Automated analysis: SDSS, BOSS, GIRAFFE Tests with MILES spectra (R~2000) from the INT (Sanchez Blazquez et al. 2006) The same code (FERRE) Fitting data calibrated in flux and continuumnormalized
Software Gaussian LSF (fiber, wavelength) Quadratic interpolation of fluxes Normalization by blocks Successful tests performed on MILES library
Fittings to MILES spectra Tuesday, August 27, 2013
Continuum on This Work MILES parameters (Cenarro et al. 2009) [Fe/H] Teff Distributions of residuals logg
Continuum off This Work MILES parameters (Cenarro et al. 2009) [Fe/H] Teff Distributions of residuals logg
Recent trends in spectroscopic studies 3D model atmospheres: a beginning full NLTE: good progress for hot stars, but Data archival: survey projects going on with massive archives that become public (low-res: SDSS, SEGUE, GALEX) (high-res: Elodie, S4N) Analysis automation: a beginning Breaking the Z barrier
The Desirable future 3D model atmospheres full NLTE A pending observational test for solar-type stars: center-to-limb variation of the solar spectrum Data archival: VOs (including both observations and models) Stronger efforts to measure/compute atomic data Stronger efforts to use the newly available atomic data Full analysis automation R an ignored variable?
Galactic Surveys Other projects and analysis strategies Carlos Allende Prieto IAC
Overview eboss, BigBOSS (DESI), HETDEX, WEAVE, 4MOST Data mashup: astrometry, photometry, spectroscopy Observing Galaxy simulations Discovery and followup of interesting targets: HVS, UMPS, CEMPS, RCrBs Tuesday, August 27, 2013
SDSS BOSS (2009-2014): redshifts for 1.5 million LRG to z=0.7 eboss (2014-2020): extension to larger redshifts using emission-line galaxies Tuesday, August 27, 2013
Kitt Peak 4-m (Mayall) at Kitt Peak, Arizona USA today... Capable of holding heavy corrector 3 deg. field 13
Mayall with BigBOSS Huge Volume of universe with 3-degree field 5000 robotic fiber positioners 10 spectrographs 15 million spectra! 10 spectrographs Dec. 6, 2011 David Schlegel P1 Science Overview
Key of BB Instrument: 5000 Spectra per pointing via Robot Positioners 5000 positioners put fiber at target position positioner focal surface fibe rs Spectrograp Focal Plane filled with positioners
HET
HETDEX
DEX stars simulation from Besançon model Tuesday, August 27, 2013
spectra Tuesday, August 27, 2013
DEX stars analysis (KM stars) Tuesday, August 27, 2013
DEX stars analysis (FG stars) Tuesday, August 27, 2013
WEAVE Tuesday, August 27, 2013
WEAVE Tuesday, August 27, 2013
4MOST Tuesday, August 27, 2013
Data flow CCDs BigBOSS GB/s HETDEX GB/s Gaia GB/s) LSST GB/s single exposure flow 10x(1+4) 0.6 GB 0.0005 75x2 0.8 GB 0.002 100 200 (2 GB) 6 GB (0.1 0.3 LSST will take 30 TB/night and 100 PB in 10
Big data and the cloud Trends are changing in observational astronomy Hand by hand, instrument development and computing are creating a new generation of machines geared to speed up discovery The biggest telescopes are, at least partially, not following that trend
Big data and the cloud Supercomputing is widely embraced for modeling and interpreting the data (plain good-old physics) Virtual observatories are developing quickly to ensure the legacy value of the data High-throughput computing widely adopted for large-scale projects, but only modest use of cloud services (e.g. HETDEX)
Data mashup Typical use of data is concentrating on a specialized data set This must change now with projects such as Gaia Analysis requires combining appropriately different types of data Bayesian statistics offers the framework to a combined analysis of astrometry, photometry and spectroscopy Tuesday, August 27, 2013
Observing simulations Tuesday, August 27, 2013
Observing simulations Tuesday, August 27, et al. 2012 2013 Rahimi
Observing simulations Tuesday, August 27, 2013
Discovery and follow-up of interesting targets HVS LMWD CEMPS RCrBs UMPS Image from J. Norris Tuesday, August 27, 2013
Structure of the halo Johnston & Bullock simulations Monolithic collapse and intensive accretion A dual halo split in rotation and [Fe/H]? Only place in the Galaxy where full orbits and chemistsry are preserved: ideal for stellar archaelogy
Chemical evolution Big bang nucleosynthesis Stellar nucleosynthesis: hydrostatic equilibrium, AGB Explosive nucleosynthesis ISM spallation Also destruction
Early Chemical Evolution Fast enrichment Extremely few stars expected (and found) at [Fe/H] < -4 Large scatter in abundances at [Fe/H]<4
Big Bang Nucleosynthesis Figure from Edward L. Wright Chemical abundances of the oldest stars constrain our model of the universe. Large discrepancy between CMB and stellar Li abundances
R-process is universal Sneden et al. 2003 Oldest stars provide pure information on single SN events Nucleosynthesis patterns then and now seem to match in cases The fraction of carbon-rich stars increases at low
Mg/Fe Ca/Fe Only two stars Abundance currently known at ratios [Fe/H]<-5!! Y/Ba Ti/Fe Frebel 2010 Chiappini et al. 2011
Discovery space Preston/Shectman/Beer s Hamburg-ESO survey (Christlieb+) SDSS (Yong et al. 2012, Bonifacio et al. 2012)
Apache Point Observatory New Mexico, USA
BOSS spectra More resolution More spectral coverage More sensitivity Deeeper into the halo Privileged access to IAC >100,000 spectra in
BOSS analysis SEGUE obtained >300,000 stellar spectra Between 2008-2010 SEGUE s pipeline can be used to compare with results from BOSS New analysis for BOSS spectra represents a significant improvement
Discovery process 1. Skim through BOSS database to pick the best candidates to have [Fe/H]<-4 2. Follow those up with higher quality: 4.2m WHT/ISIS, 10m GTC/OSIRIS 3. Investigate those that pass all the filters: 8m ESO/VLT, 10m GTC/OSIRIS, 10m HET/HRS (maybe HORUS)
GTC
HORUS
HORUS-P
HORUS-P
Led-ThAr hollow-cathode lamp 48
Summary Astrometry: Hipparcos, UCAC3, Gaia Photometry: 2MASS, UKIDSS, VHS, SDSS, PanSTARS, GALEX, WISE, LSST Spectroscopy: Various SDSS low-resolution surveys, LAMOST, ARGOS, APOGEE, RAVE, Gaia-ESO, GALAH, DESI, HETDEX, WEAVE, 4MOST, MOONS Tuesday, August 27, 2013
Summary II Data indigestion? Stress should be placed on automated analysis And on a rigurous analysis of combined data (Bayesian framework) Parametric models, increasingly sophisticated, to become a workhorse. Are they Ready? Data can be directly used to determine distribution functions across the Galaxy to be Tuesday, August 27, 2013directly compared with numerical simulations
Conclusions We are living a golden era for the study of the Milky Way Gaia flies at the end of the year, first catalog probably out around 2016 and final catalog around 2020 Now is the time to get ready to use Gaia s data, and in particular to squeeze the information content in the combination of Gaia plus the other spectroscopic surveys Now is the time to sharpen our tools for Tuesday, August 27, automatic analysis and model-based 2013