchor Hubble diagram, train
light-curve fitters, and explore systematics of SN Ia
distances
• Rolling search: determine SN/SF rates/properties vs. z,
environment
• Rest-frame u-band templates for z >1 surveys
• Large survey volume: rare & peculiar SNe, probe outliers of
population
*high-cadence, multi-band, well-calibrated
32
Spectroscopic follow-up telescope
R. Miquel, M. Molla, L. Galbany
Searching For Supernovae
Search
g
Template
Difference
• 2005
– 190,020 objects scanned
– 11,385 unique
candidates
– 130 confirmed Ia
• 2006
r
– 14,441 scanned
– 3,694 candidates
– 193 confirmed Ia
• 2007
– 175 confirmed Ia
i
•Positional match to remove movers
•Insert fake SNe to monitor efficiency
SDSS SN Photometry
Holtzman etal
(2008)
35
B. Dilday
500+ spec confirmed SNe Ia + 87 conf. core collapse
plus >1000 photometric Ia’s with host z’s
Spectroscopic Target Selection
2 Epochs
SN Ia Fit
SN Ibc Fit
SN II Fit
Sako etal 2008
Spectroscopic Target Selection
2 Epochs
31 Epochs
SN Ia Fit
SN Ia Fit
SN Ibc Fit
SN Ibc Fit
Fit with
template
library
Classification
>90%
accurate after
2-3 epochs
Redshifts
5-10%
accurate
SN II Fit
SN II Fit
Sako etal 2008
SN and Host Spectroscopy
MDM 2.4m
NOT 2.6m
APO 3.5m
Determine
NTT 3.6m
KPNO 4m
SN Type
WHT 4.2m
and
Subaru 8.2m
Redshift
HET 9.2m
Keck 10m
Magellan 6.5m
TNG 3.5m
SALT 10m SDSS 2.5m
2005+2006
Spectroscopic Deconstruction
SN model
Host galaxy model
Combined model
Zheng, et al (2008)
Fitting SN Ia Light Curves
• Multi-color Light Curve Shape (MLCS2k2)
Riess, etal 96, 98; Jha, etal 2007
• SALT-II
Guy, etal 05,08
41
MLCS2k2
Light-curve
Templates
in rest-frame j=UBVRI;
built from ~100
well-observed, nearby SNe Ia
∆ <0: bright, broad
∆ >0: faint, narrow,
redder
time-dependent model “vectors”
trained on Low-z SNe
observed
passband
i
mmod
(t  t 0 )    M j (t  t 0 )  P j (t  t0 )  Q j (t  t 0 )2
j
i
 K ij (t  t 0 )  X host
(t  t 0 )  X MW
(t  t 0 )
fit parameters
Time of maximum
distance modulus
host gal extinction
stretch/decline rate
Host Galaxy Dust Extinction
•Extinction:
 f obs () 
A  2.5log

 f true ( ) 
•Empirical Model for
wavelength dependence:
A
b(  )
 a( ) 
AV
RV
•MLCS: AV is a fit parameter,
but RV is usually fixed to a
global value (sharp prior) since
it’s usually not well
determined SN by SN
Cardelli etal 89 (CCM)
43
Host Galaxy Dust Extinction
Historically,
MLCS used
Milky Way
average of
RV=3.1
Growing
evidence that
this doesn’t
represent SN
host galaxy
population
well
Milky Way avg.
Jha
44
Extract RV by matching colors of
SDSS SNe to MLCS simulations
RV
AV

2
E(B  V )
• Use nearly complete
(spectroscopic +
photometric) sample
• MLCS previously
used Milky Way
avg RV=3.1
• Lower RV more
consistent with
SALT color law and
other recent SN RV
estimates
D. Cinabro
Carnegie Supernova Project:
Low-z
n
CSP is a follow-up project
n
Goal: optical/NIR light-curves and
spectro-photometry for
n
> 100 nearby SNIa
n
> 100 SNII
n
> 20 SNIbc
n
Filter set: BV + u’g’r’i’ + YJHK
n
Understand SN physics
n
Use as standard candles.
n
Calibrate distant SN Ia sample
CSP Low-z Light Curves
Folatelli, et al. 2009
Contreras, et al. 2009: 35 optical light curves (25 with NIR)
Varying Reddening Law?
2005A
2006X
Folatelli et al. (2009)
Local Dust?
Goobar (2008):
higher density of
dust grains in a
shell surrounding
the SN: multiple
scattering
steepens
effective dust law
(also Wang)
Two Highly Reddened SNe
Folatelli et al. (2009)
Priors & Efficiencies

2
MLCS

i
F
i
data
 Fi
mod
(t 0 ,, AV , )

2
2
i
 2lnP(AV )P()(z, AV ,)
Determine priors and efficiencies from data and Monte Carlo simulations
Inferred P()
Inferred P(AV)
Priors & Efficiencies

2
MLCS

i
F
i
data
 Fi

mod
2
i, phot
(t 0,, AV , )

2
2
i,mod
 2lnP(AV )P()(z, AV ,)
Determine priors and efficiencies from data and Monte Carlo simulations
Model Spectroscopic & Photometric Efficiency
Redshift
distribution for all
SNe passing
photometric
selection cuts
(spectroscopically
complete sample)
Data
Need to model
biases due to
what’s missing
Difficult to model
spectroscopic
selection
Model Selection Function
Include Selection Function
Monte Carlo Simulations
match data distributions
Use recorded observing conditions (local sky, zero-points, PSF, etc)
56
Show likelihood plots for MLCS
MLCS fit to one of the
ESSENCE SNe
57
Marginalized PDFs
prior
μ distribution approximated by Gaussian for cosmology
58 fit
MLCS Likelihood Contours for this object
59