Upgrade to Pro — share decks privately, control downloads, hide ads and more …

Galaxy Zoo 2 (Oxford)

Galaxy Zoo 2 (Oxford)

Given at galaxy evolution seminar at Oxford University.

Kyle Willett

May 09, 2013
Tweet

More Decks by Kyle Willett

Other Decks in Science

Transcript

  1. Galaxy Zoo 2: from clicks to morphologies for 304,122 galaxies

    in the SDSS Kyle Willett University of Minnesota
  2. Galaxy Zoo 2: from clicks to morphologies for 304,122 galaxies

    in the SDSS Kyle Willett University of Minnesota
  3. Galaxy Zoo 2: from clicks to morphologies for 304,122 galaxies

    in the SDSS Kyle Willett University of Minnesota
  4. Kyle Willett - Univ. of Minnesota University of Oxford -

    May 2013 3 Chris Lintott Brooke Simmons Kevin Schawinski Steven Bamford Karen Masters Lucy Fortson Ramin Skibba Kevin Casteels Sugata Kaviraj Bob Nichol Edd Edmondson Arfon Smith Jordan Raddick Rob Simpson and ...
  5. Kyle Willett - Univ. of Minnesota University of Oxford -

    May 2013 • Galaxy Zoo 2 is the largest ever catalog of fine morphological features, with 300,000 galaxies out to z = 0.25 • Crowdsourcing votes are reliable when compared with both expert and automated classifications • Catalog to be released at data.galaxyzoo.org and in SDSS DR10 • Early science from GZ2 includes studies of bars in disks, bulgeless AGN, and measuring the local interacting fraction 5
  6. Kyle Willett - Univ. of Minnesota University of Oxford -

    May 2013 GZ1 science: large-scale studies of galaxy populations 9 Schawinski, Urry et al. (2010) Bamford et al. (2009) Darg et al. (2010a,b) Land et al. (2008)
  7. Kyle Willett - Univ. of Minnesota University of Oxford -

    May 2013 GZ1 science: serendipitous discovery
  8. Kyle Willett - Univ. of Minnesota University of Oxford -

    May 2013 GZ1 science: serendipitous discovery Lintott et al. (2009), Keel et al. (2012a,b)
  9. Kyle Willett - Univ. of Minnesota University of Oxford -

    May 2013 GZ1 science: serendipitous discovery Lintott et al. (2009), Keel et al. (2012a,b)
  10. Kyle Willett - Univ. of Minnesota University of Oxford -

    May 2013 GZ1 science: serendipitous discovery Lintott et al. (2009), Keel et al. (2012a,b) Cardamone et al. (2009)
  11. Kyle Willett - Univ. of Minnesota University of Oxford -

    May 2013 GZ1 science: process of citizen science 11 Raddick et al. (2013) Simpson et al. (2012) Banerji et al. (2010)
  12. Is the galaxy simply smooth and rounded, with no sign

    of a disk? Could this be a disk viewed edge-on? Is there a sign of a bar feature through the centre of the galaxy? Is there any sign of a spiral arm pattern? Is there anything odd? How many spiral arms are there? How prominent is the central bulge, compared to the rest of the galaxy? How tightly wound do the spiral arms appear? Does the galaxy have a bulge at its centre? If so, what shape? How rounded is it? Is the odd feature a ring, or is the galaxy disturbed or irregular?
  13. Kyle Willett - Univ. of Minnesota University of Oxford -

    May 2013 Sample selection 14 GZ1 GZ2
  14. Kyle Willett - Univ. of Minnesota University of Oxford -

    May 2013 Sample selection 14 GZ1 GZ2
  15. Kyle Willett - Univ. of Minnesota University of Oxford -

    May 2013 15 • 14 months • 83,943 classifiers
  16. Kyle Willett - Univ. of Minnesota University of Oxford -

    May 2013 15 • 14 months • 83,943 classifiers • 16.3 million classifications
  17. Kyle Willett - Univ. of Minnesota University of Oxford -

    May 2013 15 • 14 months • 83,943 classifiers • 16.3 million classifications • 58.7 million clicks
  18. Kyle Willett - Univ. of Minnesota University of Oxford -

    May 2013 16 Courtesy Edmund Cheung, UCSC
  19. Kyle Willett - Univ. of Minnesota University of Oxford -

    May 2013 16 Bamford et al. (2009) Courtesy Edmund Cheung, UCSC
  20. Kyle Willett - Univ. of Minnesota University of Oxford -

    May 2013 17 eg, pbar = 0.40, pno bar = 0.60 eg, fbar = 0.20, fno bar = 0.80
  21. Kyle Willett - Univ. of Minnesota University of Oxford -

    May 2013 Using the catalog • “clean” samples of galaxies • very pure samples constructed by a combination of vote fraction thresholds for multiple classification tasks • raw likelihoods of galaxies • the (debiased) vote fractions can be potentially treated as probabilities, thus allowing use of the entire sample 19
  22. Kyle Willett - Univ. of Minnesota University of Oxford -

    May 2013 22 Catalog hosted on http://data.galaxyzoo.org and in SDSS CasJobs DR10 Willett et al. (2013, in prep)
  23. Kyle Willett - Univ. of Minnesota University of Oxford -

    May 2013 23 Expert and automated catalogs vs. GZ2
  24. Kyle Willett - Univ. of Minnesota University of Oxford -

    May 2013 24 0.0 0.2 0.4 0.6 0.8 1.0 1.2 tight spiral weighted fraction (GZ2) −0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 NA10 fraction per bin 0.0 0.2 0.4 0.6 0.8 1.0 1.2 tight spiral weighted fraction (GZ2) −0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 NA10 fraction per bin 488 507 565 680 761 694 692 655 356 117 E S0 Sa Sb Sc Sd 0.0 0.2 0.4 0.6 0.8 1.0 1.2 tight spiral weighted fraction (GZ2) −0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 EFIGI fraction per bin 185 175 173 203 234 227 224 217 139 41 E S0 Sa Sb Sc Sd 0.0 0.2 0.4 0.6 0.8 1.0 1.2 medium spiral weighted fraction (GZ2) −0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 NA10 fraction per bin 254 597 943 1144 1195 750 466 143 21 2 E S0 Sa Sb Sc Sd 0.0 0.2 0.4 0.6 0.8 1.0 1.2 medium spiral weighted fraction (GZ2) −0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 EFIGI fraction per bin 82 245 317 393 373 247 122 36 3 0 E S0 Sa Sb Sc Sd 0.0 0.2 0.4 0.6 0.8 1.0 1.2 loose spiral weighted fraction (GZ2) −0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 NA10 fraction per bin 3222 1041 447 290 183 103 83 55 46 45 E S0 Sa Sb Sc Sd 0.0 0.2 0.4 0.6 0.8 1.0 1.2 loose spiral weighted fraction (GZ2) −0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 EFIGI fraction per bin 981 340 190 114 64 43 36 20 16 14 E S0 Sa Sb Sc Sd 0.0 0.2 0.4 0.6 0.8 1.0 1.2 no_bulge weighted fraction (GZ2) −0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 NA10 fraction per bin 0.0 0.2 0.4 0.6 0.8 1.0 1.2 no_bulge weighted fraction (GZ2) −0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 NA10 fraction per bin 5894 583 252 145 87 57 44 28 20 10 E S0 Sa Sb Sc Sd − EFIGI fraction per bin 0.0 0.2 0.4 0.6 0.8 1.0 1.2 just_noticeable weighted fraction (GZ2) −0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 NA10 fraction per bin 453 632 785 900 1010 957 948 959 421 55 E S0 Sa Sb Sc Sd − EFIGI fraction per bin 0.0 0.2 0.4 0.6 0.8 1.0 1.2 obvious weighted fraction (GZ2) −0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 NA10 fraction per bin 941 867 772 950 1087 1014 820 530 121 18 E S0 Sa Sb Sc Sd − EFIGI fraction per bin 0.0 0.2 0.4 0.6 0.8 1.0 1.2 dominant weighted fraction (GZ2) −0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 NA10 fraction per bin 5649 928 296 148 59 31 6 2 1 0 E S0 Sa Sb Sc Sd − EFIGI fraction per bin Pitch angle (GZ2) Bulge dominance (GZ2) T-types (NA10) T-types (NA10)
  25. Kyle Willett - Univ. of Minnesota University of Oxford -

    May 2013 Automated classification vs. GZ2 26 GZ2 smooth GZ2 smooth 0.0 0.2 0.4 0.6 0.8 1.0 HC early−type probability 0.0000 0.0500 0.1000 0.1500 0.2000 0.2500 0.3000 Relative Frequency GZ2 features/disk GZ2 features/disk 0.0 0.2 0.4 0.6 0.8 1.0 HC late−type probability 0.0000 0.0500 0.1000 0.1500 0.2000 0.2500 0.3000 Relative Frequency GZ2 uncertain GZ2 uncertain 0.0 0.2 0.4 0.6 0.8 1.0 HC late−type probability 0.0000 0.0500 0.1000 0.1500 0.2000 0.2500 0.3000 Relative Frequency Huertas-Company et al. (2011) • Only capable of classifying broad morphology categories • {E, S0, Sab, Scd}
  26. Kyle Willett - Univ. of Minnesota University of Oxford -

    May 2013 Automated classification vs. GZ2 26 GZ2 smooth GZ2 smooth 0.0 0.2 0.4 0.6 0.8 1.0 HC early−type probability 0.0000 0.0500 0.1000 0.1500 0.2000 0.2500 0.3000 Relative Frequency GZ2 features/disk GZ2 features/disk 0.0 0.2 0.4 0.6 0.8 1.0 HC late−type probability 0.0000 0.0500 0.1000 0.1500 0.2000 0.2500 0.3000 Relative Frequency GZ2 uncertain GZ2 uncertain 0.0 0.2 0.4 0.6 0.8 1.0 HC late−type probability 0.0000 0.0500 0.1000 0.1500 0.2000 0.2500 0.3000 Relative Frequency Huertas-Company et al. (2011) Schawinski et al. (2009) • Only capable of classifying broad morphology categories • {E, S0, Sab, Scd}
  27. Kyle Willett - Univ. of Minnesota University of Oxford -

    May 2013 27 2,253 4,458 14,034 Experts
  28. Kyle Willett - Univ. of Minnesota University of Oxford -

    May 2013 27 2,253 4,458 14,034 Experts 304,122 Galaxy Zoo 2 Citizen science
  29. Kyle Willett - Univ. of Minnesota University of Oxford -

    May 2013 27 2,253 4,458 14,034 Experts 304,122 Galaxy Zoo 2 Citizen science 698,420 Automated
  30. Kyle Willett - Univ. of Minnesota University of Oxford -

    May 2013 Science with Galaxy Zoo 2 28
  31. Kyle Willett - Univ. of Minnesota University of Oxford -

    May 2013 Bars and disk galaxies • first statistical studies of barred galaxies with > 103 galaxies • bar fraction clearly increases with redder colors and more prominent bulges • ALFALFA-GZ2 match shows lower bar fractions in HI gas- rich galaxies • redder and longer bars occur in redder disk galaxies • colors and strengths of bars and their disks agree fairly well with simulations 29 Masters et al. (2012) Hoyle et al. (2011)
  32. Kyle Willett - Univ. of Minnesota University of Oxford -

    May 2013 Morphology and galaxy interactions •Many of the GZ2 classes correlate strongly with projected separation in galaxy pairs •The “loose winding arms” class is the strongest indicator of a physically associated pair •GZ2 data constrains the true fraction of physically interacting companions per galaxy at (0.4 - 2)% 34 Casteels et al. (2013)
  33. Kyle Willett - Univ. of Minnesota University of Oxford -

    May 2013 Morphology and galaxy interactions •Many of the GZ2 classes correlate strongly with projected separation in galaxy pairs •The “loose winding arms” class is the strongest indicator of a physically associated pair •GZ2 data constrains the true fraction of physically interacting companions per galaxy at (0.4 - 2)% 34 Casteels et al. (2013)
  34. Kyle Willett - Univ. of Minnesota University of Oxford -

    May 2013 Morphology and galaxy interactions •Many of the GZ2 classes correlate strongly with projected separation in galaxy pairs •The “loose winding arms” class is the strongest indicator of a physically associated pair •GZ2 data constrains the true fraction of physically interacting companions per galaxy at (0.4 - 2)% 34 Casteels et al. (2013)
  35. Kyle Willett - Univ. of Minnesota University of Oxford -

    May 2013 Bars and AGN • bars are theoretically predicted to channel gas to galactic center and trigger AGN activity in disks • observations of barred AGN are historically inconclusive • Oh+2012, Alonso+2013 found that AGN strength is enhanced by presence of bar and increases with stellar mass • UMN grad student Melanie Galloway, is studying bar-AGN relationships with GZ2 data for her masters’ thesis 35
  36. Kyle Willett - Univ. of Minnesota University of Oxford -

    May 2013 Galaxy Zoo and GZ2 have completed classifications for the local universe. What’s next? 36
  37. Kyle Willett - Univ. of Minnesota University of Oxford -

    May 2013 39 Galaxy Zoo Galaxy Zoo 2 Galaxy Zoo: Hubble Galaxy Zoo aka GZ1 aka GZ3 aka GZ2 aka GZ4 ~900,000 galaxies from SDSS completed in 2009 catalogs released (Lintott et al. 2011) 30+ publications ~300,000 galaxies from SDSS completed in 2010 catalog release upcoming (Willett et al.) 8 publications ~100,000 ACS images from COSMOS ~50,000 SDSS images 2010 -- 2012 analyzing and reducing data ~50,000 images from CANDELS ~230,000 images from SDSS DR8 2012 -- ? still collecting classifications
  38. Kyle Willett - Univ. of Minnesota University of Oxford -

    May 2013 40 • Galaxy Zoo 2 has 300,000+ classifications of galaxies with disks, bars, bulges, arms, and more • Averaging classifications from multiple volunteers reproduces expert results to a high degree of accuracy • Public data release out soon (data.galaxyzoo.org and in DR10) • GZ2 science: bars and gas disks, SMBH in bulgeless AGN, and constraints on local merger fraction