Search for Dark Matter Captured in the Sun with ... - KICP Workshops

Search for Dark Matter Captured in the Sun with 

the IceCube Neutrino Observatory 

Matthias Danninger for the IceCube Collaboration 

The Oscar Klein Centre for Cosmoparticle Physics, Stockholm University 

9th International Conference "Identification of Dark Matter, 2012" 

Chicago, USA on July 23-27, 2012

Results from a search 

Search for Dark Matter Captured in the Sun with 

the IceCube Neutrino Observatory 

in the 79-string configuration 

Matthias Danninger for the IceCube Collaboration 

The Oscar Klein Centre for Cosmoparticle Physics, Stockholm University 

9th International Conference "Identification of Dark Matter, 2012" 

Chicago, USA on July 23-27, 2012

Solar Dark Matter Search with IceCube 

All processes depend on WIMP mass 

Annihilation channel (branching ratios) 

Annihilation cross-section 

Capture (scattering) 

→ Scattering cross-sections (SI & SD) 

23/07/12 Matthias Danninger IDM 2012 

Proposed by: 

Silk, Olive & Srednicki '85 

Gaisser, Steigman & Tilav '86 

Freese '86 

Krauss, Srednicki & Wilzcek '86 

1








1








main analysis backgrounds: 

atm. ν∼Ο (10³ triggering events/day) 

1










atm. μ∼Ο (10⁸ triggering events/day) 

1







Striking signature: 

High-E ν excess 

over background 

from Sun direction 





1







± 23˚ 

Striking signature: 

High-E ν excess 

over background 

from Sun direction 

*Blind analysis with respect to true Sun azimuth 





1

IceCube 86-strings 

✗ 1.5 km ‐ 2.5 km deep 

✗ typically 125 m spacing between strings 

IceCube detector 

(~70 m in DeepCore,10x higher DOM density) 

✗ 60 Modules per string 

✗ 1 km ‐‐ 1 Gton instrumented volume 

✗ O(km) muon tracks from νμ CC 

✗ O(10m) cascades from νe CC, 

low energy ντ CC, and νx NC 

✗ Cherenkov radiation detected by 

3D array of optical sensors (DOMs) 

*More details in plenary talk by C. Rott 


ν 

μ 

DeepCore 

8 add. densely 

instrumented strings 

2

IceCube-79 string analysis details 

● Analysis for the whole year! Used 317 days livetime 

ν 

(151 days austral winter & 166 days austral summer) 

● more than 60 billion recorded events 

● Up-going 

IceCube 

DC 

1 

● No containment 

μ 

● Up-going 

IceCube 


ν 

DC 

2 

● strong containment 

μ 

*new* 

● At final level ~25000 signal-like events selected in 3 independent samples 

● With DeepCore, analysis reaches neutrino energies as low as 10-20GeV 

ν 

● Down-going 

IceCube 

DC 

μ 

3 

*new* 

● strong containment 

3


Dedicated online filters 

targeting: 

● muon events 

● Low energy muon 

events (IceCube) 

● DeepCore events 

● DeepCore events 

(summer) 


● By comparing signal simulation & data, cuts are 

placed that reduces the content of atmospheric 

muon events 

● Early analysis cuts: 

dedicated online-filter → offline high-level 

reconstructions → first modest cuts 

(track quality/direction + containment) 

Summer 

4



● By comparing signal simulation & data, cuts are 

placed that reduces the content of atmospheric 

muon events 

● Early analysis cuts: 

dedicated online-filter → offline high-level 

reconstructions → first modest cuts 

(track quality/direction + containment) 

● Signal event topology very different for 

low & high WIMP mass 

→ find geometrical cut to split dataset 

into 2 non-overlaping datasets 

5

Multivariate analysis step (BDT variable) 

event selection 1 (winter, high energy) event selection 3 (summer, low energy) 

data 

atm. ν atm. μ 

signal (scaled) 

“ mis” -reconstructed atm. μ background “ well” -reconstructed atm. μ background 

→ 1 separate BDT for each event selection 

→ training on off-source exp. data + separate signal simulation 


6

atm. ν 

atm. μ 

Multivariate analysis step (final cut applied) 


Total Bg 

data 



→ Optimized final cut on BDT-output: run llh-analysis for various BDT cuts, to 

determine cut value with best sensitivity (MRF & MDP) 


7

Reconstructed zenith (final analysis level) 


atm. ν 

atm. μ 

Total Bg 

data 






8

The observed angle to the 

Sun is fitted with signal and 

background pdf:s 

Maximum llh-analysis 


background expectation from data 

signal simulation (e.g. 1000 GeV) 

Angle between event track and direction 

from the Sun 

10





How many signal events can be 

consistent with the observation? 

Evaluate shape fit with log− 

likelihood rank (Feldman-Cousins) 

to construct confidence regions 

for the number of signal events μs 

where L is the pdf product over the 

final sample 





observation 


10





How many signal events can be 

consistent with the observation? 

Evaluate shape fit with log− 

likelihood rank (Feldman-Cousins) 

to construct confidence regions 

for the number of signal events μs 

where L is the pdf product over the 

final sample 





observation 


Scale to multiple datasets 

10

Unblinding results (expected sensitivity) 


11

Unblinding results (events observed) 

event selection 1 (winter, high energy) event selection 2 (winter, low energy) 

event selection 3 (summer, low energy) 


observed events 

background expectation 

Upper limit on number of 

signal events (example) 

12

Unblinding results (observed results) 


13

Systematic uncertainties 

Preliminary evaluation of systematic uncertainties 

Source of uncertainty Estimated effect 

Neutrino oscillations pending 

DOM sensetivity spread* 

pending 

Interaction cross-section 3.6 – 7.1 % 

Muon propagation in ice < 1 % 

Ice model* 

Absolute DOM efficiency* 

1 – 20 % 

14 – 51 % 

time/position callibration < 3 – 5 % 

Statistical error on signal < 2 % 

* Full analysis performed with an alternative signal simulation, 

including maximum llh-analysis (change in acceptance + PSF) 


14

Unblinding results (muon flux limit) 


15

Unblinding results (SI-cross-section limit) 


16

Unblinding results (SD-cross-section limit) 


17

Summary 

✗ New results from 79-string data (~1y livetime) 

✗ First Dark Matter analysis including DeepCore 

✗ First full year-round IceCube solar Dark Matter search 

✗ No excess of events from the Sun over expected backgrounds 

✗ New very competitive SD-cross-section limits 

→ most stringent limits in large parts of WIMP mass range 

→ new LKP limits with same search (not discussed in talk) 

✗ The near future .......... 

IceCube already took data for more than 1 year in the full 

86-string configuration 

→ 2 more DeepCore strings (even lower energy threshold) 

→ many many years of data to come.... 


18

Additional slides 

23/07/12 Matthias Danninger IDM 2012

Multivariate analysis step (BDT variable) 

event selection 2 (winter, low energy) 

data 

atm. ν 

atm. μ 

signal 




6

Reconstructed zenith & BDT output 

(final analysis level) 

event selection 2 (winter, low energy) event selection 2 (winter, low energy) 

Total Bg 

data 

atm. ν 

atm. μ 






8

Analysis: final cut on BDT-output 

1000 GeV hard 


Optimized final cut on BDT-output: 

run full llh-analysis for various BDT cuts, to 

determine the cut value with the best sensitivity: 

● each dataset individually 

● calculate MRF 

● calculate MDP (Punzi) 

● check for many mass/channel combinations 

Want to find 1 single cut per dataset 

→ (robustness rather than fine-tuning) 

Still very 

efficient

Sensitivity extends to 

WIMP masses of 20 GeV 

Only 1 year of data 

Data unblinding soon! 

also search for UED 

models (not shown here) 

IceCube 79 string sensitivity 


preliminary

New model independent method for 

theories of new physics 

(Solar Dark Matter searches) 


New SUSY analysis with IceCube 

What can the muon signal tell me? 

Roughly: 

✗ Number – how much annihilation is going on in the Sun 

⇒ info on σSD , σSI and 

✗ Spectrum – sensitive to WIMP mass mχ and branching 

fractions BF into different annihilation channels χ 

✗ Direction – how likely it is that they come from the Sun 

In model-independent analyses a lot of this information is either discarded or not 

given with final limits 

Goal: 

Use as much of this information on σSD , σSI , , mχ and BF (χ ) as possible to 

directly constrain specific points and regions in WIMP model parameter spaces 


More details: P.Scott, C.Savage, J. Edsjö 

& the IceCube Collaboration, arXiv:1207.0810

Include IceCube event level data in a global statistical fit. 

→ parameter estimation rather than model exclusion 

Composite likelihood made up of observations from all over: 

Dark matter relic density from WMAP 

Precision electroweak tests at LEP & LEP limits on sparticle masses 

B-factory data (rare decays, b → sγ) 

Muon anomalous magnetic moment 

LHC searches, direct detection (not yet included in examples) 

+ IceCube 

Global SUSY analysis with IceCube 

More details: P.Scott, C.Savage, J. Edsjö & the IceCube Collaboration, arXiv:1207.0810 

unbinned likelihood 




CMSSM, IceCube-22 

✗ Contours indicate 1σ and 2σ credible regions 

✗ Grey contours correspond to fit without IceCube data 

✗ Shading+contours indicate relative probability only, not overall goodness of fit 




CMSSM, IceCube-22 with 100x boosted effective area 

(indication for IceCube-79 and 86-string prospects) 

✗ Contours indicate 1σ and 2σ credible regions 

✗ Grey contours correspond to fit without IceCube data 

✗ Shading+contours indicate relative probability only, not overall goodness of fit

Search for Dark Matter Captured in the Sun with ... - KICP Workshops

Create successful ePaper yourself

Delete template?

Save as template?