Transcript Introductionx

Deconvolution of the energy
spectrum of atmospheric nm
Maurizio Spurio
Phone conference 14/12/2011
For reference:
 Three Amanda/IceCube papers:
1.
1 -AMANDA
2-AMANDA
2.
3.
3 – IC40
Determination of the Atmospheric Neutrino Flux and Searches for New Physics with AMANDA-II
. Physical Review D79 (2009) 102005, [arXiv:0902.0675 [astro-ph.HE]]
The Energy Spectrum of Atmospheric Neutrinos between 2 and 200 TeV with the AMANDA-II
Detector. Astroparticle Physics 34 (2010) 48-58 [arXiv:1004.2357 [astro-ph.HE]]
Measurement of the Atmospheric Neutrino Energy Spectrum from 100 GeV to 400 TeV with
IceCube. Physical Review D83 (2011) 012001[arXiv:1010.3980 [astro-ph.HE]]
 Different physics arguments involved (Lorentz invariance, diffuse flux,
prompt component, n velocity…)
 Diffuse flux requirement: blind analysis for the high-energy tail (En>10
TeV)
1 – AMANDA
forward folding
2-AMANDA
2000-2003
390 events
3 – IC40
18000 evts
Open questions in ANTARES
 Which energy estimator(s)?
 Neutrino energies difficult when Em < 2 TeV
 Data/MC comparison status
 Is the data sample selected for point-like sources (Point source search
with 2007-2010 data.- ANTARES-PHYS-2011-005) the best one?
 NO, too high fraction from atmospheric muons.
 Which runs to be used to compare with MC inputs (open sample)?
 Which kind of analysis (deconvolution?)
 Systematics
 Timescale for paper(s) preparation
1. The Energy estimators
Difficult below few TeV
“MC true”, “MC reco”, and “DATA”
“DATA” and “MC reco” should match:
Yes, if the distributions of input MC and data agree
Mandatory the use of the best/newest MC version
…and “MC reco” and “MC true” also!
 Use the best energy estimator
Recommendation #1: Data/MC
comparisons of input parameters
 If Data/MC of input parameters does not match
 MC reco will be different from MC true
 Example in diffuse flux paper
 Define the input variables for the energy estimators
 Simplest: R, Nhit or their combination
 More complex: Neural Network (ANN, ML)
 Use the most recent MC production (see Vladimir K.
presentation@Strasbourg) using –C4 option in TE
 Compare Data/MC distributions of input variables
 Define the validity range of input parameters (ex: Ri<5), if
possible
Nhit
R
 Good agreement between the variables used for the point-
like source analysis is not sufficent
 Agreement between the energy-dependent variables
Dimitris Palioselitis @ Bamberg Meet.
2.The problem of the HE “pollution”
Fn(E)En-3.7
Energy
resolution
Fn(E1)/ Fn(E1)30
E1 E0
eve
nts
Wrong energy
events
E
n
3. Atmospheric muons
 Atmospheric muon events do not have a defined “MC true” energy
 Some muons in the bundle do not fire any PMT, some produce few
hits, few produce the majority of hits
 The “MC Reco” depends on the energy estimators. It has nothing to do with
the “MC true”energy!
 However, the “MC Reco” shifted towards high value pollution
 Remove as much as possible the contamination of atmo m in the data
4. Unblinded runs
 Dimitri (Bamberg meeting): runs 034419-036791, 12 line





data 05/0810/08
Jutta: Run 36906
Bologna: used the same runs of Dimitri
COORDINATION between different analysis
We propose that at least 10-20% of the data can be used in
order to compare the output variables in data/MC
If a blind region is defined (using R, Nhit, ANN) the data
outside the blind region can be used to compare data/MC.
5. Problematic events after the unblinding
 All events candidate after unblinding must be checked
 It could happens that pathologies are present
 Pathology present (sparkling). Usually, Ri<5
Ri
6. Unfolding/systematics
 Unfolding: different works in progress (using ROOT)
 Systematic effects: other effects in addition to that used for
the diffuse flux paper? Trial on different MC samples
3 – IC40
paper
FIG. 22. Sources of uncertainty in the
unfolded energy spectrum. The solid lines are
the systematic uncertainties due to DOM
sensitivity and ice property uncertainties; the
short-dashed lines are the uncertainties
implied by zenith-dependent inconsistencies
in data/ simulation comparisons; and the
long-dashed lines are the statistical and
regularization uncertainties from toy MC
studies. Not shown is the uniform 4%
uncertainty due to miscellaneous normalization errors assumed to be independent of
energy.
7. Outlook
 The deconvolution of the atmospheric nm already published by




IceCube; other limits also presented
Note: 2-AMANDA is 10% above Bartol; 3 – IC40 is below Bartol,
with larger statistics and much higher systematic errors
Good ANTARES opportunity to produce interesting paper(s),
probably less affected by systematic uncertainties
Due to the “pollution” effect on High E bins, the most energetic
events have a certain probability to be produced by Low E nm
Consequences for the “hunt” to the highest E event for CohenGlashow neutrino velocity constraint