Table Of ContentSUSY-P5: Chargino / Neutralino Analysis in the Fully
Hadronic Final State
Daniela K¨afer1, Jenny List1, and Taikan Suehara2 ∗
1- DeutschesElektronen Synchrotron(DESY) - Hamburg, Germany
9
2- International Center for Elementary Particle Physics (ICEPP), Univ. of Tokyo- Tokyo, Japan
0
0
2 The fully hadronic final states of two signal processes from an mSUGRA inspired sce-
nario(SUSY-P5)arestudiedwithinafullsimulationoftheLDC′detectormodel. These
n are chargino pair and neutralino pair production, i.e. e+e− → χe±χe∓ → qq¯′χe0 qq¯′χe0
a 1 1 1 1
ande+e− →χe0χe0 →qq¯χe0 qq¯χe0. Both processes havetobeseparated sufficientlyfrom
J 2 2 1 1
all background to measure the respective production cross sections and extract the
0
3 massesoftheinvolvedbosinos,m(χe±1),m(χe02)andmLSP =m(χe01). Thisisachievedby
fittingtheenergy spectra of the reconstructed gauge bosons while takinginto account
] thefinitewidthofthebosonmass. Fromsimulation datacorrespondingto500fb−1 of
x luminosity, a mass resolution of about 0.5 GeV seems to beachievable.
e
-
p 1 Introduction
e
h
[ Due tothecleanenvironmentandpreciselyknowninitalstateattheILC,itwillbepossible
to measure very small signal cross sections and extract the masses of SUSY particles even
1
from fully hadronic final states. As an example, from the SUSY-P5 scenario [2], pair pro-
v
duction of charginos (χ±χ∓) and neutralinos (χ0χ0) and their subsequent hadronic decays
8 1 1 2 2
5 (qq¯χ0 qq¯χ0) are studied. To successfully extract the masses of intermediate SUSY particles
1 1
9 from a fully hadronic final state, it is mandatory to reduce the background to a minimum.
e e e e
4 Espeecialley difficult is the mutual separation of both signal processes. Four jets need to be
.
1 assigned to two gauge bosons, either two W or two Z bosons, depending on whether they
0 originated from a decaying chargino or neutralino. In addition, a considerable amount of
9
missingenergyandmomentumcarriedawaybytwoofthe lightestsupersymmetricparticles
0
(LSPs) impede a complete reconstruction of the final state.
:
v AllMonteCarlosamplesarepartoftheDESYmassproduction[3]. Theeventsaregener-
Xi atedwithWhizard[4],runthroughthefullsimulationoftheLDCPrime 02Scdetectormodel
and finally reconstructed with MarlinReco[5] using Pandora Particle Flow (PFlow) [6].
r
a
2 Selection strategies and kinematic fits
This study concentrates on the reduction of SM backgroundusing a few very basic require-
ments and a kinematic fit to improve separationof the chargino from the neutralino signal.
A rough background scan as depicted in Fig. 1 (a) shows WW and WWZ production to
be the dominant source of SM background. Especially, 6-fermion final states from WWZ
production, with an invisible Z-decay, i.e. qq¯′qq¯′νν¯, constitute a nearly irreducible back-
ground component. However,the processes shown in Fig. 1 are not normalised to the same
luminosity. While the SUSY signal processes correspond to 500 fb−1, background samples
wereonly availablefor≈20-50fb−1 forQCDmultijetproduction(qq, qqqq finalstates)and
∗AuthorsKa¨ferandListacknowledgethesupportbyDFGLi1560/1-1,andauthorSueharaacknowledges
thesupportbyKAKENHI18GS0202.
LCWS/ILC2008
1 fb−1 for γγ/eγ processes. Background from other SUSY processes has also not yet been
considered, but it is expected to not render the preliminary results of this study obsolete.
Ramaining # of events by SM suppression cuts for chargino analysis Entriehs temp 27945 w energy in chargino event w/ xn2&SM bg. EEnnttrriieessww eexxcc 2244229966
20000 Mean 2721 MMeeaann 110000..77
LDCʼ RMS 656.3 LDCʼ RRMMSS 1144..0066
18000 Chargino 600 Chargino + neutralino2
16000 (SM not included)
# of track and Evis cuts only 500
14000 + >20 GeV lepton veto
12000 bb + all jet within |cos(θ)|<0.9 400 m(χ±1)=219.7±1.54GeV
10000 m(χe01) =121.2±0.59GeV (LSP)
8000 cc WW dd 300 e SM bg (x10)
6000 200
WWZ
4000
2000 γγ γe γe WWZ 100
Neutralino2 bg
(a)00 1000 2000 3000 4000 5000 (b)060 70 80 90 100 110 120 130 140
DST file # W energy [GeV]
Important: luminosity is not normalized!
Figure(1/:e(: 1afb)-1S, qMq,qqbqqa: 2c0k-5g0 rfbo-1u, SnUdSY:c 5o00n fbt-r1)ibutions. Eventsleftafterrequiring: black: Ntracks >20
and150GeV<E <300GeV;blue: leptonvetowithE <20GeV;red: E >5GeV
vis lepton jet
and |cos(θ )| < 0.9 for all jets. (b) Empirical fit to the energy spectrum of the W boson
jet
consisting of a 3rd-order polynomial convoluted with a Gaussian (see text).
Someverybasicrequirementsreduceleptonic eventsandeventsfrommultijetQCDpro-
duction(seecaptionofFig.1),whiletheverysimilarneutralinosignalishardlyaffected. To
separate the charginoand neutralino signals from eachother, two hypotheses are compared
for the invariant di-jet masses, which can either form a W or a Z boson. The resulting
χ2-values for each hypothesis are plotted against each other, so that the final selection cri-
terium for chargino events results in:
χ2(W, j j )+χ2(W, j j )<2 with χ2(W, j j )=(m −m(j ,j ))2/(5GeV)2
1 2 3 4 k l W k l
χ2(Z, j j )+χ2(Z, j j )>4 with χ2(Z, j j )=(m −m(j ,j ))2/(5GeV)2.
1 2 3 4 k l Z k l
A moredetaileddescriptionofthe entireprocedure is givenin[7]. After the aboveselection
of chargino-typeevents,the energyspectrumof di-jet massesin Fig.1 (b) largelyresembles
that of the W bosona. The spectrum is fitted with an empirical function consisting of a
3rd-order polynomial (4+2 par. to describe the edge positions) convoluted with a Gaussian
(2 par.) yielding the following values for the bosino masses:
m(χ±)=219.7±1.54 GeV ↔ 216.5 GeV (nominal chargino mass)
1
m(χ0)=121.2±0.59 GeV ↔ 115.7 GeV (nominal neutralino/LSP mass).
1
e
The central fit values differ rather much from the nominal bosino masses in the SUSY-P5
e
scenario. However, for an analysis of real data, such a shift can be compensated by tuning
the MC distribution such that the experimental data is well reproduced.
AdifferentapproachpostponesthereductionofeventsfromSMbackgroundinfavourof
a slightly better separationof chargino and neutralino signals with the help of a 2-dim. cut
inthe(V1, V2)massplane,withV beingaW orZ boson(seeFig.2). Thecutistuned,but
not yet optimised, for a high efficiency and purity of the chargino pair production (ε×π).
aSMbackgroundismultipliedby10,sincesofartheluminosityisonlyabout50fb−1formostbackground
processes. Itisalsonotincludedinthefit,whiletheneutralinopairbackground istakenintoaccount.
LCWS/ILC2008
A simple kinematic 1C-fit, requiring equal di-jet masses (EMC), is applied to the three
di-jet combinations of all events that can be clustered into 4 jets. The fit is meant to help
find the correct jet-pairings, improve the mass and energy resolutions for the gauge bosons
and thus, ultimately, the starting point for a fit to the W (Z) boson energy spectra. The
di-jet masses populating the (V1, V2) mass plane in Fig. 2 correspond to the jet-pairings
with the best fit probability and are assumed to be the correct ones.
V]140 Charginos V]140 Neutralinos V]140 SM: qqqq+n n
e e e
G G G
s [120 s [120 s [120
s s s
a a a
m m m
Z2 100 Z2 100 Z2 100
W/ W/ W/
80 80 80
60 60 60
40 40 40
20 20 20
20 40 60 80 100 120 140 20 40 60 80 100 120 140 20 40 60 80 100 120 140
W/Z1 mass [GeV] W/Z1 mass [GeV] W/Z1 mass [GeV]
Figure 2: Selection of chargino-type events: all events between the middle (red) and the
lower (orange) lines are assumed to originate from chargino decays, while events between
the middle and upper lines are taken to be neutralino decays.
3 Physics-driven fit to the energy spectrum of the W boson
Having selected the events between the lower two lines in Fig. 2 as chargino events, the
energy spectrum of the corresponding gauge boson is then fitted with the masses of the
SUSY particles as free parameters. However, this time, not an empirical fit function is
used but a physics-driven function. Firstly, the gauge boson’s energy and momentum are
expressed in terms of the three relevant masses m = m(χ±, χ0), m = m(χ0) = m ,
2 1 2 1 1 LSP
and m =m :
V W,Z
E = 1 m2−m2−m2 and |p~ |= 1 (em2−em2−m2)2−e4m2m2
V m 1 2 V V m 2 1 V 1 V
2 2
q
(cid:0) (cid:1)
From this the inverse expression has to be calculated for either the chargino (neutralino)
mass. Next, the effect of the finite boson mass width is added by convolusionwith a Breit-
Wigner function BW m0,γ, m − ∆m, m + ∆m and, finally, some Gaussian smearing
V V 2 V 2
is added to the emerging fit function:
(cid:0) (cid:1)
∆m ∆m
E-spec′ = bin · BW m0, γ, m − , m + +E-spec
i V V 2 V 2
(cid:18) (cid:19)
bini = 21 − a2 · xEmb22−−mE22b·EpVV ! ·(cid:20)Θ“Rσj+b1”−Θ“Rσj−b1”(cid:21) + a2πb e−2(R(σ−j1b))22 −e−2(R(σ+j1b))22!,
p
no Gaussian smearing adds Gaussian smearing
| {z } | {z }
m xm − E E
2 2 b V
with: b= , R= , σ : jet energy resolution.
j
E2−m2 · p E2−m2 · p
b 2 V b 2 V
p p
LCWS/ILC2008
V V 33000000 m(~c 0)= 113.9 – 0.02 CCChhhaaarrrgggiiinnnooosss V V 11880000 m(~c 0)= 115.2 – 0.46 CCChhhaaarrrgggiiinnnooosss
1.00 Ge1.00 Ge22550000 m(~c 1–1)= 212.3 – 0.06 NNNSSSMMMeeeuuu::: tttqqqrrraaaqqqlllqqqiiinnnqqqooo+++sssnnn nnn 1.00 Ge1.00 Ge1111464600000000 m(~c 1–1)= 215.3 – 2.45 SSSNNNMMMeeeuuu::: tttqqqrrraaaqqqlllqqqiiinnnqqqooo+++sssnnn nnn
nts / nts / 22000000 nts / nts / 11220000
ee ee
vv vv
EE EE11000000
11550000
880000
11000000 660000
440000
550000
220000
00 00
(a) 4400 6600 8800 110000 112200 114400 116600 118800 220000 (b) 4400 6600 8800 110000 112200 114400 116600 118800 220000
MMCC:: WW bboossoonn eenneerrggyy [[GGeeVV]] EEMMCC:: WW//ZZ eenneerrggyy [[GGeeVV]]
Figure 3: Fit to the energy spectrum of the W boson (a) on generator level and (b) after
reconstruction and a first rough chargino selection.
Thefitvaluesforthereconstructedbosinomassesareactuallyquiteclosetothenominal
generator level values:
m(χ±)=215.3±2.45 GeV ↔ 216.5 GeV (nominal chargino mass)
1
m(χ0)=115.2±0.46 GeV ↔ 115.7 GeV (nominal neutralino/LSP mass).
1
e
Altheough the error on the chargino mass is still large, this is expected to be reduced by
a more sophisticated selection (higher ε×π) and a simultaneous fit to the cross sections of
bothsignalprocesses,i.e. charginoandneutralinopairproduction. Thecrosssectionscould
then serve as a further input to the fit of the gauge boson energy spectra.
While only the dominant SM background (WWZ → qq¯′qq¯′νν¯) is considered for the
distributionsinFig.3(a,b),signalandbackgroundprocessesarenormalisedtothesamelu-
minosityof500fb−1. Thisprovidesanestimateofthesensitivityofthechargino(neutralino)
selection to a possible contamination from SM background.
4 Conclusions and outlook
Physics analyses using a full detector simulation play an important rˆole in the ongoing
optimisationeffortfortheILCdetector. Thisstudy,althoughperformedonsimulationdata
of a previousdetector model (LDCPrime 02Sc),constitutes one partof the generaleffortfor
the ILD detectorconcept. Incases wherethe gaugebosonmassesarereconstructedfrom4-
jet finalstates,the massandenergyresolutionis expectedto improvedrasticallydue to the
interplay between a highly segmented calorimeter with tracking ability for neutral particles
andthe new approachofParticleFlowalgorithms. Togetherwithmoresophisticatedfits to
the resulting gauge boson energy spectra, mass resolutions of 0.5-1 GeV are achievable for
those SUSY particles accessible to the ILC (χ±, χ0, and χ0).
1 2 1
NextstepswillbetoincludeallSMandrelevantSUSYbackground,optimisetheselection
and fit procedures and utilise different polariesatioen scenareios to improve the S/B-ratio.
LCWS/ILC2008
References
[1] Presentation:
http://ilcagenda.linearcollider.org/contributionDisplay.py?contribId=187&sessionId=17&confId=2628
[2] M.Battaglia,T.Barklow,M.E.Peskin,Y.Okada,S.YamashitaandP.M.Zerwas,IntheProceedings
of2005InternationalLinearColliderWorkshop(LCWS2005),Stanford,California,18-22Mar2005,
pp 1602[arXiv:hep-ex/0603010].
[3] DESYMonteCarloSimulation&Reconstruction database
http://ilcsoft.desy.de/portal/data_samples, http://ilcsoft.desy.de/loi/loi_reco_tables.html
[4] http://whizard.event-generator.org/
[5] http://ilcsoft.desy.de/portal/software_packages/marlin/index_eng.html
[6] M.Thomson,J.Phys.Conf.Ser.110,092032 (2008).
[7] PresentationattheILDOptimisationMeeting,Cambridge,11-13Sep2008:
http://ilcagenda.linearcollider.org/materialDisplay.py?contribId=36&sessionId=6&materialId=
slides&confId=2813
LCWS/ILC2008
