A New Measurement of the Expansion Rate of the Universe And a Path to 1% with Gaia

Transcription

1 Johns Hopkins University Space Telescope Science Institute A New Measurement of the Expansion Rate of the Universe And a Path to 1% with Gaia based on Riess et al. 2016, ApJ, 826,56, The SH0ES Team

2 Expanding Universe reveals Composi3on, Age, Fate The Past Redshift (~a -1 ) Distance (~ct) Homogeneous, Isotropic + GRà equation of expansion a(t), scale factor Depends on present state and composition of Universe

3 Cosmology, quest for for two Two numbers (maber dominated) Cosmology:The The Quest Fundamental Numbers H 0=. a a Present rate, size, age, q 0= t=t0.. -a ah02 Deceleration by ΩM (=2q0), geometry, fate origin, viability of inflation t=t0 dense, fast big, 20 Gyr empty, slow small, 10 Gyr Scale: a 1990 s: Building of Modern SN Ia Hubble diagram

4 SN Ia and Cepheids: Standardized Candles for long range, rela%ve distances Type Ia Supernovae, Exploding Stars, 10 9 L dust Cepheids, Pulsating Stars, 10 5 L Bright=near faint=far but not all the same faint & red=not so far! SN Ia Luminosity-color-light curve shape correlations 1990 s

5 Building the Modern SN Ia Hubble Diagram; Hubble Flow relative Mid-1990 s, CCDs, light curve, reddening corrections m or magnitude = 5 log Distance + c 5 mag=100x fainter/ 10 farther Established: tight scatter (σ D ~6%), linear local expansion, H 0 ~10%-15% severely limited by luminosity calibration

6 Building the Modern SN Ia Hubble Diagram; Accelera3on! 1998: SN discoveries at z~0.5 from large format CCDs m or magnitude = 5 log Distance + c q ~100 SN Ia Established: q 0 <0!, presence of dark energy, no known unknown

7 The Modern SN Ia Hubble Diagram; Confirm, Characterize 2004-present: Massive ground surveys z<1 and z>1 with HST m or magnitude = 5 log Distance + c >1300 SN Ia 7 z=1.34 Established: not astrophysical dimming (grey dust, evolution), decelerating before accelerating, looks like lambda to ~10%

8 Why Does the Universe Appear to be Accelerating? q 0 = Ω M 2 + (1+ 3w) Ω DE 2 w p/ρ w=1 radiation w=0 matter w=-1 vacuum energy (3D) 1. Vacuum Energy, the cosmological constant: ΛCDM Constant energy of empty space (e.g.,higgs), feature of QM+GR Test: w(z)=-1 2. Dynamical dark energy A scalar field fills space (e.g., inflation-lite ), rolling down potential Test: w 0-1 or dw/dz 0 3. Modification to GR GR fails at long range (i.e., as a 1) Test: w(z) depends on scale (i.e., different in H(z) and g(z)) New Game; assume Model=plain ΛCDM and hunt for departures

9 Ul3mate End- to- end test for ΛCDM, Predict and Measure H0 The Standard Model of Cosmology, ΛCDM Big Bang CMB, z=1000 Sound σd(λcdm)=0.4% Falsifying CDM Horizon Planets 0.05% Planets+ Stars+Gas? CMB determination of matter density controls all determinations the deceleration (matter dominated) epoch 25% inplanck: h = ± ! 1.7% m 2 Distance to recombination D determined to % 0.43% ( CDM result 0.46%; h/h m h2 / m h2 ) Dark [more general: 0.11 w 0.48 ln h 0.15 ln 1.4 ln = 0 ] Expansion rate during any redshift in the deceleration epoch Energy m ΛCD determined to % epoch determined as? Distance to any redshift in the deceleration Z dz D(z) = D 70% H(z) H(z) z Gas 4% M Stars 0.5% tot z 2 d dz/h(z) Volumes determined by a combination dv = DA Structure also determined by growth of fluctuations from z Now z=0, σh=1% 0

10 A Program to Measure H 0 to percent precision The SH0ES Project (2005) (Supernovae, H 0 for the dark energy Equation of State) A. Riess, L. Macri, S. Casertano, A. Filippenko, S. Hoffman, S. Jha σ w 2 σ H (after WMAP, Planck) Measure H 0 to percent level precision through: A clean, simple ladder: Geometry Cepheids SNe Ia Reducing systematic error with better data, better collection Thorough propagation of statistical and systematic errors

11 Our route: 3 Steps to H0 1 (Kpc) 2 (Mpc) 3 (Gpc) Old vs New Parallax Limit 1:Geometryà Cepheids 2: Cepheidsà SN Ia 3:SN Ia à z,h0

12 Step 1: Extending Parallax with WFC3 Spa3al Scanning Riess, Casertano, Mackenty et al (2014) WFC3-UVIS, 0.01 pixel=0.4 mas 2 kpc Imaging, PSF σ θ =0.01 pix Scanning, σ θ =0.01/ N samples pix Max parallax Max parallax

13 Pilot Program MW Cepheid SY >2 kpc, Oct 2011 Start scan 150 scan 7x zoom Then we extract scan lines, row-by-row End scan

14 Two Features of Spa3al Scans: Repe33on and JiBer Removal Average all scan lines to produce oversampled reference line Jitter between lines is coherent, subtracted in line separations (vs time), approach is doubly differential Target scanned over ~4000 pix, Improves SNR by up to ~40 (or 10 for correlated errors on scales of 40 pix), Reaching µas

15 The millipixel (40 µas) challenge: scanning systematics millipixels Staring Scanning (Statistics) Parallax at 2 kpc Intra-scan Velocity Aberration Variable Jitter Intra-scan Rotation (FGS) Inter-scan Rotation Inter-scan Distortion ("Breathing") Total Final Precision Scanning Systematic Scanning & Staring Systematics A lot moves at the millipixel level!

16 New Anchors: Parallaxes of Milky Way Cepheids, WFC3 SS Proper Motion subtracted, Parallax measurements field stars & Cepheid Cepheid At 3 kpc, 4 years σ=27 µas Also: NGC 4258 Masers 3.0%à 2.6%, LMC DEBs, M31 DEBs, Future: 20 New MW Parallaxes, Gaia Riess et al. (2014) Casertano et al. (2016) Reduction to Absolute: e.g., Red giant at 8 Kpc, 0.4 mag spec πà 25 µas

17 Step 2: Cepheids to Type Ia Supernovae 1.Discard photographic data, SN 1895B, 1937C, 1960F 1974G, & unreliable i.e. all but 2-3 used by Key Project, Sandage et al., Freedman et al. Previously limited by 20 Mpc Cepheid range of WFPC2 SN 1960F m 20 Mpc

18 Step 2: BeBer Data- SN Ia ACS (2002),WFC3 (2009) doubled range, 8X as many possible Using only reliable SN Ia data-i.e., digital data, multiband light curves 4 bands, pre-max, normal spectra, low extinction A V <0.5). 19 m

19 Step 2: Increasing the Sample of SN Ia Calibrators New HST Program w/ WFC3 NIR+wide filters 3x speed, n=8à n=19 NEW! NEW! NEW! NEW! NEW! NEW! NEW! NEW! NEW! NEW! NEW!

20 Lower Systema3cs with Differen%al Flux Measurement Homogeneous Cepheid data for anchors and SN Hosts: same instruments, same filters, same metallicity, period range ANCHOR: NGC 4258, 2.6% geometric distance Cepheid composite LC s, > SN Ia Hosts Δ mag Δ mag Δ mag PHASE PHASE PHASE PHASE PHASE PHASE

21 Lowering Systema3cs with Near- IR Cepheid Observa3ons Inclination è crowding of more distant host Random phase HST PHAT Survey; Riess et al. 2012, Wagner-Kaiser et al 2015

22 Cepheid NIR Period- Luminosity Rela3onships: 19 hosts, 4 anchors 4 Anchors

23 Step 3: Intercept of SN Ia Hubble Diagram: Distance vs Redshie a B 0.2m 0 B

24 Simultaneous Fit: Retain interdependence of data and parameters Cepheids in SN hosts Cepheids in Anchors SN Ia Measurements = Regression Matrix Geometric Distance Priors * Free Parameters Absolute Host Distances ΔD (N4258) Cepheid Luminosity ΔD(LMC) ΔD(M31) P-L slope (P>10 days) SN Ia Luminosity Metallicity, Cepheid Zeropoint, LMC P-L slope (P<10 days) Error Matrix [ σ 2 tot,1,j,.. σ2 tot,19,j, σ2 tot,n4258,j, σ2 tot,m31,j,σ2 tot,mw,j +σ2 π,j, σ 2 tot,lmc,j, σ2 mb,1,.. σ2 mb,19, σ2 zp, σ2 µ,n4258,σ2 µ,lmc ] 5log H 0 =M B0 +5a B +25

25 The Hubble Constant in just 3 Steps..reduced systema3cs 300 SNe 19 Calibrations 15 Parallaxes + 3 anchors H0= /- 1.74, Km s-1 Mpc-1 2.4% total uncertainty

26 Analysis Variants, Anchors Variants Preferred Four anchors One anchor Optical only Anchor Value (km/sec/mpc) NGC4258+MW+LMC /-1.74 NGC4258+MW+LMC+M /-1.71 NGC 4258: Masers /-2.51 MW: 15 Cepheid Parallaxes /-2.37 LMC: 8 Late-type DEBs /-2.67 M31: 2 Early-type DEBs /-3.27 No NIR: N4258+MW+LMC /-2.49 Cardona,Kunz, Pettorino 2016 Hyper-parameter Analysis (R16) /-2.26 (2297 HP) No Hyper-parameters /-1.51

27 Analysis Variants, Addi3onal Variants Systema3cs 23 Variants: Reddening law, metallicity, SN fibers, z range, Host types, PL breaks, PL range Outliers, etc

28 Error Budget for H 0 from 2016; 2.4% uncertainty H 0 = / km/s/mpc (Riess et al. 2016) TERM KP LMC % R % R11 ALL % ANCHOR DISTANCE R16 ALL3 * % CEPHEID REDDENING, ZEROPOINTS (ANCHOR-TO-HOSTS) P-L SLOPE, D LOG P (ANCHOR-TO-HOSTS) CEPHEID METALLICITY DEPENDENCE (ANCHOR-TO-HOSTS) WFPC2 CTE, LONG-VS-SHORT ZEROPOINTS MEAN OF SN IA CALIBRATORS MEAN OF P-L IN ANCHOR MEAN OF P-L IN SN HOSTS SN IA M-Z RELATION ANALYSIS SYSTEMATICS NA TOTAL, % ERROR H How does this compare to the CMB measurements?

29 H 0, Measured vs. Predicted from Ini3al Condi3ons (CMB) New Physics? Dark Energy Dark Radiation Curved Space DM-Rad Interaction Δ σ= σ tension! - New Physics*? - Planck vs WMAP vs l? HOLiCOW-3 lenses (2.5 σ) * If a persuasive case can be made that a direct measurement of H 0 conflicts with these estimates, then this will be strong evidence for additional physics beyond the base ΛCDM model. -Planck Team Paper, 2015

30 How Big is the Difference between our H0 and Planck CMB+ΛCDM? 300 SNe 19 Calibrations H0= /- 1.74, Km s-1 Mpc-1 15 Parallaxes + 3 anchors Planck +ΛCDM Δ=0.2 mag 2.4% total uncertainty

31 Local- to- Global Varia3on in H 0? Is local H 0 (0.02<z<0.15) same as global H 0? empirically model H(z) w/ kinematic terms q 0, j 0 derived from high-z SN Ia correct z for local (peculiar) flows derived from 2M++ density field (Carrick et al. 2015) Test: explore larger volume, z min <z<z min +0.15, ΔH 0 < 0.4% N-body sims in 700 Mpc box à 0.3% (Odderskov et al. (2016) +/-0.5%

32 4 G. E. Addison et al. Evidence for a systema3c error in the Planck CMB data? MC ch2 MC ns H0 65 H Ase Ase H0 ns Amse log(10 H0 As) 2 9 m Addison, Huang, Watts, Bennett, Halpern, Hinshaw, Weiland 2016, ApJ, 818, Planck Team, arxiv: σ 0.11 like 1.8 σ for 6 parameters, but we measure H0! ch2 2 bh MC Claimed 2.5 σ Tension Between Halves of Planck CMB (WMAP) data, l>1000 vs l< s 0.13 et al. G. E. Addison TT Planck ` TT TT `< ` < 1000 Planck TT 2015Planck 1000 `TT 2508 Planck he < 1000 and 68.3% ` 1000 Planck CDM TT spectra. Here 2. ` Marginalized confidence parameter constraints from fits to the ` < 1000 and ` 1000 Planck TT spectr Figure 2. Marginalized 68.3% confidence CDM parameter constraints from fits to the ` < 1000 and `

33 What Next? Gaia will provide independent anchor and beber precision H 0 +Cepheid P-L relation (R16) predicts MW parallaxes σ ~20µas, 10x precision of Gaia DR1 for ~210 Cepheids Casertano, Riess, Bucciarelli, Lattanzi arxive: , submitted A&A (DR1+2 days) DR1 parallaxes good (zp=1±19µas), errors overestimated ~20%, average of 210à H 0 =73.0 km/sec/mpc, σ tension with Planck

34 H 0 to 1%? Anchored by 60 Best Milky Way Long Period Cepheids HST photometric system, F555W (V), F814W (I), F160W (H) F555W F814W F160W 08/07/ /26/ /08/ /07/ /09/ /27/2016 Fast Scanning 7.5 /sec exp time~0.01 sec Gaia(2022), σ π <10µas Gaia to measure 60 best MW Cepheids w/ σ π <10µas parallaxes, Net: σ D =3% per Cepheid (with scatter) à 3% / sqrt(60)=0.4%=σ H MUST measure 60 MW Cepheids on HST zpt or will add σ H =1.5%!

35 Reaching 1% in H 0...2% in w, 20% in w, in Ω K, 0.08 in N eff 1.7% 2.2% Cepheid scatter, slope Cepheid distance calibration 1.0% 1.0% 0.4% 0.4% 0.5% 0.5% 1.2% Hubble flow SNe in hosts 0.5% 0.7% 2.1% 2011: 3.3% 8 SN hosts Short-period Galactic Cepheids + NGC : 1.8% 19 SN hosts 20+ Galactic Cepheids With HST Parallax, Photometry 2022(?): 1.0% ~50 SN hosts (HST+JWST) 60+ Galactic Cepheids (Gaia astrometry + HST photometry) Powerful complement to CMB Stage IV (Manzotti et al. 2016) And stay alert for systematics along the way

36 Takeaways Universe now appears to be expanding ~9% (+/- 2.4%) faster -than-expected based ΛCDM+Planck CMB This is surprising! If not an error, could be a vital clue pertaining to the 95% of the Universe (i.e., the dark sector) we don t understand. We anticipate significant improvements in these measurements in just the next few years which may reveal the cause of the now peppy Universe. With additional measurements HST, Gaia and JWST can enable a 1% measurement of H 0