Dwarf Galaxies

The gamma-ray flux from millisecond pulsars in dwarf spheroidal galaxies

Justin Vandenbroucke, Keith Bechtol, Miles Winter (UW Madison) Gabrijela Zaharijas (University of Nova Gorica) TeVPA, CERN, September 12, 2016 arXiv:1607.06390 Outline

• Motivation: dark matter search (inc. testing the Galactic Center excess) with dwarf spheroidal galaxies • If a signal is detected from dwarf galaxies, is there an astrophysical background? • Two mechanisms of millisecond pulsar (MSP) formation • MSP luminosity function • Predicted number of MSPs and gamma-ray flux from individual dwarf galaxies • Conclusion

TeVPA 2016 Justin Vandenbroucke: millisecond pulsars in dwarf galaxies 2

The GeV excess from the Galactic center

A. Albert for LAT Collaboration Daylan et al. 2016 IAU GC Symposium 2016

Dark matter? Millisecond pulsars? Something else? See Thursday afternoon session TeVPA 2016 Justin Vandenbroucke: millisecond pulsars in dwarf galaxies 3 Latest dark matter upper limits from dwarf galaxies 23 10 Pass 8 Combined dSphs Fermi-LAT MW Halo H.E.S.S. GC Halo 24 MAGIC Segue 1 10 Abazajian et al. 2014 (1) ) 1 Gordon & Macias 2013 (2) s

3 Daylan et al. 2014 (2) 25 10 Calore et al. 2015 (2) (cm i v

h Thermal Relic Cross Section 26 (Steigman et al. 2012) 10 25 dwarf galaxies LAT collaboration, PRL 115 231301 (2015) b¯b 27 10 101 102 103 104 DM Mass (GeV/c2) • Tension between dwarf limits and Galactic Center excess • Will a signal soon emerge from dwarf galaxies? • Would it be a smoking gun?

TeVPA 2016 Justin Vandenbroucke: millisecond pulsars in dwarf galaxies 4 Additional dwarf galaxies continue to be discovered

A. Drlica-Wagner et al. ApJ 813:109 (2015) 27 previously known Discovered in past 2 years: 17 by DES and 5 by others

• DES much deeper than previous searches • DES (and SkyMapper, Pan-STARRS, LSST) will continue to provide new dwarfs, potentially with excellent J factors

TeVPA 2016 Justin Vandenbroucke: millisecond pulsars in dwarf galaxies 5 Astrophysical backgrounds must exist from dwarf galaxies: are they negligible?

• Stacked dwarf galaxy dark matter sensitivity is comparable to the Galactic Center, and growing as more galaxies are discovered • If a signal is seen from dwarf galaxies, is it confirmation of Galactic Center excess? • Or will we repeat the same discussions of dark matter vs. pulsars vs. other? • Be ready for this possibility by quantifying astrophysical backgrounds now

TeVPA 2016 Justin Vandenbroucke: millisecond pulsars in dwarf galaxies 6 How do millisecond pulsars form?

• Primordial channel – Binary is formed during stellar birth, before one becomes neutron star – Rate proportional to stellar density – Dominant in low stellar densities: Milky Way disk, dwarf galaxies • Dynamical channel – Existing neutron star is captured into orbit of companion – Rate proportional to stellar density squared – Dominant in high stellar densities: globular clusters, perhaps Milky Way bulge

TeVPA 2016 Justin Vandenbroucke: millisecond pulsars in dwarf galaxies 7 Luminosity distribution of LAT millisecond pulsars 16 2PC Distances 14 ATNF Distances

8 Counts

2 0.1-100 GeV

0 1031 1032 1033 1034 1035 1036

L (erg/s) 66 MSPs with distance estimates and not in globular clusters

TeVPA 2016 Justin Vandenbroucke: millisecond pulsars in dwarf galaxies 8 Accounting for completeness fraction in LAT detection of Galactic millisecond pulsars

Detected MSPs are nearby: Monte Carlo of true distribution:

LAT 2PC

• Model density of Milky Way MSPs as exponential in R and z

• Scale height z0 = 0.6 (+0.6, -0.3) kpc

• Scale radius R0 = 3 (+3, -1) kpc • Generate MSPs according to spatial model and determine which fraction would be detected by LAT

• Repeat for many spatial models according to uncertainty in (z0, R0) TeVPA 2016 Justin Vandenbroucke: millisecond pulsars in dwarf galaxies 9 Calculated completeness fraction for LAT millisecond pulsar detection 100

1 10

2 10

3 10

4 10 Completeness Fraction

5 10 median central 68%

10 6 1032 1033 1034 1035

L (erg/s) TeVPA 2016 Justin Vandenbroucke: millisecond pulsars in dwarf galaxies 10 Accounting for completeness fraction to determine luminosity function for Milky Way stellar mass 1039

1038

1037 (erg/s) 1036 dN/dL 2

L 1035 best-fit LF (this study) statistical uncertainty systematic uncertainty 34 10 Hooper & Mohlabeng (2016) LF: fixed lum. Hooper & Mohlabeng (2016) LF: fixed MSP LAT MSPs: 2PC dist. w/ mean completeness 1033 1031 1032 1033 1034 1035 1036

L (erg/s)

TeVPA 2016 Justin Vandenbroucke: millisecond pulsars in dwarf galaxies 11 What about the chance occurrence of a foreground MSP along the line of sight to a dwarf galaxy?

• Estimated using the same spatial distribution and Milky Way luminosity function • Used Monte Carlo to determine probability of one MSP lying along line of sight, and if it occurs what is the PDF for its flux • Most often there is no MSP along line of sight • When there is one, it is typically brighter than MSP emission from dwarf

TeVPA 2016 Justin Vandenbroucke: millisecond pulsars in dwarf galaxies 12 100 100 Segue 1 Draco Foreground Foreground 1 dSph 1 dSph 10 10 (l,b)=(220.5,50.4) (l,b)=(86.4,34.7) =0.004 =6.1

2 2 10 10

3 Underﬂow Bin 3 Underﬂow Bin 10 10

Fraction Per Bin 4 Fraction Per Bin 4 10 10

5 5 10 16 14 12 10 8 10 16 14 12 10 8 10 10 10 10 10 10 10 10 10 10 2 1 2 1 Flux > 500 MeV (ph cm s ) Flux > 500 MeV (ph cm s ) Results: Fornax 100 100 Sculptor LAT 6 year Fornax Foreground Foreground 1 dSph 1 threshold dSph 10 10 (l,b)=(287.5,-83.2) (l,b)=(237.1,-65.7) = 44.2 = 278.7

2 2 10 10

3 Underﬂow Bin 3 Underﬂow Bin 10 10

Fraction Per Bin 4 Fraction Per Bin 4 10 10

5 5 10 16 14 12 10 8 10 16 14 12 10 8 10 10 10 10 10 10 10 10 10 10 2 1 2 1 Flux > 500 MeV (ph cm s ) Flux > 500 MeV (ph cm s )

• Underflow bin: no foreground MSP • Expect on average ~300 MSPs with luminosity above 1031.5 erg/s • Brightest expected MSP flux among all dwarfs investigated • Still below current LAT sensitivity and Galactic center DM model prediction • Very small probability of a foreground source, but if there is one it is likely to be brighter than MSP signal from Fornax

TeVPA 2016 Justin Vandenbroucke: millisecond pulsars in dwarf galaxies 13 100 100 Segue 1 Draco Foreground Foreground 1 dSph 1 dSph 10 10 (l,b)=(220.5,50.4) (l,b)=(86.4,34.7) =0.004 =6.1

2 2 10 10

3 Underﬂow Bin 3 Underﬂow Bin 10 10

Fraction Per Bin 4 Fraction Per Bin 4 10 10

5 5 10 16 14 12 10 8 10 16 14 12 10 8 10 10 10 10 10 10 10 10 10 10 2 1 2 1 Flux > 500 MeV (ph cm s ) Flux > 500 MeV (ph cm s ) Results: Sculptor 100 100 Sculptor Fornax LAT 6 year Foreground Foreground 1 threshold dSph 1 dSph 10 10 (l,b)=(287.5,-83.2) (l,b)=(237.1,-65.7) = 44.2 = 278.7

2 2 10 10

3 Underﬂow Bin 3 Underﬂow Bin 10 10

Fraction Per Bin 4 Fraction Per Bin 4 10 10

5 5 10 16 14 12 10 8 10 16 14 12 10 8 10 10 10 10 10 10 10 10 10 10 2 1 2 1 Flux > 500 MeV (ph cm s ) Flux > 500 MeV (ph cm s )

• Underflow bin: probability of there being no foreground MSP • Expect on average ~40 MSPs with luminosity above 1031.5 erg/s in Sculptor • Very small probability of a foreground source, but if there is one it is likely to be brighter than MSP signal from Sculptor

TeVPA 2016 Justin Vandenbroucke: millisecond pulsars in dwarf galaxies 14 Results: Draco 100 100 Segue 1 LAT 6 year Draco Foreground threshold Foreground 1 dSph 1 dSph 10 10 (l,b)=(220.5,50.4) (l,b)=(86.4,34.7) =0.004 =6.1

2 2 10 10

3 Underﬂow Bin 3 Underﬂow Bin 10 10

Fraction Per Bin 4 Fraction Per Bin 4 10 10

5 5 10 16 14 12 10 8 10 16 14 12 10 8 10 10 10 10 10 10 10 10 10 10 2 1 2 1 Flux > 500 MeV (ph cm s ) Flux > 500 MeV (ph cm s )

• Underflow bin: probability of there being no foreground MSP (or MSP in dSph) 100 • Expect on average100 ~6 MSPs with luminosity above 1031.5 erg/s in Draco Sculptor Fornax Foreground• Moderate Poisson probability of zero MSPs in DracoForeground 1 dSph 1 dSph 10 • Very small probability10 of a foreground source, but if there is one it is likely to be (l,b)=(287.5,-83.2)brighter than MSP signal from Fornax (l,b)=(237.1,-65.7) = 44.2 = 278.7

2 2 10 10 TeVPA 2016 Justin Vandenbroucke: millisecond pulsars in dwarf galaxies 15

3 Underﬂow Bin 3 Underﬂow Bin 10 10

Fraction Per Bin 4 Fraction Per Bin 4 10 10

5 5 10 16 14 12 10 8 10 16 14 12 10 8 10 10 10 10 10 10 10 10 10 10 2 1 2 1 Flux > 500 MeV (ph cm s ) Flux > 500 MeV (ph cm s ) Results: Segue 1

100 100 LAT 6 year Segue 1 Draco threshold Foreground Foreground 1 dSph 1 dSph 10 10 (l,b)=(220.5,50.4) (l,b)=(86.4,34.7) =0.004 =6.1

2 2 10 10

3 Underﬂow Bin 3 Underﬂow Bin 10 10

Fraction Per Bin 4 Fraction Per Bin 4 10 10

5 5 10 16 14 12 10 8 10 16 14 12 10 8 10 10 10 10 10 10 10 10 10 10 2 1 2 1 Flux > 500 MeV (ph cm s ) Flux > 500 MeV (ph cm s )

• Underflow bin: probability100 of there being no foreground MSP (or MSP in dSph) 100 • Expect on average 0.004 MSPs with luminosity aboveSculptor 1031.5 erg/s in Segue 1 Fornax • Most likely to have zero MSPs above 1031 erg/s Foreground Foreground 1 dSph 1 dSph 10 10 (l,b)=(287.5,-83.2) (l,b)=(237.1,-65.7) = 44.2 = 278.7 TeVPA 2016 Justin Vandenbroucke: millisecond pulsars in dwarf galaxies2 2 16 10 10

3 Underﬂow Bin 3 Underﬂow Bin 10 10

Fraction Per Bin 4 Fraction Per Bin 4 10 10

5 5 10 16 14 12 10 8 10 16 14 12 10 8 10 10 10 10 10 10 10 10 10 10 2 1 2 1 Flux > 500 MeV (ph cm s ) Flux > 500 MeV (ph cm s ) Calculated MSP flux for 30 individual dwarf Millisecond Pulsars in Dwarf Spheroidal Galaxies 7

galaxies/candidates with good stellar mass estimatesTABLE 1 Estimated GeV Contribution from MSPs in MW dSphs

2 1 GeV2 a Galaxy D(kpc) log10(M /M )Flux> 500 MeV (ph cm s )log10 J 5 Ref. ⇤ cm Mean Stat. Syst. Pois. Total ⇣ h i⌘ +0.38 +1.42 +5.16 +2.26 +6.51 15 Segue I 23.02.53 0.20 3.36 0.93 2.01 1.29 3.00 10 19.5 0.29 1,5 +0.05 +2 .81 +10 .2 +3 .11 +11 .1 ⇥ 15 ± Tucana III 25.02.90 0.05 6.68 1.85 3.99 2.04 4.91 10 19.34 +0 .23 +1 .16 +4 .22 +0 .58 +4 .65 ⇥ 14 Ursa Major II 32.03.73 0.23 2.75 0.76 1.65 0.43 2.37 10 19.3 0.28 2,5 +0 .30 +0 .56 +2 .02 +0 .37 +2 .13 ⇥ 14 ± Reticulum II 32.03.41 0.03 1.32 0.37 0.79 0.27 0.91 10 19.33,6 +0 .39 +1 .53 +5 .58 +1 .56 +6 .82 ⇥ 15 Willman I 38.03.00 0.22 3.64 1.01 2.17 1.02 3.19 10 19.1 0.31 1,5 +0 .22 +0 .55 +1 .99 +0 .28 +2 .19 ⇥ 14 ± Coma Berenices 44.03.68 0.22 1.30 0.36 0.78 0.21 1.10 10 19.0 0.25 2,5 +0 .08 +2 .10 +7 .65 +1 .51 +8 .13 ⇥ 15 ± Tucana IV 48.03.34 0.06 4.99 1.38 2.98 1.08 3.53 10 18.74 +0 .04 +2 .67 +9 .72 +1 .61 +10 .2 ⇥ 15 Grus II 53.03.53 0.05 6.34 1.75 3.79 1.17 4.40 10 18.74 +1 .01 +1 .99 +7 .23 +1 .24 +13 .3 ⇥ 15 Tucana II 58.03.48 0.14 4.72 1.31 2.82 0.92 3.58 10 18.83,6 +0 .09 +1 .43 +5 .21 +0 .36 +5 .46 ⇥ 14 Bootes I 66.04.45 0.06 3.40 0.94 2.03 0.27 2.30 10 18.2 0.22 1,5 +0 .22 +3 .69 +13 .4 +4 .08 +15 .2 ⇥ 16 ± Indus I 69.02.90 0.22 8.76 2.43 5.24 2.63 7.75 10 18.33,6 +0 .20 +2 .06 +7 .49 +0 .14 +8 .09 ⇥ 13 Ursa Minor 76.05.73 0.20 4.88 1.35 2.92 0.11 3.93 10 18.8 0.19 2,5 +0 .10 +1 .24 +4 .51 +0 .10 +4 .73 ⇥ 13 ± Draco 76.05.51 0.10 2.94 0.81 1.76 0.09 2.06 10 18.8 0.16 2,5 +0 .21 +1.16 +4.24 +0.03 +4.59 ⇥ 12 ± Sculptor 86.06.59 0.21 2.76 0.76 1.65 0.03 2.26 10 18.6 0.18 2,5 +0 .20 +2 .07 +7 .53 +0 .12 +8 .14 ⇥ 13 ± Sextans 86.05.84 0.20 4.91 1.36 2.94 0.10 3.95 10 18.4 0.27 2,5 +0 .25 +0 .70 +2 .55 +0 .50 +2 .86 ⇥ 15 ± Horologium I 87.03.38 0.13 1.66 0.46 1.00 0.35 1.25 10 18.43,6 +0.13 +0 .52 +1 .90 +0 .40 +2 .04 ⇥ 15 Reticulum III 92.03.30 0.15 1.24 0.34 0.74 0.27 0.96 10 18.24 +0 .19 +0 .69 +2 .52 +0 .46 +2 .74 ⇥ 15 Phoenix II 95.03.45 0.11 1.64 0.45 0.98 0.32 1.20 10 18.43,6 +0 .13 +0 .45 +1 .63 +0 .13 +1 .73 ⇥ 14 Ursa Major I 97.04.28 0.13 1.06 0.29 0.64 0.10 0.78 10 18.3 0.24 2,5 +0 .11 +0 .86 +3 .12 +0 .06 +3 .27 ⇥ 13 ± Carina 105.05.63 0.09 2.03 0.56 1.22 0.05 1.40 10 18.1 0.23 1,5 +0 .14 +0 .47 +1 .72 +0 .10 +1 .82 ⇥ 14 ± Hercules 132.04.57 0.14 1.12 0.31 0.67 0.08 0.83 10 18.1 0.25 2,5 +0 .14 +2 .51 +9 .15 +0 .03 +9 .68 ⇥ 12 ± Fornax 147.07.39 0.14 5.97 1.65 3.57 0.03 4.38 10 18.2 0.21 2,5 +0 .15 +0.79 +2.89 +0.32 +3.08 ⇥ 15 ± Leo IV 154.03.93 0.15 1.89 0.52 1.13 0.24 1.42 10 17.9 0.28 2,5 +0 .20 +0 .69 +2 .50 +0 .29 +2 .71 ⇥ 15 ± Canes Venatici II 160.03.90 0.20 1.63 0.45 0.97 0.22 1.33 10 17.9 0.25 2,5 +0 .13 +4 .12 +15 .0 +1 .93 +15 .9 ⇥ 16 ± Columba I 182.03.79 0.07 9.78 2.71 5.85 1.43 6.78 10 17.64 +0.16 +2 .37 +8 .61 +1 .21 +9 .25 ⇥ 16 Indus II 214.03.69 0.14 5.62 1.56 3.36 0.89 4.22 10 17.44 TeVPA 2016 Justin Vandenbroucke: millisecond pulsars in dwarf galaxies+0 .09 +1 .41 +5 .12 +0 .12 +5 .35 ⇥ 14 17 Canes Venatici I 218.05.48 0.09 3.34 0.92 2.00 0.10 2.31 10 17.7 0.26 2,5 +0 .13 +0 .48 +1 .74 +0 .02 +1 .84 ⇥ 13 ± Leo II 233.06.07 0.13 1.14 0.31 0.68 0.02 0.82 10 17.6 0.18 2,5 +0 .13 +1 .68 +6 .11 +0 .04 +6 .45 ⇥ 13 ± Leo I 254.06.69 0.13 3.99 1.10 2.38 0.04 2.89 10 17.7 0.18 2,5 +0 .09 +1 .69 +6 .15 +0 .26 +6 .44 ⇥ 15 ± Eridanus II 330.04.92 0.07 4.01 1.11 2.40 0.20 2.73 10 17.33,6 ⇥ References.—(1)Wolfetal.(2010),(2)Kirbyetal.(2013),(3)Bechtoletal.(2015),(4)Drlica-Wagneretal.(2015),(5)Ackermann et al. (2015), (6) Drlica-Wagner et al. (2015) aFor dSphs that are kinematically conﬁrmed to be DM dominated, measured J-factors with uncertainties are quoted from Ackermann et al. (2015). For the remaining targets, we use a predicted J-factor following the distance scaling relation proposed by Drlica-Wagner et al. (2015).

Hooper, D., & Mohlabeng, G. 2016, J. Cosmology Astropart. McConnachie, A. W. 2012, AJ, 144, 4 Phys., 3, 049 Rubio-Herrera, E., & Maccarone, T. 2013, Neutron Stars and Hui, C. Y., Cheng, K. S., & Taam, R. E. 2010, ApJ, 714, 1149 Pulsars: Challenges and Opportunities after 80 years, 291, 111 Ivanova, N., Heinke, C. O., Rasio, F. A., Belczynski, K., & Schnitzeler, D. H. F. M. 2012, MNRAS, 427, 664 Fregeau, J. M. 2008, MNRAS, 386, 553 Simon, J. D., & Geha, M. 2007, ApJ, 670, 313 Kirby, E. N., Cohen, J. G., Guhathakurta, P., et al. 2013, ApJ, Spekkens, K., Urbancic, N., Mason, B. S., Willman, B., & 779, 102 Aguirre, J. E. 2014, ApJ, 795, L5 Lee, S. K., Lisanti, M., Safdi, B. R., Slatyer, T. R., & Xue, W. Walker, M. G., Mateo, M., Olszewski, E. W., et al. 2007, ApJ, 2016, Physical Review Letters, 116, 051103 667, L53 Maccarone, T. J., Kundu, A., Zepf, S. E., Piro, A. L., & Bildsten, Weisz, D. R., Dolphin, A. E., Skillman, E. D., et al. 2014, ApJ, L. 2005, MNRAS, 364, L61 789, 147 Malhotra, S., Spergel, D. N., Rhoads, J. E., & Li, J. 1996, ApJ, Wolf, J., Martinez, G. D., Bullock, J. S., et al. 2010, MNRAS, 473, 687 406, 1220 Manchester, R. N., Hobbs, G. B., Teoh, A., & Hobbs, M. 2005, Portegies Zwart, S., van den Heuvel, E. P. J., van Leeuwen, J., & AJ, 129, 1993 Nelemans, G. 2011, ApJ, 734, 55 Mateo, M. L. 1998, ARA&A, 36, 435 Calculated MSP flux for 30 different dwarf galaxies compared to point source sensitivity and DM model

TeVPA 2016 Justin Vandenbroucke: millisecond pulsars in dwarf galaxies 18 Conclusion

• Astrophysical backgrounds for dark matter signals from dwarf galaxies are said to be “negligible” • For the first time we have systematically calculated the expected background • The largest contribution is expected to be from millisecond pulsars • We determined the MSP luminosity function using Milky Way MSPs and used it to calculate the expected flux from individual dwarf galaxies • The largest expectations are an order of magnitude below current LAT sensitivity, and those from ultra-faints are even lower • An even smaller contribution is expected from diffuse cosmic-ray interaction within each dwarf galaxy • See arXiv:1607.06390

TeVPA 2016 Justin Vandenbroucke: millisecond pulsars in dwarf galaxies 19 Additional slides

TeVPA 2016 Justin Vandenbroucke: millisecond pulsars in dwarf galaxies 20 The GeV excess from the Galactic center

Daylan et al. 2016

Dark matter? Millisecond pulsars? Something else?

TeVPA 2016 Justin Vandenbroucke: millisecond pulsars in dwarf galaxies 21 Dark matter models to explain Galactic Center excess

10−25 ¯bb τ +τ − Hooper & Slatyer 2013 Huang+ 2013 Daylan+ 2014 ]

1 Abazaijan+ 2014 − s

3 Gordon+ 2014 10−26 [cm ⟩ v σ ⟨

10−27 101 102 mχ [GeV] Calore, Cholis, and Weniger JCAP03 (2015) 038

TeVPA 2016 Justin Vandenbroucke: millisecond pulsars in dwarf galaxies 22 Contributions to uncertainty in MSP flux from each dSph

1. Statistical uncertainty in luminosity function from finite number of LAT MSPs 2. Systematic uncertainty in MSP luminosity due to distance uncertainty 3. Systematic uncertainty in incompleteness correction due to spatial model and effective LAT threshold 4. Poisson fluctuations in number of MSP per dSph 5. dSph stellar mass uncertainty

TeVPA 2016 Justin Vandenbroucke: millisecond pulsars in dwarf galaxies 23 LAT MSP detection threshold

LAT 2PC

For incompleteness correction systematics, we vary the effective threshold across the limit range (10% to 90% three-year mean sensitivity)

TeVPA 2016 Justin Vandenbroucke: millisecond pulsars in dwarf galaxies 24 Formation of X-ray binaries and millisecond pulsars

Ivanova et al.

TeVPA 2016 Justin Vandenbroucke: millisecond pulsars in dwarf galaxies 25 Low-mass X-ray binaries (LMXB)

• 5 detected in Sculptor with deep Chandra search • Not detected in other dwarf galaxies • LMXB are the predecessor phase to MSP and are shorter lived (0.1-1 Gyr) • NS spin period in some LMXB measured to be millisecond scale (without radio pulsations) • LMXB number is observed to scale with stellar mass among galactic disks • When companion has donated its envelope it becomes a white dwarf and the LMXB becomes a MSP

TeVPA 2016 Justin Vandenbroucke: millisecond pulsars in dwarf galaxies 26 The “missing link” pulsar: evidence for transition from LMXB to MSP

A. Archibald et al. Science 324 (2009)1411-1414

massive star and neutron star and accretion disk and white dwarf and low-mass star low-mass star low-mass star: LMXB MSP

• 2000: identified as LMXB, with optically observed accretion disk • 2009: identified as MSP, with radio pulsations and optically identified companion, no optical accretion disk

TeVPA 2016 Justin Vandenbroucke: millisecond pulsars in dwarf galaxies 27 The “missing link” pulsar: evidence for transition from LMXB to MSP

A. Archibald et al. Science 324 (2009)1411-1414

TeVPA 2016 Justin Vandenbroucke: millisecond pulsars in dwarf galaxies 28 LAT second pulsar catalog (2PC)

TeVPA 2016 Justin Vandenbroucke: millisecond pulsars in dwarf galaxies 29 Results for four example dwarf galaxies 100 100 Segue 1 Draco Foreground Foreground 1 dSph 1 dSph 10 10 (l,b)=(220.5,50.4) (l,b)=(86.4,34.7) =0.004 =6.1

2 2 10 10

3 Underﬂow Bin 3 Underﬂow Bin 10 10

Fraction Per Bin 4 Fraction Per Bin 4 10 10

5 5 10 16 14 12 10 8 10 16 14 12 10 8 10 10 10 10 10 10 10 10 10 10 2 1 2 1 Flux > 500 MeV (ph cm s ) Flux > 500 MeV (ph cm s )

100 100 Sculptor Fornax Foreground Foreground 1 dSph 1 dSph 10 10 (l,b)=(287.5,-83.2) (l,b)=(237.1,-65.7) = 44.2 = 278.7

2 2 10 10

3 Underﬂow Bin 3 Underﬂow Bin 10 10

Fraction Per Bin 4 Fraction Per Bin 4 10 10

5 5 10 16 14 12 10 8 10 16 14 12 10 8 10 10 10 10 10 10 10 10 10 10 2 1 2 1 Flux > 500 MeV (ph cm s ) Flux > 500 MeV (ph cm s ) TeVPA 2016 Justin Vandenbroucke: millisecond pulsars in dwarf galaxies 30 Gamma-ray background due to diffuse interaction of cosmic rays within dwarf galaxies

• Uncertain CR acceleration • Uncertain CR confinement • But dSph have old stellar populations (few accelerators) and low gas content (little target material) • Scaling from SMC (SFR 0.1 solar mass per year and gamma-ray luminosity above 0.1 GeV 1037 erg/s) to Fornax (SFR 10-4 solar mass per year), expect 1034 erg/s Fornax gamma-ray luminosity, comparable to predicted MSP luminosity • But Fornax has 102 times lower gas mass (relative to stellar mass) than SMC • CR confinement time in dSph likely comparable to time between acceleration events (both ~1 Myr), so not steady state

TeVPA 2016 Justin Vandenbroucke: millisecond pulsars in dwarf galaxies 31 These are conservative predictions (upper limits)

• While Milky Way star formation is still active, dSph are not: MW MSP have ~1 Gyr characteristic age, and dSph MSP likely closer to 10 Gyr • Some MSP may have large kicks and therefore escaped the dSph or moved to its outskirts (away from the central dark matter distribution)

TeVPA 2016 Justin Vandenbroucke: millisecond pulsars in dwarf galaxies 32 Dark matter photon spectra Gamma-ray spectrum from dark matter annihilation

prompt spectra “soft” channels: 10.0000 • e+e bb produce a continuum µ+µ tt + + gamma-ray spectrum 1.0000 W W primarily from decay of neutral pions 0.1000 “hard channels”: include

• dN/dx ﬁnal state radiation 2 x 0.0100 (FSR) associated with charged leptons in the ﬁnal states 0.0010 J. Siegal-Gaskins @ SSI • line emission: γγ, Zγ, hγ 0.0001 (not shown), loop- 0.001 0.010 0.100 1.000 suppressed x=E/m Hard: e+e-, μ+μ-, τ+τ- Soft:Spectra bb, calculated tt, W with+W PPPC- 4 DM ID [Cirelli et al. 2010]

TeVPA 2016 Justin Vandenbroucke: millisecond pulsars in dwarf galaxies 33 J. Siegal-Gaskins SLAC Summer Institute | “Shining Light on Dark Matter” | August 10, 2014 19 LAT dSph sensitivity will increase by almost an order of magnitude with 15 years of LAT data and 60 dSPphs 10 22 dSphs (proj. 15 yrs 60 dpshs) Isotropic: (Proj. 15 years, syst/10) 23 10 APS: (Proj. 15 years) Clusters: (P8: 15 years)

] Unid. Sat.: (Proj. 15 years) 1 24 10 MW Center: (Proj. 15 years 1% syst)

s 3 25 10 [cm i

v 26 Thermal Relic Cross Section 10 (Steigman+ 2012) h Daylan+ (2014) 27 Gordon & Macias (2013) 10 Calore+ (2014) b¯b Abazajian+ (2014) 10 28 101 102 103 104 m [GeV] E. Charles et al. Phys Rep. 636 (2016) 1-46

TeVPA 2016 Justin Vandenbroucke: millisecond pulsars in dwarf galaxies 34 7

[7] A. W. McConnachie, Astron.J. 144,4(2012), TABLE I. Properties of Milky Way dSphs. 4 arXiv:1204.1562 [astro-ph.CO]. a a b [8] G. D. Martinez, (2013), arXiv:1309.2641 [astro-ph.GA]. Dwarf galaxy J factors Name ` b Distance log10(Jobs) Ref. 2 5 [9] A. Geringer-Sameth, S. M. Koushiappas, and M. Walker, (deg) (deg) (kpc) (log10[GeV cm ]) TABLE I. DES dSph Candidates and Estimated J-factors Bootes I 358.1 69.6 66 18.8 0.22 [39] (2014), arXiv:1408.0002 [astro-ph.CO]. ± CanesVenaticiII 113.6 82.7 160 17.9 0.25 [40] [10] M. Ackermann et al. (Fermi-LAT Collaboration), Phys. ± Rev. Lett. 107,241302(2011), arXiv:1108.3546 [astro- Name (`, b)a Distanceb log (Est.J)c Carina 260.1 22.2 105 18.1 0.23 [41] 10 ± ph.HE]. 2 Coma Berenices 241.9 83.6 44 19.0 0.25 [40] GeV ± [11] A. Geringer-Sameth and S. M. Koushiappas, Phys. Rev. deg kpc log10( cm5 ) Draco 86.4 34.7 76 18.8 0.16 [423 ] ± Lett. 107,241303(2011), arXiv:1108.2914 [astro-ph.CO]. DES J0222.7 5217 (275.0, 59.6) 95 18.3 Fornax 237.1 65.7 147 18.2 0.21 [41] ± [12] M. N. Mazziotta, F. Loparco, F. de Palma, and N. Gigli- todayDES in J0255.4 regions5406 of high (271 DM.4, 54 density.7) and 87 result 18.4in the dSphsHercules occurred for 28.7 a similar 36.9 set 132 of WIMP 18.1 characteris-0.25 [40] ± etto, Astropart. 37,26(2012),arXiv:1203.6731 [astro- DES J0335.6 5403 (266.3, 49.7) 32 19.3 Leo II 220.2 67.2 233 17.6 0.18 [43] production of energetic Standard Model particles. The tics; however, this deviation was not statistically± signifi- ph.IM]. DES J0344.3 4331 (249.8, 51.6) 330 17.3 Leo IV 265.4 56.5 154 17.9 0.28 [40] large mass of the WIMP (mDM ) permits the production cant [13]. ± [13] M. Ackermann et al. (Fermi-LAT Collaboration), Phys. DES J0443.8 5017 (257.3, 40.6) 126 18.1 Sculptor 287.5 83.2 86 18.6 0.18 [41] of gamma rays observable by the Fermi Large Area Tele- Using a new LAT event-level analysis,± known as Rev. D 89,042001(2014), arXiv:1310.0828 [astro- DES J2108.8 5109 (347.2, 42.1) 69 18.3 Segue 1 220.5 50.4 23 19.5 0.29 [44] ± ph.HE]. scopeDES (LAT), J2251.2 which5836 is (328 sensitive.0, 52 to.4) energies 58 ranging 18.8 from PassSextans 8, we re-examine 243.5 the 42.3 sample 86 of 25 18.4 Milky0.27 Way [41] ± [14] A. Geringer-Sameth, S. M. Koushiappas, and M. G. 20 MeVDES J2339.9 to > 3005424 GeV. (323.7, 59.7) 95 18.4 dSphsUrsa from Major Ackermann II 152.5 37.4et al. 32[13] using 19.3 six0. years28 of[40] ± Walker, (2014), arXiv:1410.2242 [astro-ph.CO]. Kinematic data indicate that the dwarf spheroidal LATUrsa data. Minor The Pass 105.0 8 data 44.8 benefits 76 from 18.8 an0 improved.19 [42] ± [15] B. Anderson, J. Chiang, J. Cohen-Tanugi, J. Conrad, Willman 1 158.6 56.8 38 19.1 0.31 [45] satellitea Galactic galaxies longitude (dSphs) and latitude. of the Milky Way contain a point-spread function (PSF), e↵ective area,± and energy c A. Drlica-Wagner, M. Llena Garde, and Stephan Zimmer substantialb We note DM that component typical uncertainties [6, 7]. The on the gamma-ray distances of signal dSphs reach.Bootes More II accurate353.7 Monte 68.9 Carlo 42 simulations – of the – for the Fermi-LAT Collaboration, ArXiv e-prints (2015), 2 1 Bootes III 35.4 75.4 47 – – fluxare at 10–15%.the LAT, s (phcm s ), expected from the detector and the environment in low-earth orbit have arXiv:1502.03081 [astro-ph.HE]. c J-factors are calculated over a solid angle of ⌦ 2.4 10 4 sr CanesVenaticiI 74.3 79.8 218 17.7 0.26 [40] annihilation of DM with a density distribution⇠⇢DM(⇥r)is reduced the systematic uncertainty in the LAT± instru- [16] T. Daylan, D. P. Finkbeiner, D. Hooper, T. Linden, (angular radius 0. 5). See text for more details. Canis Major 240.0 8.0 7 – – given by ment response functions (IRFs) [20]. Within the stan- S. K. N. Portillo, et al.,(2014),arXiv:1402.6703 [astro- Leo I 226.0 49.1 254 17.7 0.18 [46] ± ph.HE]. dardLeo photon V classes, 261.9Pass 58.5 8 o↵ers 178event types –,subdivi- – 1 v Emax dN [17] C. Gordon and O. Macias, Phys. Rev. D 88,083521 LAT ANALYSIS sionsPisces based II on event-by-event 79.2 47.1 uncertainties 182 in – the direc- – s(⌦)= h 2 i dE (2013), arXiv:1306.5725 [astro-ph.HE]. 4⇡ 2m dE tionalSagittarius and energy measurements, 5.6 14.2 26 which can increase – the – DM ZEmin [18] K. N. Abazajian, N. Canac, S. Horiuchi, and M. Kapling- sensitivitySegue 2 of likelihood-based 149.4 38.1 analyses. 35 In this – work we – To search for gamma-rayparticle emission physics from these new dSph hat, Phys. Rev. D 90,023526(2014), arXiv:1402.4090 (1) Ursa Major I 159.4 54.4 97 18.3 0.24 [40] candidates, we used six years of LAT data (2008 Au- use a set of four PSF event-type selections± that sub- [astro-ph.HE]. | {z2 } a [19] F. Calore, I. Cholis, and C. Weniger, (2014), ⇢DM(r)dP8R2ld⌦0SOURCE. divideGalactic the events longitude in ourand latitude. data sample according to the gust 4 to 2014 August⇥ 5) passing the event b Z⌦ Zl.o.s. qualityJ-factors of their are directionalcalculated assuming reconstruction. an NFW density In addition profile and arXiv:1409.0042 [astro-ph.CO]. class selections from 500 MeV to 500 GeV. Compared integrated over a circular region with a solid angle of [20] W. Atwood et al. (Fermi-LAT Collaboration), 2012 J factor to the improvements4 from Pass 8,weemploytheup- to the previous iteration of the LAT event-level analysis, ⌦ 2.4 10 sr (angular radius of 0.5). Fermi Symposium Proceedings,eConfC121028 (2013), datedc third⇠ LAT⇥ source catalog (3FGL), based on four Pass 8 [35] provides| significant{z improvements} in all areas dSphs below the horizontal line are not included in the arXiv:1303.3514 [astro-ph.IM]. Here, the first term is dependent on the particle physics yearscombined of Pass analysis.7 Reprocessed data, to model point-like propertiesof LATFermi analysis; — i.e., LAT thespecifically & thermally-averaged DES the 1503.02632 di↵erential annihilation point-source [21] M. Ackermann et al. (The Fermi-LAT Collaboration), sensitivity improves by 20–40% in P8R2 SOURCE V6 rela- background sources [21].Fermi Together, LAT these 1503.02641 improvements, (2015), arXiv:1501.02003 [astro-ph.HE]. cross section, v , the particle mass, mDM, and the tive to P7REPh SOURCEi V15. To remove gamma rays pro- along with an additional two years of data taking, lead [22] See Supplemental Material for more details. TeVPA 2016di↵erential gamma-ray yield perJustin Vandenbroucke: millisecond pulsars in dwarf galaxies annihilation, dN /dE , to a predicted increase in sensitivity of 70% relative to 35 [23] M. G. Walker, (2012), arXiv:1205.0311 [astro-ph.CO]. duced by cosmic-ray interactions in the Earth’s1 limb, integrated over the experimental energy range. The theFoundation, four-year analysis the Swedish of Ackermann Researchet Council al. [13] for and the theb¯b Na- [24] G. Battaglia, A. Helmi, and M. Breddels, New Astron. we rejected events with zenith angles greater than 100. second term, known as the J-factor, is the line-of-sight channeltional at Space 100 GeV. Board More (Sweden). details Science on Pass analysis 8 and other support Rev. 57,52(2013), arXiv:1305.5965 [astro-ph.CO]. Additionally, events from time intervals around bright [25] L. E. Strigari, Phys. Rep. , 1 (2013), arXiv:1211.7090 (l.o.s.) integral through the DM distribution integrated aspectsin the of operations this analysis phase can from be found INAF in (Italy) Supplemental and CNES overgamma-ray a solid angle, bursts⌦. and solar flares were removed us- [astro-ph.CO]. ing the same method as in the 4-year catalog analysis Material(France) [22]. is also gratefully acknowledged. [26] M. G. Walker, M. Mateo, E. W. Olszewski, J. Penar- Milky Way dSphs can give rise to J-factors in excess of FIG. 1. LAT counts maps in 10 10 ROI centered at each 19(3FGL)2 [36].5 To analyze the dSph candidates in Table I, ⇥ rubia, N. Evans, et al., Astrophys. J. 704,1274(2009), 10 GeV cm [8, 9], which, coupled with their lack of DES dSph candidate (white ‘ ’ symbols), for E>1GeV, we used 10 10 ROIs centered on each object. Data ⇥ arXiv:0906.0341 [astro-ph.CO]. non-thermal astrophysical processes, makes them good smoothed with a 0. 25 Gaussian kernel. All 3FGL sources in reduction was⇥ performed using ScienceTools version 09- LAT DATA SELECTION [27] J. Wolf, G. D. Martinez, J. S. Bullock, M. Kaplinghat, targets for DM searches via gamma rays. Gamma-ray the ROI are indicated with white ‘+’ symbols, and those with M. Geha, et al., Mon. Not. R. Astron. Soc. 406,1220 34-03.3 Figure 1 shows smoothed counts maps around ateststatistic> 100 are explicitly labeled. searches for dSphs yield some of the most stringent con- ⇤ [email protected] (2010), arXiv:0908.2995 [astro-ph.CO]. each candidate for energies > 1 GeV. We examine six years of LAT data (2008-08-04 to 2014- straints on v , particularly when multiple dSphs are † [email protected] [28] G. D. Martinez, J. S. Bullock, M. Kaplinghat, L. E. Stri- h i 08-05) selecting Pass 8 SOURCE-class events in the en- analyzedWe applied together the using search a joint procedure likelihood presented technique in [ Acker-10– ‡ [email protected] gari, and R. Trotta, J. Cosmol. Astropart. Phys. 0906, mann et al. [19] to the new DES dSph candidates. Specif- ergywith[1] rangeR. a small between Adam (< 10%)et 500 al. MeV energy-dependent(Planck and 500 Collaboration), GeV. correction We selected (2015), to ac- 014 (2009), arXiv:0902.4715 [astro-ph.HE]. 15]. Limits on v derived from observations of dSphs 5 thecount 500 MeVarXiv:1502.01582 for lower di↵erences limit to[astro-ph.CO] in mitigate the LAT the. response. impact of leakagePoint-like [29] J. F. Navarro, C. S. Frenk, and S. D. White, Astrophys. haveically, begun we performed to probeh i the a binned low-m maximum-likelihoodparameter space analy- for DM fromsources[2] theG. bright within Jungman, limb each of M. the ROI Earth Kamionkowski, from because the recent the and PSF 3FGL K. broad- Griest, cata- J. 490,493(1997), arXiv:astro-ph/9611107 [astro-ph]. whichsis in the 24 WIMP logarithmically-spaced abundance matches energy the bins observed and 0. DM1 spatial pixels. Data are additionally partitioned in one of enslog considerably [36Phys.] were Rep. also below267 included,195(1996) that energy. in, thearXiv:hep-ph/9506380 fit. To The further spectral avoid [hep- pa- [30] T. E. Jeltema and S. Profumo, J. Cosmol. Astropart. relic density. ph]. Phys. 0811,003(2008), arXiv:0808.2641 [astro-ph]. four PSF event types, which are combined in a joint- contaminationrameters of these from terrestrial sources were gamma fixed rays, at their events 3FGL with cata- In contrast, DM searches in the Galactic center take [3] L. Bergstrom, Rept. Prog. Phys. 63,793(2000), [31] T. Sjöstrand, S. Mrenna, and P. Z. Skands, Com- zenithlog values, angleslarger while their than normalizations 100 are rejected. were We refit also over re- the advantagelikelihood of function a J-factor when that performing is (100) the times fit larger, to each al- ROI arXiv:hep-ph/0002126 [hep-ph] . put.Phys.Commun. 178,852(2008), arXiv:0710.3820 movebroadband time intervals energy around range. bright The GRBs normalizations and solar of flares 3FGL though[19]. gamma-ray emission fromO non-thermal processes [4] G. Bertone, D. Hooper, and J. Silk, Phys. Rep. 405,279 [hep-ph]. Pass followingsources the more prescription than 5 away used for from the the 3FGL center catalog. are fixed We at [32] H. Cherno↵, Ann. Math. Statist. 25,573(1954). makesWe a used bright, a di structured↵use emission background. model based Several on the stud- (2005), arXiv:hep-ph/0404175 [hep-ph]. 7 Reprocessed model for Galactic di↵use emission,4 but extract[5] G. from Steigman, this data B. set Dasgupta, 10 10 andsquare J. F. regions Beacom, of in-Phys. [33] M. Ackermann et al. (Fermi-LAT Collaboration), Astro- ies of the Galactic center interpret an excess of gamma terest (ROIs)Rev. D in86 Galactic,023506(2012) coordinates⇥, arXiv:1204.3622 centered at [hep-ph] the po-. phys. J. 761,91(2012), arXiv:1205.6474 [astro-ph.CO]. rays with respect to modeled astrophysical backgrounds sition[6] ofM. each Mateo, dSphAnnu. specified Rev. Astron. in Table Astrophys.I. 36,435(1998), [34] A. Abramowski et al. (H.E.S.S. Collaboration), Phys. as a signal of 20 to 50 GeV WIMPs annihilating via the 5 The energy dependence of the e↵ective area and energy resolu- At a givenarXiv:astro-ph/9810070 energy, 20%–30% [astro-ph] of the events. classified as Rev. Lett. 106,161301(2011), arXiv:1103.3266 [astro- b¯b channel [16–19]. Coincidentally, the largest deviation tion is somewhat di↵erent in Pass 7 Reprocessed and Pass 8. 3 http://fermi.gsfc.nasa.gov/ssc/data/analysis/software/ photonsBecause in our the six-year Galactic diPass↵use 8 emissiondata set model are wasshared fit to withPass 7 from4 expected background in some previous studies of http://fermi.gsfc.nasa.gov/ssc/data/access/lat/ the analysisReprocessed of Ackermanndata withoutet accounting al. [13]. The for the low energy fraction dispersion, of BackgroundModels.html we have rescaled the model for this analysis. shared events can be attributed to the larger time range used for the present analysis (four versus six years), the 1 Strictly speaking, the di↵erential yield per annihilation in Equa- increase in gamma-ray acceptance of the P8R2 SOURCE tion (1)isasumofdi↵erential yields into specific final states: event class relative to P7REP CLEAN, and the migra- f dN /dE = f Bf dN /dE ,whereBf is the branching fraction of individual events across the class selection bound- tion into final state f.Here,wemakeuseofEquation(1)inthe aries caused by slight changes in the reconstructed energy P context of single final states only. and direction. Due to the relatively small overlap of the Dwarf spheroidals vs. globular clusters

12 Globular Clusters Recently Found Halo Clusters 10

(mag) a: J2204-4626

V c g b: J2356-5935

M 4 c: J0531-2801 a d d: J0002-6051 e e: J0345-6026 b f: J2337-6316 2 g: J2038-4609 f h: J0117-1725 2 2 2 Local Group Galaxies 0 h DES Y1A1 Candidates .5 mag arcsec DES Y2Q1 Candidates = 25 mag arcsec = 30 mag arcsec µ = 27 µ µ Other Candidates 2 100 101 102 103 Half-light Radius (pc) A. Drlica-Wagner et al. ApJ 813:109 (2015)

TeVPA 2016 Justin Vandenbroucke: millisecond pulsars in dwarf galaxies 36 Calculated flux PDF from individual dwarf galaxies

Fornax, M =2.5 107M 2 ⇤ ⇥ 10 Sculptor, M =3.9 106M ⇤ ⇥ Draco, M =3.2 105M ⇤ ⇥ 2 1 Segue 1, M =3.4 10 M 10 ⇤ ⇥ LF systematic uncertainty LF statistical uncertainty 100

1 10 Counts (arb. units)

2 10

3 10 17 16 15 14 13 12 11 10 10 10 10 10 10 10 2 1 Flux > 500 MeV (ph cm s )

TeVPA 2016 Justin Vandenbroucke: millisecond pulsars in dwarf galaxies 37