status on the cross check analysis for the e + +e - flux estimation villi scalzotto – luigi cossio...
Post on 16-Dec-2015
219 Views
Preview:
TRANSCRIPT
Status on the cross check analysis
for the e++e-
flux estimation
Villi Scalzotto – Luigi Cossio – Michele PalatielloUdine, MAGIC Collaboration Meeting 2012 – June,10th-16th
2
Outline
This is the status of a CROSS CHECK analysis of DANIELA BORLA TRIDON's results on the e+e- spectrum
• Quick introduction• Description of the method• Data set informations• CRAB Test of the method• Electron Flux estimation• Discussion
3
• At TeV energies, CR-electrons carry informations about nearby sources, due to energy losses via synchrotron radiation and IC-scattering.
Lifetime (propagation distance) is limited
(105lyr 1kpc) Spectrum is isotropic and steeper than hadronic CRs
(~ E-3.3 vs E-2.7)• Direct measurements at HE is then difficult, but a
chance for Cherenkov telescope:- larger coll. area than satellites (~ 105 m2)
- sufficient energy resol. to resolve the spectrum, such to discriminate among different models
CR-Electrons
OriginElectrons lose energy by:• Ionization• Synchrotron• Inverse Compton• Bremsstrahlung
Possible contributions to HE electrons flux:• Secondary electrons generated in CR interactions
with ISM• Supernova explosion (sharp cutoff in the
spectrum)• Distant sources uniformly distributed• Pulsars• Dark matter
4
MAGIC & THIS PHYSICAL CASE
Drag picture to placeholder or click icon to add
INTENT:use
MAGICto support or not
ATIC/FERMI/HESS result
PRO: No dedicated time
AGAINST:Lower sensitivity
Large MCProtons simulation needed to
match MCProtons to RealData
MAGIC
5
Villi ScalzottoFERMIATICHESS
Results by other experiments
HOW TO COUNT ELECTRONS?
Drag picture to placeholder or click icon to addGAMMAS and ELECTRONS
morphologically equivalent for MAGIC
BUT ELECTRONS are DIFFUSEALPHA parameter not useful!
All the real datacontain electrons
(both ON and OFF!)
The normalizationof the Alpha plot
allows toobtain the same
amount of electrons
in ON and in OFF:
USE OF HADRONNESS ASSIGNAL PLOT
electrons
6
7
HadronnessTheta2
0 10 0.3
• "ON" Data (with gamma rays from a pointlike source)• Non leptonic cosmic rays (p, He, ecc.) renorm.• "OFF" Data (with diffuse Electrons (and diffuse gammas)
renormalizedHadronnessTheta2
0 10 0.3
• MC Protons simulation (+He + diffuse gammas)• Real Data from a dark patch (thus diffuse electrons + non leptonic cosmic
rays)
THE METHOD
STD ANALYSIS
HADPHA METHOD
Hadpha stands for the
Hadronness parameter used as "Alpha" for the signal plot
8
No ALPHA/THETA2 plot
As signal plot
NO ALPHA/THETA2 cut!
ELECTRONS are mostly gammalikeHADRONS are mostly hadronlike
…GAMMAS…are mostly gammalike!
Hadronness plotused to find
signal!
Hadronness:our new “Alpha”
“ON” : MAGIC OFF (with electrons and no gamma contamination)
“OFF”: SIMULATION OF THE BACKGROUND(no electrons in the background)
?
THE HADPHA METHOD
9
Datasets needed (all in STEREO):
Real DataCRAB ON data (7.5h)
OFF data (14h)
SimulationMC electrons (~4e5 after stereo trigger)
MC pointlike gammas (~1.6e5 after stereo trigger)MC protons (~1.4e6 after stereo trigger)
In the future:Helium, diffuse gammas, more and more protons…
mostly given by Daniela Borla Tridon,extension thanks to Luigi Cossio and Michele
Palatiello
DATASET
10
1FGL J2347.3+0710
RA 23h47m19.9s DEC +07d10m26s
• Galactic long: 96.29 lat: -52.35 • z =N/A
• 6-10-11-13-31/10_2010, 01-25
/11_2010• Zdmin=21.5
Zdmax=30
• Data selection: bad rate (manually checked) and
cloudiness
Total effective time after cuts: 8.84h
3c454RA 22h51m34.7s DEC +18d48m40s
• Galactic long: 87.35 lat -35.65
• z =1.75• 6-8-10-11/12_2009
• Zdmin=15 Zdmax=30
• Data selection: bad rate (manually checked) and
cloudiness
Total effective time after cuts: 2.93 h
TOTAL ~ 14h
BL Lac
RA 22h00m39.3723s DEC +42d02m08.495s • Galactic long: 92.58
lat -10.44• z=0,068
• 15-17/6_2010
• Zdmin=26 Zdmax=31
• Data selection: bad rate (manually checked) and
cloudiness
Total effective time after cuts: 2.12 h
REAL "OFF" DATA
11REAL "OFF" DATA
Zenith angle distributions of the dataset
12
• RA 05h34m32.0s DEC +22d00m52s• Gal long 184.557593 lat -5.784197• 8-9-10-11-12-14/11_2010• Zdmin= 7 Zdmax=19• Data selection: bad rate (manually checked) and
cloudiness
Total effective time after cuts: 7,57
CRAB DATA
CRAB ON STEREO DATA SAMPLE:
13 MC SIMULATIONSimulated events (@ 2012):
ELECTRONSZenith 5-35 degrees
70 GeV – 7 TeV (and 30 TeV)ViewCone: 4.5 deg, PowerLaw: -2
# = O(8.107)
PROTONSZenith ~ 0-38 gradiVarious populations:
from (30,70,70) GeV to (30,20,30) TeVViewCone: 5/6 deg, PowerLaw: -1.78/-2
# = O(2.109)
Events surviving the triggering of the stereo system:
Electrons: O(4.105) (~0.4%) evtsProtons: O(1.4.106) (~0.07%) evts
Cleaning M1:6/3 time M2: 9/4.5 time
14
In addition:
HELIUMZenith 8-38 deg70 GeV - 20 TeV
ViewCone: 5/6 deg, PowerLaw: -2# = O(4.108) Post Trigger = 1.1.105 (0.03%)
DIFFUSE GAMMAS
Zenith 5-30 deg10 GeV - 30 TeV
ViewCone: 0 deg, PowerLaw: -1.6# = O(3.107) Post Trigger = 1.8.106 (6%)
POINTLIKE GAMMASZenith 5-35 deg10 GeV - 30 TeV
ViewCone: 1.5 deg, PowerLaw: -1.6# = O(2.106) Post Trigger = 1.6.105 (8%)
NOT YET USED FOR THIS ANALYSIS
MC SIMULATION
16
RF Electrons study: Train:BKG:
BL Lac 2.12h + 3c454 2.93hMC:
electrons (Daniela + Luigi)
Cuts NTree: 100
RF.trainRatio: 0.95Leakage1>0.15NumIslands>1
Energy Estimation: STD LUTSby MC RFtest sample
RF Crab Test:Train:BKG:
BL Lac 0.56h + 3c454 0.51hMC:
pointlike gammas
USED RF matrix
PARAMETERS:MaxHeight (stereo reconstructed)
Size (M1 and M2)Length (M1 & M2)
Impact par. (M1 & M2)TimeGrad (M1 & M2)
(MHillasTimeFit_*.fP1Grad)MaxHeight (M1 & M2)
Width (M1 & M2)
18
Leakage{1,2} < 0.2
Dist{1,2} <300 mm
NumIslands{1,2}<2
Size1>100-150 Size2>200-250
10-12<Impact{1,2}[m]<250-300
MuonRingCenterDist<20 Quality cutsThese cuts have been applied at the electronflux level in order to obtain a better match between MCProtons and RealData
19
PARAMETER DISTRIBUTIONS after the quality cuts
Comparison between Real Data and MC protons after the quality cuts.The MC sample is corrected considering the different power law.
20
PARAMETER DISTRIBUTIONS after the quality cuts
Parameters as function of the energy.In case of M2, the agreement is quite good.
21
PARAMETER DISTRIBUTIONS after the quality cuts
Case of M1.
22
PARAMETER DISTRIBUTIONS after the quality cuts
Quite ok.Perhaps there is a small mismatch for the distributions of L,W,M3Long.
23
Crab Nebula test
CRAB DATA
To test the performances of the Hadpha Method we use it to estimate a flux by a standard gamma source (Crab).
The method will be obviously less performing than the usual Theta2 Method.
But the spectrum of a strong gamma source should be obtained.
Standard RF, hadronness plot as significance plot
25
CRAB TESTTheta VS Hadpha
• EffArea after cuts is lower for Theta Method because it includes the hadronness plot.
• But remember the sensitivity of the Hadpha method is lower!
Theta2 Method
Hadpha Method
26
CRAB TESTTheta VS Hadpha
Theta2 Method
Hadpha Method
Good ON/OFF match!
We are using Real Data both for ON and OFF.We do expect this match is not that good in the electron case!
0.4-0.8 seems a good normalization range
27
CRAB TESTTheta VS Hadpha
Theta2 Method
Hadpha MethodThe points are consistent within errors.
But there is still a slight underestimation to be understood.
28
HADPHA METHOD SENSITIVITY
Theta2 Method
Hadpha Method
As expected, the sensitivity (in Crab units for 50hrs) is worse (~factor 4 for differential sensitivity).
The lowest points for Hadpha are not trustable (see later).
29
Theta2 Method
30
Hadpha Method
In this method it's very peculiar the selection of
the normalization region
31
11<Zd<30 // 20<Zd<30
Azimuth bins: 1 / 9
Energy Bins: 20 / 25 / 30
Hadpha Efficency: 0.55 – 0.75
Normalization region: [0.4/0.45 --- 0.7/0.85] Electronflux
optionsDifferent flux have been extracted by varying some options in a reasonable range
32
Different Hadpha normalization region possible…
Discrepancy at the highest values
Perhaps it's enough to apply a quality cut
33
Discrepancies at the lowest energy values (as in Daniela's work). Due to mismatch btw simulation and real data.
The matching improves with the energy.
34
35
Signal persists at higher energies, above 1 TeV, but these number of excesses would give huge values in the spectrum.
Lack of statistic?
To investigate.
36
??
37
e++e-
DIFFERENTIAL SPECTRUM
HadEff = 0.7-0.8HadEff = 0.55-06
E3dN/dE spectra are highly sensitive to the variation of some options in the input card
Main variations due to Zd range, Hadpha Efficiency and the normalization region
We superimpose the different result. We divided it into 2 samples, with lower and higher Hadpha Efficiency
38
e++e-
E3Spectrum
FERMIHESSMAGIC CROSS CHECK
39
e++e-
E3Spectrum
HadEff = 0.7-0.8HadEff = 0.55-06
40
Public resultShown at ICRC'11ICRC proceeding
41
• There is a mismatch at high hadronness values to be better understood (in disagreement with Daniela's result) - We used the reweighting
• The method is sensitive to the Zd selection and the normalization region
• Crab Test with diffused electrons to better understand the behaviour of the ViewAngleCone
• Behaviour at high energies? Why do the values seems to increase?
• E3dE/dN distribution is highly amplifying the fluctuations. It's a very precise work to deal with..
• The fluctuations obtained by varying the electron flux options could give an estimation of our systematics due to the analysis
Technicalconsiderations
42
• Results are consisent with Daniela's work, but the E3dN/dE spectrum looks a bit higher.
• Range 150<E<1000 GeV. Problems at higher energies.
• We have reported an overlap of spectra. No PowerLaw estimated yet.
• We cannot exclude the presence of a flux higher than expected and measured by Fermi.
• Analysis is finally working (in the past several problems with this crosscheck!)
• Need to fine tune some disagreement (large hadronness mismatch and high energy values). Larger MC production could help.
Conclusions
43
BACKUPs
44
•Consistency considerations in the mono case:
•Electrons are diffuse:
we need to integrate over the solid angle(we assume an “effective” FOV
of 0.4deg X 0.4deg = 1.5*10-4 sr)
•At 500 GeV:
CRAB = 2 10-13 cm-2 s-1 GeV-1
ELECT (Fermi) = 114/E3(GeV) m-2 s-1 GeV-1 sr-
1
= 9 10-11 * 1.5 10-4cm-2 s-1 GeV-1 sr-1
= 1.35 10-14 cm-2 s-1 GeV-1
Elect.Flux ~ 7% Crab!
Observation time: 50*(10/7)2 = 100 h
45
MC ProtonsM1 vs M2
46
MC ProtonsM1 vs M2
47
Real DataM1 vs M2
48
Real DataM1 vs M2
49
Some BACKUP slides by Daniela Borla's talk
50
51
52
53
54
55
56
57
Gamma sample (M1: PSF 11.44mm - M2: PSF 12.66mm
58
59
top related