Observational evidence for Dark Energy The historical probe : Type Ia supernovae Combined SN+BAO+CMB constraints on CDM models Looking beyond CDM SN 1994D GDR Terascale 31 March 2010

Accelerated expansion : Type Ia SNe, 1998 0.01 0.1 1 z (further back in time) FAINTER  matter only SCP: 42 high-z events + High-z search team : 10 high-z events MORE REDSHIFT (more expansion since explosion) ~ apparent magnitude = -2.5log10(L /4πdL2) S.Perlmutter et al., 1999, ApJ, 517, 565 & A.Riess et al., 1998, AJ, 116, 1009

Cosmology with type Ia SNe 1995/99: first experiments with distant SNe Ia: lack of photometric precision low statistics O(50) no way to test if SNe Ia are really standard candles 2003/10: second-generation experiments dedicated experiments with optimised detection and follow-up strategies and improved photometric precision larger statistics O(500) test "standard candle" hypothesis distant SNe: SNLS, ESSENCE, CSP… nearby SNe: CfA, CSP, SDSS, SNFactory…

From 1st to 2nd generation SN experiments High-z SN search team SNLS 03D4dh z=0.627 early detection better temporal sampling more filters more precise photometry

1st year SNLS results P.Astier et al., 2006, A&A, 447, 31 fainter residuals to (0.26,0.74): int: 0.13± 0.02 Low-z: 0.15±0.02 High-z: 0.12±0.02 Accelerated expansion of the Universe : confirmed (71 SNe)

Union compilation, 2008 SNLS 3yr, preliminary 414 published SNe Ia 307 after selection cuts M. Kowalski et al., 2008, ApJ, 686, 749 SNLS 3yr, preliminary 252 high-z SNe Ia 242 evts combined with 228 good quality low-z, SDSS, HST SNe Union compilation: blind analysis (LambdaCDM cosmology fitted, SN fluxes rescaled from not reported best-fit to arbitrarily reference cosmo 0.25/0.75). Colour, stretch and residual distributions are preserved -> set up selection criteria against outliers (large impact on fit result) t o remove contamination or unmodeled intrinsic variations -> 3sigma clipping about median -> reduces magnitude bias (below 0.015 mag) and dispersion (0.16->0.14 mag) SNLS-3: Alex july draft : 135 low-z (dominated by CfA events), 97 SDSS, 16 HST events. New high quality data at all redshifts -> restrict high quality samples only A. Conley et al., in prep.

CDM fits from SN data Low-z, SDSS, SNLS-3, HST Preliminary Contours with systematics included. Better control of systematics: light curve fitter (model, residual scatter, stat of training sample), photometric calibration, potential evolution of beta (luminosity-colour relation) evolution -> SNLS dat alone: 0.069 syst on Om (in a flat LamndaCDM fit) with 0.048 (calib) and 0.026 (model), 0.034 (stat of training sample), 0.010 (residual scatter) and 0.022 (beta evolution) Cosmlogical results from SN data: now limited by systematics (photometric calibration + understanding of SNIa diversity) Statistical precision improved Better control of systematics A. Conley et al., in prep

Combined constraints on CDM SNe Ia: SN distances dL(z) for z2 CMB: amplitude and location of acoustic peaks in CMB spectrum  distance to the surface of last scattering at z=1089 BAO: location of comoving galaxy separation at z=0.2,0.35 BAO: constrains the distance-redshift relation at multiple epochs z=0.2, 0.35. Distance is rs(zd)/Dv(z) where rs is the comoving sound horizon at the baryon drag epoch and Dv(z) is the comoving separation of pairs of galaxies related to primordial BAO oscillations Cosmology dependence: through H(z) primarily + direct dependence on Om

Generic cosmology models CDM: dark matter, curved Universe, cosmological constant  m,k (or ) wCDM: dark matter, flat Universe, dark energy with constant equation of state  m,w owCDM: similar to wCDM with curvature allowed  m ,k,w w(z)CDM or ow(z)CDM: similar to wCDM/owCDM but with time-dependent dark energy equation of state, e.g. excellent approximation to a wide variety of dark energy models (scalar fields…)

Constraints from Union ‘08 SN set M. Kowalski et al., 2008, ApJ, 686, 749 CDM wCDM SNe: Union 08 / CMB: WMAP-5 year Komatsu et al, 2008 / BAO : SDSS first measurement Eisenstein et al 2005 : stat + syst combined CMB: constraint on the reduced distance to the surface of last scattering at z=1089 assuming a constant DE EOS -> to allow for a more general reinterpretation of this constraint, correct for possible deviation in the sound horizon due to an NON matter dominated early universe SDSS: distance constraint at z=0.35, similarily corrected for a possible deviation in the sound horizon : syst assumed to be negligible / sta w= -0.969-0.063-0.066 +0.059+0.063 NB: CMB = WMAP-5, BAO = SDSS 2005

Constraints from WMAP-7 E.Komatsu et al., arXiv:1001.4538, submitted owCDM w(z)CDM WMAP-7: measurement of H0 from low-z SNe (Riess et al 2009) and measurement of rs/DV(z) at a pivot redshift of 0.275 (Percival 2010) used as priors in WMAP fits. DE fits use the constitution set of Hicken at al 2009 in addition (Union set + new redshift CfA SNe): the absolute magnitudes of SNe Ia are marginalized over with a uniform prior. Full analysis of WMAP data (Markov chain MC method). Other constraint: time-delay distance to the lens system B1608+656 owCDM: WMAP+BAO primarily constrain Ok and WMAP+SN primarily constrain w. With SN data: error in w is 4 times smaller (w=-1.44+-0.27 WMAP+BAO+H0) wCDM (flat) to owCDM: relaxing flatness hypothesis => increase of stat uncertainty on w by 0.003 Time-dependent EOS in a flat Universe: addition of BAO helps to exclude models with large negative values of w_a. The time-delay distance information helps further for w_a (gain is 0.05 in precision) wCDM: w= -0.980.053(stat) owCDM: w= -0.999-0.056 (stat) +0.057 w0=-0.930.12(stat) wa= -0.38-0.65 (stat) +0.66 NB: SN = updated Union set, BAO = SDSS 2010

Constraints beyond CDM D. Rubin et al., 2009, ApJ, 695, 391 Data sample: SNe Ia: Union 08 compilation CMB: WMAP-5year BAO: SDSS (Eisenstein 2005) A wide variety of DE models explored (extra dimensions, scalar fields, geometric dark energy…). Three examples: DGP braneworld gravity Geometric dark energy Algebraic thawing model

DGP Braneworld gravity two-parameters: m and k Poor fit to combined data (wrt CDM fit) k Acceleration caused by a weakening of gravity near the Hubble scale due to leaking into an extra dimensional bulk from our four dimension brane -> infrared moidifcation of gravity instead of a physical dark energy component Om and Ok related to the effective braneworld energy density Obw given by H0 and r_c the five dimensional crossover scale (MPl^2/2M_5^3) DGP: Dvali, Gabadadze and Porrati (2000) Poorer fit to data than LambdaCDM (Deltachi2 = +2.7 with syst, +15 without syst) Unphysical region (Ok=1-Om open, non-accelerating universe): infinite crossover scale r_c

Geometric dark energy Rlow two-parameters: m and w0 Good fit to combined data w0=-1 favored but distinct physics from CDM CMB and BAO constraints close: SNe play a valuable role Acceleration related to spacetime curvature varying with the scale factor a. In the low scalar curvature regime, variation of the reduced Ricci scalar curvature around its present value + matching at high redshift to an asymptotic matter-dominated behaviour. R0: value of the reduced Ricci scalar at present (ie for a=1) Better fit to data than LambdaCDM (Deltachi2 = -1.1 with syst)

Algebraic thawing model three-parameters: m, w0,p Good fit to combined data p[0,3] w0<-0.57 (95%CL) Models difficult to distinguish from CDM Require probes sensitive to w0 and wa Minimally coupled scalar field evolving from a matter dominated era (with a slow roll behaviour to avoid fine tuning or too steep potentials – at leading and nexto-to-leading order) The field departs from a high redshift cosmological constant behaviour, evolving toward a less negative EOS : w0 present value of EOS parameter, p sets the rapidity of the field evolution. Shaded: physically expected range 0,3 p<0: evolution of w from -1 to its leat negative value and back to -1 already over, the extreme value of w being closer to 0 as p is more negative P>0: the field takes longer to thaw, increasing its similarity to a cosmological contant unetil recently and then rapidly evolves to w0 Better fit to data than LambdaCDM (Deltachi2 = -2.3 with syst)

Conclusions Type Ia SNe: precision much improved recently (more statistics, better control of systematics) Current data (SN+BAO+CMB) are consistent with CDM and with a wide diversity of dark energy models: static dark energy is not necessarily the true answer To explain the nature of dark energy, need for next generation observations/probes sensitive to w(z)

Cosmology with SNe Ia spectroscopy triggered  type Ia confirmed, z=0.627 i’, r’, z’, g’ filters multi-band LC’s : test compatibility with SNIa model (trained on SNIa lightcurves and spectra) Accuracy of SN redshift measurements: 0.001 from host spectrum, 0.01 from SN spectrum About 40% of our detections get spectroscopic feedback, 60% of these are confirmed SNe Ia Measurement of maximum light magnitudes within 3%, points in the rise and fall of the curve to secure c and s difference wrt average light-curve modelled by two parameters : colour (intrinsic/dust extinction) and stretch (time dilation factor or shape variability) apparent B magnitude : mB*, (B-V) colour C and stretch s J.Guy et al., 2007, A&A, 466, 11

reproducible luminosity SNe Ia are (assumed to be) standard candles : reproducible luminosity mB* from B*  L(C,s) /4πdL2 with dL(z,H0,ΩM,ΩΛ,w,..) Distance estimator : B = mB* - MB + (s-1) - C reference absolute B-band magnitude apparent rest-frame maximum light magnitude (B-V) colour variability (intrinsic variation and extinction) lightcurve shape variability SNe Ia are assumed to be standard candles: objects of known luminosity at a given distance > their flux (or magnitude) reveal the geometry of the Universe Alpha (luminosity-stretch) and beta (luminosity-colour) corrections reduce the scatter from 0.5mag to 0.15mag -> distances precise to 6% + intrinsic dispersion term, int allowed in cosmological fits to account for our lack of knowledge about Sne besides alpha and beta corrections All SNe Ia treated in the same way: OK if no residual difference with redshift after correction (same  and  at all redshifts): MB, ,  assumed to be z-independent. Is that so ?  compare properties of SN sub-samples split by redshift, host activity…

B-5log10dL M.Sullivan et al., submitted to MNRAS Corrected SNIa luminosities depend on host galaxy mass: SNe Ia in massive galaxies appear brigther than those in less massive galaxies (>3)

add a host specific term Mix of SNe Ia from different hosts changes with redshift SN demographic shifts to avoid bias in cosmological results add a host specific term in cosmological fits Host galaxies at higher redshifts contain on average less stellar mass (young, less-evolved galaxies with high start forming activites) SNLS M.Sullivan et al., submitted to MNRAS

Flat CDM fits from SN data (ΩM+ΩΛ=1, w=-1): SNLS-3yr data alone: ΩM= 0.211±0.034±0.069 Better control of systematics: empirical LC model 0.026 model training stat 0.034 residual scatter 0.010 Photometric calibration 0.048 Potential  evolution 0.022 SNLS-3 Uncertainties: stat + syst due to light curve fitter (dotted), photometric calibration (dashed), model training statistical uncertainties and Malmquist bias correction uncertainties (thin solid line), possible beta evolution (thick solid line) J. Guy et al., submitted to A&A

No evolution in  (luminosity-shape relation) From the SNLS-3 sample No evolution in  (luminosity-shape relation) Possible evolution in  (luminosity-colour relation)  accounted for in SN systematics  SALT2 LC fitter o SiFTO LC fitter J. Guy et al., submitted to A&A

Testing w(z)CDM with the Union ’08 set Sharp cut-off at w0+wa=0: models above this line do NOT yield a matter-dominated early Universe, in conflict with BAO and CMB observations Inclusion of curvature does not modify the contours NB: CMB = WMAP-5, BAO = SDSS 2005

Next improvement: SNLS-3 M.Sullivan et al., in prep Preliminary wCDM w m w (stat+syst)  0.08 NB: CMB = WMAP-5, BAO = SDSS 2005

Constraints from SDSS w= -0.970.10 W.Percival et al., 2010, MNRAS, 401, 2148 WMAP5+BAO+SN owCDM w= -0.970.10 No big change vs including SDSS 2005 different statistical treatment different CDM model systematics not included WMAP5+SN SDSS 2010: constraints on the comoving separation distance at two epochs z=0.2, 0.35 Combined with constraints from SN (Union 08 set) and CMB (WMAP-5: full likelihood of WMAP used). CMB: constrains Omh^2 in all CDM models, low redshift information is needed to constrain Om and H0 separately. Allowing w diff from -1 and curvature: CMB+BAO fits degrade -> need for SN data Statistical test: likelihood analysis, no details given on how SN data are treated (priors or full info, systematics ?) WMAP5+BAO

The energy balance of the Universe Flat Universe Open Closed M+=1- k measuring the geometry of the Universe CMB BAO AT A GIVEN DISTANCE Known physical size angle depends on geometry Known luminosity flux depends on geometry SNe Ia Omega_photon ~ 2.47 10^-5 (negligible) Friedman equation cluster surveys measuring the energy content of the Universe cluster distributions DM distribution weak lensing