γγ interac*ons at KLOE Ivan Prado Longhi on behalf of the KLOE/KLOE2 Collabora*on Università “Roma Tre” and INFN Sezione ”Roma Tre” MesonNet2013 Prague, 17062013
γγ interac*ons at KLOE
Ivan Prado Longhi on behalf of the KLOE/KLOE-‐2 Collabora*on Università “Roma Tre” and INFN -‐ Sezione ”Roma Tre”
MesonNet2013 -‐ Prague, 17-‐06-‐2013
2 Ivan Prado Longhi@Mesonnet2013
Electromagne*c calorimeter
• lead/scin*lla*ng fibers • 98% solid angle coverage • σE/E = 5.7%/√(E(GeV)) • σt=57 ps/√(E(GeV)) + 100 ps • PID capabili*es
Dria Chamber
• Gas mixture 90% He + 10% C4H10 • δpT/pT < 0.4% (450<ϑ<1350) • σxy≈150 μm; σz ≈ 2 mm
MAGNETIC FIELD 0.52 T
KLOE Detector
e+e- → e+e-γ*γ* → e+e-X
X �
X: JPC = 0±+ �
( π0, η, η’, σ(500) )
} �
γγ physics @ KLOE
La Thuile 26/2/2013 S.Miscetti@LaThuile 2013
!! physics @ KLOE: "#(!!) measurement
20
KLOE: no e± tagging ! s = 1 GeV
KLOE-2: tagger to reduce background from $ and to close kinematics !s = M$
! !!!# (!0!0) studied (under-way) at KLOE
! Data sample: 240 pb%1 @ "s = 1 GeV (reduced bckg contamination from $)
! Selected channels: #"&+&%&0/&0&0&0'
! Main background: $!#! with undetected recoil photon
Analyses on L=242.5 pb-‐1 off-‐peak data (collected at √s=1 GeV): • e+e-‐ → e+e-‐η • e+e-‐ → e+e-‐π0π0 (in progress)
3 Ivan Prado Longhi@Mesonnet2013
Mo*va*ons: Γ(X→γγ), meson structure, search for σ(500)
e+e-‐ → e+e-‐ η
4 Ivan Prado Longhi@Mesonnet2013
e+e-‐ → e+e-‐ η
5 Ivan Prado Longhi@Mesonnet2013
Two channels:
η → π0 π0 π0 η → π+ π-‐ π0
• 6γ only with E>15 MeV, 230<ϑ<1570, |t-‐r/c|<3σt • no tracks in the dria chamber • γγ pairing
• kinema*c fit requiring M6γ= mη
• 2γ only with E>15 MeV, 230<ϑ<1570, |t-‐r/c|<3σt • 2 tracks with opposite charge from a cylinder ρPCA < 8 cm, |zPCA| < 8 cm, ρfirst-‐hit < 50 cm • γγ pairing to renconstruct π0 • electron-‐pion likelihood cut • kinema*c fit requiring Mππγγ= mη
!"" "#$%&%'("""""""""""""""""""""""""""""""""""""""""""""")*"#$%&")+,"-" )*")+,""!"".%',/$)%0"1%)"&,23%&%'(")+,"%'4$&%$')" "/$55")*"6,",23$7,")*")+," "/$55"
!"#
!-"$'8"*'79"-"#+*)*'5":%)+";"<=>"?,@A"BC*D D=>EFA"G)! )""HI")&$0J5"%'")+,"8&%1)"0+$/6,&"
, , " ", , " A" $'$795%5"5)&$),(9""
32
2
)(0
p a irM
MMp a ir
"! !#!$%! "! !#!"&!
B" "K8$)$L"
B" "K?M"5%('$7L"
B" "K8$)$L"
B" "K?M"5%('$7L"
Main background: e+e-‐ → ηγ with γ at small angles, outside the acceptance
!"#
! ! " "! ! " #" $%"&'(""
!"#$%&'()#!"# ##
!$%%#&'()*#+),)-#../#0'1()*-#%2/# ##0,),'0,'3)*#45565# #78%/###
0'1()*# #9#:!;8%"# #"8!2</###=+6	#$$7"=$>.%#
")! ! " "! ! # " " *+,"-!./"+"),0123"4"0125*(6("/"78"
#9:"");!./" ;$<'**"")-!.$/"
e+e-‐ → e+e-‐ η: 2D fits
6 Ivan Prado Longhi@Mesonnet2013
η → π0 π0 π0
Variables: PL and M2miss
!"#
! ! " "! ! " #" $%"&'(""
!"#$%&'()# !" #!"# ## # !"#*+#*,!"##*-*.# !"#/-/.
$%$&#'()*+#,*-*.#!"/#0(1)*+#0-*-(0-(2*+#34454# #%67/###7$/# .#$&/# .#!8/#9 9 .#!!/#9:9;.#</#3 3
0(1)*+# #=#>$&6%7# #&6$?@/###A,5'#=#$B%&A$B%7#
!""#$%&#'())#)*+),-#'.(#
")! ! " "! ! # " " *+,"-!./"+")0123"4"5160*(7("/"89"
#:;"")<!./" <$='**"")-!.$/"
e+e-‐ → e+e-‐ η: 2D fits
7 Ivan Prado Longhi@Mesonnet2013
η → π+ π-‐ π0
Variables: PT and M2miss
e+e-‐ → e+e-‐ η: results
σ(e+e-‐ → e+e-‐ η, √s = 1 GeV) = (32.0 ± 1.5stat ± 0.9syst ± 0.2BR(η → 3π)) pb
σ(e+e-‐ → e+e-‐ η, √s = 1 GeV) = (34.5 ± 2.5stat ± 1.0syst ± 0.7FF ± 0.4BR) pb
σ(e+e-‐ → e+e-‐ η, √s = 1 GeV) = (32.7 ± 1.3stat ± 0.7syst) pb
Neutral channel, ≈ 720 signal events:
Charged channel, ≈ 390 signal events:
Combined
Γ(η→γγ) extracted (see reference for details on luminosity func*on and FF parametriza*on):
Γ(η → γγ) = (520± 20stat ± 13syst) eV
8 Ivan Prado Longhi@Mesonnet2013
Most precise measurement, In agreement with PDG value
(510±26) eV
!"#
! ! " " #"$!%&'(""
")%!*"+%"+",-.%($+/.("/."(0!"1/("1-$!! ! " "! ! "! !
"" "2! ! " " 3",+."4!"
/.*!5!.*!.('6"*!$/7!*"+%"+"46"5$-*&,("-1"(0!"8+/."+.+'6%/%"/."(0!",+%!"-1"! ! " "! ! "! ! 9"6/!'*/.:" "2! ! " " 9 %;<"=!>3";"2?@A"B"C@%(+("B"@%6%("B"DEF"3"549"/."+:$!!8!.("G/(0"(0!"7+'&!"-4(+/.!*"/."(0!"*!*/,+(!*"+.+'6%/%H""
"2! ! " " 9 %;<"=!>3";"2?@D"B"?%(+("B"<C%6%("B"<<EF"3"54"$%&'#()(*+,%*&-(#./0,# #-/0((#(+-*&0'#
%2I!>3"
2.43"
!"#
! ! " " #"$!%&'(""
")%!*"+%"+",-.%($+/.("/."(0!"1/("1-$!! ! " "! ! "! !
"" "2! ! " " 3",+."4!"
/.*!5!.*!.('6"*!$/7!*"+%"+"46"5$-*&,("-1"(0!"8+/."+.+'6%/%"/."(0!",+%!"-1"! ! " "! ! "! ! 9"6/!'*/.:" "2! ! " " 9 %;<"=!>3";"2?@A"B"C@%(+("B"@%6%("B"DEF"3"549"/."+:$!!8!.("G/(0"(0!"7+'&!"-4(+/.!*"/."(0!"*!*/,+(!*"+.+'6%/%H""
"2! ! " " 9 %;<"=!>3";"2?@D"B"?%(+("B"<C%6%("B"<<EF"3"54"$%&'#()(*+,%*&-(#./0,# #-/0((#(+-*&0'#
%2I!>3"
2.43"
e+e- → ηγ @ √s = 1 GeV
9 Ivan Prado Longhi@Mesonnet2013 !"#
! ! " " #"$!%&'(""
")%!*"+%"+",-.%($+/.("/."(0!"1/("1-$!! ! " "! ! "! !
"" "2! ! " " 3",+."4!"
/.*!5!.*!.('6"*!$/7!*"+%"+"46"5$-*&,("-1"(0!"8+/."+.+'6%/%"/."(0!",+%!"-1"! ! " "! ! "! ! 9"6/!'*/.:" "2! ! " " 9 %;<"=!>3";"2?@A"B"C@%(+("B"@%6%("B"DEF"3"549"/."+:$!!8!.("G/(0"(0!"7+'&!"-4(+/.!*"/."(0!"*!*/,+(!*"+.+'6%/%H""
"2! ! " " 9 %;<"=!>3";"2?@D"B"?%(+("B"<C%6%("B"<<EF"3"54"$%&'#()(*+,%*&-(#./0,# #-/0((#(+-*&0'#
%2I!>3"
2.43"
Studied as main background for e+e-‐ → e+e-‐ η
σ(e+e-‐ → ηγ )
η→ π+ π-‐ π0, SND
η→ π0 π0 π0, SND
KLOE result
e+e-‐ → e+e-‐ π0π0
10 Ivan Prado Longhi@Mesonnet2013
Ivan Prado Longhi@Mesonnet2013 11
e+e-‐ → e+e-‐ π0π0 (work in progress)
Cut-‐based and mul*variate analysis
(possible produc*on of σ(500) as a resonant intermediate state)
ANALYSIS CUTS
To reject annihila*on processes e+ e-‐→ KSKL, ηγ, ωπ0, f0γ, a0γ, γγ
Signal e+e-‐ → e+e-‐ σ → e+e-‐ π0π0: complete matrix element, full phase space [Eur. Phys. J. C47 (2006) 65]
Selected sample contaminated by low-‐invariant mass, high-‐pt, unbalanced pL background (machine bkg)
MULTIVARIATE ANALYSIS
using TMVA package -‐> cut on the MVA output
e+ e-‐ annihila*on processes normalized according to Xsec*ons
machine background selected from data by topological criteria
track veto, 4γ only, no late-‐clusters (to reject KSKL events)
χ2ππ =
(mπ0 −mij)2
σ2ij
+(mπ0 −mkl)2
σ2kl
12
e+e-‐ → e+e-‐ π0π0: analysis cuts
π0π0 selec*on cu�ng on
1
10
10
10
M!! (MeV)
M!!
(M
eV)
1
10
10
10
M!! (MeV)
M!!
(M
eV)
0
25
50
75
100
125
150
175
200
225
250
0 25 50 75 100 125 150 175 200 225 250
0
25
50
75
100
125
150
175
200
225
250
0 25 50 75 100 125 150 175 200 225 250
χ2ππ < 4
mγ1γ2 vs mγ3γ4
ID 1000000ENTRIES 2854 0.00 0.00 0.00 0.00 0.285E+04 0.00 0.00 0.00 0.00
m4 (MeV)
p T (MeV
)
IDEntriesMeanRMSUDFLWOVFLWALLCHAN
1000000 2854
-24.11 106.6 0.000 0.000 2854.
pL (MeV)
0
50
100
150
200
250
300
350
400
450
500
200 300 400 500 600 700 800 900
0
20
40
60
80
100
120
-400 -200 0 200 400
Low m4γ, high pT
unbalanced pL
“DATA DRIVEN” machine background:
Selected studying: • d = displacement of the reconstructed photons’ origin from the IP • |t-‐r/c| of the t0 cluster
Ivan Prado Longhi@Mesonnet2013
ID 1000000ENTRIES 9788 0.00 0.00 0.00 0.00 0.979E+04 0.00 0.00 0.00 0.00
ABS(t-r/c)t0
sqrt(
d 12 +d 22 )
0
10
20
30
40
50
60
70
80
90
100
-0.2 0 0.2 0.4 0.6 0.8 1 1.2 1.4
A B
C S
“DATA DRIVEN” BKG: Normaliza*on
A B C S
Data 1418 1611 1375 6213
a b c S
KsKL 85 8 24 868
ηγ 30 3 16 345
ωπ0 27 2 23 754
f0γ -‐ -‐ -‐ 58
γγ -‐ -‐ -‐ 45
a0γ -‐ -‐ -‐ 9
# bkg events in A + B + C≈ 1418+1611+1375 = 4404 with contamina*on ≈ 218 Bkg(S) = (A-‐a)*(C-‐c)/(B-‐b)= 1048 (assuming uniformely distributed bkg in the |d| vs |t-‐r/c|t0 plane)
13
33 cm
0.5 ns
|d| vs |t – r/c|t0
Ivan Prado Longhi@Mesonnet2013
14
data
EntriesMeanRMSALLCHAN
8398 392.8 114.1 8398.
M4 (MeV)
dN/d
M4
(25
MeV
)-1
0
50
100
150
200
250
300
350
400
450
200 300 400 500 600 700 800 900
ndata= 8398
A fit to m4γ spectrum is performed using f=n/ndata as star*ng values
KSKL
ωπ0
ηγ
γγ→ π0π0
Bkg (data driven)
ε σ (nb) n = Lσε
KSKL 0.003 1.28 985
ηγ 0.006 0.284 393
ωπ0 0.006 0.550 806
f0γ, a0γ, γγ
≈ 0 -‐ 120
AFTER SELECTIONS:
e+e-‐ → e+e-‐ π0π0: cut based analysis
Ivan Prado Longhi@Mesonnet2013
Bkg (data driven) 3267
γγ→ π0π0 2830
15
MULTIVARIATE ANALYSIS WITH TMVA
• “Data driven” machine background normalized by extrapola*on • MC Distribu*ons normalized to fit results
Training
Data to be analyzed: 8398 events
Applica*on
Applica*on is performed on each MC and data-‐driven bkg sample in order to evaluate the efficiencies of cu�ng on
the MVA output
Longitudinal momentum [MeV]-400 -200 0 200 400
18.9
MeV
/ (1
/N) d
N
00.001
0.002
0.0030.004
0.005
0.0060.007
0.008
0.009Signal
Background
Longitudinal momentum [MeV]-400 -200 0 200 400
18.9
MeV
/ (1
/N) d
N
00.001
0.002
0.0030.004
0.005
0.0060.007
0.008
0.009
U/O
-flo
w (S
,B):
(0.0
, 0.0
)% /
(0.0
, 0.0
)%
Input variable: Longitudinal momentum
Transverse momentum [MeV]50 100 150 200 250 300 350 400
6.93
MeV
/ (1
/N) d
N
0
0.002
0.004
0.006
0.008
0.01
0.012
0.014
0.016
Transverse momentum [MeV]50 100 150 200 250 300 350 400
6.93
MeV
/ (1
/N) d
N
0
0.002
0.004
0.006
0.008
0.01
0.012
0.014
0.016
U/O
-flo
w (S
,B):
(0.0
, 0.0
)% /
(0.1
, 0.0
)%Input variable: Transverse momentum
t-r/c chi^2 [F]2 4 6 8 10 12
0.20
5 F
/ (1
/N) d
N
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
t-r/c chi^2 [F]2 4 6 8 10 12
0.20
5 F
/ (1
/N) d
N
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
U/O
-flo
w (S
,B):
(0.0
, 0.0
)% /
(0.0
, 10.
1)%
Input variable: t-r/c chi^2
Gaussian smearing σ = 40 MeV on px and py for γγ→ π0π0 MC
χ2t =
1
4
4�
i
�ti − ri/c
σt,i
�2PL PT
Likelihood output0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Effic
ienc
y (P
urity
)
0
0.2
0.4
0.6
0.8
1
Likelihood output0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Effic
ienc
y (P
urity
)
0
0.2
0.4
0.6
0.8
1
Signal efficiencyBackground efficiency
Signal puritySignal efficiency*purity
S+BS /
For 4886 signal and 3512 background isS+Bevents the maximum S /
65.3767 when cutting at 0.2892
Cut efficiencies and optimal cut value
Sign
ifica
nce
0
10
20
30
40
50
60
70
Ivan Prado Longhi@Mesonnet2013
M4 (MeV)
dN/d
M4
(10
MeV
)-1
M4 (MeV)
dN/d
M4
(10
MeV
) -1
0
25
50
75
100
125
150
175
200
225
200 300 400 500 600 700 800 900
0
20
40
60
80
100
120
140
160
200 300 400 500 600 700 800 900
16
KSKL
ωπ0
ηγ
γγ→ π0π0 MC
Bkg (data driven) ε σ (nb) n = Lσε
KSKL 0.003 1.28 953
ηγ 0.005 0.284 374
ωπ0 0.006 0.550 774
f0γ, a0γ, γγ
≈ 0 -‐ 110
Cu�ng on likelihood > 0.3
Subtracted spectrum
M4γ spectra
Likelihood response0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
dx / (1
/N) d
N
0
10
20
30
40
50
60 SignalBackground
Likelihood response0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
dx / (1
/N) d
N
0
10
20
30
40
50
60
U/O
-flow
(S,B
): (0
.0, 0
.0)%
/ (0
.0, 0
.0)%
TMVA response for classifier: Likelihood
5324 events
2625 events
Ivan Prado Longhi@Mesonnet2013
Bkg (data driven) 492
γγ→ π0π0 2625
Ivan Prado Longhi@Mesonnet2013 17
SUMMARY
Off-‐peak ( √s = 1 GeV) data collected by KLOE are precious to study meson produc*on in γγ interac*ons
2 Analyses:
e+e-‐ → e+e-‐ η, published [JHEP01(2013)119] • Studied in both η→π+π-‐π0 and η→π0π0π0 channels • σ(e+e-‐ → e+e-‐ η, √s = 1 GeV) evaluated, Γ(η→γγ) extracted (most precise measurement)
e+e-‐ → e+e-‐ π0π0, work in progress • excess of events just above π0π0 threshold • contamina*on from machine background handled using a mul*variate analysis approach • promising ≈ 2600 γγ→ π0π0 candidate events sample
Ivan Prado Longhi@Mesonnet2013 18
SPARES
d1 d2
χ2d1,d2
=2�
i
(r�i(d1)/c− ti)2
σ2t,i
+
2�
i
�r�j(d2)/c− tj
�2
σ2t,j
r�i,j =�
r2i,j + d21,2 + 2d1,2zi,j
r1 r2
r’2 r’1
1 2
3 4
r3 r4
r’4 r’3
MINIMIZE:
d1 ≠ 0 or d2 ≠ 0 could be a signature for machine bkg events
19
e+e-‐ → e+e-‐ π0π0: selec*ng bkg
Ivan Prado Longhi@Mesonnet2013
IDEntriesMeanRMSUDFLWOVFLWALLCHAN
6000 10042
26.47 23.05 0.000 0.000
0.1004E+05
sqrt(d12+d2
2)
IDEntriesMeanRMSUDFLWOVFLWALLCHAN
6000 33794
15.65 13.03 0.000 0.000
0.3379E+05
sqrt(d12+d2
2)
0
100
200
300
400
500
600
700
0 20 40 60 80 100 120 140 160 180 200
0
500
1000
1500
2000
2500
3000
3500
4000
0 20 40 60 80 100 120 140 160 180 200
Data
MC ηγ
|d| =�d21 + d22
CONSIDER:
20
e+e-‐ → e+e-‐ π0π0: selec*ng bkg
ID 1000000ENTRIES 2854 0.00 0.00 0.00 0.00 0.285E+04 0.00 0.00 0.00 0.00
m4 (MeV)
p T (M
eV)
IDEntriesMeanRMSUDFLWOVFLWALLCHAN
1000000 2854
-24.11 106.6 0.000 0.000 2854.
pL (MeV)
0
50
100
150
200
250
300
350
400
450
500
200 300 400 500 600 700 800 900
0
20
40
60
80
100
120
-400 -200 0 200 400
PT VS m4γ
CUTTING |d|>33 cm
Low m4γ, high pT
unbalanced pL
Ivan Prado Longhi@Mesonnet2013
Alterna*ve bkg selec*on variable
IDEntriesMeanRMSUDFLWOVFLWALLCHAN
7000 10042
-0.1382 0.4596 0.000 0.000
0.1004E+05
(t-r/c)t0
IDEntriesMeanRMSUDFLWOVFLWALLCHAN
7000 33794
-0.1051 0.2263 0.000 0.000
0.3379E+05
(t-r/c)t0
0
100
200
300
400
500
600
-1.5 -1 -0.5 0 0.5 1 1.5
0
500
1000
1500
2000
2500
3000
-1.5 -1 -0.5 0 0.5 1 1.5
Data
MC ηγ
21
ID 1000000ENTRIES 2821 0.00 0.00 0.00 0.00 0.282E+04 0.00 0.00 0.00 0.00
m4 (MeV)
p T (M
eV)
IDEntriesMeanRMSUDFLWOVFLWALLCHAN
1000000 2821
-25.77 119.1 0.000 0.000 2821.
pL (MeV)
0
50
100
150
200
250
300
350
400
450
200 300 400 500 600 700 800 900
0
20
40
60
80
100
120
-400 -200 0 200 400
t – r/c of the t0 cluster
PT VS m4γ
Low m4γ, high pT
unbalanced pL
CUTTING |t-‐r/c|>0.5 ns
Ivan Prado Longhi@Mesonnet2013
22
Longitudinal momentum [MeV]-400 -200 0 200 400
18.9
MeV
/ (1
/N) d
N
00.001
0.002
0.0030.004
0.005
0.0060.007
0.008
0.009Signal
Background
Longitudinal momentum [MeV]-400 -200 0 200 400
18.9
MeV
/ (1
/N) d
N
00.001
0.002
0.0030.004
0.005
0.0060.007
0.008
0.009
U/O
-flow
(S,B
): (0
.0, 0
.0)%
/ (0
.0, 0
.0)%
Input variable: Longitudinal momentum
Transverse momentum [MeV]50 100 150 200 250 300 350 400
6.93
MeV
/ (1
/N) d
N
0
0.002
0.004
0.006
0.008
0.01
0.012
0.014
0.016
Transverse momentum [MeV]50 100 150 200 250 300 350 400
6.93
MeV
/ (1
/N) d
N
0
0.002
0.004
0.006
0.008
0.01
0.012
0.014
0.016
U/O
-flow
(S,B
): (0
.0, 0
.0)%
/ (0
.1, 0
.0)%
Input variable: Transverse momentum
t-r/c chi^2 [F]2 4 6 8 10 12
0.20
5 F
/ (1
/N) d
N
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
t-r/c chi^2 [F]2 4 6 8 10 12
0.20
5 F
/ (1
/N) d
N
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
U/O
-flow
(S,B
): (0
.0, 0
.0)%
/ (0
.0, 1
0.1)
%
Input variable: t-r/c chi^2
smearing σ = 40 MeV on px and py for γγ→ π0π0 MC
χ2t =
1
4
4�
i
�ti − ri/c
σt,i
�2
PL PT
MULTIVARIATE ANALYSIS
INPUT DISTRIBUTIONS: “SIGNAL”: all simulated physical processes (MC γγ-‐>ππ, KSKL, ηγ, ωπ etc.)
“BACKGROUND”: Data sample selected with |d|>33 cm
All contribu*ons weighted according to previous fit results
Ivan Prado Longhi@Mesonnet2013
23
htempEntries 8398Mean 0.5671RMS 0.4197
like0 0.2 0.4 0.6 0.8 10
200
400
600
800
1000
1200
1400
htempEntries 8398Mean 0.5671RMS 0.4197
like
Likelihood response0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
dx / (1
/N) d
N
0
10
20
30
40
50
60 SignalBackground
Likelihood response0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
dx / (1
/N) d
N
0
10
20
30
40
50
60
U/O
-flow
(S,B
): (0
.0, 0
.0)%
/ (0
.0, 0
.0)%
TMVA response for classifier: Likelihood
LIKELIHOOD RESPONSE
“Training” on input distribu*ons
“Applica*on” on data sample
Ivan Prado Longhi@Mesonnet2013
24
CUTTING LIKELIHOOD OUTPUT > 0.3
Likelihood output0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Effic
ienc
y (P
urity
)
0
0.2
0.4
0.6
0.8
1
Likelihood output0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
Effic
ienc
y (P
urity
)
0
0.2
0.4
0.6
0.8
1
Signal efficiencyBackground efficiency
Signal puritySignal efficiency*purity
S+BS /
For 4886 signal and 3512 background isS+Bevents the maximum S /
65.3767 when cutting at 0.2892
Cut efficiencies and optimal cut value
Sign
ifica
nce
0
10
20
30
40
50
60
70εsig ≈ 97%
εbkg ≈ 14%
purity ≈ 90%
Ivan Prado Longhi@Mesonnet2013
25
n before TMVA n aaer TMVA Red. Factor
KSKL 985 953 0.99
ηγ 393 374 0.95
ωπ0 806 774 0.96
f0γ 61 60 0.98
γγ 50 41 0.82
a0γ 9 9 1
bkg 3512 492 0.14
γγ→ π0π0 (data) 2722 2625 0.96
γγ→ π0π0 (MC) 10184 9924 0.97
CUTTING LIKELIHOOD OUTPUT > 0.3
Ivan Prado Longhi@Mesonnet2013
26
EntriesMeanRMSALLCHAN
5324 428.6 118.2 5324.
M4 (MeV)
dN/d
M4
(25
MeV
)-1
0
25
50
75
100
125
150
175
200
225
200 300 400 500 600 700 800 900
EntriesMeanRMSALLCHAN
5884 405.8 115.0 2625.
0
20
40
60
80
100
120
140
160
200 300 400 500 600 700 800 900
CUTTING LIKELIHOOD OUTPUT > 0.3
Subtracted spectrum
γγ→ π0π0 MC
INVARIANT MASS SPECTRA
KSKL
ωπ0
ηγ
γγ→ π0π0
bkg
Ivan Prado Longhi@Mesonnet2013
27
EntriesMeanRMSALLCHAN
5323 77.25 53.24 5323.
pT (MeV)
dN/d
p T (1
0 M
eV)-1
0
100
200
300
400
500
0 50 100 150 200 250 300 350 400 450 500
EntriesMeanRMSALLCHAN
5723 87.57 65.37 2624.
PT (MeV)
0
50
100
150
200
250
300
0 50 100 150 200 250 300 350 400 450 500
CUTTING LIKELIHOOD OUTPUT > 0.3
Subtracted spectrum
γγ→ π0π0 MC
PT SPECTRA
KSKL
ωπ0
ηγ
γγ→ π0π0
bkg
Ivan Prado Longhi@Mesonnet2013
28
data
EntriesMeanRMSALLCHAN
8398 392.8 114.1 8398.
M4 (MeV)
dN/d
M4
(25
MeV
)-1
0
50
100
150
200
250
300
350
400
450
200 300 400 500 600 700 800 900
BEFORE MULTIVARIATE ANALYSIS
n = Lεσ f = n/ndata ffit
KSKL 985 0.117 0.12435
ηγ 393 0.047 0.04465
ωπ0 806 0.096 0.0768
f0γ 61 0.007 0.0056
γγ 50 0.006 0.0072
a0γ 9 0.001 0.0012
bkg 3267* 0.389 0.4182
γγ→ π0π0 2830 0.337 0.32412
ndata= 8398 A fit to m4γ spectrum is performed using f=n/ndata as star*ng values
KSKL
ωπ0
ηγ
γγ→ π0π0
Bkg (data driven)
*normalized by extrapola\on
Ivan Prado Longhi@Mesonnet2013
!"#
! ! " "! ! " #" $%"&'(""
!"#$%&'()# !" #!"# ## # !"#*+#*,!"##*-*.# !"#/-/.
$%$&#'()*+#,*-*.#!"/#0(1)*+#0-*-(0-(2*+#34454# #%67/###7$/# .#$&/# .#!8/#9 9 .#!!/#9:9;.#</#3 3
0(1)*+# #=#>$&6%7# #&6$?@/###A,5'#=#$B%&A$B%7#
!""#$%&#'())#)*+),-#'.(#
")! ! " "! ! # " " *+,"-!./"+")0123"4"5160*(7("/"89"
#:;"")<!./" <$='**"")-!.$/"
!"#
! ! " "! ! " #" $%"&'(""
!"#$%&'()#!"# ##
!$%%#&'()*#+),)-#../#0'1()*-#%2/# ##0,),'0,'3)*#45565# #78%/###
0'1()*# #9#:!;8%"# #"8!2</###=+6	#$$7"=$>.%#
")! ! " "! ! # " " *+,"-!./"+"),0123"4"0125*(6("/"78"
#9:"");!./" ;$<'**"")-!.$/"
e+e-‐ → e+e-‐ η: 2D fits
29 Ivan Prado Longhi@Mesonnet2013
η → π0 π0 π0
η → π+ π-‐ π0
Ivan Prado Longhi@Mesonnet2013 30
! "#$%&'(# #)&%*+#,-.#/0#
12-#34567894:;#63-<#4;#9243#8;87=343#>-;-?89-3#-@-;93#A492#-B8C9#589?4B#-7-5-;9#8CC:?<4;>#9:#D677#E/!:<=#F283-#3F8C-#<439?4!694:;3ӛ#?-36793#4;#92-#F?:<6C94:;#:D# #5-3:;3#A492#;:;#;->74>4!7-#9?8;3@-?3-#5:5-;965&##G--<#9:#98H-#4;9:#8CC:6;9#92-# #9?8;3494:;#D:?5#D8C9:?######
12-#D:?5#D8C9:?#43#F8?85-9?4I-<#83#####A492#
!"
J4567894:;#:D#92-#34>;87##
J7:F-#F8?85-9-?#! #98H-;#83#8@-?8>-#:D#-BF-?45-;987#5-836?-5-;93#
η TRANSITION FORM FACTOR
Ivan Prado Longhi@Mesonnet2013 31
!""#!$%&!!'())!$%&!'())!$*+!
!"#$%&& !"#''(!"#''(
!"#$%&'()*+#,"#-./&(0.12#2/'#3"#,4%25262+#70*("#849"#:;#<=>?=@#=ABC#D"#E&6+#70*("#849"#=CF+#AGC#<=>HF@#I"$&//42J+#K"I&J%'./#2/'#D"K2%1./+#LJMN"#70*("#$A;#<=>?B@#A?B#
!"#$%&&'#$()$*"#$+(,$)'-.*%(-$γγ LUMINOSITY FUNCTION