Discovery of Milky Way Dwarf Galaxies in the Dark Energy Survey

Ting Li on behalf of the DES Collaborations Milky Way Working Group

Department of Physics and Astronomy Texas A&M University APS DFP2015 Aug 4, Ann Arbor

1 Credit: Reidar Hahn, Yuanyuan Zhang The formation of Milky Way

• ELS Monolithic Collapse Model (top-down) Eggen, Lynden-Bell and Sandage 1962 Milky Way formed from the rapid collapse of a large proto-galacc nebula

• SZ Merger and Accretion Model (bottom-up) Searle & Zinn 1978 Galaxies are built up from merging or accreng smaller fragments N-body simulaons under ΛCDM context

2 Milky Way Satellite Galaxies

The Milky Way is surrounded by small satellite galaxies Segue 1 Distances ranges from 25 kpc to a few hundred kpc

Luminosities range 7 3 M. Geha from 10 L⊙ to 10 L⊙ The stars are moving too fast to be explained by visible Fornax mass ——— dark matter dominated

Astrophysically simple and clean

D. Malin 3 30Birth kpc of near-field (Bullock, Geha, Powell) cosmology Dwarf Galaxies as Cosmological Probes

• Missing Satellites Problem? • CDM simulations predict thousands of dark matter substructures • Only dozens of dwarf galaxies are found

• “Too big to fail” Problem? • Circular velocities of known dwarf galaxies are significantly smaller than predicted by simulations. • Cold dark matter? • Density profile of dark matter halo • Warm dark matter? • cusp vs core problem • Self-interacting dark matter? • Or other?

• Gamma rays from dark matter annihilation • Study the property of dark matter particles 4 Finding Milky Way Satellite Galaxies

5 Classical Dwarf Spheroidal Galaxies (dSph)

Sculptor

ESO/DSS2

6 Milky Way Satellite Galaxies Discovery Timeline

7 4 Finding Milky Way Advances in Astronomy

14 Satellite14 4 Galaxies14 14 Advances in Astronomy 20 kpc 100 kpc

16 45216 WALSH,16 WILLMAN, & JERJEN16 Vol. 137 14 14 Table 1 14 14 Angular Sizes of the Satellites Detected in SDSS 20 kpc 100 kpc Koposov et al. r 18(2008) r 18 r 18 Object r 18 rh (arcmin) Walsh et al. (2009) 16 16 Bootes¨ 12.6 16 16 Bootes¨ II 4.2 Willman et al. (2010)20 20 20 Canes Venatici20 8.9 Canes Venatici II 1.6 Coma Berenices 6.0 r Hercules 8.6 r 22 22 18 22 r 18 22 18 r 18 0.5 010.5 1.5 0.5 010.5 1.5 0.5 010.5 Leo IV1.5 0.5 010.5 2.5 1.5 − − − Leo V− 0.8 g r g r g r Leo Tg r 1.4 − − − Segue 1− 4.4 (a) 20 (b) (c) 20 Ursa Major(d) 11.3 20 20 Color-Magnitude Ursa Major II 16.0 Willman 1 2.3 DomainFigure 1: A color-magnitude (CM) filter used to suppress the noise from foreground stars while preserving the signal from dwarf galaxy stars at a specific distance. (a) and (c) CM filters for22 an old and metal-poor stellar population at22 a distance modulus of 16.5 and 20.0, respectively.22 22 The solid lines show Girardi isochrones for 8 and 14Gyr populations with [Fe/H] 1.5system and to2. avoid3. (b) projection and (d) Theseeffects. CM We then filters convolve overplotted this two- 2 0.5 010.5 1.5= − dimensional0.5− (2D)01 array0.5 with a spatial kernel1.5 corresponding0.5 to 010.5 1.5 0.5 010.5 1.5 on stars from a 1 deg field to illustrate the character− of the foreground contamination asthe a− expected function surface of dwarf density distance. profile of Data a dSph. are We from refer SDSS− to this − DR7. Figure 1. (g r, r)CMDshowingthetworeddestandtwobluesttheoreticalg r smoothed spatial arrayg as A r.Forourspatialkernelweusea g r g r isochrones for− old stellar populations ([Fe/H] 2−.27, 1.5andage 8, 14 − − − = − − = Plummer profile with a 4.′5scalelength.Thisvalueprovides Gyr) at a distance modulus of m M 16.5( 20 kpc), generated from Girardi an effective compromise between the angular scale lengths et al. (2004). The shaded region− shows= pixels∼ that pass the selection criteria. (a) of compact and/or distant(b) objects with those of closer/more (c) (d) extended objects. For reference the angular sizes of the new populated by old, metal-poor stars. Simon & Geha (2007)ob- satellites are listed in Table 1.Weusetherh values derived by Spatial tained spectraFigure of stars in eight1: A of color-magnitude the newly discovered dwarfs— (CM) filter used to suppress the noise from foreground stars while preserving the signal from dwarf galaxy stars 0.5 0.5 Martin et al.0.5 (2008)exceptforLeoV(Belokurovetal.2008). CVn, CVn II,at Com, a specific Her, Leo IV, distance. Leo T, UMa, (a) and UMa and II—and (c) CM filtersThe normalized for an signal old inand each metal-poor pixel of A,denotedby stellarS,gives population at a distance modulus of 16.5 and 20.0, respectively. Domain found mean metallicities in the range 2.29 < [Fe/H]< 1.97. the number of standard deviations above the local mean for each − − Based on thisThe result, solid we consider lines show isochrones Girardi for populations isochroneselement: for 8 and 14Gyr populations with [Fe/H] 1.5 and 2.3. (b) and (d) These CM filters overplotted with metallicities of [Fe/H] 1.5and2 2.27 (the lower = − − 0 limit in Girardion et stars al. 2004 from)andwithages8and14Gyr.Four0 = − a 1 deg−field to illustrate the character0 of theA foregroundA contamination as a function of dwarf distance. Data are from SDSS ¯ isochrones in these ranges can be used to bound the region of S − . DR7. = Aσ

dec (degrees) CMD space we are interested in, namely the four combina- dec (degrees) dec (degrees) δ δ tions of [Fe/H] 1.5and 2.27 and ages 8 and 14 Gyr. The arraysδ of running means, A,andrunningstandarddevia- Figure 1 shows these= − four isochrones− projected to a distance of ¯ 0.5 0.5 tions, Aσ ,arebothcalculatedovera00.5 ◦.9 0◦.9windowaround − 20 kpc. − − × each pixel of A.Inparticular,Aσ is given by We define the selection criteria by the CMD envelope inclu- sive of these isochrones +/ the 1σ (g r)colormeasurement 2 2 error as a function of r magnitude.− Shifting− these isochrones n(A A) B ((A A) B) 8 0.5 0 0.5 0.5 0 0.5 Aσ −0.5¯ ∗ − 0 − ¯ ∗ 0.5. over distances− between0.5m M 16.5 and 24.0 in 0.5 mag steps− = ! 0.5 n(n 1) − 0.5 δ ra (degrees)defines 16 different selection− criteria= appropriateδ ra (degrees) for old stellar δ ra− (degrees) populations between d 20 kpc and 630 kpc. We truncate B is a box filter with n elements and is the same size as (a) ∼ ! (b) (c) our color–magnitude selection template at a faint magnitude the running average window. The resulting array Aσ gives limit of r 22.0, beyond which photometric uncertainties in the standard deviation value for each pixel of A as measured = Figure 2: (a) Map of all stars inthe the colors field and around star/galaxy the Ursa separation Major limit I dwarf the ability satellite, to detectM over5.5, thed 0◦.91000◦. kpc.9spanofthefilter.Inthenextsection,wewill (b) Map of stars passing the CM V = − = × filter projected to m M 20these.0 shown populations. in Figure We also0 1(c) truncate. (c) Spatially the selection smoothed template number at define density the mapdetection of the threshold stars0 of in this (b). survey The Ursa in terms Major of S,as I 0 − = g r 1.0, as including2 redder objects adds more noise from well as in terms of the local stellar density E. dwarf galaxy has a µV,0 of onlyMW 27.5− dwarf mag= arcsecstars than[63 signal]. Data from are more from distant SDSS red giant DR7. branch dec (degrees)

3.4. dec (degrees) Detection Threshold(s)

(RGB) stars. Finally we do not include stars with δg or δr> dec (degrees) δ δ 0.3 mag in our analysis. To efficiently select stars within this In a large survey such as ours, it is critical to set detection δ CMD envelope, we treat the CMD as an image of 0.025 0.125 0.5 × thresholds strict enough0.5 to eliminate false detections but loose 0.5 (color mag) pixels− and determine which stars fall into pixels enough to retain known− objects and promising candidates. To − (iii) Identify Statisticallyclassified Significant× as “good” according Overdensities. to the selectionA criteria.possible Figure overdensities1 of point sources expected in the sky. 2 characterize the frequency and magnitude of purely random search of 10 000 deg of SDSSshows data, an example optimized of the selection for dwarfs criteria, inFor this case example, for fluctuations in stellar in the density abundant analyzed with tidal our debrisalgorithm, in we m M 16.5( 20 kpc). The shaded region highlights pixels at 16 different distances, and a single choice of stellar the Milky Way’smeasure halo the or maximum (un)bound value of S starfor 199,000 clusters 5◦.5 could3◦ simulated be that− would= be classed∼ as “good” for a system at 20 kpc. × population and scale size require evaluating the statistical0.5 ∼detected.0 It is0.5fields essential of randomly to obtain distributed follow-up stars that0.5 photometry have been smoothed0 to 0.5 0.5 0 0.5 3.3. Spatial Smoothing − as described in the previous section. The only difference is − − significance of 600 million data pixels that do not necessarily δ rafind (degrees) the color-magnitudethat there is no sequence gradient in stellar of stars density expected acrossδ ra each(degrees) for field. a δ ra (degrees) follow a Gaussian distributionAfter of signal. the photometric Setting cuts the are detection applied, we bindwarf the spatial galaxy andIn the also interest follow-up of computational spectroscopy efficiency to measure we do not the use a threshold to select candidate(R.A., dwarf decl.) positions galaxies of was the selected done stars by (a) intodark an array, massE, contentrunning (dark window matter for the is mean required and σ toof each be classified(b) simulated field. as (c) with 0◦.02 0◦.02 pixel size. We use a locally defined coordinate The field size is chosen such that 1000 fields roughly total an simulating numerous realizations of× the search, assuming a a galaxy) based on the observed line-of-sight velocities. random distribution of point sourcesFigure and permitting2: (a) Map only of all starsThis in the search field algorithm around the is very Ursa effi Majorcient. I In dwarf the WWJ satellite, M 5.5, d 100 kpc. (b) Map of stars passing the CM V = − = one completely spurious detection. Thefilter threshold projected is set to to bem Msearch,20 the.0 eleven shown strongest in Figure detections 1(c). (c) of sources Spatially unclassified smoothed number density map of the stars in (b). The Ursa Major I a function of point source number density after CM filtering. − prior= to SDSS were 112 of the 14 (probable) ultra-faint dwarf galaxy has a µV,0 of only 27.5 mag arcsec [63]. Data are from SDSS DR7. (iv) Follow-up Candidates. Regions detected above the Milky Way dwarfs. All of these but Bootes¨ II were known detection threshold are considered candidates for MW prior to the WWJ search. See references in Section 3 for dwarf galaxies. Although the threshold is set to prevent details of the follow-up observations that confirmed these the detection of any stochastic fluctuations of a randomly objects to be dwarf galaxies. Follow-up observations of distributed set of point sources [61], the detections(iii) Identify are only Statisticallyas-yet unclassified Significant SDSS dwarf Overdensities. galaxy candidatesA arepossible on- overdensities of point sources expected in the sky. “candidates” because resolved dwarf galaxiessearch are of not 10 the 000 only deg2goingof SDSS by several data, groups, optimized including for a group dwarfs at the IoAFor at example, fluctuations in the abundant tidal debris in at 16 different distances, and a single choice of stellar the Milky Way’s halo or (un)bound star clusters could be population and scale size require evaluating the statistical detected. It is essential to obtain follow-up photometry to significance of 600 million data pixels that do not necessarily find the color-magnitude sequence of stars expected for a follow a Gaussian distribution of signal. Setting the detection dwarf galaxy and also follow-up spectroscopy to measure the threshold to select candidate dwarf galaxies was done by dark mass content (dark matter is required to be classified as simulating numerous realizations of the search, assuming a a galaxy) based on the observed line-of-sight velocities. random distribution of point sources and permitting only This search algorithm is very efficient. In the WWJ one completely spurious detection. The threshold is set to be search, the eleven strongest detections of sources unclassified a function of point source number density after CM filtering. prior to SDSS were 11 of the 14 (probable) ultra-faint (iv) Follow-up Candidates. Regions detected above the Milky Way dwarfs. All of these but Bootes¨ II were known detection threshold are considered candidates for MW prior to the WWJ search. See references in Section 3 for dwarf galaxies. Although the threshold is set to prevent details of the follow-up observations that confirmed these the detection of any stochastic fluctuations of a randomly objects to be dwarf galaxies. Follow-up observations of distributed set of point sources [61], the detections are only as-yet unclassified SDSS dwarf galaxy candidates are on- “candidates” because resolved dwarf galaxies are not the only going by several groups, including a group at the IoA at Ultra-Faint Dwarf Galaxies Segue 1

Geha 9 Ultra-Faint Dwarf Galaxies Segue 1

Geha 10 Milky Way Satellite Galaxies Discovery Timeline

11 Known Milky Way Satellites after SDSS

• What will DES discover? 12 The Dark Energy Survey

5 year survey over 525 nights 5 filters: g,r,i,z,Y ~5,000 sq. degree ~24th mag in g-band with 10 tiling

13 Main survey

DES Year-One10 tilings(Y1A1) x 90 s ⇒ mag. limit 25.2 (g) .. 23.4 (z)

• ImagingMain data from survey the first year of the survey: August 2013 to February 2014 10 tilings x 90 s ⇒ mag. limit 25.2 (g) .. 23.4 (z) • Coadded image catalog covering ~1800 deg2 • ~200 deg2 overlapping with SDSS Stripe-82 • ~1600 deg2 overlapping with the South Pole Telescope

• Stellar completeness >50% down to g,r ~ 23 Eric Neilsen

iby Eric Neilsen Survey completeness after Y1 12

iby Eric Neilsen Survey completeness after Y1 12 DES Year 1 vs. SDSS on Messier 2

• A dramatic improvement in the photometric precision with Blanco+DECam!

And this is just Year 1—deeper, more precise photometry 15 will be produced throughout the five-year survey Y1A1: A First Look – 18 –

– 16 –

Reticulum II Eridanus II (DES J0335.6-5403) (DES J0344.3-4331)

14 20 m M = 17.5 m M = 22.6 DES J0344.3-4331 12 DES J0335.6-5403 (53.92, 54.05) 15 (56.09, 43.53) 0.4 0.4 10 10

8 (deg) (deg) (deg) 0.0 (deg) 0.0 2000 2000

2000 6 2000 5 4 0.4 0.4 0 2 Counts per Square Arcmin Counts per Square Arcmin 5 0 0.4 0.0 0.4 0.0 0.1 0.2 0.3 0.4 0.4 0.0 0.4 0.0 0.1 0.2 0.3 0.4 2000 (deg) Angular Separation (deg) 2000 (deg) Angular Separation (deg) 17 17 17 17 16 r < 0.03 r < 0.03 9 r < 0.10 r < 0.10 18 18 18 18 14 8 )

19 19 12 ) 7

19 19 6 10 20 20 20 20 g g 5 g

g g 8 g 21 21 21 21 4 6 22 22 3 22 22 Significance ( 4 Significance ( 2 23 23 1 23 23 2 0 0 0.5 0.0 0.5 1.0 0.5 0.0 0.5 1.0 0.5 0.0 0.5 1.0 0.5 0.0 0.5 1.0 g r g r g r g r Fig. 5.— Analogous to Figure 4 but for DES J0344.3 4331. A large number of stars, including Fig. 4.— Stellar density and color-magnitude diagrams for DES J0335.6 5403. Top left: Spatial arXiv:1503.02584 several probable horizontal branch members, are present at magnitudes fainter than the g<23 mag distribution of stars with g<24 magBechtol that are within et al. 0.1 (2015) mag of the (DES isochrone Collaboration) displayed in the threshold of our likelihood analysis. This threshold was set16 by the rapidly decreasing stellar com- lower panels. The field of view is 1.5(see1.5 centeredalso Koposov on the candidate et andal. the 2015a) stellar distribution arXiv:1503.02079 ⇥ pleteness at fainter magnitudes. However, it is likely that extending to fainter magnitudes would has been smoothed with a Gaussian kernel with standard deviation 0.027.causeTop the center best-fit: Radial distance modulus of DES J0344.3 4331 to increase. Better constraints on the distribution of stars with g r<1 mag and g<24 mag. Top right: Spatial distribution of stars properties of DES J0344.3 4331 require the stellar completeness to be robustly quantified in this with high membership probabilities within a 0.5 0.5 field of view. Small gray points indicate ⇥ regime. stars with membership probability less than 5%. Bottom left: The color-magnitude distribution of stars within 0.1 deg of the centroid are indicated with individual points. The density of the field within a 1 deg annulus is represented by the background two-dimensional histogram in grayscale. The red curve shows a representative isochrone for a stellar population with ⌧ = 13.5 Gyr and Z =0.0001 located at the best-fit distance modulus listed in the upper left panel. Bottom center: Binned significance diagram representing the Poisson probability of detecting the observed number of stars within the central 0.1 deg for each bin of the color-magnitude space given the local field density. Bottom right: Color-magnitude distribution of high membership probability stars. Reticulum II

17 DES Collaboration Reticulum II

18 DES Collaboration Eridanus II

19 Belokurov & Koposov Dwarf Galaxy Candidates

• \

20 arXiv:1503.02584 Milky Way Satellite Galaxies Discovery Timeline

21 Dwarf Galaxy Candidates

Draft version March 10, 2015 (continued…) A Preprint typeset using LTEXstyleemulateapjv.01/23/15

BEASTS OF THE SOUTHERN WILD. DISCOVERY OF A LARGE NUMBER OF ULTRA FAINT SATELLITES IN THE VICINITY OF THE MAGELLANIC CLOUDS. Draft version MarchSergey 20, E. 2015 Koposov, Vasily Belokurov, Gabriel Torrealba, andN.WynEvans Preprint typeset using LAT Xstyleemulateapjv.5/2/11 E Institute of Astronomy, Madingley Road, Cambridge CB3 0HA, UK DES Data (Dated: March 10, 2015) Draft version March 10, 2015

ABSTRACT A NEW FAINT MILKY WAY SATELLITE DISCOVERED IN THE PAN-STARRS13π SURVEY We have used the publicly released Dark Energy Survey data to hunt for new satellites of the Milky BenjaminWay P. M. in the Laevens Southern1,2,NicolasF.Martin hemisphere. Our search1,2,RodrigoA.Ibata yielded a large number1,Hans-WalterRix of promising candidates.2,EdouardJ.Bernard In this 3, Eric F. Bell4,BranimirSesar2,AnnetteM.N.Ferguson3,EdwardF.Schlafly2,ColinT.Slater4,WilliamS. paper,5 we announce the discovery6 of 9 new unambiguous6 ultra-faint objects,6 whose authenticity6 can be DraftBurgett version,KennethC.Chambers March 31, 2015 ,HeatherFlewelling,KlausA.Hodapp,NicholasKaiser,Rolf-Peter established6 with the DES7 data alone. Based on the6 morphological pro8perties, three of the new6 satellites 7 KudritzkiPreprint typeset,RobertH.Lupton using LAT Xstyleemulateapjv.5/2/11,EugeneA.Magnier,NigelMetcalfe,JeffreyS.Morgan,PaulA.Price, are dwarf galaxies,E one of which6 is located at the very outskirts6 of the Milky Way,6 at a distance of 380 kpc. The remainingJohn L. 6 Tonry objects,RichardJ.Wainscoat have sizes and luminosities,ChristopherWaters comparable to the Segue 1 satellite and Pan-STARRScan not be classified straightforwardlyDraft without version follow-up March 20, spectros 2015 copic observations. The satellites we have discovered cluster around the LMC and the SMC. We show that such spatial distribution is unlikely under the assumption of isotropy,ABSTRACT and, therefore, conclude that at least some of the new A HERO’SsatellitesWe present DARK must the HORSE: have discovery been DISCOVERY associated of a faint with Milky OF the AN Magellanic Way ULTRA-FAINT satellite, Clouds Laevens in theMILKY 2/Tr past.iangulum WAY SATELLITE II, found in IN the PEGASUS Panoramic Survey Telescope And Rapid Response System (Pan-STARRS 1) 3 π imaging data and Draft versionKeywords:Dongwon AprilGalaxy: Kim, 3, 2015 Helmut halo, galaxies: Jerjen, dwarf, Dougal globular Mackey, clusters: Gary general, S. Da galaxies: Costa,kinematics and Antonino and dynam- P. Milone PreprintResearch typesetconfirmed School using of Astronomywith LAicsT Xstyleemulateapjv.5/2/11 follow-up and Astrophysics, wide-field photometry The Australian from National the Large University, Binocular Mt Stromlo Cameras. Observatory, The stellar via sys- Cotter Rd, Weston, E ACT− 2611,± Australia +2 tem, with an absolute magnitude of MV = 1.8 0.5, a heliocentric distance of 30−2 kpc, and a half-mass radius of 34+9 pc, shows remarkableDraft version similarity March to faint, 31, 2015 nearby, small satellites such as Will- 1. INTRODUCTION−8 detectable with surveys like SDSS out toDECam a small frac- Data Currently,man 1, there Segue are 1, no Segue strong 2, and theoretical Boötes predictions II. The discoveryABSTRACTtion of Laevenof the MWs 2/Triangulum virial radius. II further As a result, populates their total as to thethe luminosity region of function parameter and space the spatial for which distribution the boundarynumber between in the dwarf Galaxy galaxies must and be reconstructed globular clusters under the HYDRAbecomes II:We A report FAINT tenuous. the AND discovery Follow-up COMPACT of spectroscopy an ultra-faint MILKY will Milky WAY ultimately Wayassumption DWARF satellite determine GALAXY of thegalaxythe shape nature in FOUND of the their of constellation this galactocentric IN new THE satellite, SURVEYof radialPegasus. dis- OF THE of the GalacticThe concentration dwarf companions. of stars was In other detected words, by to- applying our overdensity detection algorithm to the SDSS- day’s semi-analyticwhose spatial models location are hints so flexible at a possibleMAGELLANIC they can connection easily STELLAR withtribution. the complex Given HISTORY thatTriangulum-Andromeda only a handful of such stellar objects are DRstructures. 10 and confirmed with deeper photometry fromknown the today, Dark their Energy radial Camera profile is at not the observationally 4-m Blanco Nicolasproduce any F. Martin number1 of,2 Milky,DavidL.Nidever Way (MW) satellites:3,GurtinaBesla from 4,KnutOlsen5,AlistairR.Walker6,A.Katherina 6 telescope.Subject headings: FittingLocal model7 8 Group isochrones — Milky indicates Way, satellites, that5 6 constrained. this streams: object, individual: However, Pegasus III,9 if Laevens25 it features is assumed 2/T anriangulum thatold and the5 II metal-faintest 5 Vivasa few,RobertA.Gruendl tens up to several thousands, ,CatherineC.Kaleida (e.g. Koposov et al. , ,RicardoR.Munoz˜ , ,RobertD.Blum,AbhijitSaha, poor stellar10 population3 ([Fe/H] 2.1) at11,8 a heliocentricof the UFDs distance are linked of12, 20513 to,14 the20 small-mass kpc. The field new dark stellar mat-15 Blair2009). C. Similarly, Conn ,EricF.Bell a large range of,You-HuaChu spatial⇠ arrangements,Maria-RosaL.Cioni ±,ThomasJ.L.deBoer ,Carme is possible:system16, from17 has nearly an estimated isotropic18 to half-light strongly radius planar13 (e.g.of rh = 110ter (DM)6 pc and sub-halos, a19 total their luminosity distribution of MV can be4. gleaned201 0.5 Gallart ,ShokoJin1. INTRODUCTION,AndreaKunder ,StevenR.Majewskitations show that,DavidMartinez-Delgado these objects could well be the faintest,Antonela Bahl &that Baumgardt21 places it 2014). into the This domain16 is,17 why of to dwarf improve galaxies our on16from,17 the± Cosmological size–luminosity zoom-in plane.22 simulations. Pegasus III This⇠ is spatially is an± ex-4 Monachesi ,MatteoMonelli ,LaraMonteagudo DGs,NoeliaE.D.No and that tens or hundredsel¨ ,EdwardW.Olszewski of DGs with these prop-,GuyS. understandingThe lastclose couple to of the the of MW decades physics satellite ofsaw the the Pisces Universal discovery23 II. It structure is of possible nu- thatample the of two so-called might24 sub-halo be physically abundance associated,4 matching similar which Stringfellow ,RoelandP.vanderMarelerties could populate,DennisZaritsky the Milky Way halo (Tollerud et al. merousformation satellitesto the on small-scales Leo in IV the and Milky we Leo look Way V to pair. (MW) observations. Pegasus halo. While III How- is alsopredicts well aligned that the with bulk the of Vast the Galactic Polar Structure, dwarf galaxy which pop- 22 theever, Sloan to date, Digital only Sky a third Survey of the (SDSS sky has (York been et inspected.al.Draft 2000)) versionulation2008; April Hargis 3, is 2015 in objectset al. 2014). with As luminosities of yet, justM twoV < objects5. The have suggests a possible physical association. detailsbeen found of the in semi-analytic the PS1 survey calculations (LaevensDECam et may al.− vary 2014), but re- Data satelliteThe SloanSubject discoveries Digital headings: Sky have Survey providedLocal (SDSS) Group us has with – demonstrated Milky greater Way, ob- satellites: individual: Pegasus III servationalthe power of constraints deep wide-area in our imaging backyard, to fuel especially resolved stel- toABSTRACTtheinforcing result the remains tension most between astounding: theory there and ought observations to be hundreds(Klypin et of al. galaxies 1999).Only with M theV closest1, if theDGs faintest would of be the de- understandlar populationsWe the present studies faint the end of the discovery of Galactic galaxy of halo. formation a new The dwarfanalysis in the galaxy, Hydra II, found serendipitously∼− within the data preferredof the SDSS cosmological data covering paradigm only 1/5 of ofΛ celestialCDM (Belokurov sphere has UFDstected occupy with current field DM photometric sub-halos (see surveys e.g. (KoposovTollerud et et al. al. from the ongoing1. INTRODUCTION Survey of the Magellanic Stellar Historying SDSS (SMASH) catalog conducted as described with in the Walsh Dark et al.Energy (2009) and 2013),more than they doubled have also the led number to debates of the known about Galactic the na- 2008;2007; Bullock Walsh et et al. al. 2009). 2010). Spectroscopic studies do show Camera on the Blanco 4m Telescope. The new satelliteKim & is Jerjen compact (2015). (rh Briefly,=68± we11 used pc) isochrone and faint masks turedwarfFollowing of satellites the faintest the (see Sloan satellites e.g. Digital Willman (Gilmore Sky 2010; Survey Belokurov et al. (York 2007). 2013). et It al. thatIt is, the however, faint systems possible found that so the far smallestare dynamically of the UFDs hotter (MV = −4.8 ± 0.3), but well within the realm of dwarf galaxies. The stellar distribution of Hydra II 2000),hasRecently, become more both recent apparent PanSTARRS wide-field that the and imaging previously VST surveys ATLAS clear such have distinc- as ex- the maythanbased have their been on mere the acquired stellar PARSEC content via a di stellarff woulderent evolution route.imply, It hinting has models been that (Bres- in the color-magnitude diagram is well-describedsuggested by a metal-poor that some ([Fe of/ theseH] = objects−2.2) may and have old (13 been Gyr) ac- Stromlotionpanded between Milky the surveyed the Way compact Satelliteregion globular significantly, survey clusters (Jerjen but (GCs)found 2010), little and the theysan are et indeed al. 2012) DGs (Martinas a photometric et al. 2007; filter Willman to enhance et al. the isochrone and shows a distinct blue horizontal branch,creted2011;presence Simon some as part possib et of of al. old a 2011;le group and red Kirby metal-poor (see clump e.g. et al. stars, Belokurov 2013). stellar and However, populations faint2013). stars In the rela- arXiv:1503.02079v1 [astro-ph.GA] 6 Mar 2015 theof the brighter, wealth more they extended, came to seek and (Laevens dark-matter et al. domi- 2014; Dark Energy Survey (The Dark Energy Survey Collab- the case of the UFD Segue± 2 for example, the group’s−1 orationnatedBelokurov dwarfthat 2005), et are galaxies al. the suggestive 2014). Pan-STARRS (DGs), of blurs blue out 3⇡ stragglers.Survey for faint (K. systems At Cham- a heliocentriclowtive velocity distance to the dispersion Milky of134 Way of these10 foreground kpc, satellites Hydra stars. ( II< is4kms We located then), binned Thein SDSS a region discoveries of the Galactic extended halo the that dwarf models galaxy havecentralcombined suggestedthe object R.A., with may DEC.appears the host possiblypositions tomaterial have large been of from the destroyed impact the filtered leading of and stars binaries can arm and con- bers(Willman et al., & in Strader preparation), 2012). and This the is Surveyexemplified of the by Magel- the now be detected only as tidal debris in the MW halo discoveriesregimeof to the of extremely Willman Magellanic low 1 (Wil1; luminosities Stream. Willman A and comparison et sizes. al. 2005) As a and with re- N-body(McConnachievolved simulations the & density-map Côté2010), hints that with the the complexity a new Gaussian dwarf of disentan- kernel. galaxy Based lanicsult Stellar of adding History these new(SMASH; faint surface-brightness PI D. Nidever) have objects been (Belokurovgling member et al. stars 2009; from Deason foreground et al. 2014). contaminants, In this sce- and revealingSeguecould 1 (Seg1; new be Milky Belokurov or could Way have et companions al. been 2007), a satellite followed including of up the satel- by Magellanicnario,on the the Clouds. satellites density-map, of satellites we calculate survive (unlike the signal their topre- noise ra- those(known of Boötes as ultra-faint II (BooII; dwarfs, Walsh UFDs) et al. 2007), to the and panoply Segue of 2 thetios overall (S/Ns) dimness of potential of their overdensities member stars and renders measure any their liteMilky galaxiesSubject Way companions, (Koposov headings: et itdwarf al. has 2015; been galaxy: revealed Bechtol individual: that et al. there 2015; Hydraviousdefinite II — host) Local conclusion in Groupthe MW diffi —cult. halo, Magellanic but their Clouds spatial distribu- Laevens(Seg2; Belokurov et al. 2015; et al. Martin 2009), all et nearby al. 2015) satellites and star within clus- tionsignificance should differ by dramatically comparing from their that S/Ns predicted to those for of ran- 25–45appears kpc, to and be a just gap slightly in the distribution larger than of extended effective outer radii Here, we present the discovery of another faint MW ters (Belokurov et al. 2014; Laevens et al. 2014; Kim & thedom accreted clusters field sub-halos. in the residual Not only background.9 the radial density Moving the halobetween GCs. Globular At the same Clusters time, and these dwarfs systems which are extends fainter satellite,isochrone Laevens masks 2/Triangulum over1. a range II of distance, with very moduli similar (m M) arXiv:1503.05554v1 [astro-ph.GA] 18 Mar 2015 Jerjenacross 2015; a large Kim range et al. of 2015). luminosities. The new In Milky the absence Way com- of profilephotometric is modified, properties the satellites’ to Wil1,INTRODUCTION morphologicalSeg1, BooII, and proper- Seg2. than most GCs and all the other DGs. Theoretical expec- tiesbetween might have 16 been and sculpted 22 magnitudes, by the pre-processing this process (see is repeated panions,a working many definition of which of a are galaxy, in the an southern ad hoc combination sky, share the The newOur system view was of the found Milky in our Way ongoing satellite effort system to mine has thor- [email protected] e.g.with Wetzel di↵ eterent al. 2015). scales Thus, of bins the and total Gaussian numbers kernels. of the [email protected] morphological, of previously kinematic discovered and chemical ultra-faint properties stellar is sys- the Pan-STARRS 1 (PS1) 3π survey for localized stel- Observatoire1 astronomique de Strasbourg, Université de UFDsoughly would changed be biased over high the if estimated last decade, through thanks sim- to system- requiredObservatoire to be classified astronomique as one. de Therefore, Strasbourg, some Université objects de lar overdensities.With this algorithm, This letter we is recovered structured all as of follows: the previously in tems,Strasbourg, such asCNRS, low luminosities UMR 7550, 11 ( rue8 . deM l’Université,V . 1.5) F-6700 (Mar-0 ple abundanceatic CCD matching surveys calculation,of large swathes while their of the spatial sky. Explo- tinStrasbourg,thatStrasbourg, et al. live 2008) dangerously France CNRS, and UMR low close 7550, metallicities to 11 the rue notional de l’Université, [Fe/H] gap< may F-67002(Kirby risk0 sectionknown 2, we MW describe companions the PS1 in survey the SDSS along coverage with the and de- found 2Strasbourg, France anisotropyrations would of the be greatly Sloan under-estimated. Digital Sky Survey (SDSS) first misclassification.Max-Planck-Institut2 These für faintest Astronomie, of the Königstuhl UFDs are only 17, D- a few more promising candidates, one of which was re- et al.Max-Planck-Institut 2008; Norris et al. für 2010; Astronomie, Simon et Königstuhl al. 2011; 17, Koch D- & tectionOnrevealed the method sky, thatthe that remaining dwarf led to galaxies unchartedthe discovery. extend territory We to continuelies a much be- fainter 6911769117 Heidelberg, Heidelberg, Germany Germany by discussingported in follow-up Kim & Jerjen imaging◦ (2015). obtained The with new the Large object was Rich3 3 2014). neathregime declination thanδ previously= 30 and thought is currently (e.g., being Willman ex- et al. [email protected],[email protected] for of Astronomy, Astronomy, University University of Edinburgh, of Michigan, Royal 1085 Ob- S. Binoculardetected Cameras with a in significance− section 3. of We discuss7 in the the constellationnature UniversityInservatory, this letter Ave., Blackford Ann we report Hill, Arbor, Edinburgh the MI 48109-1107, detection EH9 3HJ, of USA UK the new ultra- 2005) and more than doubled⇠ the number of known 4 4 of theof Pegasus. satellite and its implication in section 4. In the faintSteward MilkyDepartment Way Observatory, of satellite Astronomy, University Pegasus University III of found Arizona, of Michigan, in SDSS 933 500 NorthData Milky Way dwarf galaxies from data mainly covering the CherryChurch Avenue, St., Ann Tucson Arbor, AZ, MI 48109, 85721 USA final section, we summarize and conclude our results. Release5 10 and confirmed with deep DECam imaging 5 NationalGMTO Optical Corporation, Astronomy 251 S. Lake Observatory, Ave, Suite 950 300, N Pasadena, Cherry Ave, InNorth3. thisFOLLOW-UP paper, Galactic magnitudes OBSERVATIONS Cap (e.g., are Belokurov dereddened AND DATA et using al. REDUCTION 2007). the Spec- (SectionsCA 91101, 2 USA & 3). Peg III appears to be located at a Tucson,6 AZ 85719, USA Schlegeltroscopic et al. surveys (1998) maps, (e.g., adopting Muñoz et the al. extinction 2006; Martin co- et al. heliocentric6 CerroInstitute Tololo distance for Inter-American Astronomy, of 205 University Observatory, kpc and of Hawaii have National at a half-light Manoa, Optical Deeper follow-up observations of the Peg III field were Honolulu, HI 96822, USA ⇠ 2007; Simon & Geha 2007) confirm the dwarf galaxy na- radiusAstronomy7 of Observatory,110 pc (Section Casilla 4). 603, In La the Serena, last section Chile we dis- 9 conducted on during the night of 17th July 2014 using Department of Astrophysical Sciences, Princeton University, Inture the absenceof these of spectroscopic systems by confirmation, demonstrating we wish them to rema toin be kine- cuss7 National the possible⇠ Center origin for Supercomputing of the new satellite Applications, galaxy 1205 and arXiv:1503.08268v1 [astro-ph.GA] 28 Mar 2015 Princeton, NJ 08544, USA agnosticthe Darkabout the Energy nature Camera of this object (DECam) and therefore at the propo 4-mse a Blanco West8 Clark St., Urbana, IL 61801, USA matically hotter than implied solely by their baryonic conclude8 Department our results. of Physics, Durham University, South Road, doubleTelescope name. For located future reference at Cerro in this Tololo paper, Inter-American we abbreviate to Ob- DurhamDepartment DH1 3LE, of Astronomy, UK University of Illinois, 1002 West Laecontent 2/Tri II. — which can be as low as only a few hundred Green St., Urbana, IL 61801, USA servatorysolar luminosities (CTIO) in (Martin Chile. et DECam al. 2008). imager is equipped 9 2. DISCOVERY with a focal plane array containing sixty-two 2k 4k Departamento de Astronom´ıa, Universidad de Chile, Camino More recently, a cohort of faint satellites, most of them delThe Observatorio SDSS is 1515, a photometric Las Condes, and Santiago, spectroscopic Chile survey CCD detectors with a wide field of view (3.0 square⇥ 10 Gemini Observatory, Recinto AURA, Colina El Pino s/n, likely dwarf galaxies, was found within the first year in the ugriz photometric bands to a depth of r 22.5 degrees) and a pixel scale of 000. 27 (unbinned). Un- La Serena, Chile. ⇠ data from the Dark Energy Survey (Bechtol et al. 2015;

arXiv:1503.06216v2 [astro-ph.GA] 2 Apr 2015 magnitudes11 (York et al. 2000). Data Release 10 (DR10), der photometric conditions, we obtained 840 s expo- Institute of Astronomy and Astrophysics, Academia1 Sinica, Koposov et al. 2015). Although these data only cover publiclyNo.1, Sec. available 4, Roosevelt on Rd, the Taipei SDSS-III 10617, Taiwan, Web site R.O.C.,covers sures in g2and 1050 s in r band, divided over dithered 14,12555Universität deg2 mostly Potsdam, around Institut the north für Physik Galactic und Astronomie pole (Ahn, single1,800 deg exposuresof the of southern 210 s. Galactic The average cap to seeing the north dur- of Karl-Liebknecht-Str. 24/25, 14476 Potsdam, Germany ingthe the Magellanic observing Clouds, was 1. they3inthe revealedg and 9 new 1. 1inthe stellar sys-r et13 al. 2014). 00 00 TheLeibniz-Institut new object was für first Astrophysics flagged by Potsdam our detection (AIP), An algo- der band.tems. The The majority single exposure of these images new systems were fully have reduced distances Sternwarte 16, 14482 Potsdam Germany that hint that they likely are, or were, associated with rithm14 University in the search of Hertfordshire, for stellar overdensities Physics Astronomy over theand exist- Math- and stacked through the DECam community pipeline ematics, Hatfield AL10 9AB, United Kingdom (Valdesthe Clouds. et al. 2014). We carried out weight-map com- [email protected] Institute of Astronomy, University of Cambridge, Mading- bination,Within source the ΛCDM extraction cosmology, and PSF the photometry Milky Way with is ex- ley1 Road,http://www.sdss3.org/dr10/ Cambridge CB3 0HA, UK 16 thepected use to of grow WeightWatcher hierarchically (Marmo by accreting & Bertin groups 2008) of Instituto de Astrof´ısica de Canarias, La Laguna, Tenerife, smaller galaxies over time (e.g., Li & Helmi 2008) and Spain 17 Departamento de Astrof´ısica, Universidad de La Laguna, it is therefore not unexpected that the Large Magel- Tenerife, Spain lanic Cloud (LMC) would be in the process of shedding 18 Kapteyn Astronomical Institute, University of Groningen, its satellites in the Milky Way halo (D’Onghia & Lake P.O. Box 800, 9700 AV Groningen, The Netherlands 2008). In fact, comparisons with simulations show that 19 Department of Astronomy, University of Virginia, Char- lottesville, VA 22904, USA the majority of current Milky Way satellites, especially 20 Astronomisches Rechen-Institut, Zentrum für Astronomie the fainter ones, could have been accreted as satellites of der Universität Heidelberg, Mönchhofstr. 12-14, 69120 Heidel- larger dwarf galaxies (Wetzel et al. 2015). These group berg, Germany – 14 –

Dwarf Galaxies or Globular Clusters?

12 Globular Dwarf 10 Clusters Galaxies

8 J0344.3-4331

6 (mag)

V J0335.6-5403

M 4 J2339.9-5424 J2251.2-5836 J0443.8-5017 J0255.4-5406 J0222.7-5217 2 J2108.8-5109 Globular Clusters 2 2 2 MW Satellite Galaxies 0 M31 Satellite Galaxies .5 mag arcsec Local Group Galaxies = 25 mag arcsec = 30 mag arcsec µ = 27 µ µ DES Candidates Brighteness 2 100 101 102 103 Half-light Radius (pc) Physical Size 23 arXiv:1503.02584 Fig. 3.— Local Group galaxies (McConnachie 2012a) and globular clusters (Harris 1996, 2010 edition) occupy distinct regions in the plane of physical size and absolute luminosity. The majority of DES satellite candidates (red dots) are more consistent with the locus of Local Group galaxies (empty blue shapes) than with the population of Galactic globular clusters (black crosses). Dashed lines indicate contours of constant surface brightness at µ = 25, 27.5, 30 mag arcsec 2. { } – 16 – Reticulum II: Newest Dwarf Galaxy?

Membership probabilities will play an important role targeting spectroscopy 14 1.0 m M = 17.5 0.2 12 DES J0335.6-5403 (53.92, 54.05) 0.8 0.4 10

8 0.6 (deg) 0.0 (deg) 0.0

2000 6 2000 0.4

4 0.4 0.2 2 0.2 Membership Probability Counts per Square Arcmin 0 0.0 0.4 0.0 0.4 0.0 0.1 0.2 0.3 0.4 0.2 0.0 0.2 2000 (deg) Angular Separation (deg) 2000 (deg) 17 17 16 17 1.0 r < 0.10 r < 0.10 pi = 308.0 18 18 14 18 0.8

12 )

19 19 19 10 20 20 20 0.6

g g 8 g 21 21 21 6 0.4 22 22 22 4 Signiﬁcance ( 0.2 23 23 23

2 Membership Probability 0 0.0 0.5 0.0 0.5 1.0 0.5 0.0 0.5 1.0 0.5 0.0 0.5 1.0 g r g r g r 24 arXiv:1503.02584 Fig. 4.— Stellar density and color-magnitude diagrams for DES J0335.6 5403. Top left: Spatial distribution of stars with g<24 mag that are within 0.1 mag of the isochrone displayed in the lower panels. The field of view is 1.5 1.5 centered on the candidate and the stellar distribution ⇥ has been smoothed with a Gaussian kernel with standard deviation 0.027. Top center: Radial distribution of stars with g r<1 mag and g<24 mag. Top right: Spatial distribution of stars with high membership probabilities within a 0.5 0.5 field of view. Small gray points indicate ⇥ stars with membership probability less than 5%. Bottom left: The color-magnitude distribution of stars within 0.1 deg of the centroid are indicated with individual points. The density of the field within a 1 deg annulus is represented by the background two-dimensional histogram in grayscale. The red curve shows a representative isochrone for a stellar population with ⌧ = 13.5 Gyr and Z =0.0001 located at the best-fit distance modulus listed in the upper left panel. Bottom center: Binned significance diagram representing the Poisson probability of detecting the observed number of stars within the central 0.1 deg for each bin of the color-magnitude space given the local field density. Bottom right: Color-magnitude distribution of high membership probability stars. Reticulum II: Spectroscopy Campaign

Magellan/M2FS Gemini/GMOS VLT/GIRAFFE

Magellan/M2FS VLT/GIRAFFE 25 arXiv:1504.02889 Simon et al. 2015 (DES Collaboration) Figure 5. (Left) Magellan/M2FS spectra in the Mg b triplet region for three Ret II member stars covering a range of line strengths. From top to bottom, the stars are DES J033556.28 540316.3, DES J033454.24540558.0, and DES J033457.57540531.4. These stars span only 0.1 mag in luminosity and 0.08 mag in g r color, so their e↵ective temperatures and surface gravities should be very similar. Any di↵erences in line strength therefore translate directly into chemical abundance di↵erences. The apparent emission features near 5182 Ain˚ the spectrum of DES J033454.24540558.0 are contamination by the Littrow ghost (Burgh et al. 2007). (Right) VLT/GIRAFFE spectra of the bluer two CaT lines for the same stars.

SMC stars. If they are at the distance of the SMC, they within an angular cone of radius 0.2, and log10(J)= 2 5 are at projected separations of 27 kpc, indicating that 18.9 0.6 GeV cm within 0.5. This latter value as- they have likely been tidally stripped. The higher veloc- sumes± that the dark matter halo extends beyond the ra- ity stars have very similar velocities to the Magellanic dius of the outermost spectroscopically confirmed star, Stream gas a few degrees away from Ret II, and could but truncates within the estimated tidal radius for the therefore represent the stellar counterpart of the Stream. dark matter halo. The quoted uncertainties are 1, and are estimated by modeling the posterior probability den- 4.5. J-Factor sity function of log10(J) as a Gaussian. Note that the It is posited that dark matter particles could self- uncertainty obtained by modeling this individual system annihilate to produce gamma rays (e.g., Gunn et al. 1978; is larger than is obtained by modeling the entire popula- Bergström& Snellman 1988; Baltz et al. 2008). The tion of dSphs (Martinez 2013). large dark matter content, relative proximity, and low Several previously known ultra-faint dwarf galaxies astrophysical foregrounds of dwarf galaxies make them possess larger mean J-factors than Ret II, most notably promising targets for the detection of these gamma rays. Segue 1, Ursa Major II, and Coma Berenices (Acker- The predicted signal from the annihilation of dark mat- mann et al. 2014; Geringer-Sameth et al. 2015a; Conrad ter particles is proportional to the line-of-sight integral et al. 2015). Though the velocity dispersions of Ret II through the square of the dark matter density (e.g., Baltz and Segue 1 are consistent within uncertainties, Ret II et al. 2008), is more distant (32 kpc compared to 23 kpc) and has a larger half-light radius as measured along the major axis 2 (55 pc compared to 29 pc). The larger distance and larger J(⌦)= ⇢DM(r)ds d⌦0. (2) half-light radius imply a reduced mean J-factor relative ⌦ l.o.s. Z Z to Segue 1. In comparison to Ursa Major II, Ret II is Here, ⇢DM(r) is the dark matter particle density, and the at a similar distance, but has a velocity dispersion that integral is performed over a solid angle ⌦. The J-factor is smaller by roughly a factor of two. The larger dis- is derived by modeling the velocities using the spheri- persion, and hence mass, accounts for the larger J-factor cal Jeans equation, with assumptions on the theoretical of Ursa Major II. Coma Berenices is more distant than priors for the parameters that describe the dark matter Ret II (44 kpc compared to 32 kpc); however, the larger halo (e.g., Strigari et al. 2008; Essig et al. 2009; Charbon- velocity dispersion of Coma Berenices implies a slightly nier et al. 2011; Martinez 2013; Geringer-Sameth et al. larger mean J-factor. 2015a). Here, we model the dark matter halo as a gen- Since Segue 1, Ursa Major II, and Coma Berenices eralized Navarro-Frenk-White (NFW) profile (Navarro all possess larger J-factors than Ret II, we expect dark et al. 1997), and we use flat, ‘uninformative’ priors matter annihilation to produce a larger gamma-ray flux on the dark matter halo parameters (see Essig et al. from these objects. However, no gamma-ray excess has 2009). Using this procedure, we find an integrated J- been associated with any of the previously known dwarf 2 5 factor for Ret II of log (J) = 18.8 0.6 GeV cm galaxies (Ackermann et al. 2015). Given comparable 10 ± Reticulum II: Newest Dwarf Galaxy 5 –1–

Table 1. Reticulum II

Figure• Velocity 1. (a) DES peak color-magnitude indicative diagram of ofa Reticulumgravitationally II. Stars within 14.650 of the center of Ret II are plotted as small black dots, and stars selected for spectroscopy with M2FS, GIRAFFE, and GMOS (as describedQuantity in 2.1) are plotted as Value filled gray circles. Points surroundedbound by blackobject outlines represent the stars for which we obtained successful velocity measurements,§ and those we identify as Ret II members are filled in with red. The four PARSEC isochrones used to determine membership probabilities are displayed as black lines. (b) Spatial• Dynamical distribution of mass the observed calculated stars. Symbols from are as inthe panel width(a).Thehalf-lightradiusofRetIIfromBechtoletal.(2015)isoutlined of the 1 as a black ellipse. (c) Radial velocity distribution of observed stars, combining all threeSystemic spectroscopic Velocity data sets.v The=62 clear.8 narrow0.5kms peak of velocity dispersion1 stars at v 60 km s highlighted in red is the signature of Ret II. The hatched histogram indicates stars that are not members± of Ret II; ⇠ 1 1 note that there are two bins containing non-member stars near v =70kms that areVelocity over-plotted Dispersion on top of the red=3 histogram...3 0.7kms • Every measured characteristic of Reticulum is v ± twilight sky spectra. We fit these twilight spectra with of absolute V magnitude from Carrera et al. (2013). We consistent with the known population of dwarf Metallicity [Fe/H] = 2.65 0.07 a high resolution solar template spectrum. The scatter convert the DES g and r magnitudes to the SDSS± photo- galaxies 1 in velocity from fiber to fiber was 0.20 km s ,so metric system,Metallicity and then Dispersion use the relations[Fe/H] =0 for.28 metal-poor0.09 we conclude that the internal velocity errors over short stars from Jordi et al. (2006) to transform to± V .We 5 timescales on an individual frame (incorporating, e.g., determine absoluteDynamical magnitudes Mass assumingM1/2 =5.6 a distance2.4 10 ofM ± ⇥ any fiber-to-fiberSimon et al. systematics) 2015 (DES are Collaboration) negligible. However, 32 3 kpc (Bechtol et al. 2015) and a V -band extinction over multiple science exposures spanning several hours, of A± =0.05Mass-to-Light mag (Schlafly Ratio & FinkbeinerM/L =470 2011).210 M / L (see also Walker et al. 2015, Koposov et al 2015b) V ± this is not necessarily the case (see above). J-Factor (0.2 )logJ =18.8 0.6GeV2 cm26 5 In order to verify the reliability of our velocity zero 10 arXiv:1504.02889 arXiv:1504.03060 arXiv:1504.079163.3. Spectroscopic Membership Determination± point, we also measured the velocity of the radial ve- 2 5 J-Factor (0.53.3.1.)logM2FS 10 J =18.9 0.6GeV cm locity standard star CD 432527 by fitting it with the ± HD 122563 template, exactly as we did for the science Out of the 185 M2FS fibers placed on stars, we suc- spectra. For the two exposures on CD 432527 , we find cessfully measured velocities for 52, including a large ma- 1 1 v = 19.6 0.1kms and v = 19.9 0.1kms , jority of the observed targets brighter than g = 20.6. hel ± hel ± compared to the cataloged velocity of vhel = 19.7 The remaining stars had S/N ratios too low for spec- 1 ± 0.9kms (Udry et al. 1999). tral features to be confidently detected in the data. The velocity measurements and other properties of the stars 3.2. Metallicity Measurements are listed in Table 1. The velocity distribution we measure from the M2FS spectra exhibits a strong peak at We calculated metallicities for 16 Ret II RGB stars 1 a velocity of 60 km s (see Fig. 1), as is character- with the CaT calibration of Carrera et al. (2013). As ⇠ recommended by Hendricks et al. (2014), we measured istic of a gravitationally bound system. Approximately the equivalent widths (EWs) of the CaT lines in the same half of the stars for which we measure velocities are con- tained in this peak, with the remainder spread across a way as Carrera et al., fitting each of the three lines with a 1 wide range from heliocentric velocities from 0kms Gaussian plus Lorentzian profile. Also following Carrera 1 ⇠ et al. (2013), we adopt the line and continuum regions to 330 km s . For⇠ a large majority of the observed stars, the member- defined by Cenarro et al. (2001), except for the 8498 A˚ 1 ship status is unambiguous; stars with vhel > 90 km s line. Cenarro et al. employed a continuum bandpass of 1 ˚ and vhel < 40 km s are clearly not related to the peak 8474 8484 A for this line, but the blue limit of the associated with Ret II, while those very near the mean ˚ GIRAFFE spectra is 8482 A, so we instead use a region velocity of the system and close to the central position on the red side of the line from 8513 8522 A.˚ This spatially are almost certainly members. However, to en- wavelength range may be modestly a↵ected by two weak sure that the member sample is defined optimally we 1 Fe I lines at 8514 A˚ and 8515 A,˚ but at the metallicity carefully examine all stars within 20 km s of the mean of typical ultra-faint dwarf stars any depression of the velocity of Ret II, considering their velocities, positions in continuum should be negligible over a 9 A˚ band. the color-magnitude diagram, spatial locations, member- CaT metallicity measurements usually use the horizon- ship probabilities from Bechtol et al. (2015), and spectral tal branch (HB) magnitude to correct for the dependence features. Below we discuss the individual stars whose of the CaT EWs on stellar luminosity. The horizontal membership is not immediately obvious. branch magnitude of Ret II, however, is not well deter- Three stars in our sample have velocities of 1 mined because the galaxy contains so few HB stars. We vhel 50 km s , just to the left of the Ret II peak in ⇠ 1 therefore rely on the calibration of CaT EW as a function Fig. 1c, and about 15 km s away from the systemic Milky Way Satellite Galaxies Discovery Timeline

27 4

TABLE I. DES dSph Candidates and Estimated J-factors a b c Name (`, b) Distance log10(Est.J) GeV2 deg kpc log10( 5 ) Darkcm Matter Searches DES J0222.7 5217 (275.0, 59.6) 95 18.3 DES J0255.45406 (271.4, 54.7) 87 18.4 in Gamma Rays DES J0335.65403 (266.3, 49.7) 32 19.3 Gamma-ray Counts • Search for discrete gamma-ray DES J0344.3 4331 (249.8, 51.6) 330 17.3 Map (E > 1 GeV) sources coincident with the DES DES J0443.8 5017 (257.3, 40.6) 126dwarf galaxy candidates 18.1 Reticulum II DES J2108.85109 (347.2, 42.1)• 69No significant gamma-ray 18.3 sources detected over DES J2251.2 5836 (328.0, 52.4) 58background 18.8 DES J2339.9 5424 (323.7, 59.7)• 95Most significant 18.4 excess coincident with Reticulum II • LAT Collaboration, Pass 8: local pvalue = 0.06 (1.5σ) a Galactic longitude and latitude. • Geringer-Sameth+, Pass 7: b We note that typical uncertainties on the distanceslocal pvalue of = dSphs 0.01 (2.3σ) are 10–15%. c J-factors are calculated over a solid angle of ⌦ 2.4 10 4 sr • How does⇠ the expected⇥ dark (angular radius 0. 5). See Section 4 for morematter details. annihilation signal from Drlica-Wagner et al. (2015) Reticulum II compare to other (LAT & DES Collaboration) (see also Geringer-Sameth et al. 2015) dwarf galaxies? 28 arXiv:1503.02632 arXiv:1503.02320 LAT ANALYSIS

To search for gamma-ray emission from these new dSph candidates, we used six years of LAT data (2008 Au- gust 4 to 2014 August 5) passing the P8R2 SOURCE event class selections from 500 MeV to 500 GeV. Compared to the previous iteration of the LAT event-level analysis, Pass 8 [35] provides significant improvements in all areas of LAT analysis; specifically the di↵erential point-source sensitivity improves by 20–40% in P8R2 SOURCE V6 relative to P7REP SOURCE V15. To remove gamma rays produced by cosmic-ray interactions in the Earth’s limb, we rejected events with zenith angles greater than 100. Additionally, events from time intervals around bright gamma-ray bursts and solar flares were removed using the same method as in the 4-year catalog analysis FIG. 1. LAT counts maps in 10 10 ROI centered at each (3FGL) [36]. To analyze the dSph candidates in Table I, DES dSph candidate (white ‘ ’⇥ symbols), for E>1GeV, we used 10 10 ROIs centered on each object. Data ⇥ ⇥ smoothed with a 0. 25 Gaussian kernel. All 3FGL sources in reduction was performed using ScienceTools version 09- the ROI are indicated with white ‘+’ symbols, and those with 34-03.3 Figure 1 shows smoothed counts maps around ateststatistic> 100 are explicitly labeled. each candidate for energies > 1 GeV. We applied the search procedure presented in Acker- mann et al. [19] to the new DES dSph candidates. Specif- with a small (< 10%) energy-dependent correction to ac- 5 ically, we performed a binned maximum-likelihood analy- count for di↵erences in the LAT response. Point-like sources within each ROI from the recent 3FGL cata- sis in 24 logarithmically-spaced energy bins and 0.1 spatial pixels. Data are additionally partitioned in one of log [36] were also included in the fit. The spectral pa- four PSF event types, which are combined in a joint- rameters of these sources were fixed at their 3FGL cata- likelihood function when performing the fit to each ROI log values, while their normalizations were refit over the [19]. broadband energy range. The normalizations of 3FGL We used a di↵use emission model based on the Pass sources more than 5 away from the center are fixed at 7 Reprocessed model for Galactic di↵use emission,4 but

5 The energy dependence of the e↵ective area and energy resolution is somewhat di↵erent in Pass 7 Reprocessed and Pass 8. 3 http://fermi.gsfc.nasa.gov/ssc/data/analysis/software/ Because the Galactic di↵use emission model was ﬁt to Pass 7 4 http://fermi.gsfc.nasa.gov/ssc/data/access/lat/ Reprocessed data without accounting for the energy dispersion, BackgroundModels.html we have rescaled the model for this analysis. Dark Matter Searches in Gamma Rays

20.0 Ultra-faint dSph 19.5 Classical dSph ) 5 19.0 cm 2

18.5

(J) (GeV 18.0 10 log 17.5

17.0 1.0 0.5 0.0 0.5 Residual 1.0 50 100 150 200 250 300 350 arXiv:1503.02632 Distance (kpc) 29 Drlica-Wagner et al. (2015) (LAT & DES Collaboration) –1–

–1–

Table 1. Reticulum II

Dark Matter SearchesQuantity Value

1 Systemic Velocity v =62.8 0.5kms in Gamma Rays ± 1 Velocity Dispersion =3.3 0.7kms v ± Metallicity [Fe/H] = 2.65 0.07 ± TableMetallicity 1. Reticulum Dispersion II =0.28 0.09 [Fe/H] ± 20.0 1 Dynamical Mass M1/2 =5.6 2.4kms Ultra-faint dSph± Mass-to-LightQuantity Ratio ValueM/L =470 210 M / L 19.5 Classical dSph± 2 5 ) J-Factor (0.2)logJ =18.8 0.6GeV cm 5 10

± 1 Systemic Velocity v =62.8 0.5kms 2 5 19.0 J-Factor (0.5)logJ =18± .9 0.6GeV cm cm 10 ± 1 2 Velocity Dispersion =3.3 0.7kms v ± 18.5 Metallicity [Fe/H] = 2.65 0.07 The expected dark matter ± signal Metallicity Dispersion [Fe/H] =0.28 0.09 (J) (GeV 18.0 from Ret II is smaller than± that 10 1 Dynamicalexpected Mass fromM some1/2 =5.6 other2.4kms dwarf

log ± 17.5 Mass-to-Lightgalaxies Ratio M/L =470 210 M / L Reticulum II ± 2 5 J-Factor (0.2)log10 J =18.8 0.6GeV cm 17.0 Unlikely to see a dark matter± 2 5 1.0 J-Factor (0.5)logJ =18.9 0.6GeV cm signal from Ret 10II without± also 0.5 seeing it from other galaxies. 0.0 0.5 Residual 1.0 50 100 150 200 250 300 350 30 Simon et al. 2015 (DES Collaboration) Distance (kpc) (see also Bonnivard et al. 2015) arXiv:1504.02889 arXiv:1504.03309 More Confirmed Dwarf Galaxies

arXiv:1504.07916

31 arXiv:1506.01021 Looking Forward

• More spectroscopic follow-ups on the dwarf galaxy candidates • Magellan • VLT • AAT • Y2 data are coming! Total sky coverage >4,000 deg2 —— Find MORE candidates! • Y3 observing starts TODAY! • Dwarf galaxy is not the only interesting science in Milky Way working group. More results coming soon!

32 Summary

• Dwarf Galaxies are important cosmological probes for studying dark matter

• More ultra faint dwarf galaxy candidates have been discovered using DECam

• Coordinating follow-up • Spectroscopic confirmation • Obtaining dynamical mass and J-factors • Fermi-LAT collaborations

• More results coming from the DES Y2+ 33