The MEDIDO Project: Surprising Result for Leo I

The MEDIDO project: surprising result for Leo I

Eva Noyola McDonald Observatory

Majo Bustamante-Rosell, Karl Gebhardt (UT, Austin) Constraints on α Shape of dark matter haloes in dwarf galaxies The Astrophysical Journal Letters,775:L30(5pp),2013September20 Jardel & Gebhardt Given the freedom to choose a dark matter profile of any NGC 959 shape, it is immediately apparent that our models have chosen a variety of shapes for the dSphs. Draco appears the most similar to the NFW profile while Sculptor most closely resembles a broken power law that becomes shallower toward its center. The other galaxies host profiles that resemble neither cores nor cusps: Carina’s profile appears flat where we have kinematics but then displays a possible up-bending inside of this region. Sextans has asteeperslopethantheNFWprofileuntilitsoutermostpoint where it suddenly becomes flat. These sharp differences among dSph dark matter profiles demonstrate the variety of profile shapes in the Local Group. Unfortunately, due to a lack of central stellar velocities in the Walker et al. (2009a)data,thecentralprofilesofthedSphswe model become increasingly uncertain there. This is evidenced by the larger error bars on our gray points in Figure 1 where we have no kinematics coverage. However, we do have some constraint from projection effects and radial orbits in our models Figure 2. Combined dark matter density profiles of all the dSphs plotted on that have apocenters at radii where we do have data. the same axes. Each galaxy’s profile is plotted with the same colored points. Adams et al, 2013 Uncertainties on theseJardel points are& the Gebhardt,∆χ 2 1 uncertainties 2013 from Figure 1.We 4.1. Fornax plot the derived best-fit line with slope α = 1.2 0.5asadashedlineaswell as the NFW profile with α 1.0 as a red dashed= ± line. A fit excluding the points One galaxy must be strongly cuspy. One= other is Fornax• is an especially difficult case for non-parametric where we have no kinematics available is shown as a dotted line. The individual modeling because,Core/cusp compared to theproblem: other dSphs, it ismeasure relatively profilesthe have central been scaled to ashape common height. of DM halos marginally consistent with being(A color versionfully of this cuspy. figure is available inThe the online journal.) baryon-dominated. Our imprecise determination of M /LV in Fornax causes ρ (r)tobegreaterthanthetotalmodeleddensity∗ remainder∗ seem to require modification away from NFW. at some radii, making ρ (r)negative.InouranalysisofFornax, 1 • MissingDM Satellite problem: measurethat are different accurate from the mean velocitiesr− profile. Our interpretation for we do not plot the radial range over which this occurs as it is of this observation is that variations in their individual formation unphysical.dwarf Instead, in Figuregalaxies1 we overplot and the stellar test density DM contenthistories cause for galaxies globular to scatter from theclusters average profile. Only in red to illustrate why the subtraction is difficult in Fornax. In when multiple galaxies are averaged together does it become all other panels, ρ (r) ρDM(r)andisnotplotted. clear they follow a combined r 1 profile. This single power- ∗ ≪ − There is strong evidence from multiple studies using indepen- law profile compares well with the predicted NFW profile in dent methods that suggests that Fornax has a dark matter profile the inner portion of the plot. However, at larger radii (>1kpc that is not cuspy like the NFW profile. (Goerdt et al. 2006; in dwarf galaxies) the NFW profile becomes steeper than∼ r 1 Walker & Penarrubia˜ 2011;Jardel&Gebhardt2012). Each of − (Springel et al. 2008). More data are needed at both large and these studies only contrasts between cored and cuspy profiles small radii to further explore this. or uses a single slope to characterize the profile. It is therefore interesting to explore the non-parametric result we obtain. Even K.G. acknowledges support from NSF-0908639. This work though we cannot determine ρDM where the stellar density is greater than the total density, we can still place an upper limit would not be possible without the state-of-the-art supercomput- ing facilities at the Texas Advanced Computing Center (TACC). on ρDM such that it must not be greater than ρ or the red band in Figure 1.Giventhisconstraint,wecanseethattheouterprofile∗ We also thank Matt Walker and the MMFS Survey team for of Fornax is flat, while the inner portion rises more steeply than making their radial velocities publicly available. 1 r− .PastdynamicalstudiesofFornaxonlycomparedgeneric cored and NFW profiles and did not test this up-bending profile, REFERENCES therefore it is difficult to compare to their results. Arraki, K. S., Klypin, A., More, S., & Trujillo-Gomez, S. 2012, arXiv:1212.6651 4.2. A Common Halo? Blumenthal, G. R., Faber, S. M., Flores, R., & Primack, J. R. 1986, ApJ, 301, 27 Despite the differences in the individual profiles of the dSphs, Borriello, A., & Salucci, P. 2001, MNRAS, 323, 285 when we plot them on the same axes they appear to follow a Breddels, M. A., Helmi, A., van den Bosch, R. C. E., van de Ven, G., & 1 combined r− profile with scatter. We plot this combined profile Battaglia, G. 2013, MNRAS, 433, 3173 in Figure 2 with each galaxy’s profile as a separate color. The Burkert, A. 1995, ApJL, 447, L25 uncertainties on the points are the ∆χ 2 1uncertaintiesfrom Coleman, M. G., Da Costa, G. S., Bland-Hawthorn, J., & Freeman, K. C. Figure 1.Wehavescaledeachgalaxy’sprofilerelativetoan= 2005, AJ, 129, 1443 1 de Blok, W. J. G., McGaugh, S. S., Bosma, A., & Rubin, V. C. 2001, ApJL, arbitrary r− profile. In this way, the shape of each profile is 552, L23 preserved and only the height has been adjusted to reduce the Gao, L., Navarro, J. F., Cole, S., et al. 2008, MNRAS, 387, 536 scatter. We fit a line to the log ρDM profiles and determine that Gebhardt, K., Bender, R., Bower, G., et al. 2000, ApJL, 539, L13 the slope α 1.2 0.5. We also restrict our fit to only points in Gebhardt, K., Richstone, D., Ajhar, E. A., et al. 1996, AJ, 112, 105 the profile where= ± we have kinematics (dotted line in Figure 2) Gebhardt, K., Richstone, D., Tremaine, S., et al. 2003, ApJ, 583, 92 and find a similar slope of α 0.9 0.5. Gebhardt, K., & Thomas, J. 2009, ApJ, 700, 1690 = ± Goerdt, T., Moore, B., Read, J. I., Stadel, J., & Zemp, M. 2006, MNRAS, We conclude from Figure 2 that the average dark matter 368, 1073 1 profile in the dSphs is similar to an r− profile. However, when Irwin, M., & Hatzidimitriou, D. 1995, MNRAS, 277, 1354 we model each galaxy individually, we find a variety of profiles Jardel, J. R., & Gebhardt, K. 2012, ApJ, 746, 89

4 MEDIDO (MEasuring Dynamics In Dwarf Objects) Sample a) UGC2245 c) UGC2245 SDSS i Unresolved dwarf galaxies

M79 M79 b) DSS d)

Science Figures

a) UGC2245 c) UGC2245 SDSS i Resolved dwarf galaxies

M79 M79 b) DSS d)

Galactic globular clusters

Figure 1: Finder charts and preliminary kinematic maps for the globular cluster M79 and the low surface brightness galaxy UGC2245 that we observed during the late December/early January run. In the left two panels the VIRUS-W footprint is represented by blue regions. The upper two panels on the right show the mean line of sight velocity and velocity dispersions that we derived for UGC2245. The lower panel shows the line of sight velocity. We propose similar observations for the other targets of this proposal. For the nearby dwarfs and dwarf spheroidals, the foucus is on the central region for the core/cusp problem and a more robust analysis for the total mass in regards to the missing satellite problem. For the globular clusters, the focus on the outer region for theConstraints presence of dark matter, on and hence α have multiple pointings. Central pointings will be added from the project Central Rotation and Velocity Dispersion of Milky Way Globular Clusters (PI: Rukdee). The Astrophysical Journal Letters,775:L30(5pp),2013September20 Jardel & Gebhardt Given the freedom to choose a dark matter profile of any NGC 959 shape, it is immediately apparent that our models have chosen a variety of shapes for the dSphs. Draco appears the most similar to the NFW profile while Sculptor most closely resembles a broken power law that becomes shallower toward its center. The other galaxies host profiles that resemble neither cores nor cusps: Carina’s profile appears flat where we have kinematics but then displays a possible up-bending inside of this region. Sextans has asteeperslopethantheNFWprofileuntilitsoutermostpoint where it suddenly becomes flat. These sharp differences among dSph dark matter profiles demonstrate the variety of profile shapes in the Local Group. Unfortunately, due to a lack of central stellar velocities in the Walker et al. (2009a)data,thecentralprofilesofthedSphswe model become increasingly uncertain there. This is evidenced by the larger error bars on our gray points in Figure 1 where we have no kinematics coverage. However, we do have some constraint from projection effects and radial orbits in our models Figure 2. Combined dark matter density profiles of all the dSphs plotted on that have apocenters at radii where we do have data. the same axes. Each galaxy’s profile is plotted with the same colored points. Figure 2: Results from previous studies showing thatUncertainties the data on these is points better are the ∆χ fit2 1 byuncertainties dark frommatter Figure 1.We cusps. LEFT) 4.1. Fornax plot the derived best-fit line with slope α = 1.2 0.5asadashedlineaswell Delta chi-square vs.One central galaxy density must slope be from strongly the sample cuspy.as the of NFW Adams profile withOneα et1.0 al.otheras a red (2013). dashed= ± line.is A fit Published excluding the points results prefered Fornax is an especially difficult case for non-parametric where we have no kinematics= available is shown as a dotted line. The individual a core (slope equalmodeling zero) because, that compared are to based the other dSphs, on gas it is relatively kinematics.profiles have The been scaled stellar to a common analysis height. is more consistent with the marginally consistent with being(A color versionfully of this figurecuspy. is available in The the online journal.) theoretical expectations,baryon-dominated. but Our still imprecise su↵ determinationer from of smallM /LV statistics.in RIGHT) Mass density vs. radius for the dwarf Fornax causes ρ (r)tobegreaterthanthetotalmodeleddensity∗ remainder∗ seem to require modification away from NFW. at some radii, making ρ (r)negative.InouranalysisofFornax, 1 spheroidals as published in JardelDM & Gebhardt. Previousthat are results different from also the prefered mean r− profile. a Ourcore, interpretation whereas our average we do not plot the radial range over which this occurs as it is of this observation is that variations in their individual formation profile is very closeunphysical. to theInstead, theoretical in Figure 1 we overplot expectation the stellar density of slopehistories equal cause galaxies -1. Our to scatter goal from theis average to include profile. Only more dwarfs and obtain better estimatesin red to illustrate for why individual the subtraction isdwarfs. difficult in Fornax. In when multiple galaxies are averaged together does it become all other panels, ρ (r) ρDM(r)andisnotplotted. clear they follow a combined r 1 profile. This single power- ∗ ≪ − There is strong evidence from multiple studies using indepen- law profile compares well with the predicted NFW profile in dent methods that suggests that Fornax has a dark matter profile the inner portion of the plot. However, at larger radii (>1kpc that is not cuspy like the NFW profile. (Goerdt et al. 2006; in dwarf galaxies) the NFW profile becomes steeper than∼ r 1 Walker & Penarrubia˜ 2011;Jardel&Gebhardt2012). Each of − (Springel et al. 2008). More data are needed at both large and these studies only contrasts between cored and cuspy profiles small radii to further explore this. or uses a single slope to characterize the profile. It is therefore interesting to explore the non-parametric result we obtain. Even 3 K.G. acknowledges support from NSF-0908639. This work though we cannot determine ρDM where the stellar density is greater than the total density, we can still place an upper limit would not be possible without the state-of-the-art supercomput- ing facilities at the Texas Advanced Computing Center (TACC). on ρDM such that it must not be greater than ρ or the red band in Figure 1.Giventhisconstraint,wecanseethattheouterprofile∗ We also thank Matt Walker and the MMFS Survey team for of Fornax is flat, while the inner portion rises more steeply than making their radial velocities publicly available. 1 r− .PastdynamicalstudiesofFornaxonlycomparedgeneric cored and NFW profiles and did not test this up-bending profile, REFERENCES therefore it is difficult to compare to their results. Arraki, K. S., Klypin, A., More, S., & Trujillo-Gomez, S. 2012, arXiv:1212.6651 4.2. A Common Halo? Blumenthal, G. R., Faber, S. M., Flores, R., & Primack, J. R. 1986, ApJ, 301, 27 Despite the differences in the individual profiles of the dSphs, Borriello, A., & Salucci, P. 2001, MNRAS, 323, 285 when we plot them on the same axes they appear to follow a Breddels, M. A., Helmi, A., van den Bosch, R. C. E., van de Ven, G., & 1 combined r− profile with scatter. We plot this combined profile Battaglia, G. 2013, MNRAS, 433, 3173 in Figure 2 with each galaxy’s profile as a separate color. The Burkert, A. 1995, ApJL, 447, L25 uncertainties on the points are the ∆χ 2 1uncertaintiesfrom Coleman, M. G., Da Costa, G. S., Bland-Hawthorn, J., & Freeman, K. C. Figure 1.Wehavescaledeachgalaxy’sprofilerelativetoan= 2005, AJ, 129, 1443 1 de Blok, W. J. G., McGaugh, S. S., Bosma, A., & Rubin, V. C. 2001, ApJL, arbitrary r− profile. In this way, the shape of each profile is 552, L23 preserved and only the height has been adjusted to reduce the Gao, L., Navarro, J. F., Cole, S., et al. 2008, MNRAS, 387, 536 scatter. We fit a line to the log ρDM profiles and determine that Gebhardt, K., Bender, R., Bower, G., et al. 2000, ApJL, 539, L13 the slope α 1.2 0.5. We also restrict our fit to only points in Gebhardt, K., Richstone, D., Ajhar, E. A., et al. 1996, AJ, 112, 105 the profile where= ± we have kinematics (dotted line in Figure 2) Gebhardt, K., Richstone, D., Tremaine, S., et al. 2003, ApJ, 583, 92 and find a similar slope of α 0.9 0.5. Gebhardt, K., & Thomas, J. 2009, ApJ, 700, 1690 = ± Goerdt, T., Moore, B., Read, J. I., Stadel, J., & Zemp, M. 2006, MNRAS, We conclude from Figure 2 that the average dark matter 368, 1073 1 profile in the dSphs is similar to an r− profile. However, when Irwin, M., & Hatzidimitriou, D. 1995, MNRAS, 277, 1354 we model each galaxy individually, we find a variety of profiles Jardel, J. R., & Gebhardt, K. 2012, ApJ, 746, 89

4 Global trLeoend I among of dSphMW dwarf haloes spheroidals

' !$ F00/9< 23450

.0 6(4 6(7 # !$ A437;4

..+..*(+,- C;DBE 89:/4;<

./0/1) C;DBB ,90BB =03;4:

*(+,- ! ,90B !$ Mateo et al. 1998 MateoWilkinson et al.et al(1998), 2006 85>?@/03 WilkinsonGilmore etet al. al. 2007 (2006),

$ Gilmore et al. (2007) !$ !& !% !!$ !!# !!" ( ) • Majority of current data consistent with cored dark matter • Leo I is almost the richest MW dwarf galaxy, and it has a low7 distributions andmeasured a mass M/L scale (interior to light) of 3 10 M ⇥ 3 2 3 • Mean dark matter density 0.1 M pc (5GeV/c cm ) ⇥ 3 2 3 • NFW halo: (r < 10 pc) 60 M pc (2 TeV/c cm ) ⇥ VIRUS-W coverage

• Previously only a couple dozen stars measured in the central region by Mateo et al. (1998) Updated density proﬁle

• Central surface brightness proﬁle shows a shallow cusp, both for ground based and HST data 5

Table 1. Leo I Kinematic Datasets Properties

Bustamante et al. Mateo et al.

Instrument VIRUS-W HECTOCHELLE Number of Fibers 477 328 Angular Coverage [2.57”,1’32.84”] [7.82”,14’23.16”] Radial Velocity Range [217.91,350.99] km/s [260.10,311.10] km/s Error Range [0.38,97.74] km/s1 [1.60,7.60] km/s

Note—1: Within our sample, 224 ﬁbers are below Mateo’s maximum error. We keep the higher error measurements since their contribution gets properly weighted in our maximum likelihood estimation of Leo’s LOSVD parameters. This is explained in further detail in the a following section.Kinematics with virus-w IFU

Figure 3. Leo I data set (excluding points with errors higher than Mateo’s dataset to ease visualization.) Figure 4. UNRESOLVED CROWDING Illustrated: The contribution from many gaussians appears as one gaussian velocity The unresolved flux contribution by j fainter stars to centered closer to their mean when the gaussian width is close to the deviation of their centers. one given spectrum i, whose individual velocities vij are randomly distributed around some mean valuev ¯,will • of 6800s (GO-10520; PI: Smecker Hane). The second set have the e↵ect of biasingCrowding the measured is an velocity issue of such in central regions includes F555W imaging from the Wide Field Camera added spectrum v towardsv ¯. i 3 taken on January 2011 with an exposure time of 880s If such resulting• spectrum is to be confused with that We can’t use individual velocity(GO-12304; measurements PI: Holtzman). We create catalogs from of an individual star due to unresolved crowding, the both sets of imaging using daophot (REFERENCE). e↵ective standard deviation of the measured velocities Within our 3.2” diameter fibers we find a median of v will be lowered with respect to that of the actual i 20 stars in our deep HST catalogs. The brightest star in stellar velocities v (just as the standard deviation of the ij each fiber has a median contribution of 30% of the total mean of samples is smaller than the standard deviation flux. of the entire sample). Thus, no single star dominates the light of our fibers Particularly dense and faint environments are good and rather we are capturing the light from small popu- candidates for this confusion. lations on the order of tens of stars each o↵set by their In this section, we investigate the e↵ects of unresolved own individual radial velocity. crowding on the measurements of radial velocities for in- In light of this unresolved crowding issue, we choose dividual fibers on LEO I’s most central regions. We use not to measure radial velocities from individual VIRUS- two sets of deep, high resolution Hubble Space Telescope W fibers but rather stack fibers within radial bins and (HST) imaging to quantify the number of stars and their infer the velocity dispersion from the profiles of the Mg relative flux contribution to each fiber. The first set in- b triplet absorption features. Our kinematic measure- cludes F435W imaging from the Advanced Camera for ments are discussed in more detail in X. Surveys taken on February 2006 with an exposure time § Updated velocity dispersion profile

• Central Mateo points suffer from crowding effects • VIRUS-W measures a slight velocity rise towards the center Results from Schwarzschild modeling

• Best ﬁt black hole mass is > 106 Msun

• Best ﬁt M/L is lower than for previous models. The shape of the dark matter halo is unconstrained. Conclusions

• Leo I shows a central density cusp. Steeper than predictions from DM cosmological simulations

• The central velocity dispersion shows a clear rise towards the center

• Results from Schwarzschild modeling point to a large central BH and less dark matter than previously measured