Dark Matter Annihilation in the Galactic Center

DARK MATTER ANNIHILATION IN THE GALACTIC CENTER

Dan Hooper – Fermilab and the University of Chicago UCLA Dark Matter Conference February 18, 2016 10 Dan Hooper – DM Annihilation in the GC

Sorry if I sound like a broken record… 10

Reason 1: Overwhelming Statistical

Significance and Detailed Information

4 ! This excess consists of ~10 Thephotons Dark per square Matter meter, Interpretation per year (>1 GeV, within 10° of the Galactic Center) ! The spectral shape of the excess can be well fit by a dark matter Raw Mapparticle with a massResidual in the Map range (x3) of 7 to 12 GeV (similar to that required Dark Matter In The Galactic by CoGeNT, DAMA, and CRESST), annihilating primarily to τ+τ- Center Region (possibly among other leptons) ! The spectrum contains a bump-like feature at ~1-5 GeV

! The angular distribution of the ! Can be fit quite well by a simple 25-30 GeV dark matter particle, in a cusped signal is well fit by a halo profile withdistribution (γ~1.1), annihilating to bb with σv ~ 9 x 10-26 cm3/s -γ ρ(r) ~ r , with γ ~ 1.25 to 1.4 (in good agreement with expectations from simulations)

! The normalization of the signal requires the dark matter to have an annihilation cross section within a factor of a few of the value predicted for a simple thermal FIG. 9: The raw gamma-ray maps (left) andrelic the ( residualσv ~ 3x10 maps-26 after cm subtracting3/s) the best-fit Galactic di↵use model, 20 cm template, point sources, and isotropic template (right), in units of photons/cm2/s/sr. The right frames clearly contain a significant central and spatially extended excess, peaking at 1-3 GeV. Results are shown in galactic coordinates, and all maps ⇠ haveUCLA been smoothed DM by a 0.252014 Gaussian. of the Galactic Plane, while values greater Hooper than one and are Linden,not onlyPRD, elongated arXiv:1110.0006 along or perpendicular to the Galac- preferentially extended perpendicular to the plane. In tic Plane, but along any arbitrary orientation. Again, each case, the profile slope averaged over all orientations we find that that the quality of the fit worsens if the the is taken to be =1.3 (left) and 1.2 (right).UCLA From this DMtemplate 2012 is significantly elongated either along or per- figure, it is clear that the gamma-ray excess prefers to pendicular to the direction of the Galactic Plane. A mild be fit by an approximately spherically symmetric distri- statistical preference isDan found, Hooper however, - The for Hunt a morphology For Dark Matter L. Goodenough, D. Hooper, arXiv:0910.2998 bution, and disfavors any axis ratio which departs from with an axis ratio of 1.3-1.4 elongated along an axis ro- ⇠ FIG.unity 9: by The more raw than gamma-ray approximately maps (left) 20%. and the residual mapstated after subtracting35 counterclockwise the best-fit Galactic from the di↵ Galacticuse model, Plane 20 cm in template, point sources, and isotropic template (right), in unitsgalactic of photons/cm⇠ coordinates2/s/sr. (a similarThe right preference frames clearly was also contain found a In Fig. 11, we generalize this approach within our significant central and spatially extended excess, peaking at 1-3in GeV. our Results Inner Galaxy are shown analysis). in galactic While coordinates, this may and be aall statis- maps ⇠ UCLA DM 2010 haveGalactic been Center smoothed analysis by a 0.25 to testGaussian. morphologies that are of the Galactic Plane, while values greater than one are not only elongated along or perpendicular to the Galac- preferentially extended perpendicular to the plane. In tic Plane, but along any arbitrary orientation. Again, each case, the profile slope averaged over all orientations we find that that the quality of the fit worsens if the the is taken to be =1.3 (left) and 1.2 (right). From this template is significantly elongated either along or per- figure, it is clear that the gamma-ray excess prefers to pendicular to the direction of the Galactic Plane. A mild be fit by an approximately spherically symmetric distri- statistical preference is found, however, for a morphology bution, and disfavors any axis ratio which departs from with an axis ratio of 1.3-1.4 elongated along an axis ro- ⇠ unity by more than approximately 20%. tated 35 counterclockwise from the Galactic Plane in galactic⇠ coordinates (a similar preference was also found In Fig. 11, we generalize this approach within our in our Inner Galaxy analysis). While this may be a statis- Galactic Center analysis to test morphologies that are Dan Hooper – DM Annihilation in the GC

The Evolving Nature of the Galactic Center Debate

Dan Hooper – DM Annihilation in the GC

The Evolving Nature of the Galactic Center Debate UCLA 2010 There is no Galactic Center excess (“what are you smoking?”)

Dan Hooper – DM Annihilation in the GC

The Evolving Nature of the Galactic Center Debate UCLA 2010 There is no Galactic Center excess (“what are you smoking?”)

UCLA 2012 Sure there seems to be a Galactic Center excess, but 1) Are we sure that it is spatially extended? 2) Are we mismodeling standard diffuse emission mechanisms? 3) Is there really a Galactic Center excess?

Dan Hooper – DM Annihilation in the GC

The Evolving Nature of the Galactic Center Debate UCLA 2010 There is no Galactic Center excess (“what are you smoking?”)

UCLA 2014 What is generating this excess? 1) A large population of centrally located millisecond pulsars? 2) A series of recent cosmic ray outbursts? 3) Annihilating dark matter? Dan Hooper – DM Annihilation in the GC

The Evolving Nature of the Galactic Center Debate UCLA 2010 There is no Galactic Center excess (“what are you smoking?”)*

The Basic Features of the GeV Excess 7

! Bright and highly statistically significant ! Distributed with spherical symmetry about the Galactic Center with a flux that falls as ~r -2.4, between ~0.06° and ~10° (if interpreted as annihilating dark matter, -1.2 ρDM ~ r between ~10-1500 pc)

! The spectrum of this excess peaks at 10 10 ~1-3 GeV, and is in good agreement with

FIG. 6: Left frame: The spectrum of the dark matter component, extracted from a fit in our standard ROI (1 < b < 20, | | that predictedl < 20from) for a templatea ~35-50 corresponding GeV to a generalizedWIMP NFW halo profile with an inner slope of =1.18 (normalized to the | | flux at an angle of 5 from the Galactic Center). Shown for comparison (solid line) is the spectrum predicted from a 43.0 GeV 26 3 3 2 annihilating darkto matterbb (for particle example) annihilating to b¯b with a cross section of v =2.25 10 cm /s [(0.4GeV/cm )/⇢local] . Right frame: ⇥ ⇥ as left frame, but for a full-sky ROI ( b > 1), with =1.28; shown for comparison (solid line) is the spectrum predicted from | | 26 3 3 2 a 36.6 GeV dark matter particle annihilating to b¯b with a cross section of v =0.75 10 cm /s [(0.4GeV/cm )/⇢local] . ! To normalize the observed signal with ⇥ ⇥

annihilating ofdark the Galactic matter, plane; masking a cross the region section with b < 2 sistent with that extracted at higher latitudes (see Ap- | | -26changes3 the preferred value to =1.25 in our default pendix A).7 Shown for comparison (as a solid line) is the of σv ~ 10 ROI, cm and/s =1 is. 29required over the whole sky. In contrast to spectrum predicted from (left panel) a 43.0 GeV dark Ref. [8], we find no significant di↵erence in the slope pre- matter particle annihilating to b¯b with a cross section 26 3 3 2 ferred by the fit over the standard ROI, and by a fit only of v =2.25 10 cm /s [(0.4 GeV/cm )/⇢local] , over the southern half (b<0) of the ROI (we also find and (right panel)⇥ a 36.6 GeV dark⇥ matter particle anni- DH, Goodenough (2009, 2010), DH, Linden (2011),26 no significant di↵erence between the fit over the full sky hilating to b¯b with a cross section of v =0.75 10 and the southern half of the full sky). ThisAbazajian can be seen, Kaplinghatcm3/s [(0.4 GeV(2012),/cm3)/⇢ Gordon,]2. The spectra Macias extracted⇥ (2013), ⇥ local directly from Fig. 5, where the full-skyDaylan and southern- et al (2014),for this component Calore are, Cholis in moderately, Weniger good agreement (2014) sky fits for the same level of masking are found to favor with the predictions of the dark matter models, yielding quite similar values of (the southern sky distribution fits of 2 = 44 and 64 over the 22 error bars between 0.3 is broader than that for the full sky simply due to the and 50 GeV. We emphasize that these uncertainties (and di↵erence in the number of photons). The best-fit values the resulting 2 values) are purely statistical, and there for gamma, from fits in the southern half of the standard are significant systematic uncertainties which are not ac- ROI and the southern half of the full sky, are 1.13 and counted for here (see the discussion in the appendices). 1.26 respectively. We also note that the spectral shape of the dark matter template is quite robust to variations in ,withinthe In Fig. 6, we show the spectrum of the emission cor- range where good fits are obtained (see Appendix A). related with the dark matter template in the default ROI and full-sky analysis, for their respective best-fit values of =1.18 and 1.28.6 We restrict to energies In Fig. 7, we plot the maps of the gamma-ray sky 50 GeV and lower to ensure numerical stability of the in four energy ranges after subtracting the best-fit dif- fit in the smaller ROI. While no significant emission is fuse model, Fermi Bubbles, and isotropic templates. In absorbed by this template at energies above 10 GeV, the 0.5-1 GeV, 1-3 GeV, and 3-10 GeV maps, the dark- ⇠ a bright and robust component is present at lower en- matter-like emission is clearly visible in the region sur- ergies, peaking near 1-3 GeV. Relative to the analy- rounding the Galactic Center. Much less central emission ⇠ sis of Ref. [8] (which used an incorrectly smoothed dif- is visible at 10-50 GeV, where the dark matter compo- fuse model), our spectrum is in both cases significantly nent is absent, or at least significantly less bright. harder at energies belowFIG. 1 10: GeV, The rendering raw gamma-ray it more maps (left)con- and the residual maps after subtracting the best-fit Galactic di↵use model, 20 cm template, point sources, and isotropic templateFIG. 10: (right),The raw in gamma-ray units of photons/cm maps (left)2/s/sr. and the The residual right mapsframes after clearly subtracting contain a the best-fit Galactic di↵use model, 20 cm 2 significant central and spatially extendedtemplate, excess, peaking point sources,at 1-3 GeV. and isotropicResults are template shown in (right), galactic in coordinates, units of photons/cm and all maps/s/sr. The right frames clearly contain a significant central and⇠ spatially extended excess, peaking at 1-3 GeV. Results are shown in galactic coordinates, and all maps have been smoothed by a 0.25 Gaussian. ⇠ have been7 An smoothed earlier by version a 0.25 ofGaussian. this work found this improvement only in 6 AcomparisonbetweenthetwoROIswith held constant is the presence of the CTBCORE cut; we now find this hardening presented in Appendix A.ing to a statical preference for such a componentindependent at the ofas the shown CTBCORE in the right cut. frame of Fig. 2, respectively). The level of 17. In Fig. 8, we show theing spectrum to a statical of the preferencenormalizations for such a of component the Galactic at the Centeras and shown Inner in Galaxy the right frame of Fig. 2, respectively). The ⇠ level of 17. In Fig. 8, we show the spectrum of the normalizations of the Galactic Center and Inner Galaxy dark-matter-like component, for values of ⇠=1.2 (left signals are compatible (see Figs. 6 and 8), although the frame) and =1.3 (right frame). Showndark-matter-like for compari- component,details for of thisvalues comparison of =1.2 depend (left onsignals the precise are compatible mor- (see Figs. 6 and 8), although the son is the spectrum predicted from aframe) 35.25 GeV and WIMP =1.3 (rightphology frame). that isShown adopted. for compari- details of this comparison depend on the precise mor- annihilating to b¯b. The solid line representsson is the the contribu- spectrum predicted from a 35.25 GeV WIMP phology that is adopted. annihilating to b¯b. The solidWe line note represents that the theFermi contribu-tool gtlike determines the tion from prompt emission, whereas the dot-dashed and We note that the Fermi tool gtlike determines the tion from prompt emission,quality whereas of the the fit dot-dashed assuming a and given spectral shape for dotted lines also include an estimate for the contribution quality of the fit assuming a given spectral shape for dotted lines also includethe an dark estimate matter for template, the contribution but does not generally provide from bremsstrahlung (for the z =0.15 and 0.3 kpc cases, the dark matter template, but does not generally provide from bremsstrahlung (fora model-independent the z =0.15 and 0.3 spectrum kpc cases, for this or other compo- a model-independent spectrum for this or other compo- Dan Hooper – DM Annihilation in the GC

8 10 0.5-1 GeV residual 1-3 GeV residual 20 20 20 12

15 10 10 10 10 10 -6 -6 counts/cm 8 counts/cm 10 6 0 0 4 2 2 5 /s/sr /s/sr -10 -10 2 0 0 -20 -20 -2 20 10 0 -10 -20 20 10 0 -10 -20

3-10 GeV residual 10-50 GeV residual 20 20 1.5 4 10 10 10 10 -6 -6

3 counts/cm 1.0 counts/cm

0 2 0 0.5 2 2 1 /s/sr /s/sr -10 -10 0 0.0 -20 -20 20 10 0 -10 -20 20 10 0 -10 -20

FIG. 10: The raw gamma-ray maps (left) and the residual maps after subtracting the best-fit Galactic di↵use model, 20 cm FIG. 7: Intensity maps (in galactic coordinates) after subtracting the point source model and best-fit Galactic di↵usetemplate, model, point sources, and isotropic template (right), in units of photons/cm2/s/sr. The right frames clearly contain a Fermi bubbles, and isotropic templates. Template coecients are obtained from the fit including these three templatessignificant and central and spatially extended excess, peaking at 1-3 GeV. Results are shown in galactic coordinates, and all maps ⇠ a =1.3 DM-like template. Masked pixels are indicated in black. All maps have been smoothed to a common PSFhave of been 2 smoothed by a 0.25 Gaussian. degrees for display, before masking (the corresponding masks have not been smoothed; they reflect the actual masks used in the analysis). At energies between 0.5-10 GeV (i.e. in the first three frames), the dark-matter-like emission is clearly visible ⇠ around the Galactic Center. Daylaning to, a staticalFinkbeiner preference for such, DH, a component Linden, at the as shownSlatyer in the right, Rodd frame of Fig. (2014) 2, respectively). The level of 17. In Fig. 8, we show the spectrum of the normalizations of the Galactic Center and Inner Galaxy dark-matter-like⇠ component, for values of =1.2 (left signals are compatible (see Figs. 6 and 8), although the V. THE GALACTIC CENTER 10.0 GeV. Included in the fit is a model for theframe) Galac- and =1.3 (right frame). Shown for compari- details of this comparison depend on the precise mor- tic di↵use emission, supplemented by a model spatiallyson is the spectrum predicted from a 35.25 GeV WIMP phology that is adopted. tracing the observed 20 cm emission [45], a modelannihilating for to b¯b. The solid line represents the contribu- We note that the Fermi tool gtlike determines the In this section, we describe our analysis of the Fermi the isotropic gamma-ray background, and all gamma-raytion from prompt emission, whereas the dot-dashed and quality of the fit assuming a given spectral shape for data from the region of the Galactic Center, defined as dotted lines also include an estimate for the contribution sources listed in the 2FGL catalog [46], as well as the the dark matter template, but does not generally provide b < 5, l < 5. We make use of the same Pass 7 data from bremsstrahlung (for the z =0.15 and 0.3 kpc cases, | | | | two additional point sources described in Ref. [47]. We a model-independent spectrum for this or other compo- set, with Q2 cuts on CTBCORE, as described in the pre- allow the flux and spectral shape of all high-significance vious section. We performed a binned likelihood analysis (pTS > 25) 2FGL sources located within 7 of the to this data set using the Fermi tool gtlike,dividing Galactic Center to vary. For somewhat more distant or the region into 200 200 spatial bins (each 0.05 0.05), lower significance sources ( =7 8 and pTS > 25, and 12 logarithmically-spaced⇥ energy bins between⇥ 0.316- Dan Hooper – DM Annihilation in the GC

Spectrum of the Residuals7

FIG. 6: Left frame: The spectrum of the dark matter component,! extracted from a fit in our standard ROI (1 < b < 20, | | l < 20) for a template corresponding to a generalized NFW halo profile with an inner slope of =1.18 (normalized to the | | flux at an angle of 5 from the Galactic Center). Shown forInner comparison Galaxy (solid line) - is The the spectrum DM predictedtemplate from anaturally 43.0 GeV picks up the following spectral 26 3 3 2 dark matter particle annihilating to b¯b with a cross section of v =2.25 10 cm /s [(0.4GeV/cm )/⇢local] . Right frame: shape -⇥ the normalization⇥ of the NFW template is allowed to float as left frame, but for a full-sky ROI ( b > 1), with =1.28; shown for comparison (solid line) is the spectrum predicted from | | 26 3 3 2 a 36.6 GeV dark matter particle annihilating to b¯b with a cross section of v =0.75 10 cm /s [(0.4GeV/cm )/⇢local] . independently ⇥in every energy⇥ bin ! Daylan, Finkbeiner, DH, Linden, Slatyer, Rodd (2014) of the Galactic plane; masking the region with b < 2 sistent with that extracted at higher latitudes (see Ap- changes the preferred value to =1.25 in our| default| Galacticpendix A). Center7 Shown for - comparisonVarious (asinitial a solid seeds line) is the for the dark matter spectrum, the ROI, and =1.29 over the whole sky. In contrast to spectrum predicted from (left panel) a 43.0 GeV dark Ref. [8], we find no significant di↵erence in the slope pre-bestmatter fit spectrum particle annihilating is then to bcalculated¯b with a cross section and fed back into the fitting algorithm, 26 3 3 2 ferred by the fit over the standard ROI, and by a fit onlythe processof v =2.25 is 10repeated cm /s [(0iteratively.4 GeV/cm ) /until⇢local] ,a best fit solution is reached. We over the southern half (b<0) of the ROI (we also find and (right panel)⇥ a 36.6 GeV dark⇥ matter particle anni- 26 no significant di↵erence between the fit over the full skyfind hilatingthe final to b¯b withspectrum a cross section to be of vindependent=0.75 10 of the initial seed. 3 3 2 ⇥ and the southern half of the full sky). This can be seen cm /s [(0.4 GeV/cm )/⇢local] . The spectra extracted directly from Fig. 5, where the full-sky and southern- for this⇥ component are in moderately good agreement sky fits for the same level of masking are found to favor with the predictions of the dark matter models, yielding quite similar values of (the southern sky distribution fits of 2 = 44 and 64 over the 22 error bars between 0.3 is broader than that for the full sky simply due to the and 50 GeV. We emphasize that these uncertainties (and di↵erence in the number of photons). The best-fit values the resulting 2 values) are purely statistical, and there for gamma, from fits in the southern half of the standard are significant systematic uncertainties which are not ac- ROI and the southern half of the full sky, are 1.13 and counted for here (see the discussion in the appendices). 1.26 respectively. We also note that the spectral shape of the dark matter template is quite robust to variations in ,withinthe In Fig. 6, we show the spectrum of the emission cor- range where good fits are obtained (see Appendix A). related with the dark matter template in the default ROI and full-sky analysis, for their respective best-fit values of =1.18 and 1.28.6 We restrict to energies In Fig. 7, we plot the maps of the gamma-ray sky 50 GeV and lower to ensure numerical stability of the in four energy ranges after subtracting the best-fit dif- fit in the smaller ROI. While no significant emission is fuse model, Fermi Bubbles, and isotropic templates. In absorbed by this template at energies above 10 GeV, the 0.5-1 GeV, 1-3 GeV, and 3-10 GeV maps, the dark- ⇠ a bright and robust component is present at lower en- matter-like emission is clearly visible in the region sur- ergies, peaking near 1-3 GeV. Relative to the analy- rounding the Galactic Center. Much less central emission ⇠ sis of Ref. [8] (which used an incorrectly smoothed dif- is visible at 10-50 GeV, where the dark matter compo- fuse model), our spectrum is in both cases significantly nent is absent, or at least significantly less bright. harder at energies below 1 GeV, rendering it more con-

7 An earlier version of this work found this improvement only in 6 AcomparisonbetweenthetwoROIswith held constant is the presence of the CTBCORE cut; we now ﬁnd this hardening presented in Appendix A. independent of the CTBCORE cut. Dan Hooper – DM Annihilation in the GC

An Excess Relative to What? Although it is clear at this point that Fermi has observed an excess relative to standard astrophysical background models, it is important and reasonable to be asking to what extent we can trust and rely upon the predictions of such background models

Are there any viable astrophysical models that can explain the excess?

Do variations in the background model significantly impact the characteristics of the residual excess? FERMILAB-PUB-14-289-A

Prepared for submission to JCAP Dan Hooper – DM Annihilation in the GC

Background model systematics for the Fermi GeV excess arXiv:1409.0042 Francesca Calore,a Ilias Cholisb and Christoph Wenigera Highly Recommended!

aGRAPPA, University of Amsterdam, Science Park 904, 1090 GL Amsterdam, Netherlands bFermi National Accelerator Laboratory, Center for Particle Astrophysics, Batavia, IL 60510, !USA First comprehensive study of the systematic uncertainties on the E-mail:[email protected] astrophysical, [email protected], [email protected] backgrounds Abstract.! ConsideredThe possible gamma-ray a very excess wide in the range inner Galaxy of and models, the Galactic center with (GC) extreme variation in suggested by Fermi-LAT observations has triggered a large number of studies. It has been interpretedcosmic as a variety ray ofsource di↵erent phenomena distribution such as a signaland from injection, WIMP dark matter gas distribution, diffusion, annihilation, gamma-ray emission from a population of millisecond pulsars, or emission from cosmicconvection, rays injected in a sequence re-acceleration, of burst-like events or continuouslyinterstellar at the GC.radiation We present and magnetic fields the ﬁrst comprehensive study of model systematics coming from the Galactic di↵use emission in! the Not inner only part of does our Galaxy the and theirexcess impact onpersist the inferred for properties all such of the excess background models, the emission at Galactic latitudes 2 < b < 20 and 300 MeV to 500 GeV. We study both | | theoreticalspectral and empirical and model morphological systematics, which we deduceproperties from a large rangeof the of Galactic excess are “remarkably di↵use emission models and a principal component analysis of residuals in numerous test regionsstable” along the to Galactic these plane. variations We show that the hypothesis of an extended spherical excess emission with a uniform energy spectrum is compatible with the Fermi-LAT data in our! regionThe of excess interest at 95% does CL. Assuming not thatappear this excess to is the be extended the counterpart result ofof the the mismodeling of one seen in the inner few degrees of the Galaxy, we derive a lower limit of 10.0 (95% CL) on its extensionstandard away from astrophysical the GC. We show that, emission in light of the large processes correlated uncertainties that a↵ect the subtraction of the Galactic di↵use emission in the relevant regions, the energy spectrum of the excess is equally compatible with both a simple broken power-law of break energy E =2.1 0.2 GeV, and with spectra predicted by the self-annihilation of dark arXiv:1409.0042v1 [astro-ph.CO] 29 Aug 2014 break ± ¯ +6.4 matter, implying in the case of bb ﬁnal states a dark matter mass of m = 49 5.4 GeV. Dan Hooper – DM Annihilation in the GC 5 10 GC excess spectrum with 60 GDE models

] stat. and corr. syst. errors 5 10 1

GCsr excess spectrum with

60 GDE models 1 ] stat. and corr. syst. errors s

1 6

2 10

5 5 10 10 ⇥ ⇥ sr 1.2 1 I II

1.0

s 6

2 10 0.8

0.6

[GeV cm 0.4 7 0.2 10 0.0 0.2 [GeV cm dN/dE 6 6 7 2 10 10 10 4 ⇥ ⇥ E III IV ssr)] 3 2 dN/dE 2

(cm 2 8 / E 10 0 1 1 2

10 [GeV 10 10 dE dN 0

8 2 E [GeV]

10 E 0 1 2 1 10 10 10 10 6 10 6 E [GeV] ⇥ ⇥ 1.5 V VI

Figure 14. Spectrum of the GCE emission for1.0 model F (black dots) together with statistical and

systematical (yellow boxes, cf. ﬁgure 12) errors.0 We.5 also show the envelope of the GCE spectrum for

Figure 14. Spectrum of the GCE emission forall model 60 GDE F (black models dots ()blue together dashed with line, statistical cf. figure and70)..0 systematical (yellow boxes, cf. figure 12) errors. We also show the envelope of the GCE spectrum for0.5 6 6 all 60 GDE models (blue dashed line, cf. figure 7). 20 10 10 X 3.0 ⇥ ⇥ 2.5 VII VIII

15 XI ssr)] 20 V 2 2.0

(cm 1.5

X 10 / 15 XI III 1.0

[GeV 0.5 V #ROI Deﬁnition ⌦ROI [sr] 5 VII VIII 0.0 dE 10 I dN 2 0.5 3 III E 2 2 0 I, II p` + b < 5, b> ` 6.0 10

[deg] 1.0 5 VII VIII #ROI Definition ⌦ROI [sr] 6 ± | | 6 ⇥ 2 b 10 2 2 10 I II III,1. IV5 ⇥ 5 < p` + b < 10, b>⇥ ` 1.78 10 5 2 2 3 ± | | ⇥ 2 0 I, II p` + b < 5, b> ` 6.0 10V, VI 10 < p`2 + b2 < 15,IX b> ` 2.93 10 X [deg] 1.0 IV± | | ⇥ 2 ± | | ⇥ b 2 2 2 II III, IV10 5 < p` + b < 10, b> ` 1.78 10VII, VIII 5 < p`2 + b2 < 15, ` > b 3.54 10 5 0.5 2 2 ± | | ⇥ 2 ± | | ⇥ 1 V, VI 10 < p` + b ` 2.93 10 IX 15 < p`2 + b2 < 20 1.51 10 IV 15 ± | | ⇥ 0.0 10 p 2 2 2 2 2 ⇥ 1 VII, VIII 5 < ` + b < 15, ` > b 3.54 10 X20 < p` + b 1.01 10 ± | | ⇥ 1 0.5 ⇥ VI IX20 15 < p`2 + b2 < 20 1.51 10 15 20 15 10 5 0 5 10 15 20 p 2 2 ⇥ 1 1.0 X20 < ``[deg]+ b 1.01 10 100 101 102 100 101 102 20 ⇥ 20 15 10 5 0 5 10 15 20 E [GeV] E [GeV] ` [deg] Figure 15. Geometry of the ten GCE Table 3. Definition of the ten GCE segments that are segments used in our morphology anal- Figureshown 16 in. Same figure as figure15, as14, function but from a fit of with Galactic the segmented latitude GCEb templateand as illustrated in Figure 15.Calore Geometry of, Cholis the ten GCE, WenigerTable 3. Definition, arXiv:1409.0042 of the ten GCE segments that are ysis, see table 3. figurelongitude15. We` show, together results for with GDE theirmodel F angular (black dots size), as⌦ wellROI as. the envelope for all 60 GDE segments used in our morphology anal- shown in figure 15, as function of Galactic latitudemodelsb and (blue dotted lines) and the systematic errors that we derived from fits in 22 test regions along the Galactic disk (yellow boxes, in analogy to figure 12). See figure 28 below for the spectra of all ysis, see table 3. longitude `, together with their angular size ⌦ROIcomponents.. the fit. The definition of the segments aims at studying the symmetries of the GCE around the GC: Allowing regions in the North (I, III, and V) and South (II, IV, and VI) hemisphere, the fit. The definition of the segments aims at studying the symmetries of the GCE around as well as in the West (VII) and East (VIII) ones, to vary independently, we can test the the GC: Allowing regions in the North (I, III, and V) and South (II, IV, and VI) hemisphere, spectrum absorbed by the GCE template in the di↵erent regions of the sky. Moreover, with as well as in the West (VII) and East (VIII) ones, to vary independently, we can test the – 32 – spectrum absorbed by the GCE template in the di same↵erent segments, regions we of the can sky. investigate Moreover, its with the extension in latitude. the same segments, we can investigate its the extensionTo facilitate in latitude. the study of morphological properties of the excess, we furthermore allow To facilitate the study of morphologicaladditional properties latitudinal of the excess, variations we furthermore in the ICS allow components of the individual GDE models. We additional latitudinal variations in the ICS componentssplit our ICS of the component individual into GDEnine models. ICS segments We , corresponding to 9 latitude strips with boundaries at b =2.0 ,2.6 ,3.3 ,4.3 ,5.6 ,7.2 ,9.3 , 12.0 , 15.5 and 20 . We then allow split our ICS component into nine ICS segments, corresponding| | to 9 latitude strips with boundaries at b =2.0 ,2.6 ,3.3 ,4.3 ,5.6 ,7the.2 normalization,9.3 , 12.0 , 15 of.5 theand ICS 20 strips. We to then vary allow independently, though we keep the normalization | | the normalization of the ICS strips to vary independently, though we keep the normalization

– 30 – – 30 – Dan Hooper – DM Annihilation in the GC

10 26 25 10 4.0 ⇥ ¯bb ¯bb + ! ⌧ ⌧ 3.5 m =49GeV Hooper & Slatyer 2013 Huang+ 2013 3.0 Daylan+ 2014 ] ]

1 Abazaijan+ 2014 1

s s 2.5

3 Gordon+ 2014 3 26 10 [cm [cm

i i 2.0 v v h h

1.5

1.0

27 10 0.5 101 102 0.91.01.11.21.31.41.5 Calore, Cholis, McCabe, Weniger, 1411.4647; m [GeV] Calore, Cholis, Weniger, arXiv:1404.0042 Figure 18. Left panel: Constraints on the v -vs-m plane for three di↵erent DM annihilation channels, from a fit to the spectrum shown in figureh i 14 (cf. table 4). Colored points (squares)referto best-fit values from previous Inner Galaxy (Galactic center) analyses (see discussion in section 6.2). Right panel: Constraints on the v -vs- plane, based on the fits with the ten GCE segments. h i

10 25 1.0 ⇥ I(0.89) VI (0.64) II (0.15) VII (0.55) 0.8 III (0.15) VIII (0.04) IV (0.90) IX (0.19) V(0.20) X(0.31) ] 1

0.6 s 3 95% CL [cm i v 0.4 h

0.2

0.0 10 20 30 40 50 60 70 80 90 100 m [GeV]

Figure 19. Constraints on the v -vs-m plane at 95% CL, individually for the GCE template segments shown in figure 15, for theh channeli ¯bb. The cross indicates the best-fit value from a fit ! 26 3 1 to all regions simultaneously (m 46.6 GeV, v 1.60 10 cm s ). Note that we assume a NFW profile with an inner slope of' =1.28. Theh i' individual⇥ p-values are shown in the figure legend; the combined p-value is 0.11.

mass fixed at 49 GeV. This plot is based on the fluxes from the segmented GCE template, see figure 16. As expected, the cross-section is strongly correlated with the profile slope. We

– 35 – Dan Hooper – DM Annihilation in the GC

The Fermi Collaboration has recently presented their first paper on this subject (arXiv:1511.02938), reporting an excess with a similar spectrum 24 Fermi–LAT Collaboration and morphology to that reported by previous groups tions made in the transformation for the site lines over the 15 15 region about the GC have an impact on the interstellar⇥ emission and point sources in the maximum-likelihood fitting and consequently the spatial distribution of residuals. Approximations made interpolating the gas column density across the l 10 range can result in an incorrect gas density distribution± along the line-of-sight. Spurious point sources in the analysis and structure in residuals can result from this because a higher/lower CR intensity compared to where the gas should be placed is used in creating the interstellar emission templates. The scaling procedure for the IEM then adjusts the individual annuli potentially producing low-level artifacts due to a combination of the effects described above. To obtain an estimate of the uncertainties associated with misplacement of the gas new maps of the column density per annuli are created. 10% of the H I gas column density is randomly displaced over the annuli and recombined with the ⇡0-decay emissivity 29 in each annulus to create modified Figure 18. Same as in Figure 13, but with the spectrum of the NFW profile intensity maps for this process, which are summed to pro- modeled with a power-law per energy band over the 1 100 GeV range. The envelopes include the fit uncertainties for the normalisation and spectral duce new fore-/background intensity maps. The 68% frac- indices. tional change per pixel from 100 such realisations for each Simona Murgia, et al. (Fermi Collaboration)IEM is compared with the fore-/background resulting from through the line-of-sight to the GC. the scaling procedure (Sec. 3.1). Depending on the IEM and The IEM fitting interior to the solar circle uses the tangent energy range, variations from 1% to 15% in the intensity per ranges for positive and negative longitudes to obtain parame- pixel for the fore-/background from the structured interstel- ters for the annuli 2 4 (Table 5). To examine the effect of lar emission across the 15 15 region are obtained, with the azimuthal averaging, fits to the tangent ranges were made the largest for OBstars index-scaled⇥ and smallest for the Pul- for positive and negative longitudes to gauge the difference in sar intensity-scaled IEM, respectively. Because of the some- the parameters for the IEMs obtained when considering each what arbitrary choice of the precise fraction of H I column separately. The scaling factors for annulus 4 obtained when density30 that is redistributed over the annuli these variations fitting negative and positive longitude ranges were statistically are illustrative rather than providing a true ‘systematic uncer- consistent 28 with those found when fitting both ranges com- tainty’ associated with the gas misplacement. Note that the bined. For annuli 2 and 3 the fits to the positive and nega- uncertainty is maximised toward the GC because it is furthest tive tangent longitude ranges result in scaling parameters that away from the gas column density interpolation base points at differ by factors up to 2 from each other, which is well l 12. beyond the statistical uncertainty;⇠ the average value obtained ⇠ ± by fitting both tangent ranges together is approximately in- 6. SUMMARY between for the intensity-scaled IEMs over annuli 2 and 3. The analysis described in this paper employs specialised For the index-scaled IEMs the spectral parameters are harder IEMs that are fit to the -ray data without reference to the or softer than the average when using the positive/negative 15 15 region about the GC. Finding point-source seeds tangent ranges individually for annuli 2 4. However, there ⇥ for the same region using a method that does not rely on de- is no clear trend and the over/under-prediction is not confined tailed IEMs, the source-seeds and IEMs are combined in a to a particular energy interval. maximum-likelihood fit to determine the interstellar emission The uncertainty for the IEM fore-/background flux toward across the inner 1 kpc about the GC and point sources the GC due to the azimuthally averaged IEMs is difficult to over the region. The⇠ overwhelming majority of -ray emis- quantify precisely. A minimal estimate can be made from the 0 sion from the 15 15 region is due to interstellar emission statistical uncertainty for the annulus 4 ⇡ -decay flux for each and point sources.⇥ To summarise the results for these aspects IEM, because the fit results for the combined tangent ranges of the analysis: are within these uncertainties when fitted to the positive and 8 negative ranges individually. Above 1 GeV this is 4 10 15 15 2 1 ⇠ ⇥ The interstellar emission over the region is ph cm s for the 15 15 region about the GC across all • 85% of the total. For the case of fitting⇥ only ‘stan- IEMs. This is comparable⇥ to the fitted flux from annulus 1 ⇠ 0 dard’ interstellar emission processes and point sources ⇡ -decay or the TS < 25 point sources over the same region. the fore-/background is 80% with the remaining Any analysis employing the Galactocentric annulus decom- 20% mainly due to IC⇠ from the inner region. The position for the gas column densities is subject to the loss of contribution⇠ by the ⇡0-decay process over the inner re- kinematic resolution for sight lines within l 12 of the ⇠ ± gion is much less than the IC, with the relative contri- GC/anti-GC. Appendix B of Ackermann et al. (2012a) details butions by the H I- and CO-related emission suppressed the transformation of H I and CO gas-survey data into the col- compared to the GALPROP predictions. umn density distributions over Galactocentric annuli used in this analysis, and employed by many others. The assump- 29 The contribution by CO-related ⇡0-decay emission is the same as that obtained from the scaling procedure. 28 The average statistical uncertainty for the normalisation of each inter- 30 Similar modifications of the CO column density distribution are not stellar emission component per annulus is 10%, except for annuli 2 and 3; explored because the detailed knowledge to make a truly informed estimate see Appendix A. ⇠ is not available. Dan Hooper – DM Annihilation in the GC

What Produces the Excess?

! Non-standard cosmic-ray models? ! A large population of centrally located millisecond pulsars? ! Annihilating dark matter? Dan Hooper – DM Annihilation in the GC Beyond Standard Diffuse Backgrounds

! Although Calore, Cholis and Weinger showed that the excess is robust to standard variations in the diffuse emission model, one might wonder whether a less standard scenarios might work ! To accommodate the morphology of the observed excess, one needs a very strong and steep concentration of cosmic ray sources and/or gas distributed symmetrically about the Galactic Center

! Two main difficulties: 1) Diffusion broadens the profile of cosmic rays, making it difficult for steady-state models to account for the innermost 1-2° of the excess 2) The gamma-ray spectra that result from cosmic ray processes are less sharply peaked than the observed excess

! Together, these considerations make it very difficult for steady-state cosmic ray scenario to account for the excess (although such models could plausibly alter the inferred characteristics of the excess) COSMIC RAY PROTONS IN THE INNER GALAXY AND … PHYSICAL REVIEW D 90, 023015 (2014) In Fig. 3 we investigate the overall spatial distribution of As long as the injection episode is recent enough, the the emission from a new population of cosmic ray protons morphology primarily traces the distribution of cosmic ray injected in the Galactic Center region. The figure shows the protons, and is relatively insensitive to the details of the gamma-ray flux associated with a central proton source target gas density distribution—the diametrically opposite COSMIC RAY PROTONS IN THE INNER GALAXY AND … PHYSICAL REVIEW D 90, 023015 (2014) for benchmark impulses of age 0.5, 2.5, and 19 Kyr (upper regime from what is assumed in the diffuse Galactic In Fig. 3 we investigate thepanels) overall spatial and of distribution 100 Kyr, of 2 Myr,As long as well as the as a injection continuous episode isemission recent enough, background the models of Refs. [20,21]. the emission from a new populationsource of (lower cosmic panels). ray protons We usemorphology a linear scale primarily in the traces three the distributionIt is of evident cosmic ray that the sub-Myr simulations show a injected in the Galactic Center region. The figure shows the protons, and is relatively insensitive to the details of the gamma-ray flux associated withupper a panels central to proton help sourcethe readertarget visually gas compare density distribution our results—thesignificant diametrically degree opposite of spherical symmetry outside the for benchmark impulses of agewith 0.5, what 2.5, and is shown 19 Kyr e.g. (upper in Fig.regime 9, right from panels, what of isRef. assumed[21]. inmasked the diffuse regions. Galactic Also, an excess with the same morpho- panels) and of 100 Kyr, 2 Myr,To the as end well of as emphasizing a continuous the emission background outside the models Galactic of Refs.logical[20,21] aspect. as in Fig. 9, right panels, of Ref. [21] can be source (lower panels). We useplane, a linear we scale instead in theemploy three a logarithmicIt is evident scale forthat the the older sub-Myreasily simulations reproduced show aby young or very young sources, as upper panels to help the readerbursts visually and compare continuous our results sourcessignificant in the lower degree panels. of In spherical each symmetryshown in outside the three the upper panels. As the diffusion time with what is shown e.g. in Fig.case, 9, right the panels, fluxes of are Ref. rescaled[21]. suchmasked that regions. the maximum Also, an fluxexcess withincreases the same to morpho- several Myr, the emission profile becomes To the end of emphasizing theequals emission unity. outside The the Galactic Galactic planelogical mask aspect ( b < as1 in) Fig.is bounded 9, right panels,more of Ref. elongated[21] can and be spherical symmetry is degraded. At easily reproduced∘ by young or very young sources, as plane, we instead employ a logarithmicby white lines scale (or for is the masked older out) and referencej j reticles have higher latitudes ( b 2 ), most of the spherical symmetry bursts and continuous sources in the lower panels. In each shown in the three upper panels. As the diffusion timej j ≳ ∘ case, the fluxes are rescaled suchbeen that overlaid the maximum at radial flux incrementsincreases of 2°. to several Myr, the emissionis, however, profile becomes restored as the molecular and atomic gas equals unity. The Galactic plane maskThe top( b < three1 ) is panels bounded show thatmore a elongated recent (from and a spherical fraction symmetrydistributions is degraded. fall off, At and the ionized component produces a j j ∘ by white lines (or is masked out)of and a Kyr reference to tens reticles of have Kyr) impulsivehigher latitudes cosmic ( b ray2∘ proton), most of themore spherical isotropic symmetry emission. In the template analyses of been overlaid at radial incrementsinjection of 2°. event in the Galactic Centeris, however, region restored yieldsj j ≳ a as highly the molecularRefs. and[20,21] atomic, a gas portion of this residual ridge emission The top three panels showspherically that a recent symmetric (from a fraction and concentrateddistributions source, fall off, with and the mor- ionizedmay component also be produces absorbed a by the Fermi diffuse model, although of a Kyr to tens of Kyr) impulsivephological cosmic properties ray proton very closelymore resembling isotropic emission. and match- In theit template is difficult analyses to exactly of pinpoint this effect without repeat- injection event in the Galacticing Center those region found yields in the a highly GalacticRefs. Center[20,21] analysis, a portion of Ref. of[21] this residualing the ridge full maximum emission likelihood analysis. It is also evident spherically symmetric and concentrated source, with mor- may also be absorbed by the Fermi diffuse model, although phological properties very closely(see their resembling Fig. 9, and right match- panels),it as is well difficult as in to the exactly GCE pinpoint source thisthat effect gas without structure repeat- is mostly washed out for recent impulsive ing those found in the Galacticresiduals Center analysis shown of in Ref. the[21] bottoming panels the full of Fig. maximum 1 in Ref. likelihood[20], analysis.sources, It is and also that evident it becomes increasingly more prominent (see their Fig. 9, right panels),and as well in the as in residual the GCE found source in Ref.that[19] gas structureand shown is mostly in Fig. washed3. outfor for older recent sources impulsive and for the continuous emission cases. residuals shown in the bottom panels of Fig. 1 in Ref. [20], sources, and that it becomes increasingly more prominent and in the residual found in Ref. [19] and shown in Fig. 3. for older sources and for the continuous emission cases.

Dan Hooper – DM Annihilation in the GC

A Series of Cosmic Ray Outbursts? ! To address these challenges, it has been proposed that recent burst-like injection of cosmic rays might be responsible for the excess ! Hadronic scenarios predict a signal that is more disky than spherical; incompatible with the data ! In more generality, the small- scale structure of the excess does not correlate with the distribution of gas (Daylan et al. 2014), disfavoring a hadronic cosmic ray origin for the excess

FIG. 3 (color online). Hadronic gamma-ray flux density at 2 GeV from an approximately central source of high-energy protons integrated over the line ofFIG. sight. 3 We (color show online). impulsive sourcesHadronic of increasing gamma-ray age flux in all densitypanels with at the 2 exceptionGeV from of the an bottom approximately central source of high-energy right which shows a continuouslyprotons emitting integrated source in steady over state.the line For of each sight. map, We the show fluxes impulsive are normalized sources to the of maximum. increasing For age the inease all panels with the exception of the bottom Carlson, ofProfumo comparing, PRD, the morphologyarXiv:1405.7685, of the claimed GCE in Ref. [21] shown in their fig. 9, we employ a linear scale in the three upper panels. Petrovic, TheSerpico three, lowerZaharijas panels, arXiv:1405.7928 employ, instead,right which a logarithmic shows a scale continuously to enhance emitting the features source of the in steady emission state. outside For the each Galactic map, the plane fluxes region. are normalized to the maximum. For the ease of comparing the morphology of the claimed GCE in Ref. [21] shown in their fig. 9, we employ a linear scale in the three upper panels. Also overlaid are reference reticles in increments of two degrees and indicators of the Galactic plane mask b < 1∘. All maps have been smoothed by a Gaussian of widthTheσ three0.25 lowerto match panels Ref. employ,[21]. instead, a logarithmic scale to enhancej j the features of the emission outside the Galactic plane region. ¼ ∘ Also overlaid are reference reticles in increments of two degrees and indicators of the Galactic plane mask b < 1∘. All maps have been smoothed by a Gaussian of width σ 0.25 to match Ref. [21]. j j ¼ ∘ 023015-7

023015-7 Dan Hooper – DM Annihilation in the GC

Cholis, Evoli, Calore, Linden, Weniger, DH, arXiv:1506.05104 Dan Hooper – DM Annihilation in the GC

Angle from the Galactic Center

Cholis, Evoli, Calore, Linden, Weniger, DH, arXiv:1506.05104 Dan Hooper – DM Annihilation in the GC

Angle from the Galactic Center

Cholis, Evoli, Calore, Linden, Weniger, DH, arXiv:1506.05104 Dan Hooper – DM Annihilation in the GC

Angle from the Galactic Center

Cholis, Evoli, Calore, Linden, Weniger, DH, arXiv:1506.05104 Dan Hooper – DM Annihilation in the GC

A Series of Cosmic Ray Outbursts? ! Leptonic outburst scenarios (Petrovic et al.) are more difficult to rule out ! After exploring a wide range of leptonic outburst scenarios, there appear to be two main challenges (among others): 1) The morphology from a given ~1048 erg, ~103 yr outburst is “convex”, whereas the Softer Spectra data is “concave” – to fit the data, ~10 49 erg, ~104 yr we need several outbursts, with ~1050 erg, ~105 yr highly tuned parameters 2) The gamma-ray spectrum is ~10 51 erg, ~106 yr approximately uniform across the Hard Spectrum Inner Galaxy, but energy losses LogFlux] [ should lead to softer emission from the outer regions – to fit the data, we need the older outbursts to Angle from the Galactic Center inject electrons with higher energies than more recent outbursts

Cholis, Evoli, Calore, Linden, Weniger, DH, arXiv:1506.05104 Dan Hooper – DM Annihilation in the GC

Gamma-Rays From Millisecond Pulsars 2

! Although noneered have in Ref. [47]been do not yieldidentified spectra that arenear compatible with the observed emission) [3, 4, 6]. In the case the Galactic Center,of a burst dominated it seems by high-energy plausible cosmic ray electrons, in contrast, such an event could potentially yield that large numbersa somewhat of more MSPs spherically could symmetric distributionexist in of gamma-rays (due to their inverse Compton scattering this region with radiation rather than with the disk-like distribution of gas) [50], although the accompanying bremsstrahlung emission would be disk-like. It is very difficult, however, ! Fermi has observedto simultaneously gamma-ray account for the observed emission spectrum and morphology of the gamma-ray excess in such a scenario. from ~75 knownFurthermore, MSPs the energy-dependance of diffusion would lead to a more spatially extended distribution at higher ! The average observedenergies, in contrast spectra to the energy-indepenent from these morphology reported in Ref. [1].2 sources is similarThe second to that category of of proposedthe Galactic astrophysical explanations for the gamma-ray excess are scenarios involving Center excessalargepopulationofunresolvedgamma-raysources.Mil- – this is the main reason FIG. 1: The measured spectral shape (blue error bars) and lisecond pulsars (MSPs) are known to exhibit a spectral best fit parameterizaation (blue dashed) of the stacked emis- shape that is similar to that of the observed excess, and sion from 61 millisecond pulsars observed by Fermi [54] (black that MSPs have been considered as a dashed) compared to that of the observed gamma-ray ex- have thus received some attention within this context [3– cess [1] (black error bars). Also shown is the spectral shape possible explanation8, 53]. In this for letter, the we discuss excess what is known about from the stacked emission from 36 globular clusters (red er- the spectrum, luminosity function, and spatial distribu- ror bars) [54], and the spectrum predicted from a 35.5 GeV tion of millisecond pulsars in the Milky Way, and use WIMP annihilating to b¯b (black solid). ! The luminositythis function information to evaluateof MSPs whether theyhas might been be able measured fromto account the forobserved the observed gamma-ray population excess. The Measured Spectra of Millisecond Pulsars: We have that of an unbiased sample of MSPs. The spectra ob- recently reported measurements of the gamma-ray spec- served from Fermi’s globular clusters (shown in Fig. 1 (both for thosetra MSPs of 61 MSPs in observed the byfield the Fermi of the Gamma-Ray as red error bars [54]) is even softer than that from Space Telescope, using data collected over a period of MSPs [54], however, and provides a very poor fit to the Galaxy and within5.6 years globular [54]. The best-fit clusters) spectrum of this collection observed excess. of (stacked) sources is shown in Fig. 1, and compared to Prior to the study of Ref. [1] and their application the spectrum of the observed gamma-ray excess. Over- of cuts to CTBCORE [46], significant systematic uncer- Cholis, DH, Linden,all, thearXiv:1407.5625, spectral shape of the gamma-ray 1407.5583 excess is fairly tainties complicated the determination of the low-energy similar to that observed from MSPs, and this comparison spectrum of the gamma-ray excess (for an illustrative ex- DH, Mohlabeng, arXiv:1512.04966has motivated an unresolved population of such sources ample, see Fig. 10 of Ref. [8]). After cutting on CTB- as a possible source of the Galactic Center gamma-ray CORE, however, the shape of the low-energy spectrum excess. At energies below 1GeV,however,thespec- ⇠ is much more robust to variations in analysis procedure. trum observed from MSPs is significantly softer than is And while imperfections in the diffuse emission model exhibited by the excess. used may impact the spectral shape of the excess, the At this time, a few comments are in order. First, if variations considered in Ref. [51] do not favor the possi- the observed catalog of gamma-ray MSPs is not repre- bility of a significantly softer low-energy spectrum than sentative of the overall population, it is possible that was found in Ref. [1]. the stacked spectrum could differ from that produced The Observed Distribution of MSPs in the Milky Way: by a large and unbiased collection of such objects. The Along with many MSP detections made at radio wave- gamma-ray emission from globular clusters is dominated lengths, Fermi has reported the observation of gamma- by MSPs, and their spectra has often been presented as rays from 62 MSPs. While most of these objects have been found in or around the disk of the Milky Way, some have also been observed to reside within globular clusters. In the left frame of Fig. 2, we plot the distribu- 2 When considering models which invoke extreme physical conditions to account for the excess at the Galactic Center, it may be tion of Fermi’s MSPs on the sky. This population has necessary to reevaluate the contributions from pion production, been shown to be well described by a thick disk-like dis- bremsstrahlung, and inverse Compton emission. In the forthcom- tribution, with an exponential scale height of 0.5-1.0 ⇠ ing study of Calore et al. [51], a wide range of diffuse emission kpc [56, 57]. In the right frame of Fig. 2, we use a MSP models are considered, accounting for a wide variety of physi- thick-disk distribution model fit to this population to cal conditions in the inner region of the Galaxy, finding that a spherical excess with a profile similar to that predicted by dark estimate the morphology predicted from the unresolved matter annihilations is preferred by the data in all models (see members of this population (solid contours). This pre- also Ref. [52]). diction is very elongated along the disk, and does not Dan Hooper – DM Annihilation in the GC Could Millisecond Pulsars Generate the Galactic Center Excess? ! From the measured luminosity function, we find that approximately 2000 to 15,000 MSPs would be required within a few kpc of the Galactic Center to account for the excess ! Taking into account Fermi’s detection threshold, the 3FGL catalog should contain ~20-40 MSPs from this region (in fact, only a few candidates exist) ! Estimates based on the numbers of bright LMXBs observed in globular clusters and in the Galactic Center lead us to expect that MSPs might account for ~1-5% of the observed excess ! If MSPs account for this signal, the luminosity function of this population is very different from that observed elsewhere in the Milky Way, without many bright members ! Perhaps young MSPs continue to be spun-up in both the field of the Milky Way and in globular clusters, but not in the Galactic Center, allowing the Galactic Center population to contain fractionally fewer bright MSPs

Cholis, DH, Linden, arXiv:1407.5625, 1407.5583 DH, Mohlabeng, arXiv:1512.04966 4

Isotropic Point Sources • The source-count function for the isotropic PS template is also modeled as a broken power law, as in (S8), except (p) with Ap µp,iso ✏ . As above, the isotropic PS template can be specified by either its overall normalization / iso / iso parameter APS or the pixel-averaged intensity IPS . For the IG analysis, the contribution from the isotropic PS component is subdominant and thus poorly con- strained. Scanning over the isotropic PS parameters slows down the NPTF, so for the majority of the IG analyses, we fix the source-count–function parameters to the best-fit values found in the high-latitude analysis. Removing this constraint leaves the results unchanged, as discussed in Sec. IV A.

Given these templates and their associated generating functions, the overall photon-count probability distribution (p) pk (✓) can be written as a function of the 12 parameters

✓ = A ,A ,A ,A ,A ,S ,n ,n ,Aiso,Siso,niso,niso . (S12) { iso di↵ bub NFW PS b 1 2 PS b 1 2 }

Then, for a data set d consisting of the set of np photon counts in each pixel p, the likelihood function for observing a particular photon-count distribution over all{ pixels} in the ROI is

p(d ✓, )= p(p)(✓) . | M np (S13) p Y With the priors speciﬁed above, this likelihood function can be used in the standard framework of Bayesian inference to compute both the posteriors and the evidence for models that include various subsets of the parameters ✓.We use the MultiNest package for the Bayesian calculations [25,M 26].

Dan Hooper – DM Annihilation in the GC B. Data Selection Criteria

The NPTF analysis was performed using the Extended Pass 7 Reprocessed Fermi data from August 4, 2008 ⇠ toEvidenceDecember 5, 2013 made For available byUnresolved [11]. Ultraclean front-converting Point events with zenithSources? angle less than 100 and⇠ “DATA QUAL==1 && LAT.CONFIG==1 && ABS(ROCK.ANGLE) < 52” are selected, and a Q2 cut on the CTBCORE parameter is used to remove events with poor directional reconstruction. The main body of the Report focused ! Twoprimarily recent on two regionsstudies of interest: find a that high-latitude ~1-10 analysis GeV with photonsb 30 and from an IG analysis the direction that included allof pixels the | | within 30 of the GC, with b 2. These regions are shown in Fig. S1. When masking identiﬁed PSs from the Inner Galaxy are more| | clustered than expected, suggesting that the GeV Fermi 3FGL catalog [15], all pixels within 5 0.198 of the source are excluded. This mask is suciently large to excesscompletely containmight the be ﬂux generated from the majority by⇥ of thea population PSs; the results doof not unresolved qualitatively change point as the sources mask size is varied, for example, to 7 0.198. ⇥

50 50

counts counts latitude [degrees]

0 0

counts longitude [degrees]

FIG. S1: The counts map for the high-latitude ( b 30) analysis (clipped at 15 counts)counts is shown in the left panel. The IG | | 0 50 analysis focuses on the region within 30 of the GC, with b 2.Theassociatedcountsmap(clippedat50counts)isshown | | in the right panel. All pixels within 1 of known Fermi 3FGL sources are masked. ⇠ Lee, Lisanti, Safdi, Slatyer, Xue, arXiv:1506.05124 (see also Bartels, Krishnamurthy, Weniger, arXiv:1506.05104)

90 0 90 0

-1 -1 45 45 -2 -2 0 0 -3 -3 -45 -45 -4 -4

-90 -5 -90 -5 180 90 0 -90 -180 180 90 0 -90 -180

90 0

Dan Hooper – DM Annihilation in the GC -1 45 Evidence For Unresolved Point Sources? ! Lee et al. use smooth and point source population templates that trace the following morphologies: 1) The dark matter density squared (tracing the excess) 2) The (default) Fermi diffuse model 3) The Galactic Disk 18 18 -2

50 6 30 50

50 6 50

counts counts counts counts counts 0 counts

latitude [degrees] latitude [degrees] latitude [degrees] latitude [degrees] 0 0 0 0 0 0 0 0

longitudelongitude [degrees] [degrees] longitudelongitude [degrees] [degrees] counts counts ! The question theirFIG. analysis S14:FIG. (Left) S14: The (Left)0 Fermi asks The0p6v11Fermi diisp6v11↵use this:counts backgrounddi↵use50 backgroundWhich in50 the IG in region theof IG these (smoothed0 region (smoothed0 usingdistributionsFermi usingcountstools),Fermi50 withtools),b50 with

Lee, Lisanti, Safdi, Slatyer, Xue, arXiv:1506.05124 (see also Bartels, Krishnamurthy, Weniger, arXiv:1506.05104)

-45 -4

FIG. S15:FIG. Best-fit S15: Best-fitsource-count source-count function function for PSs forwithin PSs 10 within of the 10 GCof the and GCb and2,b with2 the, with 3FGL the sources 3FGL unmasked,sources unmasked, for for | | models withmodels both with an both NFW an PS NFW population PS population (green band) (green and band) a di↵ and-corr a di PS↵-corr population PS population (blue| | band). (blue For band). this For analysis, this analysis, the NPTF the NPTF includesincludes an additional an additional template template corresponding corresponding to di↵use-correlated to di↵use-correlated PSs. This PSs. new This template new template either has either support has support in the inner in the 10 inner 10 (left) or(left) over the or over full ROI the full (30 ROIfrom (30 the from GC with the GCb with2)b (right).2) (right).As for the As standard for the standard IG analyses, IG analyses, the isotropic the isotropic PS parameters PS parameters | | | | are fixedare to fixedtheir tobest-fit their values best-fit at values high latitudes. at high latitudes. Figure S16Figure shows S16 the shows results the ofresults this analysis of this analysis when no when sources no sources are masked, are masked, in the form in the of form the best-fit of the best-fit source-count source-count functionsfunctions for the NFWfor the PS NFW (green PS band) (green and band) disk and PS disk (blue PS band) (blue populations. band) populations. Here the Here disk-correlated the disk-correlated PS template PS template accountsaccounts for the forobserved the observed 3FGL sources, 3FGL sources,10 but the10 but NFW the PS NFW population PS population is still strongly is still strongly preferred preferred by the data. by the The data. The source-countsource-count function function of the NFW of the PS NFW template PS template is similar is insimilar shape in to shape our earlier to our results earlier from results the from analysis the analysis with known with known sourcessources masked masked (as shown (as in showne.g. Fig. in e.g. 2),Fig. with 2), a steep with cutoa steep↵ just cuto below↵ just the below source the sensitivity source sensitivity threshold; threshold; the parameters the parameters +13.6 +13.6 +0.95 +0.95 +0.72 +0.7210 10 2 2 of the brokenof the powerbroken law power are lawn1 = are 29n.21 =14. 292,.n22 14=.2, 0n.244= 1.020.,44 and1.02Fb,=2 and.72Fb =20.48.72100.48 photons/cm10 photons/cm/s. The/s. best-fit The best-fit +0.05 +0.05⇥ ⇥ -90 slope ofslope the disk-correlated of the disk-correlated PS template PS template below the below break the is breakn2 =1 is.47n2 =1..47 It is worth. It is noting worth that noting in thisthat case in this there case is there is -5 0.03 0.03 no externallyno externally imposed imposed threshold, threshold, as no sources as no sources are masked. are masked. This supports This supports the idea the that idea the that low-flux the low-flux sources sources absorbed absorbed by the NFWby the PS NFW template PS template represent represent a separate a separate population, population, with a cuto with↵ ain cuto the↵ source-countin the source-count function function slightly slightly below below the currentthe current source sensitivity source sensitivity threshold. threshold. +0.76 +0.76 In this case,In this the case, fraction the fraction of flux that of flux is absorbedthat is absorbed by the NFW by the PS NFW template PS template is 9.23 is0.78 9.%23 (in0.78 the% (ininner the 10 inner region 10 region with b with> 2b),> while2), the while DM the contribution DM contribution is consistent is consistent with zero. with When zero. the When NFW the PS NFW template PS template is omitted, is omitted, the the | | | | +0.75 +0.75 fractionfraction of flux absorbedof flux absorbed by the diskby the PS disk population PS population remains remains unchanged, unchanged, and the and DM the template DM template absorbs absorbs 8.39 0.68 8.%39 0.68% 180 90 0of the flux.of the These flux. results These areresults consistent are consistent with the with flux the attributed flux attributed to the DM to the template DM template in [5]. Thein [5]. normalization The-90 normalization of of -180

10 The source-count10 The source-count function functionfor the disk for PSs the diskshould PSs not should be trusted not be much trusted below much threshold, below threshold, as the fit as is clearlythe fit is being clearly driven being by driven the high-flux by the high-flux PSs. PSs.

FIG. 4: The spatial templates (in galactic coordinates) for the Galactic di↵use model (upper left), the Fermi bubbles (upper right), and dark matter annihilation products (lower), as used in our Inner Galaxy analysis. The scale is logarithmic (base 10), normalized to the brightest point in each map. The di↵use model template is shown as evaluated at 1 GeV, and the dark matter template corresponds to a generalized NFW proﬁle with an inner slope of =1.18. Red dashed lines indicate the boundaries of our standard Region of Interest (we also mask bright point sources and the region of the Galactic plane with b < 1). | | we show the PSF for front-converting, Ultraclean events, IV. THE INNER GALAXY at three representative energies, for di↵erent cuts on CTBCORE (all events, Q2, and Q1). Such a cut can In this section, we follow the procedure previously pur- be used to mitigate the leakage of astrophysical emis- sued in Ref. [8] (see also Refs. [41, 42]) to study the sion from the Galactic Plane and point sources into our gamma-ray emission from the Inner Galaxy. We use the regions of interest. This leakage is most problematic at term “Inner Galaxy” to denote the region of the sky that low energies, where the PSF is quite broad and where the lies within several tens of degrees around the Galactic CTBCORE cut has the greatest impact. These new event Center, excepting the Galactic Plane itself ( b < 1 ), classes and their characterization are further detailed in which we mask in this portion of our analysis.| | [40], and accompanied by a data release of all-sky maps for each class, and the instrument response function ﬁles Throughout our analysis, we make use of the Pass 7 necessary for use with the Fermi Science Tools. (V15) reprocessed data taken between August 4, 2008 Throughout the remainder of this study, we will em- and December 5, 2013, using only front-converting, Ul- ploy the Q2 event class by default, corresponding to the traclean class events which pass the Q2 CTBCORE cut top 50% (by CTBCORE) of Fermi’s front-converting, Ul- as described in Sec. III. We also apply standard cuts to traclean photons, to maximize event quality. We select ensure data quality (zenith angle < 100, instrumental Q2 rather than Q1 to improve statistics, since as demon- rocking angle < 52, DATA QUAL = 1, LAT CONFIG=1). strated in Fig. 3, the angular resolution improvement in Using this data set, we have generated a series of maps moving from Q2 to Q2 is minimal. In Appendix A we of the gamma-ray sky binned in energy. We apply the demonstrate that our results are stable upon removing point source subtraction method described in Ref. [42], the CTBCORE cut (thus doubling the dataset), or ex- panding the dataset to include lower-quality events.1

ground model. When the background model is smoothed cor- rectly, we ﬁnd results that are much more stable to the choice of CTBCORE cut, and closely resemble the results previously 1 An earlier version of this work found a number of apparent obtained with Q2 events. Accordingly, the CTBCORE cut ap- peculiarities in the results without the CTBCORE cut that pears to be e↵ective at separating signal from poorly-modeled were removed on applying the cut. However, we now attribute background emission, but is less necessary when the background those peculiarities to an incorrect smoothing of the di↵use back- is well-modeled. Dan Hooper – DM Annihilation in the GC

Disk-Like Population

Lee, Lisanti, Safdi, Slatyer, Xue, arXiv:1506.05124 (see also Bartels, Krishnamurthy, Weniger, arXiv:1506.05104)

Dan Hooper – DM Annihilation in the GC

Evidence For Unresolved Point Sources? Lee et al.’s Conclusions include the following: 1) The brightest sources (including those contained in source catalogs) are distributed along the disk – not tracing the excess 2) The fit suggests that the GeV excess could be generated by ~103 unresolved sources, most with a flux that is just slightly below Fermi’s threshold for point source detection Excess-Like Population

Disk-Like Population

Lee, Lisanti, Safdi, Slatyer, Xue, arXiv:1506.05124 (see also Bartels, Krishnamurthy, Weniger, arXiv:1506.05104)

Dan Hooper – DM Annihilation in the GC Evidence For Unresolved Point Sources? A few comments of my own: ! It is difficult to tell whether these clustered gamma-rays result from unresolved sources, or from backgrounds that are less smooth than are being modeled ! Keep in mind that these clusters consist of only a few photons each, on top of large and imperfectly known backgrounds ! These studies do not make use of any spectral information (they use only a single energy bin); whether these putative sources have a spectrum that matches that of the excess will be an important test

Lee, Lisanti, Safdi, Slatyer, Xue, arXiv:1506.05124 (see also Bartels, Krishnamurthy, Weniger, arXiv:1506.05104)

Dan Hooper – DM Annihilation in the GC Could These Be Millisecond Pulsars?

! The measured luminosity function of MSPs extends well above the threshold for detection by Fermi; very different than this new putative source population ! Where are all of the bright MSPs? (bright sources are disk-like, not DM-like) ! If these are point sources, they are very weird point sources ! A new class of standard candles?! MSP – 68% possess luminosities within Luminosity a factor of 2 (ΔM ~ 0.4) Function

Lee, Lisanti, Safdi, Slatyer, Xue, arXiv:1506.05124 (see also Bartels, Krishnamurthy, Weniger, arXiv:1506.05104)

12 Fermi–LAT Collaboration

Table 2 Fluxes of the Components1 for 15 15 region about GC. ⇥ Interstellar Energy Band Annulus 1 Annulus 1 Point Sources Fore-/background Isotropic Model Data Emission Model (GeV) ⇡0-decay IC ⇡0-decay IC Brem Total Pulsars intensity-scaled 1.00 3.16 6.1 1.1 32.5 0.6 36.3 1.2 135 23 24 2.3 259 3 251 132 3.16 10.00 1.0±0.2 7.1 ±0.1 7.3 ±0.2 21 5.4 1.7 0.6 44.1 ±0.5 44± 3 10.00 31.62 0.13±0.02 1.41±0.03 0.81±0.04 2.9 1.2 0.14 0.17 6.7 ±0.1 6.8 ±0.7 ± ± ± ± ± 31.62 100.00 0.023 0.243 0.11 0.01 0.4 0.2 0.01 0.04 1.04 0.02 1.2 0.2 ± ± ± Pulsars index-scaled 1.00 3.16 2.1 1.1 35.5 0.6 37.9 1.5 127 25 254 3 3.16 10.00 0.3±0.2 7.8 ±0.1 6.6 ±0.2 25 6 48 ±0.5 10.00 31.62 0.05±0.02 1.54±0.03 0.62±0.03 4.2 1.3 8.0±0.1 ± ± ± ± 31.62 100.00 0.013 0.33 0.07 0.01 0.75 0.23 1.37 0.01 OBstars ± ± intensity-scaled 1.00 3.16 1.3 0.5 47.0 0.6 35.7 1.2 128 23 21 2.6 259 2 3.16 10.00 0.2±0.1 9.1 ±0.1 7.3 ±0.2 19 5.1 1.4 0.7 43.3 ±0.4 10.00 31.62 0.02±0.01 1.62±0.02 0.8±0.1 2.6 1.1 0.12 0.16 6.4 ±0.1 ± ± ± ± 31.62 100.00 –3 0.253 0.11 0.01 0.4 0.2 0.01 0.03 0.97 0.02 ± ± OBstars index-scaled 1.00 3.16 1.0 0.5 40.9 0.6 38.3 1.3 135 19 257 2 3.16 10.00 0.14±0.07 7.9 ±0.1 6.8 ±0.2 24 4.3 45.6 ±0.4 10.00 31.62 0.02±0.01 1.41±0.02 0.69±0.04 3.9 0.9 7.2 ±0.1 ± ± ± ± 31.62 100.00 –3 0.223 0.08 0.01 0.6 0.2 1.15 0.01 ± ± 1 8 2 1 Units: 10 ph cm s . 2 The errors are dominated by systematic uncertainties from the effective area, see Ackermann et al. (2012b) for details. Dan Hooper – DM Annihilation in the GC 3 10 2 1 Flux and/or statistical uncertainty below 10 ph cm s .

100 100

5 5

0 10 0 10 Galactic Latitude (degrees) Galactic Latitude (degrees)

5 5

1 1 5 0 5 5 0 5 Galactic Longitude (degrees) Galactic Longitude (degrees)

Figure 7. Point sources for 3FGL (left panel) and 1FIG (right panel, for Pulsars intensity-scaled IEM) overlaid on the total1FIG counts Catalog, for the 15 Fermi15 Collaborationregion (Murgia, et al.) ⇥ about the GC. Left panel symbol key: filled squares, ‘flagged’ 3FGL sources; filled triangles, other 3FGL sources; upright crosses,arXiv:1511.02938 3FGL sources with a multi- wavelength association. Right panel symbol key: filled circles, 1FIG sources with TS 25; angled crosses, 1FIG source candidates with TS < 25; upright crosses, as in left panel. Colour scale is in counts per 0.052 degree pixel.

It is probable that there is some misattribution of interstellar tion. Figure 9 shows all point sources and candidates with a 9 2 0 emission to low-flux (e.g., less than few 10 ph cm TS < 50 overlaid on the fitted ⇡ -decay annulus 1 template 1 ⇠ ⇥ s > 1 GeV) point sources. The low-flux sources are all rel- for the Pulsars intensity-scaled IEM. The point sources are atively low-significance sources and modelled using power- coded according to the spectral indices: circles show those law spectra (Section 3.2.1). The distribution of their spec- with indices > 2.5, while triangles show those with indices tral indices over the 15 15 region may provide some in- 2.5. There is no clear trend of softer spectrum point sources ⇥  formation: softer spectral indices (e.g., & 2.5 in spectral in- tracing the structured emission, nor one where the harder dex) can indicate that the low-flux sources are more likely as- spectrum point sources have a high density out of the plane. It sociated with the structured/gas-related interstellar emission, is difficult to identify the exact fraction of the emission, or to while harder indices can indicate a more “IC-like” distribu- what component (gas-related, IC), the low-flux point sources 12 Fermi–LAT Collaboration

100 100

5 5

0 10 0 10 Galactic Latitude (degrees) Galactic Latitude (degrees)

5 5

1 1 5 0 5 5 0 5 Galactic Longitude (degrees) Galactic Longitude (degrees)

100 100 6 N+S Sources 34 W+E Sources 5 5

-Fermi’s resolved sources are not at all spherical, but are 0 10 0 10 instead concentrated along the disk

-If anything, one

Galactic Latitude (degrees) Galactic Latitude (degrees) expects that sources 5 5 in the plane would be more difficult to detect

1 1 5 0 5 5 0 5 Galactic Longitude (degrees) Galactic Longitude (degrees)

Center don’t look like pulsars – instead look like power-law 10sources 1

J1703.6-2850 J1740.5-2642 s

2 9

10 10 GeV cm

1 J1808.3-3357

J1740.8-1933 s

2 9

10 10 GeV cm

1 J1808.4-3703

J1808.4-3519 s

2 9

10 10 GeV cm 1

J1820.4-3217 s

2 9

10 10 Stacked Spectrum GeV cm 100 101 100 101 Energy (GeV) Energy (GeV)

FIG. 6: The -ray spectrum obtained for 7 of the -ray point sources indicated as possible pulsars in the analysis of [27]. The T. Linden, arXiv:1509.02929proceedure utilized in this paper provides a more detailed spectral model of -ray sources compared to that utilized for the 3FGL catalog. We ﬁnd that some sub-population of these sources have spectral characteristics consistent with MSPs, while others clearly have excessive levels of low-energy emission.

Appendix A: An Analysis of the Gamma-Ray Point (notably J1808.4-3519 and J1703.6-2850) appear to have Sources Identified as Possible MSPs by Bartels et al. spectra which are consistent with the MSP population, (2015) while othes (such as J1808.3-3357 and J1820.4-3217) appear to have power-law best fit spectra and contain sig- An analysis by [27] analyzed the point sources in the nificantly more low-energy emission compared to MSP GC region, and indicated 13 known 3FGL sources that models. In the bottom right of Figure 6 we show the stacked spectrum from the seven -ray sources, showing have MSP-like spectra, 7 of which lie within 10 of the GC (and are covered within the ROI of our analysis). emission which is generally consistent with power-law be- While the 3FGL spectral measurements are consistent havior. It remains possible that a subset of these systems with a MSP origin for these systems, we note that this is does in fact have an MSP origin, a result which would not based on intensity measurements of each source in only be in tension either MSP or dark matter explanations for 5 energy bins, utilizing 4 years of -ray data [28]. We the -ray excess. We note that none of these 13 sources examine this finding by utilizing the spectral fitting al- lie within 0.1 of any ATNF pulsar, but two sources (of the 6 outside our 10 ROI), lie at 0.25 from a known gorithm described in Section II B in order to calculate ⇠ the spectrum of each source in much greater detail. In ATNF pulsar (3FGL J1744.8-1557 and 3FGL J1837.3- Figure 6 we show the resulting spectra obtained for each 2403). 3FGL source, finding varied results. Some 3FGL sources, In theory, a similar test could be run on the posi- Dan Hooper – DM Annihilation in the GC What’s Next? ! After ~7 years of investigation, the origin of the Galactic Center excess remains unclear – it looks a lot like annihilating dark matter, but we can’t rule out other possibilities ! How do we go from establishing an intriguing signal, to being able to claim discovery? (or to rule out a dark matter interpretation)

Dan Hooper – DM Annihilation in the GC

Dwarf Galaxies ! The constraint from observations of dwarf galaxies remains compatible with a dark matter interpretation of the Galactic Center excess ! That being said, if the Galactic Center signal is coming from annihilating dark matter, one should expect gamma rays from dwarfs to be detected soon

Fermi Collaboration, 1503.02641

(see Alex Drlica-Wagner’s talk for updates) Dan Hooper – DM Annihilation in the GC

New Dwarf Galaxies! ! In the past year, 23 new dwarf galaxy candidates have been discovered! (most from Dark EnergyMilky Survey Way data, Companions but also SDSS, and Pan-STARRS) ! Particularly exciting are Reticulum II, Tucana III, and Cetus II which are each nearby (~25-30Found kpc in) andTwo attractive Years targetsof DES for Datadark matter searches Blue = Known prior to 2015 Stellar density ﬁeld from Red triangles = DES Y2Q1 candidates SDSS and DES Red circles = DES Y1A1 candidates Green = Other new candidates

Drlica-Wagner et al. 2015 arXiv:1508.03622 2 DES footprint in Galactic coordinates (~5000 deg ) 12 Dan Hooper – DM Annihilation in the GC Significance Scans over Fermi’s View of the New Dwarf Galaxies ! This spring,Annihilation three groups reported Channel an excess andfrom Reticulum Mass II, but with only 2.4-3.2σ significance, (Geringer-Sameth et al. Drlica-Wagner, et al, DH & Linden) No statistically significant signal towards any individual confirmed or candidate dSph! ! No papersNo statisticallyon Tucana significant III or the signalother mostfound recentlyin joint likelihood discovered analysis dwarfs! yet, but Fermi hasPeak recently local significances begun presenting of 2 to 3 ! for preliminary a few of the resultsnew targets in talks:

Expected ±1! Gray curves for other dSph targets Black curves for joint likelihood Expected ±2!

PRELIMINARY

Most of the new targets have not yet been conﬁrmed as DM dominated dSphs From⇒ Keith Use distanceBechtol’s scaling talk, relation TAUP for2015 provisional (for the J-factor DES andestimates Fermi with Collaborations) uncertainty 0.4 dex 20

Dan Hooper – DM Annihilation in the GC Significance Scans over Fermi’s View of the New Dwarf Galaxies ! This spring,Annihilation three groups reported Channel an excess andfrom Reticulum Mass II, but with only 2.4-3.2σ significance, (Geringer-Sameth et al. Drlica-Wagner, et al, DH & Linden) No statistically significant signal towards any individual confirmed or candidate dSph! ! No papersNo statisticallyon Tucana significant III or the signalother mostfound recentlyin joint likelihood discovered analysis dwarfs! yet, but Fermi hasPeak recently local significances begun presenting of 2 to 3 ! for preliminary a few of the resultsnew targets in talks:

Expected ±1! Gray curves for other dSph targets Black curves for joint likelihood Expected ±2!

PRELIMINARY

Signiﬁcance Scans over Annihilation Channel and Mass

No statistically significant signal towards any individual confirmed or candidate dSph! No statistically significant signal found in joint likelihood analysis! Peak local significances of 2 to 3 ! for a few of the new targets Dan Hooper – DM Annihilation in the GC

Expected ±1! Gray curves for other dSph targets Black curves for joint likelihood Expected ±2!

PRELIMINARY

Most of the new targets have not yet been conﬁrmed as DM dominated dSphs From Keith Bechtol’s⇒ talk, Use TAUP distance 2015 (for scaling the DES relation and Fermi for Collaborations) provisional J-factor estimates with uncertainty 0.4 dex 20

Signiﬁcance Scans over Annihilation Channel and Mass

FavoredExpected by ±1! Galactic Center Gray curves for other dSph targets Black curves for joint likelihood Expected ±2!

PRELIMINARY

Dan Hooper – DM Annihilation in the GC Summary ! The Galactic Center’s GeV excess is particularly compelling: highly statistically significant, robust, distributed spherically out to at least 10° from the Galactic Center, and difficult to explain with known or

proposed astrophysics ! The spectrum and angular distribution of this signal is very well fit by a ~35-50 GeV WIMP; future observations of dwarf galaxies will provide a critical test of a dark matter origin for this signal ! Fermi’s observations of dwarf galaxies – both old, new, and yet to be discovered – seem to be the most promising way forward -Fermi’s current sensitivity to dwarfs is roughly that needed to test dark matter interpretations of Galactic Center excess -A modest excess does exists -New dwarf discoveries (DES, LSST), combined with more data from Fermi, could provide significant support for this interpretation or essentially rule out dark matter as the origin of the Galactic Center