constraining supersymmetry with fermi and segue 1patscott/talks/ps_leicester.pdf · constraining...

Constraining supersymmetry with Fermi andSegue 1

Pat Scotton behalf of the Fermi-LAT Collaboration

Oskar Klein Centre for Cosmoparticle Physics (OKC) &Department of Physics, Stockholm University

With: Jan Conrad, Joakim Edsjö, Lars Bergström, Yashar Akrami(OKC/Stockholm) & Christian Farnier (Montpellier II,

LPTA/CNRS-UM2)

Based on arXiv:0909.3300 (JCAP in press)Slides available from www.fysik.su.se/̃ pat

Pat Scott – Jan 18 – Leicester Constraining supersymmetry with Fermi and Segue 1

http://www.arxiv.org/abs/0909.3300

http://www.fysik.su.se/~pat

The Fermi Large Area Telescope (LAT)

• Overall modular design.

• 4× 4 array of identical towers (each one including a tracker and a calorimeter module).

• Tracker surrounded by Anti-Coincidence Detector (ACD)


The Fermi Large Area Telescope (LAT)

• Overall modular design.

• 4× 4 array of identical towers (each one including a tracker and a calorimeter module).

• Tracker surrounded by Anti-Coincidence Detector (ACD)

Tracker• Silicon strip

detectors, Wconversion foils; 1.5radiation lengthson-axis.

• 10k sensors, 80 m2

of silicon active area,1M readout channels.

• High-precisiontracking, short deadtime.

Anti-CoincidenceDetector• Segmented (89 tiles) to

minimize self-veto athigh energy.

• 0.9997 averagedetection efficiency.

Calorimeter• 1536 CsI(Tl) crystal; 8.6 radiation

lengths on-axis.

• Hodoscopic, 3D shower profilereconstruction for leakage correction.


Fermi-LAT Instrument Response Functions (IRFs)

Effective areaPoint-Spread Function (PSF)Energy dispersion


Fermi-LAT Instrument Response Functions (IRFs)

Effective areaPoint-Spread Function (PSF)Energy dispersion


FLATLIB - fast convolution library for processor-intensivemodel scansSource freely downloadable fromwww.fysik.su.se/̃ pat/flatlib

http://www.fysik.su.se/~pat/flatlib

Gamma-rays from neutralino dark matter2 photons (or Z+photon):

monochromatic lines

χ 0

1

χ 0

1

γ

γ/Z

Internal bremsstrahlung:

hard gamma-ray spectrum

Secondary decay:

soft(er) continuum spectrum

χ 0

1

χ 0

1

γ

χ 0

1

χ 0

1

SM

SM

SM

SM

SM

SM

SM

SM

π

γ

γ

γ

Neutralinos: linear combinations ofsuperpartners of γ, Z and H0

Specific example of WIMP darkmatter

Neutral, carries SU(2)L charge,stable if R-parity conserved

3 main gamma-ray channels:monchromatic linesinternal bremsstrahlungcontinuum from secondary decay

Φ ∝ annihilation rate ∝ ρ2DM


Likely targets

Galactic centreGalactic haloclusters & extragalactic diffusedark clumpsdwarf galaxies


Likely targets

Galactic centreGalactic haloclusters & extragalactic diffusedark clumpsdwarf galaxies


7 × 7◦ around the GC

V. Vitale, 2009 Fermi Symposium Poster

Likely targets

Galactic centreGalactic haloclusters & extragalactic diffuse (coming soon)dark clumpsdwarf galaxies


7 × 7◦ around the GC

V. Vitale, 2009 Fermi Symposium Poster

Dark clumps: Ultracompact minihalos

Small-scale, large amplitudedensity perturbations in the earlyUniverse can create ultracompactminihalos (Ricotti & Gould, arXiv:0908.0735)

Known phase transitions couldhave generated the enhancedperturbationsClump mass is set by horizonscale at time of transition =⇒specific clump mass scaleAlso excellent indirect detectiontargets

(PS & Sivertsson, Phys. Rev. Lett. 2009)


➤

Inte

grat

edflu

xab

ove

thre

shol

d(c

m−

2s−

1)

Energy threshold (GeV)

e +e −

annihilation epoch

QCDphase transition

electroweak phase transition

Scott & Sivertsson 2009

bb̄: no boostµ+µ−: boost = 100

CTA/AGIS

VERITAS+ MAGIC

+ HESS

Fermi -LAT

0.1 1 10 100 103 104

10−

20

10−

15

10−

10

10−

5

bb̄, mχ = 100 GeV

bb̄, mχ = 5 TeV

µ+µ−, mχ = 100 GeV

µ+µ−, mχ = 5 TeV

Dwarf galaxies - constraints on ‘normal’ models

Why dwarfs?Very high mass-to-lightratios=⇒ lots of DM, little BGHigh latitude =⇒ low BG=⇒ arguably the besttargets for WIMP gammas


C. Farnier, 2009 Fermi SymposiumLAT Collaboration, ApJ submitted

Dwarf galaxies - constraints on boosted models


C. Farnier, 2009 Fermi SymposiumLAT Collaboration, ApJ submitted

Segue 1 dwarf galaxy

Why Segue 1?Close(ish) – 23 kpcM/L ∼ 1300 (large)The best S/N dwarf forWIMP gammasLeading the pack in Fermidwarf upper limit analysis

Purposes:see if Segue observations do/willimpact real models at all,considering ‘soft bounds’

attempt to validate dwarf ULanalysis via an independent,rather different analysis


Segue 1100 MeV–300 GeV

Scanning supersymmetric parameter spaces

Goal: given a particular version of SUSY, determine whichparameter combinations fit all experiments, and how well

Issue 1: Combining fits to different experimentsEasy – composite likelihood (L1 × L2 ≡ χ2

1 + χ22)

dark matter relic density from WMAPprecision electroweak tests at LEPLEP limits on sparticle massesB-factory data (rare decays, b → sγ)muon anomalous magnetic moment

Issue 2: Finding the points with the best likelihoodsTough – grid scans, MCMCs, nested sampling or geneticalgorithms (see e.g. arXiv:0910.3950 for genetic)

Model: We focus on the Constrained MSSM (CMSSM)GUT boundary conditions on soft SUSY breaking parameters such thatonly 4 free parameters and 1 sign remainincorporates the simplest implementation of mSUGRA



Scanning supersymmetric parameter spaces

Goal: given a particular version of SUSY, determine whichparameter combinations fit all experiments, and how well

Issue 1: Combining fits to different experimentsEasy – composite likelihood (L1 × L2 ≡ χ2

1 + χ22)

dark matter relic density from WMAPprecision electroweak tests at LEPLEP limits on sparticle massesB-factory data (rare decays, b → sγ)muon anomalous magnetic moment

Issue 2: Finding the points with the best likelihoodsTough – grid scans, MCMCs, nested sampling or geneticalgorithms (see e.g. arXiv:0910.3950 for genetic)

Model: We focus on the Constrained MSSM (CMSSM)GUT boundary conditions on soft SUSY breaking parameters such thatonly 4 free parameters and 1 sign remainincorporates the simplest implementation of mSUGRA


m0 scalar mass parameterm 1

2gaugino mass parameter

tan β ratio of Higgs VEVsA0 trilinear couplingsgn µ Higgs mass parameter

(+ve in our scans)


Including Segue 1 in SUSY scans

Same cuts as dwarf UL analysis“DIFFUSE” event class

105◦ zenith angle cut

10◦ ROI

14 energy bins from 100 MeV–300 GeV

Binned Poissonian likelihood (similar to dwarf UL analysis)

Spatial-spectral fit to inner 6× 6 bins of 64× 64 ROI (dwarf ULanalysis assumes point source)

Segue 1 halo profile from best fit Einasto profile by Martinez etal. (2009; JCAP 6:14) (NFW in dwarf UL analysis)


Including Segue 1 in SUSY scans

Galactic diffuse BG from preliminary Fermi all-skyGALPROP fitsIsotropic powerlaw extragalactic BG (as seen by EGRET)BG normalisations from dwarf UL fits (i.e. full 10◦ × 10◦)Fast integration over energy-dependent IRFs (P6v3) withFLATLIB – (dwarf UL analysis skips energy dispersion)Inclusion of systematic errors from effective area andtheoretical calculations – (dwarf UL analysis skipssystematics)Integration into SUPERBAYES, upgraded with DARKSUSY5 (including internal bremsstrahlung), bug fixes, etc.515 data points in new global fit, vs 11 previously withSUPERBAYES 1.35 (admittedly not such a fair comparison)


Results - Segue 1 only

mχ1

0 (TeV)

log 10

[ <

σ v>

(cm

3 s−

1 ) ]

Profile likelihood

Flat priors

CMSSM, µ>0

Phys + Nuis only

Scott et al. 2009

0.5 1 1.5

−26

−25

−24

mχ1

0 (TeV)

log 10

[ <

σ v>

(cm

3 s−

1 ) ]

Profile likelihood

Flat priors

CMSSM, µ>0

Segue 9 mth, BF=1

Scott et al. 2009

0.5 1 1.5

−26

−25

−24

mχ1

0 (TeV)

log 10

[ <

σ v>

(cm

3 s−

1 ) ]

Profile likelihood

Flat priors

CMSSM, µ>0

Segue 9 mth, BF=50

Scott et al. 2009

0.5 1 1.5

−26

−25

−24

mχ1

0 (TeV)

log 10

[ <

σ v>

(cm

3 s−

1 ) ]

Profile likelihood

Flat priors

CMSSM, µ>0

Segue 5 yr, BF=1

Scott et al. 2009

0.5 1 1.5

−26

−25

−24

mχ1

0 (TeV)

log 10

[ <

σ v>

(cm

3 s−

1 ) ]

Profile likelihood

Flat priors

CMSSM, µ>0

Segue 5 yr, BF=50

Scott et al. 2009

0.5 1 1.5

−26

−25

−24


Results - all observables + Segue 1

mχ1

0 (TeV)

log 10

[ <

σ v>

(cm

3 s−

1 ) ]

Profile likelihood

Flat priors

CMSSM, µ>0

All other constr.

Scott et al. 2009

0 0.5 1−30

−28

−26

−24

mχ1

0 (TeV)

log 10

[ <

σ v>

(cm

3 s−

1 ) ]

Posterior pdf

Flat priors

CMSSM, µ>0

All other constr.

Scott et al. 2009

0 0.5 1−30

−28

−26

−24

Relative P

robability Density

0

1

mχ1

0 (TeV)

log 10

[ <

σ v>

(cm

3 s−

1 ) ]

Profile likelihood

Flat priors

CMSSM, µ>0

Segue 9 mth + All, BF=50

Scott et al. 2009

0 0.5 1−30

−28

−26

−24

mχ1

0 (TeV)

log 10

[ <

σ v>

(cm

3 s−

1 ) ]

Posterior pdf

Flat priors

CMSSM, µ>0

Segue 9 mth + All, BF=50

Scott et al. 2009

0 0.5 1−30

−28

−26

−24

Relative P

robability Density

0

1

mχ1

0 (TeV)

log 10

[ <

σ v>

(cm

3 s−

1 ) ]

Profile likelihood

Flat priors

CMSSM, µ>0

Segue 5 yr + All, BF=50

Scott et al. 2009

0 0.5 1−30

−28

−26

−24

mχ1

0 (TeV)

log 10

[ <

σ v>

(cm

3 s−

1 ) ]

Posterior pdf

Flat priors

CMSSM, µ>0

Segue 5 yr + All, BF=50

Scott et al. 2009

0 0.5 1−30

−28

−26

−24

Relative P

robability Density

0

1


Using extreme (∼95% C.L. excluded) DM profile

mχ1

0 (TeV)

log 10

[ <

σ v>

(cm

3 s−

1 ) ]

Profile likelihood

Flat priors

CMSSM, µ>0

9 mth + All, BF=50, max halo

Scott et al. 2009

0 0.5 1−30

−28

−26

−24

mχ1

0 (TeV)

log 10

[ <

σ v>

(cm

3 s−

1 ) ]

Posterior pdf

Flat priors

CMSSM, µ>0

9 mth + All, BF=50, max halo

Scott et al. 2009

0 0.5 1−30

−28

−26

−24

Relative P

robability Density

0

1

mχ1

0 (TeV)

log 10

[ <

σ v>

(cm

3 s−

1 ) ]

Profile likelihood

Flat priors

CMSSM, µ>0

5 yr + All, BF=50, max halo

Scott et al. 2009

0 0.5 1−30

−28

−26

−24

mχ1

0 (TeV)

log 10

[ <

σ v>

(cm

3 s−

1 ) ]

Posterior pdf

Flat priors

CMSSM, µ>0

5 yr + All, BF=50, max halo

Scott et al. 2009

0 0.5 1−30

−28

−26

−24

Relative P

robability Density

0

1


Conclusions

Existing 9 month dataset does constrain the CMSSM byitself, but only weakly5 years of data will provide significantly better constraints,but...Not quite good enough to impact models which are notalready disfavoured by other constraints (eg relic density)In the (unlikely) event of a later signal from Segue 1, wecan zero in on the preferred CMSSM model andcross-section very quickly, and provide confidence intervalsConsistent with limits found in the dwarf upper limitanalysisFLATLIB source freely available fromwww.fysik.su.se/̃ pat/flatlib


http://www.fysik.su.se/~pat/flatlib

constraining supersymmetry with fermi and segue 1patscott/talks/ps_leicester.pdf · constraining...

Documents