particle flow calorimetry: experimental status and ...indico.ihep.ac.cn/event/910/session/17/... ·...
TRANSCRIPT
Particle Flow Calorimetry:Experimental Status and Technical
DevelopmentsFelix Sefkow
Friday, May 14, 2010
MC
Particle Flow Calorimetry: Experimental Felix Sefkow CALOR 2010, Beijing, May 10-14, 2010
What we learnt from CALICE:
• Introduction: – testing PFLOW calorimetry
• In real terms:– Validate simulation– test algorithms– test technologies and establish
feasibility
2
Friday, May 14, 2010
MC
Particle Flow Calorimetry: Experimental Felix Sefkow CALOR 2010, Beijing, May 10-14, 2010
Understand particle flow performance
• Particle flow is always better– even at high jet energies
• HCAL resolution does matter– also for confusion term
• Leakage plays a role, too
3
ARTICLE IN PRESS
neutral hadrons being lost within charged hadron showers. For alljet energies considered, fragments from charged hadrons, whichtend to be relatively low in energy, do not contribute significantlyto the jet energy resolution.
The numbers in Table 5 can be used to obtain an semi-empirical parameterisation of the jet energy resolution:
rms90E
!21!!!E
p " 0:7" 0:004E" 2:1E
100
" #0:3
%
where E is the jet energy in GeV. The four terms in the expression,respectively, represent: the intrinsic calorimetric resolution;imperfect tracking; leakage and confusion. This functional formis shown in Fig. 10. It is worth noting that the predicted jet energyresolutions for 375 and 500GeV jets are in good agreement withthose found for MC events (see Table 3); these data were not usedin the determination of the parameterisation of the jet energyresolution.
For a significant range of the jet energies relevant for the ILC,high granularity PFlow results in a jet energy resolution which isroughly a factor two better than the best achieved at LEP(sE=E! 6:8% at
!!s
p!MZ). The ILC jet energy goal of sE=Eo3:8%
is reached in the jet energy range 40–420GeV.Fig. 10 also shows a parameterisation of the jet energy
resolution #rms90$ obtained from a simple sum of the total
calorimetric energy deposited in the ILD detector concept. Thedegradation in energy resolution for high energy jets is due tonon-containment of hadronic showers. It is worth noting thateven for the highest energies jets considered, PFlow reconstruc-tion significantly improves the resolution compared to the purelycalorimetric approach. The performance of PFlow calorimetry alsois compared to 50%=
!!!!!!!!!!!!!!!E#GeV$
p" 3:0% which is intended to give an
indication of the resolution which might be achieved using atraditional calorimetric approach. This parameterisation effec-tively assumes an infinitely deep HCAL as it does not correctlyaccount for the effect of leakage (which is why it deviatessignificantly from the ILD Calorimetric only curve at highenergies).
8. Dependence on hadron shower modelling
The results of the above studies rely on the accuracy of the MCsimulation in describing EM and hadronic showers. The Geant4MC provides a good description of EM showers as has beendemonstrated in a series of test-beam experiments [27] using aSilicon–Tungsten ECAL of the type assumed for the ILD detector
Table 5The PFlow jet energy resolution obtained with PandoraPFA broken down into contributions from: intrinsic calorimeter resolution, imperfect tracking, leakage andconfusion.
Contribution Jet Energy Resolution rms90#Ej$=Ej
Ej ! 45GeV Ej ! 100GeV Ej ! 180GeV Ej ! 250GeV
Total (%) 3.7 2.9 3.0 3.1Resolution (%) 3.0 2.0 1.6 1.3Tracking (%) 1.2 0.7 0.8 0.8Leakage (%) 0.1 0.5 0.8 1.0Other (%) 0.6 0.5 0.9 1.0Confusion (%) 1.7 1.8 2.1 2.3(i) Confusion (photons) (%) 0.8 1.0 1.1 1.3(ii) Confusion (neutral hadrons) (%) 0.9 1.3 1.7 1.8(iii) Confusion (charged hadrons) (%) 1.2 0.7 0.5 0.2
The different confusion terms correspond to: (i) hits from photons which are lost in charged hadrons; (ii) hits from neutral hadrons that are lost in charged hadron clusters;and (iii) hits from charged hadrons that are reconstructed as a neutral hadron cluster.
EJET/GeV
rms 9
0/Eje
t [%
]
0
1
2
3
4 TotalResolutionConfusion
OtherLeakage
0 50 100 150 200 250
Fig. 9. The contributions to the PFlow jet energy resolution obtained withPandoraPFA as a function of energy. The total is (approximately) the quadraturesum of the components.
Ejet/GeV
rms 9
0/Eje
t [%
]
0
2
4
6
8
10Particle Flow (ILD+PandoraPFA)Particle Flow (confusion term)Calorimeter Only (ILD)
E(GeV) ! 3.0 %50 % /
0 100 200 300 400 500
Fig. 10. The empirical functional form of the jet energy resolution obtained fromPFlow calorimetry (PandoraPFA and the ILD concept). The estimated contributionfrom the confusion term only is shown (dotted). The dot-dashed curve shows aparameterisation of the jet energy resolution obtained from the total calorimetricenergy deposition in the ILD detector. In addition, the dashed curve,50%=
!!!!!!!!!!!!!!!E#GeV$
p" 3:0%, is shown to give an indication of the resolution achievable
using a traditional calorimetric approach.
M.A. Thomson / Nuclear Instruments and Methods in Physics Research A 611 (2009) 25–4034
ARTICLE IN PRESS
neutral hadrons being lost within charged hadron showers. For alljet energies considered, fragments from charged hadrons, whichtend to be relatively low in energy, do not contribute significantlyto the jet energy resolution.
The numbers in Table 5 can be used to obtain an semi-empirical parameterisation of the jet energy resolution:
rms90E
!21!!!E
p " 0:7" 0:004E" 2:1E
100
" #0:3
%
where E is the jet energy in GeV. The four terms in the expression,respectively, represent: the intrinsic calorimetric resolution;imperfect tracking; leakage and confusion. This functional formis shown in Fig. 10. It is worth noting that the predicted jet energyresolutions for 375 and 500GeV jets are in good agreement withthose found for MC events (see Table 3); these data were not usedin the determination of the parameterisation of the jet energyresolution.
For a significant range of the jet energies relevant for the ILC,high granularity PFlow results in a jet energy resolution which isroughly a factor two better than the best achieved at LEP(sE=E! 6:8% at
!!s
p!MZ). The ILC jet energy goal of sE=Eo3:8%
is reached in the jet energy range 40–420GeV.Fig. 10 also shows a parameterisation of the jet energy
resolution #rms90$ obtained from a simple sum of the total
calorimetric energy deposited in the ILD detector concept. Thedegradation in energy resolution for high energy jets is due tonon-containment of hadronic showers. It is worth noting thateven for the highest energies jets considered, PFlow reconstruc-tion significantly improves the resolution compared to the purelycalorimetric approach. The performance of PFlow calorimetry alsois compared to 50%=
!!!!!!!!!!!!!!!E#GeV$
p" 3:0% which is intended to give an
indication of the resolution which might be achieved using atraditional calorimetric approach. This parameterisation effec-tively assumes an infinitely deep HCAL as it does not correctlyaccount for the effect of leakage (which is why it deviatessignificantly from the ILD Calorimetric only curve at highenergies).
8. Dependence on hadron shower modelling
The results of the above studies rely on the accuracy of the MCsimulation in describing EM and hadronic showers. The Geant4MC provides a good description of EM showers as has beendemonstrated in a series of test-beam experiments [27] using aSilicon–Tungsten ECAL of the type assumed for the ILD detector
Table 5The PFlow jet energy resolution obtained with PandoraPFA broken down into contributions from: intrinsic calorimeter resolution, imperfect tracking, leakage andconfusion.
Contribution Jet Energy Resolution rms90#Ej$=Ej
Ej ! 45GeV Ej ! 100GeV Ej ! 180GeV Ej ! 250GeV
Total (%) 3.7 2.9 3.0 3.1Resolution (%) 3.0 2.0 1.6 1.3Tracking (%) 1.2 0.7 0.8 0.8Leakage (%) 0.1 0.5 0.8 1.0Other (%) 0.6 0.5 0.9 1.0Confusion (%) 1.7 1.8 2.1 2.3
(i) Confusion (photons) (%) 0.8 1.0 1.1 1.3(ii) Confusion (neutral hadrons) (%) 0.9 1.3 1.7 1.8(iii) Confusion (charged hadrons) (%) 1.2 0.7 0.5 0.2
The different confusion terms correspond to: (i) hits from photons which are lost in charged hadrons; (ii) hits from neutral hadrons that are lost in charged hadron clusters;and (iii) hits from charged hadrons that are reconstructed as a neutral hadron cluster.
EJET/GeV
rms 9
0/Eje
t [%
]
0
1
2
3
4 TotalResolutionConfusion
OtherLeakage
0 50 100 150 200 250
Fig. 9. The contributions to the PFlow jet energy resolution obtained withPandoraPFA as a function of energy. The total is (approximately) the quadraturesum of the components.
Ejet/GeV
rms 9
0/Eje
t [%
]
0
2
4
6
8
10Particle Flow (ILD+PandoraPFA)Particle Flow (confusion term)Calorimeter Only (ILD)
E(GeV) ! 3.0 %50 % /
0 100 200 300 400 500
Fig. 10. The empirical functional form of the jet energy resolution obtained fromPFlow calorimetry (PandoraPFA and the ILD concept). The estimated contributionfrom the confusion term only is shown (dotted). The dot-dashed curve shows aparameterisation of the jet energy resolution obtained from the total calorimetricenergy deposition in the ILD detector. In addition, the dashed curve,50%=
!!!!!!!!!!!!!!!E#GeV$
p" 3:0%, is shown to give an indication of the resolution achievable
using a traditional calorimetric approach.
M.A. Thomson / Nuclear Instruments and Methods in Physics Research A 611 (2009) 25–4034
Resolution Tracking Leakage Confusion
Friday, May 14, 2010
MC
Particle Flow Calorimetry: Experimental Felix Sefkow CALOR 2010, Beijing, May 10-14, 2010
How to test it experimentally?
• “Jets” from thin targets?– Would require magnet
spectroscopy and large acceptance ECAL + HCAL
• Simulation study
– Multi-million $ experiment– and still inconclusive
• need to control target losses and acceptance losses at 1-2% level
• model dependence
4
20 GeV pion, 0.8 T
• Factorize the problem: check the ingredients– simulation– algorithms– technical performance
Friday, May 14, 2010
MC
Particle Flow Calorimetry: Experimental Felix Sefkow CALOR 2010, Beijing, May 10-14, 2010
Critical questions
• Are the basic detector performance predictions confirmed?
• Are the shower parameters well enough simulated to predict PFLOW?
• Is the substructure actually there and well modeled?• Can one realize the potential of software
compensation for gain and linearity?• Can we verify the "double track resolution" of a
tracking calorimeter?• Are detector effects under control?• Can we calibrate millions of cells and control
stability?• Can we build the detector without spoiling it by dead
material everywhere?• What are the relative merits of different
technologies for PFLOW?
5
The PandoraPFA Algorithm M
ark Thomson
12
! High granularity Pflow
reconstruction is highly non-trivial !
PandoraPFA consists of a many com
plex steps (not all shown)
Clustering
Topological Association
30 GeV
12 GeV
18 G
eV
Iterative Reclustering
9 GeV
9 GeV
6 GeV
Photon ID Fragm
ent ID
Calor 2010, Beijing, 11/5/2010
MT, N
IM 611 (2009) 24-40
Friday, May 14, 2010
MC
Particle Flow Calorimetry: Experimental Felix Sefkow CALOR 2010, Beijing, May 10-14, 2010
Technology tree
6
• mostly ILD, SiD• ILC, CLIC
Friday, May 14, 2010
MC
Particle Flow Calorimetry: Experimental Felix Sefkow CALOR 2010, Beijing, May 10-14, 2010
Overall status
• Major test beam campaigns at DESY, CERN and Fermilab• 1st generation “physics” prototypes• Mostly combined set-ups ECAL-HCAL
• Si W ECAL 2005-08• Scint W ECAL 2007-09• Scint Fe HCAL 2006-09• RPC Fe HCAL to start end 2010
• 2nd generation “technical” prototypes: construction and commissioning ongoing, single or few layers
• Complete detectors to start with RPC-Fe HCAL 2011• ECAL, Scint Fe HCAL later
7
Friday, May 14, 2010
Validation of the simulationsdetector performance
shower models
8
Friday, May 14, 2010
MC
Particle Flow Calorimetry: Experimental Felix Sefkow CALOR 2010, Beijing, May 10-14, 2010
ECAL options
• W Si or Sci: common mechanics, similar electronics
9
!"#$%&'()*$+,- ./010123#! !"#$1#&",4&56$3/010123#$10$#7158647"$1"#$%&'()*$9#574846412:
! %5,-;4&"$71/<&1</#$;41"$7&4,1466510/=71/437$>?$@@A$5,-$1<,B71#,$652#/7$>?:C$@@A:
! 'D1/<-#-$7&4,1466510/$5,-$1"#$E..($5/#$9<662$5-031#-:
! %1/437$5/#$0/1"0B0,56$4,$561#/,51#$652#/7:FG$&@
! %1/437$5/#$0/1"0B0,56$4,$561#/,51#$652#/7:! H+$71/437$D$?I$652#/7$J$+FKI$&"5,,#67:FG$&@
!"#$%&'()#*#(%)+#"#$%,'(! "#$%&'()*%*+),%')%$-+./+*$%*#$%012345%6$(&7(8+91$%&7(%$.$1*(79):! "#$%$.$1*(79;*/9$<%=$+8%179*+'9)%)78$%!; >%"; 179*+8'9+*'79):! 3$($9,7-%17/9*$(%)'?9+.)%#+-$%=$$9%/)$<%&7(%*#$%@4A%*('??$(B#7C$-$(%)*'..%7&&.'9$%$-$9*%)$.$1*'79%')%9$1$))+(D%*7%6/('&D%*#$$.$1*(79%)+86.$:! 2-$9*%)$.$1*'79%')%<79$%=+)$<%79%E; 579?'*/</9+.%>%.+*$(+.%)#7C$(%)#+6$; !; >%"; -$*7%=D%*#$%F345%)'?9+.%.71+*$<%+*%<7C9)*($+8%
!"#$!%&'()*+,+-.(/
; !; >%"; -$*7%=D%*#$%F345%)'?9+.%.71+*$<%+*%<7C9)*($+8%
!"#$%$&
$&
!! "#$%&'(")*+"(,-.-$/")0
! 12-,-3)*45*!! "(64$%3"#63-4$*5"4.*7".-893*2(**#%(5#,*34*-.+"4'(*:(3*($("8)*"(%4,#3-4$;
! <($("/3(*!! 2)*+#33-$8*-"4$*4$*2(/.,-$(
/$=*-$:(63-$8*>?@A7*<(B !@ 2(/.;
! C")*"(64$%3"#63-4$*45*39(*8($("/3(=*!! D-39*E6-$3-,,/34"@FG1H;
!!!!! "#$""""
!! =(3(63-4$*%#66(%%5#,I
Friday, May 14, 2010
MC
Particle Flow Calorimetry: Experimental Felix Sefkow CALOR 2010, Beijing, May 10-14, 2010
Pions in the SiW ECAL
• test Geant 4 predictions with 1 cm2 granularity
• sensitive to shower decomposition• favor recent G4 physics lists• certainly not perfect - certainly not bad
either!
10
Shower Components:
- electrons/positrons knock-on, ionisation, etc.- protons from nuclear fragmentation- mesons- others- sum
Friday, May 14, 2010
MC
Particle Flow Calorimetry: Experimental Felix Sefkow CALOR 2010, Beijing, May 10-14, 2010
Fe Scint tile HCAL
• Present-day simulation quality requires good detector understanding to discriminate
• Fluctuations also well reproduced
11
Frank Simon ([email protected])Particle Showers in a Highly Granular HCALCALOR2010, Beijing, China
Imaging of Hadronic Showers
• Highly granular calorimetry motivated by the Particle Flow Concept:
Separation of particle showers within hadronic jets
2
• Physics prototypes in test beams provide unprecedented 3D information of the
structure of hadronic showers
! Excellent possibilities for the validation and the further development of
hadronic shower models in simulation codes (GEANT4)!
Frank Simon ([email protected])Particle Showers in a Highly Granular HCALCALOR2010, Beijing, China
Shower Start & Shower Profiles
• Identification of the shower start point:
Increase of activity in the detector
7
!"#$$%
&'&&
(&&)&&
*&& +"#$$%(&&&(,&&
)&&&
"-.-/0
!"(
1$
$23
-/-!45"6-/7205"
8*'&
8)'&
8('&
8''&
9:;<9="3!->2$2/?!5
• Full profile of showers with high
resolution:
Perfect for detailed studies and
MC comparisons
10 GeV !-
Frank Simon ([email protected])Particle Showers in a Highly Granular HCALCALOR2010, Beijing, China
Mean Shower Radius
• Mean radius, energy weighted
9
!"#$%"&"'()%*+",-./ 0/ 1/ 2/ 3/ 4/ 5/ 6/
7"#&%89:;"'%'#<=>8%*$$-
3/
33
4/
43
5/
53
6/
63
?/
?3
.//
@ABC@D%E'"F=$=&#') G%@DHI%0//5J!
K#L#
BMDN
O+PNQ!DHR
O+PNQSRSNQ!DHR
!"#$%"&"'()%*+",-
./ 0/ 1/ 2/ 3/ 4/ 5/ 6/
7"#&%89:;"'%'#<=>8%*$$-
3/
33
4/
43
5/
53
6/
63
?/
?3
.//
@ABC@D%E'"F=$=&#') G%@DHI%0//5J!
K#L#
M+N@OM+N@
M+N@O@PCQN
M+N@O!DHR
M+N@O!C@
!"#$%"&"'()%*+",-./ 0/ 1/ 2/ 3/ 4/ 5/ 6/
7"#&%89:;"'%'#<=>8%*$$-
3/
33
4/
43
5/
53
6/
63
?/
?3
.//
@ABC@D%E'"F=$=&#') G%@DHI%0//5J!
K#L#
MNMOP!DHN
MNMOP!DHNPNH,
MNMP!C@
2D profile from starting point
Friday, May 14, 2010
MC
Particle Flow Calorimetry: Experimental Felix Sefkow CALOR 2010, Beijing, May 10-14, 2010
Shower fine structure
• Could have the same global parameters with “clouds” or “trees”• Powerful tool to check models• Surprisingly good agreement already
12
Frank Simon ([email protected])Particle Showers in a Highly Granular HCALCALOR2010, Beijing, China
Digging Deeper: 3D Substructure - Particle Tracks
11
Beam25 GeV !-
ECAL upstream
identified tracks
• Imaging capability of detector
allows the identification of
individual MIP-like tracks
within hadronic showers
Frank Simon ([email protected])Particle Showers in a Highly Granular HCALCALOR2010, Beijing, China
Track Segments in Hadronic Showers: Length
• Track length and slope well described by all models:
• Beam composition well modeled, satisfactory inclusion of detector noise
• High energy cross sections well described
12
) [HCal Layer])!cos(
lengthReal Track Length (
5 10 15 20 25 30 35
MC
/Da
ta
0.8
1
1.2
1.4
1.65 10 15 20 25 30 35
En
trie
s (n
orm
aliz
ed
to
no
. E
ven
ts)
-210
-110
data_rec_v0408-"
LHEP-"
QGS_BIC-"
QGSP_BERT_TRV-"
QGSP_BERT-"
FTF_BIC-"
FTFP_BERT-"
CALICE preliminary
Energy [GeV]10 20 30 40 50 60 70 80
MC
/Da
ta
0.96
0.97
0.98
0.99
1
1.01
1.02 10 20 30 40 50 60 70 80
Me
an
Re
al T
rack
Le
ng
th [
hca
l la
yers
]
10.5
11
11.5
12
12.5
data_rec_v0408-!
LHEP-!
QGS_BIC-!
QGSP_BERT_TRV-!
QGSP_BERT-!
FTF_BIC-!
FTFP_BERT-!
CALICE preliminaryµ component in beam
20 GeV
Frank Simon ([email protected])Particle Showers in a Highly Granular HCALCALOR2010, Beijing, China
Track Distributions: Angles & Multiplicities
• Large discrepancy between different models
• Best agreement with QGSP_BERT
• LHEP, QGS_BIC have too small angles and too small multiplicity: Insufficient
production of high-energy secondaries at large angles
13
]°Track Angle [0 10 20 30 40 50 60
MC
/Da
ta
0.5
0.6
0.7
0.8
0.9
1
1.1
1.20 10 20 30 40 50 60
En
trie
s (n
orm
aliz
ed
to
In
teg
ral =
1)
0
0.05
0.1
0.15
0.2
0.25
0.3
data_rec_v0408-!
LHEP-!
QGS_BIC-!
QGSP_BERT_TRV-!
QGSP_BERT-!
FTF_BIC-!
FTFP_BERT-!
CALICE preliminary
Track Multiplicity / event0 1 2 3 4 5 6 7 8
MC
/Da
ta
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6 0 1 2 3 4 5 6 7 8
En
trie
s (n
orm
aliz
ed
to
In
teg
ral =
1)
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4 data_rec_v0408-!
LHEP-!
QGS_BIC-!
QGSP_BERT_TRV-!
QGSP_BERT-!
FTF_BIC-!
FTFP_BERT-!
CALICE preliminary
20 GeV
Friday, May 14, 2010
MC
Particle Flow Calorimetry: Experimental Felix Sefkow CALOR 2010, Beijing, May 10-14, 2010
Summary on validation:
• The particle flow detectors perform as expected– support predictions for full-scale detector
• Geant 4 simulations not perfect, but also not as far off as feared a few years ago– fruitful close cooperation with model builders ongoing
• Predicted shower sub-structure is seen– detailed checks possible, benefits for all calorimeters
13
Friday, May 14, 2010
Test the algorithms with real data
14
Friday, May 14, 2010
MC
Particle Flow Calorimetry: Experimental Felix Sefkow CALOR 2010, Beijing, May 10-14, 2010
Software compensation
• Poor man’s dream• Significantly improved resolution AND linearity• High granularity - many possibilities
15
beam Energy [GeV]
0 10 20 30 40 50 60 70 80 90
E/E
!
0
0.05
0.1
0.15
0.2
0.25
c GeV/E " b " EFit: a/
0.353 [GeV]±0.73% c = 0.172±0.4% b = 0.00±a = 64.3
0.055 [GeV]±0.39% c = 0.966±1.0% b = 1.97±a = 45.3
test beam data:
constant cluster weight
neural network
Energy resolution - FTF_BIC trainingCALICE Preliminary
Frank Simon ([email protected])Particle Showers in a Highly Granular HCALCALOR2010, Beijing, China
Software Compensation: Global Method
• Cluster finding in HCAL and TCMT to determine properties of the shower:
total energy, volume, longitudinal structure...
! Used as input for a neural net, training of the NN with simulations (quasi-
continuous energy)
! No prior knowledge of the beam energy needed for application of method
16
HCAL+TCMTNN trained with FTF_BIC
Resolution improved by ~25%(~15% at 10 GeV, ~20% at 15 GeV)
Resolution given by Gaussian sigma / mean of a fit to the distribution within 1.5" of peak
!"#$%&'()*&+,-./0&12345
6 766 866 966 :66 ;666 ;766
!"#$%&'()*&+,-./0&<,&(-%%=&>&8?@&234AB-%%&12345
6
766
866
966
:66
;666
;766
+C>#D.-=D"%EFA+C#"#$%F&G&;
+C>#D.-=
D"%EFA+C#
"#$%F&G&6?
H
!"#
$%&'$()*+,-./.0"+176&I-J&K<",=L&,"&M-</D#<,/
!"#$%&'()*&+,-./0&12345
6 766 866 966 :66 ;666 ;766
!"#$%&'()*&+,-./0&<,&(-%%=&>&8?@&234AB-%%&12345
6
766
866
966
:66
;666
;766
+C>#D.-=D"%EFA+C#"#$%F&G&;
+C>#D.-=
D"%EFA+C#
"#$%F&G&6?
H
!"#
$%&'$()*+,-./.01+276&I-J&K<",=L&-,-./0&E-,=<#0&M-</D#<,/
Frank Simon ([email protected])Particle Showers in a Highly Granular HCALCALOR2010, Beijing, China
Energy Reconstruction & Software Compensation
• The CALICE HCAL is non-compensating: e/! ~ 1.3 (energy dependent)
• High granularity provides detailed information for software compensation:
• Electromagnetic energy deposits tend to be denser than hadronic ones
" Improvement studied on the cell (local) and on the cluster (global) level
14
Local method: apply weight to cells according to their energy, lower weight for cells with
higher energy content, weights are determined with a minimization technique
weighting
10 20 30 40 50 60 70 80 90
reco
nst
ruct
ed
En
erg
y [G
eV
]
10
20
30
40
50
60
70
80
90
CALICE Preliminary
test beam data
constant cluster weight
neural network
Linearity of detector response - FTF_BIC training
beam Energy [GeV]10 20 30 40 50 60 70 80 90
beam
)/E
beam
- E
rec
(E -0.1
0
0.1
Frank Simon ([email protected])Particle Showers in a Highly Granular HCALCALOR2010, Beijing, China
Software Compensation: Linearity of Response
• Unweighted reconstruction
shows typical non-linear
behavior: Increased response at
high energies
• Software compensation
recovers linearity within < 2%
from 10 to 80 GeV
17
Friday, May 14, 2010
MC
Particle Flow Calorimetry: Experimental Felix Sefkow CALOR 2010, Beijing, May 10-14, 2010
Two-particle separation
• The “double-track resolution” of an imaging calorimeter • Small occupancy: use of event mixing technique possible• Important: agreement data - simulation
– sharing the same limitations
16
!"
!!!!!!!!!!!!!!"#$%&'()(*+*,-$'.$&#/'0#&*12$,"#$$34$5#6$1#7,&)+$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$$")8&'1$#1#&2-$9*,"*1$:$;+#.,<$)18$=$;&*2",<$>,)18)&8$8#0*),*'1>$.&'?$*,>$&#)+$#1#&2-$0#&>7>$,"#$8*>,)1/#$.&'?$,"#$/")&2#8$34$5#6$;/'1,*17'7>$+*1#<$)18$=4$5#6$;8)>"#8$+*1#<$%*'1$.'&$,#>,$(#)?$8),)$;<$)18$.'&$(',"$@ABC$;(+7#<$)18$D5ECFGBH!$;2&##1<$%"->*/>$+*>,>$
!"#$!%&&'()*+,+-.(/ !"#$!%&&'()*+,+-.(/
"
to be done with photons, too
Friday, May 14, 2010
MC
Particle Flow Calorimetry: Experimental Felix Sefkow CALOR 2010, Beijing, May 10-14, 2010
Leakage estimation
• Infer leakage from seen part of shower topology and energy• multivariate techniqes; striking potential• implications for detector optimization: implement in Pandora
17
!
!"#$%&'()*+&,-
!!"#!$%&!!'#!$%&!!(#!$%&!!)#!$%&!!*#!$%&
./0-.1(-231(.4-(%.5.5/%)6(-+.7489+).:*/%*;,1<(/-=>$%&'()*+&,3(7-
!
!""#$%&'$()*+*,
!"#$%&$'($)&**()%(+,$-.*/01-
2)%(330%'04 5(330%'04
6&7864089:406;8036<6<&3%=08$6";>#$%6)(%(33-.:0"8*&)4*%(33-.:0"8
Simulation, to be confirmed
ongoing thesis work I.Marchesini
Friday, May 14, 2010
MC
Particle Flow Calorimetry: Experimental Felix Sefkow CALOR 2010, Beijing, May 10-14, 2010
Summary on algorithms
• Granularity is extremely powerful
• Energy resolution and imaging capabilities verified with data at sub-structure level– the main drivers of PFLOW
performance
• Leakage estimation and software compensation not yet implemented in present Pandora
18
ARTICLE IN PRESS
neutral hadrons being lost within charged hadron showers. For alljet energies considered, fragments from charged hadrons, whichtend to be relatively low in energy, do not contribute significantlyto the jet energy resolution.
The numbers in Table 5 can be used to obtain an semi-empirical parameterisation of the jet energy resolution:
rms90E
!21!!!E
p " 0:7" 0:004E" 2:1E
100
" #0:3
%
where E is the jet energy in GeV. The four terms in the expression,respectively, represent: the intrinsic calorimetric resolution;imperfect tracking; leakage and confusion. This functional formis shown in Fig. 10. It is worth noting that the predicted jet energyresolutions for 375 and 500GeV jets are in good agreement withthose found for MC events (see Table 3); these data were not usedin the determination of the parameterisation of the jet energyresolution.
For a significant range of the jet energies relevant for the ILC,high granularity PFlow results in a jet energy resolution which isroughly a factor two better than the best achieved at LEP(sE=E! 6:8% at
!!s
p!MZ). The ILC jet energy goal of sE=Eo3:8%
is reached in the jet energy range 40–420GeV.Fig. 10 also shows a parameterisation of the jet energy
resolution #rms90$ obtained from a simple sum of the total
calorimetric energy deposited in the ILD detector concept. Thedegradation in energy resolution for high energy jets is due tonon-containment of hadronic showers. It is worth noting thateven for the highest energies jets considered, PFlow reconstruc-tion significantly improves the resolution compared to the purelycalorimetric approach. The performance of PFlow calorimetry alsois compared to 50%=
!!!!!!!!!!!!!!!E#GeV$
p" 3:0% which is intended to give an
indication of the resolution which might be achieved using atraditional calorimetric approach. This parameterisation effec-tively assumes an infinitely deep HCAL as it does not correctlyaccount for the effect of leakage (which is why it deviatessignificantly from the ILD Calorimetric only curve at highenergies).
8. Dependence on hadron shower modelling
The results of the above studies rely on the accuracy of the MCsimulation in describing EM and hadronic showers. The Geant4MC provides a good description of EM showers as has beendemonstrated in a series of test-beam experiments [27] using aSilicon–Tungsten ECAL of the type assumed for the ILD detector
Table 5The PFlow jet energy resolution obtained with PandoraPFA broken down into contributions from: intrinsic calorimeter resolution, imperfect tracking, leakage andconfusion.
Contribution Jet Energy Resolution rms90#Ej$=Ej
Ej ! 45GeV Ej ! 100GeV Ej ! 180GeV Ej ! 250GeV
Total (%) 3.7 2.9 3.0 3.1Resolution (%) 3.0 2.0 1.6 1.3Tracking (%) 1.2 0.7 0.8 0.8Leakage (%) 0.1 0.5 0.8 1.0Other (%) 0.6 0.5 0.9 1.0Confusion (%) 1.7 1.8 2.1 2.3
(i) Confusion (photons) (%) 0.8 1.0 1.1 1.3(ii) Confusion (neutral hadrons) (%) 0.9 1.3 1.7 1.8(iii) Confusion (charged hadrons) (%) 1.2 0.7 0.5 0.2
The different confusion terms correspond to: (i) hits from photons which are lost in charged hadrons; (ii) hits from neutral hadrons that are lost in charged hadron clusters;and (iii) hits from charged hadrons that are reconstructed as a neutral hadron cluster.
EJET/GeV
rms 9
0/Eje
t [%
]
0
1
2
3
4 TotalResolutionConfusion
OtherLeakage
0 50 100 150 200 250
Fig. 9. The contributions to the PFlow jet energy resolution obtained withPandoraPFA as a function of energy. The total is (approximately) the quadraturesum of the components.
Ejet/GeVrm
s 90/E
jet [
%]
0
2
4
6
8
10Particle Flow (ILD+PandoraPFA)Particle Flow (confusion term)Calorimeter Only (ILD)
E(GeV) ! 3.0 %50 % /
0 100 200 300 400 500
Fig. 10. The empirical functional form of the jet energy resolution obtained fromPFlow calorimetry (PandoraPFA and the ILD concept). The estimated contributionfrom the confusion term only is shown (dotted). The dot-dashed curve shows aparameterisation of the jet energy resolution obtained from the total calorimetricenergy deposition in the ILD detector. In addition, the dashed curve,50%=
!!!!!!!!!!!!!!!E#GeV$
p" 3:0%, is shown to give an indication of the resolution achievable
using a traditional calorimetric approach.
M.A. Thomson / Nuclear Instruments and Methods in Physics Research A 611 (2009) 25–4034
Friday, May 14, 2010
Test the technologies and establish feasibility
19
Friday, May 14, 2010
MC
Particle Flow Calorimetry: Experimental Felix Sefkow CALOR 2010, Beijing, May 10-14, 2010
Calibration
• Study triggered by review of LC detector LOI• Can you calibrate millions of channels and maintain stability?
– not really a worry for Si, but could be an issue for scintillator
• 1. Simulate impact of statistic (uncorrelated) and systematic (correlated) calibration errors, find ∫L for in-situ calibration– PFLOW performance VERY robust w.r.t. channel-to-channel variations;
coherent effects easy to control
20
• 2. Exercise in-situ methods (SiPM auto-calib, track segments) with test beam data from CERN and FNAL - transport calibration
across the ocean and restore performance
Friday, May 14, 2010
MC
Particle Flow Calorimetry: Experimental Felix Sefkow CALOR 2010, Beijing, May 10-14, 2010
Integration
• Sensor technology, precision mechanics
• Next: system engineering
• Industrialized ASIC development using common building blocks
• New operational challenges– power pulsing– on-detector zero suppression– real-time threshold monitoring– time measurement
21
spin-off
SDHCAL Scint E/HCAL
Si ECAL
Friday, May 14, 2010
MC
Particle Flow Calorimetry: Experimental Felix Sefkow CALOR 2010, Beijing, May 10-14, 2010
Digital calorimetry
• MAPS DECAL: Tera-pixel– 1st sensor tests in e
showers
• Digital and semi-digital hadron calorimeter– even higher granularity– suppress dE/dx fluct.– reduced n sensitivity– limited at high E?
• Small RPC proto successful• Full-size RPC based
prototypes underway
• Promising tests of GEM and MicroMEGAS based read-out modules
22
Frank Simon ([email protected])CALICE
68. DESY PRC
Pushing the Limits of Granularity: A Digital ECAL
• Extreme resolution needed to resolve every single particle within an
electromagnetic shower: Densities of up to 100 particles / mm2 expected
! Readout granularity of 50 x 50 µm2 required
28
go from here...
to there...
Frank Simon ([email protected])CALICE
68. DESY PRC
DECAL Technology
• A challenging idea: A complete ILC detector would be a Tera-pixel Calorimeter!
• The technology: MAPS - A standard CMOS product
• Potentially significant price advantage over high resistivity Si diodes
• Tests of sensor prototypes at CERN in 2009: 8.4 x 8.4 mm2 sensitive area
29
!"#$%&'(%$)#&"*'+,-./010 2
!"#$%&'(#)%"
! !"#$%&'()*+*+,(-
" *)")+,-./01+
! ./$0'1'%2342'567'89:;'*<
'='4>?'4-@@2
! %2340'A1'$B566-@2
! C)3DD-)'E';'2436+3@@5+*)2
" 2+*3+4
! =11'%2342
! !5>-'+)3DD-)36D
'2,2+->
17
Square meter plane with Readout boards
Frank Simon ([email protected])CALICE
68. DESY PRC
A Digital HCAL Physics Prototype
17
• The concept: Active layers of glass RPCs
with 1 cm2 pads, one bit readout per channel
• Proof of principle measurement at Fermilab:
• small prototype: 20 x 20 cm2 active area,
8 layers (6 read out)
• 1.2 X0 Steel/Cu absorber per layer
positron shower in the prototype
Friday, May 14, 2010
MC
Particle Flow Calorimetry: Experimental Felix Sefkow CALOR 2010, Beijing, May 10-14, 2010
High energy
• Particle flow also a promising option for CLIC energies
• Leakage expected to limit PFLOW performance– need 1 λ ECAL + 7 λ HCAL
• Tungsten absorber cost-competitive with larger coil - and less risky
• Test beam validation with scintillator and gas detetctors
• More neutrons:– different model systematics– timing measurements
23
81
A
A!"#$ %$&$'()#$*+
•!"#$ "% &'&()* +&,"#-$&+&.(/*
• 0/&.( +-%%&(&'1& 2&$3&&'#$&&/ .'+ $4')#$&' #&$4,#
05 6(&%&
7895 :,&1;<.*&(= >0?:7@A@= B&-C-')DE78E7@A@
Friday, May 14, 2010
MC
Particle Flow Calorimetry: Experimental Felix Sefkow CALOR 2010, Beijing, May 10-14, 2010
Summary on technologies
• a leap in several orders of magnitude in channel count
• new sensor technologies, new integration concepts– the latter is part of the feasibility demonstration
• progress towards realism:– realistic designs– realistic simulations – realistic cost– realistic proposal
• Digital calorimetry ready for exploration
24
Friday, May 14, 2010
MC
Particle Flow Calorimetry: Experimental Felix Sefkow CALOR 2010, Beijing, May 10-14, 2010
Conclusion
• Particle flow calorimetry does not solve the inherent problems of hadron calorimeters
• But it holds the promise of providing a highly performant work-around
• Substantiated by test beam data• Can be built
25
Friday, May 14, 2010