Cosmology with TypeIa Supernovae Ramon Miquel Lawrence Berkeley
Cosmology with Type-Ia Supernovae Ramon Miquel Lawrence Berkeley National Laboratory and ICREA / IFAE, Barcelona IRGAC, July 11 -15 2006, Barcelona July 13, 2006 Ramon Miquel IRGAC 2006
Outline • • • Type-Ia SNe as cosmological tools Cosmological analysis Systematic uncertainties Current surveys: SNLS Future surveys: SNAP Summary July 13, 2006 Ramon Miquel IRGAC 2006 2
Universe Constituents and Dynamics Type-Ia SNe probe dark energy through the history of the expansion rate • Friedmann-Lemaître Equations (GR + homogeneity and isotropy): a : scale factor r : energy density p : pressure k : curvature • After specifying a equation of state p = p(r) for each component: H 2(z) = H 20. [ M (1+z)3 + DE (1+z)3(1+w)] matter • dark energy , a = (1+z)-1 flat universe, constant w = p/r Measuring the history of the expansion rate, H(z), we can learn about the universe constituents, M, DE, w. July 13, 2006 Ramon Miquel IRGAC 2006 3
Probing Dark Energy with Type-Ia SNe • Standard candles provide a measurement of the luminosity distance as a function of redshift: f : flux L : intrinsic luminosity d. L : luminosity distance r(z) : co-moving distance (geometric test of dark energy) • Astronomers measure the apparent magnitude and redshift: – – M is the (assumed unknown) absolute magnitude of a type-Ia SN. H 0 d. L does NOT depend on H 0 July 13, 2006 Ramon Miquel IRGAC 2006 4
Type-Ia Supernovae (I) • Defined empirically as supernovae without Hydrogen but with Silicon in spectrum. • Progenitor understood as a white dwarf accreting material from a binary companion. • As the white dwarf approaches Chandrasekhar mass, a thermonuclear runaway is triggered. • A naturally triggered and standard bomb. July 13, 2006 Ramon Miquel IRGAC 2006 5
Type-Ia Supernovae (II) • General properties: – Homogeneous class of events: luminosity, color, spectrum at maximum light. Only small (correlated) variations – Rise time: ~ 15 – 20 days – Decay time: ~ 2 months – Bright: MB ~ – 19. 5 at peak • No hydrogen in the spectra: – Early spectra: Si, Ca, Mg, . . . (absorption) – Late spectra: Fe, Ni, …(emission) • SN Ia found in all types of galaxies, including ellipticals – Progenitor systems must have long lifetimes July 13, 2006 Ramon Miquel IRGAC 2006 6
Discovering Supernovae July 13, 2006 Ramon Miquel IRGAC 2006 7
Are Type-Ia SNe Standard Candles? apparent magnitude → distance → time July 13, 2006 Ramon Miquel IRGAC 2006 8
Type-Ia SNe as Standardizable Candles • Nearby (z < 0. 1) supernovae used to study SNe light curves • Brightness not quite standard • Intrinsically brighter SNe last longer • Correction needed Peak-magnitude dispersion of 0. 25 – 0. 30 mag • After correction, standard ~ 0. 10 – 0. 15 mag dispersion candles in optical region (5 – 7% precision in distance) (at least). July 13, 2006 Ramon Miquel IRGAC 2006 9
Near-Optical Bands g July 13, 2006 r Ramon Miquel i IRGAC 2006 z 10
Near-Optical Bands l → l·(1+z) z = 0. 5 U B z = 1. 0 July 13, 2006 V U Ramon Miquel R B IRGAC 2006 I V R 11
Type-Ia SN Spectral Features • Spectra at near maximum light are used to determine type of SN (Si-II feature) • And to measure the redshift, z, by observing the shift in the spectrum Si-II July 13, 2006 Ramon Miquel IRGAC 2006 12
SN Analysis Light Curves Images Redshift & spectral properties M and L w and wa Spectra Data July 13, 2006 Analysis Ramon Miquel IRGAC 2006 Science 13
Hubble Diagram July 13, 2006 Ramon Miquel IRGAC 2006 14
Discovery of Acceleration July 13, 2006 Ramon Miquel IRGAC 2006 15
D(m-M) (mag) High-z Results redshift Riess et al. 2004; also Knop et al. 2003 • Expansion went from deceleration to acceleration • Exclude simple gray dust models July 13, 2006 Ramon Miquel IRGAC 2006 16
Current Surveys (300) July 13, 2006 Ramon Miquel IRGAC 2006 17
Systematic Errors • Statistical error is dominated by intrinsic SN peak magnitude dispersion sint = 0. 10– 0. 15 • Many systematic errors will be totally correlated for SNe at similar redshifts – Current and near-future surveys will have O(100) SNe for Dz = 0. 1 redshift bin. – Therefore, systematic errors of order sint/√NSN = 0. 01– 0. 02 will already become important or even dominant. July 13, 2006 Ramon Miquel IRGAC 2006 18
Sources of Systematic Errors Error Source Control Host-galaxy dust extinction* Wavelength-dependent absorption identified with high S/N multi-band photometry. Supernova evolution * Supernova sub-classified with high S/N light curves and peak-brightness spectrum. Flux calibration error * Program to construct a set of 1% error flux standard stars. Malmquist bias Supernova discovered early with high S/N multi-band photometry. K-corrections * Construction of a library of supernova spectra. Gravitational lensing Measure the average flux for a large number of supernovae in each redshift bin. Non-Type-Ia contamination Classification of each event with a peak-brightness spectrum. Kim, Linder, Miquel, Mostek, MNRAS 347 (2004) 909 July 13, 2006 Ramon Miquel IRGAC 2006 19
Extinction by Dust (I) Dust in the path between the SN and the telescope attenuates the amount of light measured • Milky Way dust is well measured and understood (Schlegel, Finkbeiner & Davis 1998) 2600 citations • Host galaxy extinction leads to reddening of supernova colors: AV = RV · E(B-V) AV : increase in magnitude in V band E(B-V): excess in B-V color over expected RV ≈ 3. 1 in nearby galaxies • In another band j, the extinction is (Cardelli, Clayton & Mathis 1989) 1400 citations known (≈ 0 -0. 10) What is the value of RV in distant galaxies? July 13, 2006 Ramon Miquel IRGAC 2006 20
Extinction by Dust (II) Different approaches to RV determination: • Riess et al. 2004 (HZT) assume RV = 3. 1, as it appears to be in the local universe. Include exponential prior on AV. Bias? • Astier et al. 2006 (SNLS) instead determine one RV value for all their high-z SNe, coming up with a much lower value RV = 0. 57 ± 0. 15 – Their RV effectively includes any other effect that might correlate SN color and magnitude. • SNAP will determine RV for each SN independently. – Needs at least 3 bands for each SN July 13, 2006 Ramon Miquel IRGAC 2006 21
Dust Biases No extinction (e. g. only SNe in ellipticals) Extinction corrected With AV bias With AV and RV biases Current data quality Linder & Miquel 2004 w(z) = w 0 + (1 -a) wa Linder 2003 July 13, 2006 Ramon Miquel IRGAC 2006 22
Gray Dust? Simple gray dust models excluded Some contrived models are just indistinguishable from LCDM D(m-M) (mag) • Gray dust would be dust that does not lead to any measurable reddening (equivalently, RV → ∞) • Therefore, it’s not correctable with the usual methods. • “Natural” models would lead to a dimming of SNe at all redshifts. Riess et al. 2004; also Knop et al. 2003 July 13, 2006 Ramon Miquel IRGAC 2006 23
K-corrections l → l·(1+z) z = 0. 5 U B V R I • At high z, one needs to relate measured fluxes in, say, R, I, z filters with fluxes in SN rest frame B, V, R bands. ≈ O(0. 5 mag) • Good empirical model for SN spectrum from B to z is needed. July 13, 2006 Ramon Miquel IRGAC 2006 24
Calibration • Calibration ≡ determining the “zero-points” f 0, j of each filter j • Overall normalization is irrelevant • Relative filter-to-filter normalization is crucial (K-corrections, dustextinction corrections) Standard scal = 0. 005 Standard scal = 0. 001 Self scal = 0. 005 Alternative procedure using also SN data themselves achieves a large degree of self-calibration wa Standard procedure uses wellunderstood stars to get scal = 0. 01 at best * (Kim & Miquel 2006) Example for SNF + SNAP (300 + 2000 SNe up to z = 1. 7) 68% CL contours w 0 July 13, 2006 Ramon Miquel IRGAC 2006 Kim & Miquel 2006 25
Current SNe Surveys SNLS ESSENCE SDSS-II / SNe Super. Nova Factory July 13, 2006 Ramon Miquel IRGAC 2006 26
The Super. Nova Legacy Survey (SNLS) • Ongoing (2003 -2008) SN survey using CFHT (Mauna Kea): – 3. 6 m aperture – 1 deg 2 field of view – 328 Megapixel camera (Mega. Cam) • Photometry for 40 nights/yr during 5 years. – 4 -night cadence rolling search in four 1 -deg 2 fields in g, r, i, z bands. • Expect to discover 500 -700 type-Ia SNe up to z = 1. • Spectroscopic follow-up of most good SN candidates in VLT, Gemini, Keck… July 13, 2006 Ramon Miquel IRGAC 2006 27
SNLS Data z = 0. 36 z = 0. 91 day z = 0. 285 July 13, 2006 Ramon Miquel IRGAC 2006 28
SNLS Analysis • Light-curve fit performs K-corrections and returns mi. B at peak (SN rest frame), stretch si and color excess Ei(B-V). • Every available filter is used in fit, provided it corresponds to U, B, V, R in SN rest frame. • At least two filters are required. • The cosmology fit then proceeds as: x : free parameter q : cosmological params. • 44 published nearby (z < 0. 1) SNe and 73 new high-z SNe are used in the fit. • Statistical errors dominate now, but systematic errors will dominate with final sample. – Main systematic error: calibration. July 13, 2006 Ramon Miquel IRGAC 2006 29
SNLS Hubble Diagram magnitude (B-band) + constant First-Year SNLS Hubble Diagram 73 high-z SNe sint = 0. 12 mag redshift July 13, 2006 Ramon Miquel IRGAC 2006 Astier et al. 2006 30
SNLS Results SNLS + BAO (Einsestein et al. 2005) SNLS + WMAP-3 (Spergel et al. 2006) Flat universe assumed July 13, 2006 Ramon Miquel IRGAC 2006 31
Next Generation SNe Surveys from the Ground (2008 -2012) Pan-STARRS DES LSST July 13, 2006 Ramon Miquel IRGAC 2006 32
SNe without spectroscopy? • Next generation SN surveys from the ground will gather about 2000 type-Ia SNe with redshifts up to z = 1. 2. – Practically impossible to get spectroscopy for all those SNe. • Is it possible to do cosmology with type-Ia SNe without spectroscopy? – Redshift determination • Photometric redshifts • Host galaxy redshift? – Typing • Typing from goodness of light-curve fit. – Systematic tests: ? ? ? July 13, 2006 Ramon Miquel IRGAC 2006 33
SNLS Photo-z’s and Photo-z Typing Sullivan et al. 2006 ◊ Fail c 2 cut – Photo-z’s: • <|zphot - zspec|> = 0. 03*(1+z) assuming cosmology known. • But small dependency on the assumed cosmological values. M=0. 25, L=0. 75 – Photo-typing: • 90% purity using a real-time analysis of pre-maximum light curves. • Presumably, it can be improved using all light-curve information. July 13, 2006 Ramon Miquel M = 1, L = 0 IRGAC 2006 34
Future SNe Surveys from Space (2013 -2016) JDEM/Destiny JDEM/SNAP JDEM/JEDI DUNE July 13, 2006 Ramon Miquel IRGAC 2006 35
Why Space? • Precision on wa increases by going to z > 1 SNAP simulation • Window into deceleration (z > 1) era can help with syst. errors. • For z > 1 -1. 2, rest-frame B band redshifts into observer IR region (l > 1. 2 mm) • Atmospheric absorption is large in IR region Need space-based telescope Miquel 2004 July 13, 2006 Ramon Miquel IRGAC 2006 36
The SNAP Satellite • 2 m-class wide-field telescope with state of the art optical and NIR camera and spectrograph – Collect about 2000 type-Ia SNe with z < 1. 7 – Study weak lensing from space • Could fly in ~2013. Part of JDEM (DOE/NASA) competition. July 13, 2006 Ramon Miquel IRGAC 2006 37
D=56. 6 cm (13. 0 mrad) 0. 7 square degrees Guider NIR SNAP Focal Plane Visible Fixed filters atop the sensors Focus star projectors Integral Field Spectrograph port July 13, 2006 Calibration projectors Ramon Miquel IRGAC 2006 38
SNAP (and DES) Optical Detectors • New LBNL technology: thick back-illuminated CCD detector. • Better red response (up to l = 1 mm) than “thinned” CCDs devices in use at most telescopes. • High-purity silicon has better radiation tolerance for space applications. July 13, 2006 Ramon Miquel IRGAC 2006 39
(Some) SNAP Systematics • Dust extinction: – Measure each SN in 9 → 3 (low to high z) filters – Can determine AV and RV for each SN independently. • Evolution: – Properties of SNe that correlate with luminosity can change with z – Get precise spectrum at maximum light for all SNe – Classify SNe according to sub-type. This needs a large database of nearby SNe with good photometry and spectra (SNF) – Perform cosmology fits within sub-types including low- and high-z SNe (“like-to-like” comparison). – In practice, allow for several Mi in cosmology fit (one for each subtype). – Statistical degradation because of extra parameters is only few % (Kim, Linder, Miquel, Mostek 2004) July 13, 2006 Ramon Miquel IRGAC 2006 40
SNAP Reach • For a fiducial LCDM model – w 0 measured to 10% w’ ≈ wa / 2 – w’ ( ≈ wa / 2) to 10% • Better for most other models (more sensitive to late-time dark-energy) • Big improvement after adding weak lensing w 0 July 13, 2006 Linder 2005 Ramon Miquel IRGAC 2006 41
Summary • Type-Ia SNe provided the “smoking gun” for acceleration. • Mature technique still being perfected. • Control of systematic errors key to future improvements. • Vigorous current and future program: • Low-z from ground: SNF, SDSS-II/SNe, Cf. A, Carnegie… • Medium- to high-z from ground: Essence, SNLS, DES, Pan. STARRS, LSST • High-z from space: HST, JDEM, DUNE Expect more insight on the nature of Dark Energy from type-Ia SNe studies July 13, 2006 Ramon Miquel IRGAC 2006 42
- Slides: 42