A Search for the Rare Decay D - Recinto Universitario de...
Transcript of A Search for the Rare Decay D - Recinto Universitario de...
A Searchfor theRareDecayD��� �������
by
Hugo R. HernandezMora
Thesissubmittedin partialfulfillment of therequirementsfor thedegreeof
MASTER OFSCIENCE
in
Physics
UNIVERSITY OFPUERTO RICO
MAYAGUEZ CAMPUS
2001
Approvedby:
AngelM. LopezBerrıos,Ph.D. DatePresident,GraduateCommittee
HectorMendez,Ph.D. DateMember, GraduateCommittee
Dorial Castellanos,Ph.D. DateMember, GraduateCommittee
WolfgangA. Rolke,Ph.D. DateRepresentativeof GraduateStudies
Dorial Castellanos,Ph.D. DateChairpersonof theDepartment
L. Antonio Estevez,Ph.D. DateDirectorof GraduateStudies
Abstract
This thesispresentsa searchfor theraredecayD � �� ���� basedon datacollected
by the Fermilabfixed target photoproductionexperimentcalled FOCUS(FNAL-E831).
This decayis an exampleof a flavor-changingneutralcurrentprocesswhich, according
to the StandardModel, should occur at a very low branchingratio of at most � � ����� .A rate higher than this would be an indication of new physics. Using a blind analysis
methodwhich consistsof the observation of events outsidea definedsignal area, the
optimumselectioncriteria for candidateeventswasdetermined.This methodminimizes
biasby basingthe choiceon the backgroundobserved outsidethe signal region andnot
on maximizing the signal understudy. The ratio of backgroundoutsideto background
insidethesignalregion wasdeterminedusinganindependentrealdatasample.Theresult
for thebranchingratio of theD ��� ���� decayat 90%ConfidenceLevel upperlimit is��� ��� � � �"! whichreducesthepublishedvalueby 50%.No evidencewasfoundfor this rare
decay.
ii
Resumen
Estetrabajodetesispresentala busquedadel decaimientoraroD ��� ���� basado
en datoscolectadospor el experimentode fotoproduccion de blanco fijo de Fermilab
llamado FOCUS (FNAL-E831). Este decaimientoes un ejemplo de un procesode
cambio de sabora traves de corrientesneutrales(FCNC) el cual, de acuerdocon el
ModeloEstandar, deberıa ocurrir a unamuy bajaprobabilidada un ordende � � ����� . Una
probabilidadmayora estapodrıa serunaindicacion de nueva fısica. Usandoun metodo
deanalisisa ciegasqueconsisteenla observacion deeventosfueradeunaregion definida
comoareadesenal,sedetermino uncriterio optimodeseleccionenla busquedadeeventos
candidatos.El usodeestemetododeanalisispermitio minimizarel prejuiciopresenteen
la eleccion de nuestrocriterio cual estababasadoen observar eventosde ruido fueradel
areadesenalconla ideadenomaximizarla senalbajoestudio.La razonderuidofueradel
areadesenal al ruido dentrodeestasedetermino usandounamuestrareal independiente.
El resultadoparala razon deporcentajeparael decaimientoD �� ���� mostro un lımite
superior(con un nivel de confianzade 90%) de��� �#�$� � �"! , el cual reduceen un 50% el
valorpublicadohastaahora.No seencontro evidenciaalgunaparaestedecaimientoraro.
iii
To God,
for mycountry:COLOMBIA ,
to mybeautifulcity BARRANQUILLA ,
to mywifeDiane,
to myparentsFanny andAnthony and,
to mybrothersMadelaine andJorge Luis.
Los quiero muchısimo!!!
To thethousandsof countrymenwhohavefallenwhile
I have beenin the “Isla del Encanto”becauseof theabsurd
conflict that lives within the boundariesof my harmed
country, speciallyto apersonwhowasanexampleto meand
my professionalguide, ProfessorLisandroVargasZapata,
cowardly murderedby gunmenon february23, 2001at his
doorstepin Barranquilla.
iv
Acknowledgments
There are so many people that in one way or anotherhave contributed in my
developmentas a humanbeing and as a professional. To mention them all would be
impossible. It is very difficult to even mentionsomebeforeothers. The most important
thing is thatall of thosewho believedin me,will alwaysholda key to my heart.
Firstof all. I thankGod,thatentitywhohasgivenuslife andwith that,agreatnumber
of questionsthatwetry to answerdayby day. I thankHim for giving metheopportunityof
beingborn in COLOMBIA , a countrythatevennow, whenit is goingthrougha crisis,has
a faceunknown to many. Thefaceof a countrythatalwaysfought for progress,of a hard
working countrythat is proudof its roots,thatcriesfor thosewho have fallenandmisses
thosewhoaremissing.A countrythatscreamsfor freedom,but aboveall, thereis acry of
hopethatthearmedconflict thatharmsuswill end. %'&�(*),+.-�)0/�12&3&547698.+.-:1<;0/�6=1>),?@? !I don’t want to leave behindmy city, BARRANQUILLA , my “Currambala Bella”.
Thecity thatwatchedmegrow up. On her streetsI learnedto live life to theextreme. It
wastherewheremy life is centeredandthe crib of my professionalfuture. I have great
expectationsfor my professionaldevelopmentandI wish to offer thoseto my lovedcity.
v
I thankFanny Mora, my mother, for makingme into what I amtoday. Without her
unconditionalhelp I wouldn’t heve gottento this point. To her above all, I dedicatethis
triumph.To my brothers,MadelaineandJorgeLuis, for believing in meandfor takingcare
of my motherwhile I wasabsent.To my fatherAnthony Valle, for alwaysbeingtherein
thegoodtimesaswell asin thebadtimes.
PUERTO RICO is a countrythat I will alwayshave with me. A master’s degree,a
marriageanda family. All this I take from you “Isla del Encanto”.I thankmy wife Diana
for herunconditionalhelp.Thankyou for sharingmy goodtimesandmy sorrow.
An immensurablethanksto the friendship,professionalandeconomicsupportthat
Dr. Angel M. Lopezgaveme. Thankyou for everythingyou did duringthetime I worked
for you, andfor alwaysbelieving in meandtrustingme. To professorsWill E. Johnsand
HectorMendezfor helpingmeout without askingandalsoto Dr. WeiJunXiong. Thanks
to Dr. WolfgangRolke for your cooperationin thestatisticalanalysisfor this work. Your
contribution wasof greatimportancefor me. Overall, to all of thosewho work with meat
theHEPlabatMayaguez.Thosefrom thepast:AlejandroMirles, CarlosRivera,Mauricio
Suarez,SalvadorCarrillo, FabiolaVazquezandKennieCruz;andthosefrom thepresent:
JoseA. Quinones,EduardoLuiggi, Alexis Parıs,Luis A. NincoandCarlosR.Perez.Thank
youguys!!!
To thePhysicsDepartmentat MayaguezCampusandits directorwhenI becamea
partof theUniversity, Dr. Dorial Castellanos(January, 1998),whogavemechanceto make
it happen.I thatall of thoseat theFOCUS Collaboration for giving metheopportunity
vi
to work on their experiment.To all thosepeoplewho work at EPSCoRbut mostof all to
SandraTrochewho wasalwaystherefor me.
To the greatfriendshipsI madein PR.To JuanCarlosHernandez,CarlosJose and
speciallyto their parentsJose “Machito” HernandezandJuanitaGonzalez. You werethe
family I neededthroughtoutthesethreeyears.
Thanksto Mr. JoeOrtız for helpingmein my collegestudies,thankyou for helping
mepursuethegroundsof PR.I alsothankMr. Felix Rivera,andtheRomeroValle family
aswell asValle Mirandafamily. Thanksfor everything. I thankMr. Rafael Sanjuanand
his family for beingtherewhenI neededit. A specialthanksto Mrs. MarthaCastano,may
Godblessyournoblesoul.
I thank my professorsat the Universidad del Atlantico - Barranquilla: Arnaldo
CohenandOswaldoDede,for makingme look at science.To my colleaguesandfriends
Arnulfo Barrios, OscarMontesinos,Harkccynn Torregroza,Ericka Navarro and Rafael
Manga. I also thank my High Schoolteachersfrom the BienestarSocial de la Policıa
- Barranquilla: Blas Torres, Carlos Bray and Orlando Ortız, specially Mr. Armando
Consuegra(Barrabas).
To all of thesewhomI have mentionedaswell asthosewhomI have not, otherwise
I wouldneverbeableto finish...
Mil y mil gracias!!!
vii
Agradecimientos
Sonmuchaslaspersonasquedeunau otra formahancontribuido enmi desarrollo
comoserhumanoy comoprofesional.Enumerarlasa todasresultarıa imposible.Inclusive
mencionaralgunasprimero que a otrasresultadifıcil paramı. Lo masimportanteque
quieromanifestaresquea todasestaspersonasquecreyeronenmı por siemprelasllevare
enmi corazon.
Inicialmentequisieradarlelasgraciasa Dios, a eseserquenoshaotorgadola vida
y con ella un sinnumerode interrogantesquedıa a dıa tratamosdecontestarnos.A El le
agradezcoel habermedadola oportunidaddenacerdondenacı, mi patriaCOLOMBIA , un
paıs queauncuandoenestosmomentosatraviesapor momentosdecrisisposeeunacara
quemuchosdenosotrosdesconocemos.La caradeun puebloquepor siemprehaluchado
por salir adelante,deun pueblotrabajadory orgullosodesusraices,quellora a sushijos
caidosy extrana a sushijos ausentes.Un puebloque lanzaa todo pulmon un grito de
libertad,perosobretodoungrito deesperanzapor la culminaciondeesteconflictoarmado
quenosacongoja.Queremosun paıs librepara todos!
No quierodejarde ladoa mi BARRANQUILLA del alma,a mi Currambala Bella.
La ciudadque me vio crecer. En suscallesaprendı a vivir la vida en cadauno de los
viii
extremosqueestanospresenta.Y esallı dondesecentrami vida y mi futuro profesional.
Resultaninmensaslasexpectativasdepoderdesarrollarmecomoprofesionaly ofrecermis
serviciosa la ciudaddemisamores.
Le agradezcoami madre,Fanny Mora,por hacerdemı lo quehoy endıasoy. Sinel
apoyo incondicionaldepartedeellanohubiesesidoposiblela consecusiondecadaunode
mis exitos. A ella masquea nadiededicoestetriunfo quehoy dıa estoy consiguiendo.A
mis hermanos,Madelainey JorgeLuis, por creerenmı y por cuidardemi madredurante
mi ausencia.A mi padre,Anthony Valle, por estarallı siempre,tantoenlasbuenascomo
enlasmalas.
PUERTO RICO esunatierraquellevareconmigoporsiempre.Unamaestrıa,muchos
amigos,unmatrimonioy unafamilia. Todoestomellevo detussuelos“Isla delEncanto”.
Agradezcoenormementela ayudaincondicionalrecibidade partede Diana, mi esposa.
Graciaspor compartira mi lado mis alegrıasy soportarademaslas tristezasy los malos
ratos.
Un agradecimientomasqueenormea la amistad,el apoyo profesionaly economico
queme brindo el Dr. Angel M. Lopez. Graciaspor todo lo quehizo por mı duranteel
periodoenel queestuve trabajandoa sulado,por creersiempreenmı y por brindarmesu
confianza.A losprofesoresWill E. Johnsy HectorMendezpor la ayudaquemebrindaron
sin nuncarecibir unanegativadesuparte.Al Dr. WeiJunXiong por sucolaboracion y su
apoyo a mi trabajo.Graciaspor la ayudabrindadapor partedel Dr. WolfgangRolke. Su
colaboracion en el analisis estadıstico paraestetrabajode tesisfue de granimportancia.
En generala todoslos companerosdel laboratoriodeHEPenMayaguez.Los del pasado:
ix
AlejandroMirles, CarlosRivera,Mauricio Suarez,SalvadorCarrillo, FabiolaVazquezy
KennieCruz; los del presente:Jose A. Quinones,EduardoLuiggi, Alexis Parıs, Luis A.
Nincoy CarlosR. Perez.Muchachos,gracias!!!
Al Departamentode Fısica del RUM y su directoren el periodo(Enerode 1998)
en que inicie estudiosgraduadosen UPR. A ellos les agradezcoel habermebrindadola
oportunidadparaque esto fueseuna realidad. Agradecimientosmuy especialesa toda
la gentede FOCUS Collaboration por darmela oportunidadde trabajaren estegran
experimento.A todaslaspersonasquetrabajanen EPSCoRen especiala SandraTroche
por suincondicionalayuday por suconstantecolaboracion.
A lasgrandesamistadesconstruidasdurantemi permanenciaenPR.A JuanCarlos
Hernandez,CarlosJose y en especiala suspadresJose “Machito” Hernandezy Juanita
Gonzalez.Ustedesfueronel soportefamiliarquetantaayudamehizoenestosultimostres
anostiempoqueduro mi maestrıa.
Graciasespecialesa JoeOrtız por apoyarmeincondicionalmenteen mis estudios
universitarios,graciaspor tu patrocinioenmi empresahaciaPR.Igualmentele agradezco
a Felix Rivera,a la FamiliaRomeroValle al igual quea la Familia Valle Miranda.Gracias
por su patrocinio. Agradezcoa Rafael Sanjuany familia por siempreestarcuandolos
necesitabamos.Gracias.A MarthaCastano, un mill on debendicionesparaustedmi bella
dama.
Agradezcoamisprofesoresdela UniversidaddelAtlantico- Barranquilla: Arnaldo
Coheny Oswaldo Dede,por inculcarmelos deseosde hacerciencia. A mis colegasy
x
amigosdeuniversidad,Arnulfo Barrios,OscarMontesinos,HarkccynnTorregroza,Ericka
Navarroy RafaelManga.Igualmenteagradezcoa todosmis profesoresdeBachilleratoen
el BienestarSocialdela Policıa - Barranquilla: BlasTorres,CarlosBrayy OrlandoOrtız,
especialmenteaArmandoConsuegra(Barrabas).
A todaslas personasque he mencionadoal igual que a las muchasque deje de
mencionarpuesnuncaterminarıaestedocumento...
Mil y mil gracias!!!
xi
Contents
List of Tables xviii
List of Figures xix
List of Abbreviations xxi
1 Intr oduction 1
1.1 TheStandardModel . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.1.1 FundamentalParticles . . . . . . . . . . . . . . . . . . . . . . . . 3
Bosons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
Fermions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
1.1.2 Interactions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
Long andShortDistanceEffects . . . . . . . . . . . . . . . . . . . 9
FeynmanDiagrams. . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.2 ConservationLaws . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
xii
1.2.1 Conservationof AngularMomentumandHelicity Suppression. . . 11
1.2.2 Conservationof theelectricalcharge . . . . . . . . . . . . . . . . . 12
1.2.3 LeptonNumberConservation . . . . . . . . . . . . . . . . . . . . 13
1.2.4 BaryonNumberConservation . . . . . . . . . . . . . . . . . . . . 14
1.2.5 Flavor Conservation . . . . . . . . . . . . . . . . . . . . . . . . . 15
1.3 CabibboTheory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
a)Cabibbo-Favored . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
b) Cabibbo-Suppressed. . . . . . . . . . . . . . . . . . . . . . . . . . . . 20
c) Doubly-Cabibbo-Suppressed. . . . . . . . . . . . . . . . . . . . . . . . 21
1.4 GIM Mechanism . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
1.5 RareandForbiddenDecays. . . . . . . . . . . . . . . . . . . . . . . . . . 24
a)LeptonFamily NumberViolation . . . . . . . . . . . . . . . . . . . . . 24
b) LeptonNumberViolating . . . . . . . . . . . . . . . . . . . . . . . . . 24
c) Flavour-ChangingNeutral-Currents. . . . . . . . . . . . . . . . . . . . 24
1.6 SpecialRelativity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
1.7 ConfidenceIntervalsandConfidenceLevels . . . . . . . . . . . . . . . . . 29
2 Previous Works 30
2.1 FCNCDecayTheory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
xiii
2.2 PreviousSearchesfor RareCharmDecays. . . . . . . . . . . . . . . . . . 32
2.2.1 TheHadroproductionWA92 Experiment . . . . . . . . . . . . . . 33
2.2.2 TheHadroproductionE791Experiment . . . . . . . . . . . . . . . 34
2.2.3 ThePhotoproductionE687Experiment . . . . . . . . . . . . . . . 36
2.3 FOCUS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
2.3.1 Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
2.3.2 TheFOCUSPhotonBeam . . . . . . . . . . . . . . . . . . . . . . 39
2.3.3 InsidetheFOCUSExperiment. . . . . . . . . . . . . . . . . . . . 40
3 Objectives 44
4 Procedure 46
4.1 DataFormat . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
4.2 ReconstructionProcess. . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
4.3 DataSelection. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
4.3.1 SelectionVariables . . . . . . . . . . . . . . . . . . . . . . . . . . 50
a) Vertexing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
1. ACB<D . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
2. PrimaryVertex Multiplicity . . . . . . . . . . . . . . . . . . . 52
3. Vertex ConfidenceLevels . . . . . . . . . . . . . . . . . . . . 52
xiv
4. Vertex z Positions . . . . . . . . . . . . . . . . . . . . . . . . 52
5. Vertex Isolations . . . . . . . . . . . . . . . . . . . . . . . . 53
b) ParticleIdentification . . . . . . . . . . . . . . . . . . . . . . . 53
1. Muon Identification . . . . . . . . . . . . . . . . . . . . . . . 53
i. KaonConsistency for Muons . . . . . . . . . . . . . . . . . 54
ii. InnerMuons . . . . . . . . . . . . . . . . . . . . . . . . . 55
A) Numberof MissedMuonPlanes . . . . . . . . . . . . . 56
B) InnerMuonConfidenceLevel . . . . . . . . . . . . . . 56
iii. OuterMuons. . . . . . . . . . . . . . . . . . . . . . . . . 56
A) OuterMuonConfidenceLevel . . . . . . . . . . . . . . 56
2. HadronIdentification . . . . . . . . . . . . . . . . . . . . . . 56
i. Kaonicity . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
ii. PionConsistency . . . . . . . . . . . . . . . . . . . . . . . 57
3. Momentum . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
4. DimuonicInvariantMass . . . . . . . . . . . . . . . . . . . . 58
4.3.2 FOCUSDataSelectionStages. . . . . . . . . . . . . . . . . . . . 59
a) Pass1. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
b) Skim1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
c) Skim2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
xv
d) Skim3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
4.4 NormalizingMode . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
4.5 MonteCarloSimulation . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
a) D ��� ���� Events . . . . . . . . . . . . . . . . . . . . . . . . 66
b) D K �FE Events. . . . . . . . . . . . . . . . . . . . . . . . 66
c) D E E � andD K K � Events . . . . . . . . . . . . . . 67
4.6 AnalysisMethodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
4.6.1 Blind Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
4.6.2 RolkeandLopezLimits . . . . . . . . . . . . . . . . . . . . . . . 67
4.6.3 Determinationof Sensitivity andConfidenceLimits . . . . . . . . . 69
4.6.4 SidebandsBackgroundandSignalArea . . . . . . . . . . . . . . . 71
4.7 BackgroundStudy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
4.7.1 Introduction. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
4.7.2 MisID Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
4.7.3 BackgroundRatioStudy . . . . . . . . . . . . . . . . . . . . . . . 74
4.8 Optimizationof theSelectionCriteria . . . . . . . . . . . . . . . . . . . . 78
4.8.1 FixedCutsandVariableCut Ranges. . . . . . . . . . . . . . . . . 79
4.8.2 OptimizationBasedonSensitivity . . . . . . . . . . . . . . . . . . 80
xvi
5 Resultsand Conclusions 81
5.1 MisID . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
5.2 BackgroundRatio . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
5.3 SkimmingResults. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
5.4 CutOptimizationResults . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
5.5 ConfidenceLimits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
5.6 Conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
Bibliography 92
xvii
List of Tables
2.1 Chronologicalbranchingratiosfor the GIHKJL� weakneutral-currentdecayD 5��� ���� . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
4.1 Descriptionof Superstreams. . . . . . . . . . . . . . . . . . . . . . . . . 61
4.2 Descriptionof Superstream1 . . . . . . . . . . . . . . . . . . . . . . . . . 62
4.3 Sensitivity numberasperRolkeandLopez . . . . . . . . . . . . . . . . . 69
4.4 An exampleof RolkeandLopez90%ConfidenceLimits . . . . . . . . . . 70
4.5 DimuonicInvariantMassCutsfor SidebandsandSignalRegions . . . . . . 79
4.6 Cutvaluesfor eachvariableconsideredin theoptimizationprocess. . . . . 80
5.1 BestCutCombination. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
5.2 Resultsof Cut Optimization . . . . . . . . . . . . . . . . . . . . . . . . . 88
xviii
List of Figures
1.1 TheStandardModel for ElementaryParticles . . . . . . . . . . . . . . . . 5
1.2 Transformationspermmitedby theStandardModel . . . . . . . . . . . . . 8
1.3 Feynmandiagramaweakprocessmediatedby avirtual bosonexchange.. . 10
1.4 A chargedcurrentinteraction . . . . . . . . . . . . . . . . . . . . . . . . . 12
1.5 Feynmandiagramfor aweaktransition . . . . . . . . . . . . . . . . . . . 17
1.6 Representationfor theCKM matrixelements . . . . . . . . . . . . . . . . 20
1.7 Feynmandiagramfor aCabibbo-Favoredprocess . . . . . . . . . . . . . . 20
1.8 Feynmandiagramfor aCabibbo-Suppressedprocess . . . . . . . . . . . . 21
1.9 Feynmandiagramfor aDoubly-Cabibbo-Suppressedprocess. . . . . . . . 21
1.10 Weakinteractionsin neutral-currentprocesses. . . . . . . . . . . . . . . . 22
2.1 Feynmandiagramsfor thedecayD ��� ���� . . . . . . . . . . . . . . . . 31
2.2 ProtonsynchrotronandbeamlinesatFermilab . . . . . . . . . . . . . . . . 38
2.3 TheE831PhotonBeam. . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
xix
2.4 TheFOCUSSpectrometer . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.1 TrajectoriesthroughtheFOCUSspectrometer. . . . . . . . . . . . . . . . 49
4.2 Productionanddecayverticesfor theD ���� ���� process . . . . . . . . . 51
4.3 Analysisoverview of FOCUS . . . . . . . . . . . . . . . . . . . . . . . . 60
4.4 MonteCarlogeneratedprocessschematizationfor thedecayD �� ���� . 66
4.5 D ��� ���� InvariantMassdistributions . . . . . . . . . . . . . . . . . . 71
4.6 DimuonicInvariantMasscomparisonof MonteCarlodatafromD E E �andD �� .��� . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
4.7 Dimuonicinvariantmassof D K ��E generatedby MonteCarlo . . . . 76
4.8 BackgroundcomparisonbetweenrealandMonteCarlodata . . . . . . . . 77
5.1 MisID asa functionof momentum . . . . . . . . . . . . . . . . . . . . . . 82
5.2 Typical InvariantMassDistribution Usedin Calculatingthe BackgroundRatio . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
5.3 InvariantMassDistributionsfor Skim3Cuts . . . . . . . . . . . . . . . . . 84
5.4 InvariantMassDistributionsfor Skim4Cuts . . . . . . . . . . . . . . . . . 85
5.5 Massdistributionsfor thebestcut combination. . . . . . . . . . . . . . . . 87
5.6 Invariant Mass Distribution for BackgroundRatio using the best cutcombination. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
5.7 Eventcandidatesfor theD �� � � process . . . . . . . . . . . . . . . . 90
xx
List of Abbreviations
BeO: Beryllium Oxide.
BR: BranchingRatio.
CERN: OrganisationEuropeennepourla RechercheNucleaire.
CITADL: Cerenkov IdentificationThroughA Digital Likelihood.
CKM Matrix: Cabibbo-Kobayashi-MaskawaQuark-MixingMatrix.
CP Violation: Charge-Parity Violation.
CL: ConfidenceLevel.
CLP: PrimaryVertex CL.
CLS: SecondaryVertex CL.
E687: FermilabExperiment687.
E831: FermilabExperiment831.
E791: FermilabExperiment791.
Fermilab: FermiNationalAcceleratorLaboratory.
FNAL-E831: FermilabExperiment831.
FCNC: Flavor-ChangingNeutral-Current.
FOCUS: FotoproductionOf Charmwith anUpgradedSpectrometer.
GIM Mechanism: Glashow-Iliopoulos-MaianiMechanism.
GeV: GigaElectronVolt.
IMU: InnerMuonSystem.
IMUCL: InnerMuonCL.
ISOP: PrimaryVertex Isolation.
ISOS: SecondaryVertex Isolation.
K/ M : KaonConsistency to separatekaonsfrom muons.
LFNV: LeptonFamily NumberViolation.
LNV: LeptonNumberViolation.
L/ N : Separationbetweentheproductionanddecayverticesdividedby
theerror, D , on thatmeasurement.
MeV: MegaElectronVolt.
MisID: Percentof eventsmisidentifiedwithin any specificanalysis.
MISSPL: Numberof MissedMuonPlanes.
M1: MagnetNumber1.
xxi
M2: MagnetNumber2.OQPRP: LorentzDimuonicInvariantMass.
OMU: OuterMuonSystem.
OMUCL: OuterMuonCL.
PAW: PhysicsAnalysisWorkstation.
Pass1: Basiceventreconstructionstage.
PWC: ProportionalWire Chambers.
Skim1: Firstdatareconstructionselectionstage.
Skim2: Seconddatareconstructionselectionstage.
Skim3: Third datareconstructionselectionstage.
Skim4: Fourthdatareconstructionselectionstage.
SSD: SiliconMicrovertex Detectors.
ZVPRIM: PrimaryVertex z Position.
ZVSEC: SecondaryVertex zposition.SW( T*U ): Kaonicity.T�VXWZY : PionConsisitency.NC[7\ : Dif ferencebetweenthePrimaryVertex z positionandthez target
edgedividedby theuncertainty.NC[^] : Dif ferencebetweentheSecondaryVertex z position and the z
targetedgedividedby theuncertainty.
xxii
Chapter 1
Intr oduction
This thesisis a researchwork in a topicof highenergy physics.It is alsoa formative
work and could be usedas a sourceof referencefor studentsand peopleinterestedin
knowing moreaboutthephysicsrelatedwith themicroscopicworld aroundus.
The StandardModel of electroweak interactionsis usedto understandthe decays
of heavy quarkswhich are known to us. A significant observation of charm Flavor-
ChangingNeutral-Currentdecays,predictedto occur rarely (in the StandardModel the
neutral-currentinteractionsdo not changeflavor), would imply new physicsbeyond the
StandardModel.
Rare decaymodesare probesof particle statesor massscaleswhich cannotbe
accesseddirectly. One of the most interestingprocessesto study is the D �� .���decay(all referencesto this decayalso implies the correspondingcharge-conjugatestate_ ���F�� ). Themaingoalof thework reportedin this thesiswasthemeasurementof
1
`1.1 TheStandardModel 2
the branchingratio for this decayanda comparisonto the actual90% ConfidenceLevel
upperlimit a � �0�b� � �"! .This chapterwill display somebasicand interestingconceptsusedfor this study.
First, anintroductionto theStandardModel of elementaryparticlesis done,next someof
the conservation laws arediscussed.The CabibboTheoryandthe GIM Mechanismwill
alsobediscussedin this chapter. BasicinformationaboutRareandForbiddendecayswill
be shown defining threeimportantrelatedprocesses:LeptonFamily NumberViolation,
LeptonNumberViolation andFlavor-ChangingNeutral-Currents.A shortrepresentation
of SpecialRelativity will bediscussedandfinally someusefulconceptsaboutConfidence
Intervalswill bedefined.
1.1 The Standard Model
At thebeginningof the�2�<ced
century, physiciststhoughtthatall matterwascomposed
of atomsasthe fundamentalbuilding block. As the centuryenteredadolescence,it was
discoveredthatatomsareconstructedfrom electronsandasmallpositivechargednucleus.
This nucleuswassubsequentlyfound to be composedof protonswhich have a positive
charge, and neutrons(except the Hydrogennucleus),which are uncharged. However,
duringthelatterhalf of thecentury, physicistsdeterminedthatneitherprotonsnorneutrons
arefundamental.A betterdescriptionaboutthe matterin the universewasobtainedand
now weknow thatall matteris madefrom oneor moreof sixteenelementaryparticles.
To explain the natureof theseelementaryparticlesandthe forcesbetweenthem,a
`1.1 TheStandardModel 3
theoryhasbeendevelopedcalledtheStandardModel[1–3]. Themodeldescribesall matter
andforcesin theuniverse(exceptgravity) andit is a packageof two theories:theUnified
Electroweak[4, p.95]andQuantumChromodynamics[4, p.85].
1.1.1 FundamentalParticles
TheStandardModel dividestheelementaryparticlesinto two groupscalled fg62h369+�hand /�&i1<j kl6=+�h whicharedifferentiatedaccordingto thebehavior of multiparticlesystems.
Bosonsobey theBose-Einsteinstatisticsandfermionsobey theFermi-Diracstatistics.The
Bose-Einsteinstatisticsmeansthat any numberof identicalbosonscanexist in the same
quantumstate.This is very essentialin shapingthestatisticalbehavior of a largenumber
of bosons.On theotherhand,in theFermi-Diracstatisticstwo fermionscannotcoexist in
thesamequantumstate.Whicheverstatisticsa particlefollows will determinethe m wave
function’s symmetryunderinterchangeof particles. If thesetwo particleshave identicaln m n o (wave-functionsquared),theprobabilityof particle1 beingin onespotandparticle2
at another, mustbeequalto theprobability that thefirst particleis in thesecondspotand
particle2 is in thefirst. Therefore,underinterchangethewave-function mIp � qr�s JKt0m#p qr�swhere
qr is therelativeposition.
For identical bosonsthe Bose-Einsteinsymmetryprinciple statesthat the wave-
functionunderinterchangemustgo to plusitself
mIp � qr�s JumIp qr�s (1.1)
and, for identical fermions the Pauli principle states that the wave-function under
`1.1 TheStandardModel 4
interchangemustgo to minusitself
mIp � qr�s J � mIp qr�s (1.2)
For nonrelativistic speeds,thetotal wave-functionof thepair canbeexpressedasa
productof functionsdependingonspatialcoordinatesandspinorientation:
n mIp qr�swv Jux�pyhwzF),4Z& s:{ pyhwz�kl+ s (1.3)
wherethespatialpart, x , will describetheorbitalmotionof oneparticleabouttheother. It
canhappenthat x is symmetricor antisymmetricunderinterchangeandthespinfunction {maybesymmetric(spinsparallel)or antisymmetric(spinsantiparallel)underinterchange.
Bosons . They are:photon, gluon, W | andZ . Theseparticleshaveanintegralspin
(0, 1, 2,...) andareknown asthe intermediaryparticlesbecausethey areresponsiblefor
carryingtheforceinteractions.Thephotonis masslesswith spin1 andis theintermediary
of the electromagneticinteraction. Thereis onekind of photonandtwo photonsdo not
directly interactwith oneanother. Thegluonis masslesswith spin1 andis theintermediary
of thestronginteraction.Thereareeight typesof gluons,eachdifferentfrom oneanother
only by a quantumnumbercalledcolor. The gluonsinteractwith oneanother. The W |(m } 80.4 ~��Z� ) andZ (m } 91.2 ~��Z� ), all of spin1, aretheintermediariesof theweak
interaction.TheW | andZ donot interactwith oneanother.
Fermions . They have a half-integral spin ( �o , �o ,...) andconstitutethe fundamental
familiesof matter. Thesearedivided into quarkswith fractionalelectriccharges(+o� n & n
and� �� n & n ) andleptonswith integercharge(
� � n & n ) or no charge(neutrinos). At thesame
`1.1 TheStandardModel 5
time, quarksare subdivided into six different particles: up (m } 1 to 5 ���7� ), charm
(m } 1.15to 1.35 ~��7� ) andtop (m } 169to 179 ~��Z� ) with electricchargesof� o� n & n and
down(m } 3 to 9 ���7� ), strange (m } 75 to 170 ���Z� ) andbottom(m } 4.0 to 4.4 ~��7� )
with� �� n & n . Thesix typesof quarksaresometimescalledflavors. Leptonsaresubdivided
into threedifferentelectricallycharged particles,the electron, & (m } 0.511 ���7� ), the
muon, � (m } 105.7 ���Z� ), andthe tau, � (m } 1.777 ~��7� ), with their respectiveneutral
neutrinos,�=� (m � 3 eV), �9� (m � 0.19 ���7� ) and �9� (m � 18.2 ���Z� ). Eachparticle,
quarkor lepton,hasits respectiveantiparticlewith thesamemassbut with oppositeelectric
charge. Figure1.1shows thefundamentalparticlesof matterarrangedinto familiesby the
StandardModel.
Figure1.1: TheStandardModel for ElementaryParticles.
`1.1 TheStandardModel 6
Eachof the threecolumnscorrespondsto a “f amily”. The relationshipsamongthe
particlesin afamily arethesameasthosein theothertwo families.In awaywehavethree
copiesof thesamestructure.In eachof thefamiliesthereis a leptonanda quarkdoublet.
The charge of the membersof the doubletsdiffer by one & unit. The main difference
betweenfamiliesarethemassesof their members.Thefirst family membersarelow mass
while themassesof thethird family aremuchhigher. This largedifferencein massis one
of theprincipalopenmysteriesin elementaryparticlephysics.
1.1.2 Interactions
Of theknown four forces,two, gravityandelectromagnetism, have a long rangeand
are commonin every day life. The remainingtwo forces,the strong and weaknuclear
forces,have sucha small rangethat it is frequentlycomparedto the size of an atomic
nucleus. Due to its low strength,gravity is still not a completepart of the Standard
Model. Another interactionthat hasan infinite rangeis the electromagneticinteraction
whoseeffects have beenmeasuredthroughoutimmensescales. The strong interaction
holdstogetherthenucleusworking againsttheelectrostaticrepulsionthatoccursbetween
protons. Lastly, the weak interactionis known to be responsiblefor a large variety of
radioactivedecays.Also, it is responsiblefor theinteractionsthatoccurbetweenneutrinos
andmatter.
Besideselectriccharge,quarksandgluonshave anothertypeof charge. It is called
the color charge. The effectsof the strongnuclearforce or color force areonly felt by
quarks. It is a very strongforce that bindsquarkstogetherforming particles. Thereare
`1.1 TheStandardModel 7
three“colors” for quarkswhich areusuallycalled“red”, “blue” and“yellow”. If you put
togethera redquarkwith anantiredantiquark,theproductis colorless.If you put together
a red,ablueandayellow quark,theproductis alsocolorless.
Due to thenatureof thecolor interaction,combinationswith a netcolor chargeare
highly unstable.If onestartswith a systemwith no netcolor charge,it will never separate
into color chargedpieces. The reasonfor this is believed to be that, dueto the fact that
gluonshave a color charge,thereis an“anti-screening”effect wherebytheeffective force
betweentwo quarkskl+�4g1<&3)�h3&9h with distance.This increasedpotentialenergy is converted
to quark-antiquarkpairsin sucha way that the separatedpiecesareeachcolorless.This
is believed to be the reasonno systemwith a net color charge has ever beendirectly
observed, including both quarksor gluons. However, throughthe useof deepinelastic
scatteringexperiments,the interactionsof gluonsor quarkswith incomingprojectileshas
beeninferred.
Only groupsof two quarks(mesonsformedof a quarkandan antiquark)or three
quarks(baryonsformedof threequarksor threeanti-quarks),havebeenseen.Collectively,
thesegroups are called hadrons. On the other hand, leptons can exist without the
companionshipof otherparticlessincethey have no color charge. Thenet forcebetween
color neutralsystems(thestrongforce)actually �,&34g12&3)"h3&9h with distance.
Theweakinteractionis felt by bothquarksandleptons.Thereis a deeprelationship
betweenthe weakinteractionandthe family structureof quarksandleptons. In the case
of theleptons,it is only possibleto changeleptonswithin thesamefamily into eachother.
Thus,theelectroncanbeturnedinto �9� but not into a muon.Theweakinteractioncanact
within the leptonfamilies,but not betweenthem. Therearecomplicationswhenit comes
`1.1 TheStandardModel 8
to the quarks.Herealsotheweakinteractioncanturn oneflavor into another. However,
it is not trueto saythat theweakforcecannotactacrossfamilies. It can,but with a much
reducedeffect.
Quarks�u
�d
��
��
��
�c
�s
��
��
��
�t
�b
����� �� � ������ �� � ��� ��� �� � ������ �� � ���
Leptons�e
��� ���
�g�
��� ���
�7�
��� ���
(a) (b)
Figure1.2: Transformationspermmitedby theStandardModel. (a) Theeffect of theweak force on the quarks. Transformationssuchas b � u are also possible. (b)The effect of the weak force on the leptons. Transformationswithin families arerepresentedby vertical arrows andacrossfamiliesarerepresentedby horizontalanddiagonalarrows.
For quarks,theelectriccharge changingtransformationswithin familiesandacross
families are dominant(the vertical transformationsare more probableand the diagonal
transformationsare less probable)as seen in Figure 1.2. This does not mean that
transformationsdueto neutralcurrentscannotoccur. Quark transitionssuchasc u,
in which thecharge is thesame,couldoccurthrougha neutralcurrent . Whathappens
is that transitionssuchasthis onehave a smallprobabilityof occurrencecomparedto the
electricchargechangingtransitions.In Section1.3,this will beexplainedin moredetail.
As mentionedbefore,theweakinteractionsof quarksaswell asleptonsaremediated
by theW , W � andZ . In theStandardModel leptonscannothavecrosstransformations
directly. Nevertheless,theStandardModelpermitsprocessessuchas &3 �&<����� ���� where
the & & � first convertsto a Z which thenturnsinto � � � . It is importantto notethatall
`1.1 TheStandardModel 9
particlesthathaveanelectricchargeincludingtheW feel theelectromagneticforcebut the
gravity forcedoesnothaveanobservableeffect in particlephysics.
Shor t and Long Distance Effects . Quantummechanicallyinteractionsaredue
to the exchangeof virtual bosons. The fundamentalcolor interactionbetweenquarks
is mediatedby virtual gluons. Effects due to such fundamentalprocessesinvolving a
finite numberof quark-gluonverticesare called short distanceeffects. The effective
strengthof this fundamentalinteractionis relatively small and Feynmandiagrams(See
next section.)andperturbationtheorycanbeusedto calculatethem.However, thegluons
(unlike thephotonandtheintermediariesof theweakinteraction)carrycolor charge. The
consequencesof this color charge are profound. Gluon-gluoninteractionsare possible.
Theseleadsto an ”antiscreening”effect where the effective color interactionbetween
quarksactuallyincreaseswith distanceinsteadof decreasingasin theelectromagneticand
weak interactions. The result is that stronginteractionsbetweenhadronsdecreasewith
energy sinceat higherenergy thecomponentquarksinteractat shorterdistances.At lower
energies and longer distancesthe effective interactionis larger and it is not possibleto
calculatesuch”long distanceeffects”usingperturbationtheory.
In hadrondecaythe fundamentaldecaymechanismmaywell bea weakinteraction
betweenthe constituentquarksbut as the decayproductsseparatelong distanceeffects
due to the color interactionwill play an important role. Somecharacteristicsof the
decaywill be determinedby its fundamentalweak naturebut other characteristicswill
be strongly affected by the long distanceeffects. For example, parity violation in a
decayis an indication of its fundamentalweak nature. However, the precisevaluesof
lifetimes andbranchingratioscannot be calculatedwith pureperturbationtheory. Our
`1.2 ConservationLaws 10
currentbestunderstandingof suchdecaysusesour knowledgeof the weak interaction
in perturbationtheory calculationstogetherwith phenomenologicalmodelsfor the long
distanceeffectsof thecolor interaction.In the future it is hopedthat latticegaugetheory
numericalcalculationsof thecolor interactionwill achieve sufficient accuracy asto make
phenomenologicalmodelsunnecessary. Thesedo not useperturbationtheoryto calculate
color interactioneffects.
Feynman Diagrams . A Feynmandiagramis a representationof thewayparticles
interactwith oneanotherby theexchangeof bosons.They areextremelyusefulin making
perturbationtheorycalculationsbecauseeachrepresentationis associatedwith formalrules
for assigningvertex couplings,propagatorterms,etc.
%e
e
e
e
Figure 1.3: Feynman diagram for a weak processmediatedby a virtual bosonexchange.
1.2 Conservation Laws
Amongthemostsignificantaidsto understandingwhattakesplacein theuniverseare
theconservationlaws,giventhatthey identify thosetypesof interactionsthatcannotoccur.
`1.2 ConservationLaws 11
It is a trustworthy rule of thumbin physicsthatwhatever thing is not definitelyprohibited
will occur.
In classicalmechanics,the invariantsare the conservation of the electric charge,
the conservation of linear momentum,the conservation of energy and mass,and the
conservationof the angularmomentum.In thequantumworld, thereareotherinvariants
aswell that arepreserved in sometypesof interactionsbut not others. For example,the
isotopicspin is preserved in thestronginteractionsbut not in interactionsconcerningthe
electroweakforce.
1.2.1 Conservation of Angular Momentum and Helicity Suppression
Thehelicityof aparticleis definedasits spinprojectionalongits directionof motion.
A particlewith spinvector N andmomentump hashelicity
¡ J N �£¢n N n¤n ¢ n (1.4)
For spin �o thereareonly two possiblehelicity states,tI� . If thehelicity of a particle
is positive, it is called right-handed,and left-handedif the helicity is negative. The
correspondingstatesaredenotedas ¥§¦ and ¥©¨ respectively.
Due to the natureof the weak interaction, lepton statesfrom weak decaysare
predominantlyleft-handed. This helicity predominanceis strongerthe lower the mass
of the lepton. In the caseof the neutrino, it is total. Only left-handed � have been
observed. For otherleptons,the termhelicity suppressionis used.This meansthat right-
`1.2 ConservationLaws 12
handedleptonsandleft-handedantileptonsareproducedwith muchlowerprobability. The
D hasspin zeroso the two muonsin D ��� ���� musthave oppositespin to conserve
angularmomentum.But they alsohaveoppositemomentumsothey havethe h3),j�& helicity.
Howeveroneis anantilepton.Thus,theweakdecayD ��� � � is helicity suppressed.
1.2.2 Conservation of the electrical charge
Whichever reaction,theentirechargeof all theparticlesbeforethereactionhasgot
to beidenticalto thewholechargeof all theparticlesfollowing thereaction.For instance,
in a betadecay, a neutrondecaysto form a protonandan electronplus an antineutrino.
This beginswith a zerochargeandfinishesonepartnegativeandonepartpositivecharge.
Figure1.4shows thecontributionof chargefor eachfermioninvolvedin thereaction.
+ z � & � ��� (1.5)
with charges: 0 +1n & n -1
n & n 0
W n
p
���
e
Figure1.4: A chargedcurrentinteraction.Theneutronhasaquarkcontentof udd andits charge is ª5«9¬i,® ¯<®<°²±7¬i,® ¯2®<°²±7¬i,® ¯<®�³$´"® ¯<® . Theprotonquarkcontentis uud andits charge is ªµ«9¬i,® ¯<®¶ª·«9¬i,® ¯<®<°¸±7¬i,® ¯<®,³¹±=® ¯<® . Theelectronhas °µ±=® ¯<® aschargeandthe� � has ´"® ¯<® ascharge.
`1.2 ConservationLaws 13
Conservationof electricalchargepreventsthecolor forcefrom materializingquarks
suchasu andd from energy (thetotal charge is not zero). However, a uc materialization
would conserve charge. Yet, suchmaterializationis not seenin color forcereactions.The
propertiesof thecolor forcedictatethat thequarkandantiquarksinvolvedmustbeof the
sameflavor. It is not possibleto materializea u anda c by the color force but you can
materializeau andau.
Flavor-changingprocessescannot occurvia the color force or the electromagnetic
force, only via the weak force. Sincethe color andelectromagneticforcessatisfymore
conservation laws thanthe weakforce,many particlescanonly decayweakly. This fact
allowsusto studytheweakforceby studyingthosedecays.
1.2.3 Lepton Number Conservation
A leptonnumberis a tagusedto indicatewhichparticlesareleptonsandwhichones
arenot. Every leptonhasa leptonnumberof � , which is split into threedifferent“values”:
Aº� , A§� and A§� . On the otherhand,every antileptonhasa leptonnumberof� � . Other
particleswill havea leptonnumberof�. A»� , A§� and A§� arecalledleptonfamily numbers.
Both leptonnumberandleptonfamily numberarealwaysconservedaccordingto the
StandardModel. To understandthedifferencebetweenthem,considertheprocess
D � � & � (1.6)
wheretheleptonnumbersandleptonfamily numbersare:
`1.2 ConservationLaws 14
A : 0 �1 1Aº� : 0 0 1A§� : 0 �1 0
For this reactionwe canseethat the leptonnumberis conservedbut not the lepton
family number. It is due to the fact that for this processin the initial statewe do not
have any leptonbut in thefinal statewe have two leptonsof differentfamilies.Thedecay
D ��� | &3¼ is aLeptonFamily NumberViolationprocesswhich is notpermittedto occur
within theStandardModel.
A processpermittedby theStandardModel is
�9� � & � � �9� (1.7)
wheretheleptonnumbersare
A : 1 1 1 1A»� : 0 1 0 1Aº� : 1 0 1 0
For the processabove the lepton family numberaswell as the leptonnumberare
conserved.
1.2.4 Baryon Number Conservation
Analogousto leptons,thebaryonnumberis a tagusedto indicatewhichparticlesare
baryonsandwhich onesarenot. All baryonshave a baryonnumberof � . Eachantibaryon
`1.2 ConservationLaws 15
hasa baryonnumberof� � . Every quarkhasa baryonnumberof �9B2½ . Antiquarkshave
a baryonnumberof� �9B2½ and,therefore,eachmesonhasa baryonnumberof
�. Every
leptonhasa baryonnumberof�. In all reactions,thetotal baryonnumberof theparticles
prior to the reactionis the sameasthe total baryonnumberfollowing the reaction. For
example,
+ z � & � ��� (1.8)
wherethebaryonnumbersare
B: � � � �Lepton and baryonnumberconservation are ad hoc assumptionsin the Standard
Model in contrastto otherconservationlawssuchaschargeconservationwhicharerelated
to the natureof the interactions. This brings up the possibility that lepton and baryon
numberconservationmaybeviolatedin some(asyetunobserved)processes.
1.2.5 Flavor Conservation
In particle physicsanotherfrequentlyusedterm is flavor. Quarkshave different
flavorssuchas:up,down, strange,etc. Therearesix flavorsof quarks.Theflavor number
for eachquark in the order of its massis the following: ¾ J � , _ J � � , H J � ,¿ J � � , ÀÁJÂ� and ÃÄJ � � .Thestrongandelectromagneticinteractionsconservethequarkflavor numberswhile
the weak interactiondoesnot. Flavor changeoccursmostly throughthe charged weak
`1.3 CabibboTheory 16
interactionssincetheflavor-changingneutral-currentweakinteractionis suppressedby the
GIM mechanism.This is explainedin detail in Section1.4.
1.3 Cabibbo Theory
Transitions between quarks of different flavors occur only due to the weak
interaction. For the electromagneticandcolor interactionsall familiesbehave the same.
The only differencesbetweenthem are their masses. When we talk about quarksof
different flavors, we are referring to masseigenstates.However, thesestatesare +�63-eigenstatesof thetotal Hamiltonian(which includestheweakinteraction)because,if they
were,flavor wouldbeconservedin weakdecays.
The weakdecayprocesscanbe describedasfollows. At -ÅJ � , a quarkis created
in a mass(flavor) eigenstatevia an electromagneticor color interaction. This statecan
bewritten asa linearsuperpositionof weakeigenstateseachof which hasa well-defined
anddifferentlifetime. It is saidthat thequarkis in a “mixed” state.As time evolves,the
weakeigenstatemix changesbecauseof thedifferentlifetime evolutions.At a latertime - ,thestatehasacquiredcomponentsof othermasseigenstates(otherflavors)thusexplaining
flavor transitions.
Only quarkswith the samecharge canmix. Due to the explicit form of the weak
interaction, it is only necessaryto considermixing of the� �9B2½ quarks(this induces
transitionsinvolving the� � B2½ quarksalso). Themixing is treatedmathematicallyvia the
unitaryCabibbo-Kobayashi-Maskawa(CKM) [5,6] matrixwhoseelementsare
`1.3 CabibboTheory 17ÆÇÇÇÇ
Èd ÉsÉb É
ʶËËËËÌ J
ÆÇÇÇÇÈÍ�Î7ÏÐÍ�ÎZÑÒÍÓÎ7ÔÍFÕ@Ï ÍFÕ�Ñ Í�Õ@ÔÍ c Ï Í c Ñ Í c Ô
ʶËËËËÌÆÇÇÇÇÈ
d
s
b
ʶËËËËÌ (1.9)
In thelenguageof Feynmandiagramsthechargedweaktransitionoccursat a vertex
suchasthatshown in Figure1.5whereq is a� � B2½ quarkandq É is a
� �9B2½ quark.
%' q q É
Figure1.5: Feynmandiagramfor aweaktransition.
Mixing is treated mathematicallyby taking the coupling at that vertex to be
proportionalto Ö Í�×Ø×lÙ where Ö is the universalweak-couplingstrengthwhich is the same
for all threefamilies. Notice the reasonfor the namesof the CKM matrix elementsÍ�×Ø× Ù
sincethey areassociatedwith thecouplingbetween� � B<½ and
� �9B2½ quarks.For example,
amplitudesfor theprocessesb c andb u areproportionaltoÍ�Õ@Ô
andÍ�Î7Ô
, respectively.
TheCKM matrix is in generalcomplex but, dueto theunitarity condition,it canbe
shown thatthemostgeneralform containsonly four realparameterswhich canbechosen
to be threeangles( Ú � o9Û Ú o � Û Ú � � ) andonephaseÜ � � [7–10]. The anglesÚ¶ÝßÞ arerelatedto
the mixing betweenthe families k and à . Defining H�ÝßÞáJâ476<hÅÚ¶ÝßÞ and¿ ÝßÞ Jãh3ky+²Ú¶ÝßÞ this
parametrizationgives:
`1.3 CabibboTheory 18
Í JÆÇÇÇÇÈ
H � o H � � ¿ � o H � � ¿ � � & � Ý�äØåçæ� ¿ � o H o � � H � o ¿ o � ¿ � � & Ý�äØåçæ H � o H o � � ¿ � o ¿ o � ¿ � � & Ý�äØåçæ ¿ o � H � �¿ � o ¿ o � � H � o H o � ¿ � � & Ý�äØåçæ � H � o ¿ o � � ¿ � o H o � ¿ � � & Ý�äØåçæ H o � H � �
ʶËËËËÌ (1.10)
Thosematrix elementswith a simple form can be directly measuredin a decay
processandare found in the first row and third column. H � � is known to deviate from
unity only in the sixth decimalplace. Therefore,Í�Î7Ï JâH � o3Û Í�ÎZÑ J ¿ � o3Û Í�Õ@Ô J ¿ o � andÍ c Ô JèH o � to anexcellentapproximation.The phaseÜ � � lies in the range
�·é Ü � � � � E ,
with non-zerovaluesbreakingCPinvariancefor theweakinteractions.
An alternative approximateparametrizationis given by Wolfenstein[11] usingthe
fact that¿ � o*ê ¿ o � ê ¿ � � areall small. Thesizeof theCabibboangle1 is setas ë�ì ¿ � o
andthentheotherelementsareexpressedin termsof leadingpowersof ë up to the third
order:
Í JÆÇÇÇÇÈ
� � ë o B � ë í�ë � p�î � klï s� ë � � ë o B � íðë oíðë � p:� � î � kyï s � í�ë o �
ʶËËËËÌ (1.11)
having í Û î and ï asrealnumbersthatareof orderunity.
Transitions between quarks for the two family case are described by the
transformation1 ñóò�ôöõ¤÷5ø7ù�ú�ò�ôöõû÷5ø^üIý þ9ÿ���� whereø^ü is theCabibboangle( ø^ü ý�þ=ÿ ��� rad).
`1.3 CabibboTheory 19ÆÇ
È d ÉsÉʶËÌ J
ÆÇÈ 4762h5Ú�� hikl+ Ú��� hikl+ Ú�� 476<h�Ú��
ʶËÌÆÇÈ d
s
ʶËÌ (1.12)
In this approximation,the transitionc s andu d areproportionalto 4Z6<h o ��while the transitionsc d and s u are proportionalto h3ky+ o �� . This is a good
approximationsince¿ � o Juh3ky+áÚ�� ê ¿ o � ê ¿ � � .
A relation betweenthe two smallestelementsof the CKM matrixÍÓÎ7Ô
andÍ c Ï is
obtainedby applyingtheortogonalityconditionto thefirst andthird columns:
ÍÓÎ7϶Í��Î7Ô � Í�Õ�Ï7Í��Õ@Ô � Í c ϶Í��c Ô J � (1.13)
whereeachtermin thesumis of theorder ë � . In theparametrizationgivenabove,ÍFÕ@Ô Û ÍFÕ@Ï
andÍ c Ô arereal,andby using
ÍÓÎZÏ } Í c Ô }Ä� andÍ�Õ@Ï � � weobtain
Í �Î7Ôn Í�Õ@Ï7ÍFÕ@Ô n � Í c Ïn Í�Õ@Ï7ÍFÕ@Ô n JK� (1.14)
This equationis representedgeometricallyby an “unitarity” triangle [7] in the
complex plane. The lengthsof the two uppersidesareproportionalto themagnitudesof
theleastwell known elementsof theCKM matrix,ÍÓÎZÔ
andÍ c Ï asshown in theFigure1.6
TheStandardModelof CPviolationcanbetestedby measuringthesidesandangles
of theunitarity triangleto testwhetherthey really form a triangle.
a) Cabibbo-F avored . Thediagonalelementsof theCKM matrix aremuchlarger
`1.3 CabibboTheory 20
A
C BVs12 cb
ubV Vtd*
*
γ β
α
Figure1.6: Representationfor the CKM Matrix elements.An unitary trianglein thecomplex planeis formedby theCKM matrixelements� �Î7Ô , � c Ï , and � o � �Õ@Ô .
thantheoff-diagonalelements.Thustransformationswithin familiesaremuchmorelikely.
Transitionsproportionalto 4762h o �� areknown asCabibbo-Favored.
D� W
u
c
u
s
d
u �
K�
Figure 1.7: Feynman diagramfor a Cabibbo-Favored process. A charm quark istransformedinto a strangequark. The other transformationof a W into u and dquarksin thesecondvertex is alsoCabibbo-Favored.
If weconsidertheprocessin whichaD is decayinginto K � E shown in Figure1.7,
a charmquarkof the D particle is transformedinto a strangequark. This is Cabibbo-
Favored.In thesecondvertex theW is decayingintoaud. Thisis alsoaCabibbo-Favored
processbecauseu andd arequarksof thesamefamily.
`1.3 CabibboTheory 21
b) Cabibbo-Suppressed . Transitionsproportional to hikl+ o �� are known as
Cabibbo-Suppressed.Thesearetransformationsbetweenthefirst andsecondfamilies.
D� W
u
c
u
d
d
u �
�
Figure 1.8: Feynmandiagramfor a Cabibbo-Suppressedprocess. In this processacharmquarkis changedto a down quark. TheW decayvertex is Cabibbo-Favoredjustasin D � K ��� (Figure1.7).
In Figure1.8acharmquarkturnsinto adown quarkin a D processdecayinginto a
E E � . Thesecond =8.)R1�� -W vertex is aCabibbo-Favoredprocess.
c) Doub ly-Cabibbo-Suppressed . Processesin which two =8.),1�� -W vertices
areof theCabibbo-Suppresedtypeareknown asDoubly-Cabibbo-Suppressed.Thedecay
modeD K �E � is shown asanexampleof theseprocesses(Figure1.9).
D� W
u
c
u
d
s
uK�
�
Figure 1.9: Feynmandiagramfor a Doubly-Cabibbo-Suppressed process. For thisprocessboth vertices, the W- ��������� and the ��������� -W, are Cabibbo-Suppressedprocesses.
`1.4 GIM Mechanism 22
1.4 GIM Mechanism
The processD ��� ���� is a flavor changingneutral current processwhich is
suppressedin the StandardModel due to the GIM mechanism,proposedin a famous
paper[12] by Glashow-Iliopoulus-Maianiin which they explain thesuppressionof flavor-
changingneutral-currents.They suggestedthat,insteadof atriplet of quarks(u, d, s), there
weretwo doubletsof quarks(u, d) and(c, s). Thissuggestionrequiredanew quarkrespect
to theCabibbomodel,thecharmquark.Thispredictionwasmadebeforetheexperimental
discoveryof this quark.
Z
Z 0
0 +
+
θ(d cos + s sinθc c )
(d cos + s sinθc )θc
_
_u
u
c
c
θ )c
θc )
θ
θc
c
(s cos − d sin
(s cos − d sin
_
__
_
(a)
(b)
Figure1.10: Weakinteractionsin neutral-currentprocesses.(a)Feynmandiagramsfora neutralcurrentin thethree-quarkmodel. (b) Includingthecharmquark,extra termsareadded.
If wehavea ascarrierof aweakinteractionasshown in Figure1.10wehave that
`1.4 GIM Mechanism 23
theeigenstatesof theHamiltonianfor theweakinteractionare:
ÆÇÈ u
d ÉʶËÌ Û
ÆÇÈ c
sÉʶËÌ Û
ÆÇÈ t
b ÉʶËÌ (1.15)
In thethree-quarkmodel,mixing thed ands quarksby theCabibboangle,oneobtain
d É�J d 4Z6<h5Ú�� � s hiky+áÚ���� sÉ J s 4Z6<h�Ú�� � d hiky+áÚ�� (1.16)
whered É andsÉ areanorthogonaltransformof d ands. Thentheneutral-currentcoupling
will beof theform
¥ Í ¥ � J uu� p dd 4Z6<h o Ú�� � ss hiky+ o Ú�� s� �� !"$#�%
� p sd�
sd s h3ky+ ���4762h5��� �� !"$#�% �(1.17)
Experimentally, very small rates for flavor-changing( G ¿ J � ) neutral-current
processesareobserved but eq.(1.17)saysthat G ¿ J � shouldbe significant. This was
theproblemwith thethree-quarkmodelwhich wassolvedby thecharmquarkprediction.
Includinga doubletcorrespondingto thecharmquarkwe obtainfor theweak-interaction
neutral-currentmatrixelement
¥ Í ¥ � J uu�
cc� p dd
�ss s 4Z6<h o �� � p ss
�dd s hikl+ o ��� �� !"$#�%
� p sd�
sd�
sd�
sd s hikl+áÚ���4Z6<h5Ú��� �� !"$#�% �(1.18)
`1.5 RareandForbiddenDecays 24
wherethestrangeness-changingpartvanishesdueto theeffectof thecharmquark.To first
order, therearenostrangeness-changingneutral-currentsandsimilarly nocharm-changing
neutral-currents.Thiscancellationof flavor-changingneutral-currentprocessesoccursonly
in thefirst orderFeynmandiagrams(treelevel). Flavor-changingneutral-currentscanoccur
in the StandardModel throughhigherorderterms(loop diagrams)which aresuppressed
dueto thesmallnessof theweakinteractioncoupling.
1.5 Rareand Forbidden Decays
Rareandforbiddendecaysaredivided into threedifferentgroups: LeptonFamily
NumberViolating(LFNV), LeptonNumberViolating(LNV) andFlavor ChangingNeutral
Current(FCNC).
a) Lepton Famil y Number Violation. Processesof this type are forbidden
becausethey do not conserve the leptonfamily number. For instance,processessuchas
D � | & ¼ and_ & Ï(' Ñ*) +� �� | & ¼ and
_ & Ï(' Ñ*) +��.�� �&3 , where + is E or , and
the leptonsbelongto differentfamilies,areexamplesof this mode. Thesedecayswould
suggesttheexistenceof heavy neutralleptonswith non-negligible couplingsto & and � .
b) Lepton Number Violating. Sincethereis noexplanationin theStandardModel
for theleptonnumberconservation,we canexpecta violation at somepresentlyunknown
energy range.Processessuchas_ & Ï(' Ñ*) + �.-¶ /-¶ in which leptonscomefrom thesame
family andareof samechargeareexamplesof LNV modes.
c) Flavor-Changing Neutral-Current. In theStandardModel theneutral-current
`1.6 SpecialRelativity 25
interactionschangeof the flavor is at a very low level. Lower limits for charm-changing
neutral-currentare expected. Such decaysare expectedonly in second-orderin the
electroweakcouplingin theStandardModel. Processesin the_
systemreferto thedecays
D � ? ? � and_ & Ï0' Ñ1) + - - � , etc.Testfor charmFCNCarelimited to hadrondecays
into leptonpairs.
1.6 SpecialRelativity
We will go through a brief review of specialrelativity to establishnotation and
whateverpointsarenecessaryfor this work. If wehaveaparticlethatis moving anywhere
closeto light speed,therelationsof its energy andmomentumwill bedifferentfrom those
at a lowerspeed:
qz J32�j q4 Û (1.19)
and
5 J32.j�4 o Û (1.20)
where
2�J �6 � � { o and { J 44 Û (1.21)
`1.6 SpecialRelativity 26
4 is the velocity of light in vacuum,j is the massof the particleat the rest. From eq’s.
(1.19),(1.20)and(1.21):
5 o J p qz�4 s o � p�j�4 o s o Û { J zF45 Û 2�J 5j�4 o (1.22)
Note that if 4 J � Û 5 Jãj�4 o . Specialunits areusedin particlephysicssuchthat
4óJ � and 7 J � . Energy, momentumandmassarethenall in unitsof ~��7� , andtime and
lengthwill bein unitsof ~��7� ��� .Let us considerthe displacement,a four-vector which is generalizedby special
relativity asthedifferencebetweentwo “events”,whereboththespatialdisplacementand
timeof occurrenceareconsidered.
8í¹J p�4g- Û r Û ; Û(9 s JÄp@í Û í � Û í o¶Û í � s J p@í Û qí s Û (1.23)
Theinvariantdot productconceptis generalizedaswell
8í � 8ÃÄJ í à � í � à � � í o à o � í � Ã;:µJ í à � qí � qà (1.24)
This an invariant underLorentz transformationsfor any pair of vectors8í and
8Ãwhich transformlike “event” vectors. Energy andthree-momentumform sucha vector.
This four-momentumvectoris
8í¹J p 5 Û z.< Û z.= Û z?> s (1.25)
Considerthe situationwherea primedsystemmoveswith the constantvelocityq4
`1.6 SpecialRelativity 27
with respectto the unprimedsystem.q4 is in the directionof
� r andwe will use { for
thefractionalvelocity of thecoordinatesystem.Considera systemwith momentumqz and
energy5
. TheLorentztransformationequationsare:
zÓÉ= J z.= Û zÓÉ > J z?>�� (1.26)
z É< J@2©p�z.< � { 5 s � (1.27)
5 É J@2©p 5 � { z�< s (1.28)
Undera Lorentztransformation,thescalarproductof two four-vectorsis invariant.
Momentumandenergy form a four-vectorjust like positionandtime. The ? kX/�&¶-:kyj�& ( � )of an unstablemoving particle is calculatedto be equalto 2F� where � is the particle’s
lifetime asmeasuredin its restframe.
TheLorentzinvariantmassof any systemis its energy in a framewhereit is at rest.
In our analysiswe aretrying to identify decayprocessesof a short-livedparticleinto two
oppositechargedparticles. The decayingparticle travels a negligible distancemaking it
impossibleto determinethe identity of the particlebasedon its trajectory. Therefore,an
importantquantityin our work is theLorentzinvariantmasswhich is usedto determineif
two particles(whosemomenta,qz � and
qz o areknown) arethedaughtersfrom thedecayof a
particleof massA into productsof massj � and j o .Conservationof energy andmomentumis usedin orderto reconstructA o from the
`1.6 SpecialRelativity 28
candidatedecayproducts:
qz?BÄJ qz � � qz oiÛ 5 BÂJ 5 � � 5 o � (1.29)
WecandefinetheLorentzinvariantquantity
A o� o JÄp 8z � � 8z o s o then, A o� o JÄp 5 � � 5 o s o � p qz � �Qqz o s o (1.30)
where5 Ý�J 6 jáÝ o � z�Ý o .
Noticethat A � o canbecalculatedfromtheassumedmassesj � , j o andthemeasured
momentaqz � , qz o in any frame. A � o is calledtheinvariant( j � , j o ) massof thetwo-particle
system.If thesetwo particlesdid indeedhavethemassesj � and j o andif they did indeed
comefrom the decayof a particle A , then A � o will be equalto A within experimental
errors.Thecalculationof invariantmassis amajortool in identifyingdecays.
If we considertheD ��� � � process,j � J j o J j�} � �DC��FE�E MeV [4, p.23]
andqz � J qz�� å , qz o J qz��HG , thus
5 o�=å J j o�9å � z o�9å Û 5 o� G J¹j o� G � z o� G (1.31)
andtheLorentzinvariantmassfor theD will begivenby
A o�7� J � j o� � �.IKJ pez o�9å � j o� s p�z o� G � j o� s � zÓ�9å zÓ� G 4762h5Ú¶�7��L (1.32)
wherez��=å Û zÓ� G and Ú¶�Z� aremeasuredquantities.
`1.7 ConfidenceIntervalsandConfidenceLevels 29
1.7 ConfidenceInter valsand ConfidenceLevels
Severalconceptsin statisticswill beusefulin our work. Confidencelimits [13] refer
to thosevaluesthat definean interval rangeof confidence(lower andupperboundaries)
for an unknown parameter. If independentsamplesare taken repeatedlyfrom the same
populationanda confidenceinterval calculatedfor eachsample,thena certainpercentage
of theintervalswill includetherealvalueof theparameter.
A differentconceptis thatof ConfidenceLevel. This statisticis usedin hypothesis
testing where one calculatesa M o from the deviations of the observed data from the
hypothesis.If the hypothesisis correct,high valuesof M o areunlikely. The confidence
level is definedas
H0A J NO�PRQTS;U1V�WYXZV (1.33)
whereQTS[U\VDW is the ]_^ probabilitydensitywhichdependsonthenumberof observations, .
Theconfidencelevel (CL) is arandomvariable.If thehypothesisis correct,it hasauniform
probabilitydensityandall valuesof CL areequallylikely. This mayseemto make it not
very useful. However, if the hypothesisis not correct,high valuesof ] ^ (corresponding
to low valuesof CL) will be likely. Onecanstatisticallydifferentiatethecaseswherethe
hypothesisis fulfilled by requiringaminimumCL value.
Chapter 2
Previous Works
2.1 FCNC DecayTheory
The fundamentalwork concerningFCNC decaysis the famouspaperby Glashow,
Iliopoulos & Maiani in which they proposetheexistenceof thecharmquarkanddiscuss
how it canexplain the suppressionof FCNC decays[12] throughwhat is now calledthe
GIM mechanism.Thishasbeendiscussedin Section1.4.
With respectto D acb d$efdhg which is a charmFCNC decay, the GIM mechanism
leadsto suppressionof thetree-level diagramssuchasFigure2.1.a.In a 1997paper[14],
Pakvasapresentedthe latestcalculationfor this process.He found that theshortdistance
effects are dominatedby internal s-quark loop diagramswhich are suppressedby the
smallnessof thes quarkmassandby helicity considerations(Section1.2.1).
30
i2.1 FCNCDecayTheory 31
j ac u
d gd$e
(a)SuppressedDiagramof First Order k
j ac u
d gdle
(b) PenguinDiagram
X?mon�m0pW g
qW e
u
c
d g
d e
(c) Box Diagram
Figure2.1: Feynmandiagramsfor thedecayD asrut e t g . (b)and(c)aresecondorderdiagramswith quarkmassesd, s, andb within the loopsthatgive a rateproportionalto vxwy , being v y themassof thestrangequark.(a) representsasuppresseddiagramoffirst order.
Pakvasacalculatedthe shortdistanceeffectsto be of the orderof z�{ g?|\} . However,
the long distanceeffectsare large bringing Pakvasa’s calculationof the StandardModel
branchingratio to zH{ g?|\~ for D a b d e d g .
Thelong distanceeffectsaredueto intermediatestatessuchas � a , � a , � a , ï , ïZ� or
�l� and � � a [14].
i2.2 PreviousSearchesfor RareCharmDecays 32
2.2 PreviousSearchesfor RareCharm Decays
Many experimentshave reportedlimits for the branchingratio of the rare decay
D a b�d e d g asis shown in Table2.1publishedby ParticleDataGroupin 2000[4, p.566].
Table2.1: Chronologicalbranchingratiosfor the ������z weakneutral-currentdecayD acb d$efd�g .
Value Experiment Comment Year(atCL 90%) Name�������x� z�{ g�� E791 � g 500 ����� 2000� z ����� z�{ g�~ E789 � nucleus800 ����� 2000����� z � z�{ g�� Beatrice � g Cu,W 350 ����� 1997�������x� z�{Zg�� E771 � Si, 800 ����� 1996�3���F��� z�{Zg�~ CLEO ��el��g¡ £¢ (4S) 1996��¤������ z�{ g�� Beatrice � g Cu,W 350 ����� 1995�����F��� z�{ g�~ E653 � g emulsion600 ����� 1995�3��� z � z�{ g�~ E789 ¥ �.� z§¦ �.�F¨ events 1994��¤�� { � z�{ g�~ ARGUS � e � g 10 ����� 1988� z � z � z�{ g�~ SPEC �?© g W 225 �ª��� 1986�3���F��� z�{Zg w EMC Deepinelasticdhg N 1985
In this sectionwe will discussthe most recentexperiments. We will alsodiscuss
theFOCUSpredecessorexperiment,E687,which searchedfor otherFCNCcharmdecays
but not D a b d e d g . The first experimentis the onethat hasobtainedthe lowest limit,
theCERN1 WA92 BEATRICE experiment.Thenwe presentFermilab’s2 E791andfinally
E687.1CERN-OrganisationEuropeennepourla RechercheNucleaire2Fermilab-FermiNationalAcceleratorLaboratory
i2.2 PreviousSearchesfor RareCharmDecays 33
2.2.1 The Hadroproduction WA92 Experiment
During the 1992-93 data-taking period of the BEATRICE Collaboration, the
hadroproductionWA92 experiment[15] at CERN SuperProtonSynchrotronwascarried
out. In the interactionsof 350 ������� e particlesin a W target during 1992(W92 runs),
charmedparticleswereproducedaswell asin theCu targetof 1992and1993(Cu92and
Cu93runsrespectively). Theapparatuswasmadeup of a 2 mm thick target followedby
a beamhodoscopeandby an in-target counter(IT), a high-resolutionsilicon-microstrip
detector(SMD), a largemagneticspectrometerandamuonhodoscope.
Candidateevents D a b�d e d g were required to be in a mass range between
z �F¨ {«¥�z �F¬D� ����� . Dimuoniceventswererequiredandwereselectedby a dimuontrigger.
Eventsthat camefrom target interactionsaswell asthosein IT counterswereaccepted.
Dimuoniceventsassociatedto a simplesecondaryvertex weredifferentiatedfrom events
where both muonscamefrom different secondaryvertices. The D a b K g �_e decay
processwasthenormalizingmodewhichwasdetectedin runsof Cu(W)applyingthesame
selectioncriteriafor thedimuoniceventsexcepttheonesfor leptonidentification.
Monte Carlo simulation generatorsused Phytia 5.4 and Jetset7.3 [16–19] to
understandthe hard processesand quark fragmentation. Fluka [20, 21] was used to
determinethe characteristicsof all other interactionproducts. Monte Carlo simulations
generatedevents that containeda pair of charmedparticles. Dimuonic eventscoming
from the Monte Carlo simulatedD a b�d e d g decayswere treatedin the sameway as
the experimentaldatapermittingthe determinationof the efficienciesandthe acceptance
ratios.
i2.2 PreviousSearchesfor RareCharmDecays 34
No D ab d$efdhg candidatewasfound [15]. This resultled to anupperlimit on the
branchingfraction B(D a b�d e d g ) of �.� z � z�{ g�� at 90% confidencelevel. Systematic
errorson thewholeanalysisproceduredid notsignificantlyalterthis value.
2.2.2 The Hadroproduction E791Experiment
Eventscollectedin theE791experiment[22] wereproducedby a500 �ª���®�$e beam
in five target foils. This was a hadroproductionexperimentin which track and vertex
reconstructionwereprovidedby 23 silicon microstripplanesand45 wire chamberplanes,
plus two magnets. Muon identification was obtainedfrom two planesof scintillation
counters. The experimentalso includedelectromagneticand hadroniccalorimetersand
two multi-cell Cerenkov countersthat provided ��¯�� separationin the momentumrange� ¥ � {[����� .
In theanalysisstage,to separate“good” eventsfrom backgroundnoise,a separation
of the productionvertex from the decayvertex wasrequiredby morethan z �±°?² , where°?² is thecalculatedlongitudinalresolution.Also thesecondaryvertex wasrequiredto be
separatedfrom theclosestmaterialin the target foils by morethan ��°´³² , where °´³² is the
separationuncertainty.
E791usedasaMonteCarlosimulationgeneratorPythia/Jetset[23,24] andmodeled
theeffectsof resolution,geometry, magneticfields,multiple scattering,interactionsin the
detectormaterial,detectorefficienciesandtheanalysiscuts. As normalizingmodein the
studyof theraredecaysD a b d e d g , D a b � e � g andD a b�d$µl��¶ , theCabbibo-Favored
modeD a b K g � e wasused.
i2.2 PreviousSearchesfor RareCharmDecays 35
Before the branchingfractionswere calculatedfor eachof the channelsstudied,
E791 useda p�· ©1¸ X analysistechnique. All the eventswithin a ��¹»º rangearoundthe
D a masswere“masked” in orderto avoid biasin thecut selectiondueto thepresenceor
absenceof a possiblesignal.Thecut selectionwasbasedon thestudyof eventsgenerated
by Monte Carlo for signal events and real data for backgroundevents. Background
eventswerechosenandstudiedin masswindows �¼¹½º beforeandafter the signalarea.
The signal areawas chosenas z �F¨��¾� ¹ U1¿ a W � z �F¬ { . The 90% upper limit was
calculatedusing the methodof Feldman& Cousins[25] to accountfor background,
and then correctedfor systematicerrors by the methodof Cousins& Highland [26].
The upperlimit wasdeterminedfrom the numberof candidateeventsand the expected
numberof backgroundeventswithin thesignalregion. Only afterthecutswereoptimized
was the signal areaunmasked and then the numberof eventswithin the window was
`«À*Á y �£`º�Ã�Ä�Ū`ÂÆÇà y\ÈYÉ Å�`«Ê�ËTÁ , whereº�Ã�Ä correspondsto thenumberof “good” candidates
of D a b d e d g , `«ÆÇÃ y1ÈÌÉ correspondsto hadronic decayswith pions misidentified as
muons,and `«Ê�ËTÁ correspondsto combinatoricbackgroundarisingprimarily from false
verticesandpartially reconstructedcharmdecays.
No evidencefor the raredecayD a b�d e d g wasfound [22]. The 90% confidence
level branchingfraction limit was ������ z�{ g�� . Systematicerrorsin this analysisincluded
statisticalerrorsfrom thefit in thenormalizationsample;statisticalerrorsonthenumberof
MonteCarloeventsfor both thenormalizingmodeandtheprocessstudied;uncertainties
in thecalculationof MisID background;anduncertaintiesin therelativeefficiency for each
mode.
i2.2 PreviousSearchesfor RareCharmDecays 36
2.2.3 The Photoproduction E687Experiment
Production and decaysof charm particles from high energy collisions with a
high intensity photonbeamand a beryllium target were studiedby the FermilabE687
experiment[27] usingamultiparticlespectrometerfor particleidentificationandvertexing
of charged hadrons and leptons. From the Tevatron beam, a photon beam from
bremsstrahlungof secondaryelectrons( Î*ϱÐ[� 350 ����� ) hit a beryllium target. Charged
particleswere tracedby four silicon microstrip detectorstations. Thosedetectorswere
very efficient for separatingprimaryandsecondaryvertices.Chargedparticlemomentum
was determinedfrom deflectionsin two analysismagnetsof oppositepolarity with five
stationsof multiwire proportionalchambers.In orderto identify electrons,pions,kaons
and protons, three multicell thresholdCerenkov counterswere used. There were two
electromagneticcalorimeters.Particleswhich passedtheaperturesof both magnetswere
detectedby the innercalorimeterwhich coveredthe forward solid angle. Thoseparticles
thatonly passedthefirst magnetweredetectedby theoutercalorimetercoveringtheouter
annularregion. Two sectionsof thick steelmuon filters divided four proportionaltube
planeswhich wereusedto identify muonsin theforwardsolid angle[28].
The selectionof candidatesfor D e b Ñ?µ · ¶ · e and D e b � g � e � g startedby
making three-trackcombinationsof an event with the right particle identification. The
combinationof theresultingthreetrackswerefit to acommonsecondaryvertex.
In orderto detectmuonswith amomentumhigherthan � �ª��� , planesof scintillator
materialin additionto four stationsof proportionaltubes,altogetherknown astheMuon
i2.2 PreviousSearchesfor RareCharmDecays 37
System,wereusedby theE687Collaboration.A muonConfidenceLevel wascalculated
whenclassifyingcandidatemuontrajectories.At least1% wasrequiredfor theCL.
In orderto identify electrons,informationof tracksassociatedwith electromagnetic
showersin eitherof thetwo electromagneticcalorimeterswasusedif they wereconsistent
with theelectronhypothesisin theCerenkov counters.
Thesensitivity to themeasuredbranchingratioswasoptimizedfor eachdecaymode
studiedin E687. This sensitivity dependedon the efficiency relative to the normalizing
mode.ThiswascalculatedusingaPythia[29] MonteCarlosimulationgeneratoralongwith
a detailedspectrometersimulation,the amountof normalizingmodeevents(determined
from a fit to the invariantmassplot), andthe amountof backgroundin the signalregion
(determinedusingsidebandsbeforeandafterthedefinedsignalregion).
Monte Carlo signalshapesandbackgroundshapesderived from datawereusedto
calculatethe90%CL upperlimit onthenumberof signaleventsin thecorrespondingmass
plot. Misidentificationof hadronswastheprimarybackgroundin theE687analysis.Three
track combinationswith no leptonidentificationwereusedto determinethe shapeof the
background.The probability of misidentifyinga hadronasa leptonwasintroducedasa
weighton eachleptoncandidate.
No evidencefor the fourteenexclusive modesof rare and forbiddendecayswas
observed in this experiment [27]. However, 90% CL upper limits on their absolute
branchingfractionsin the U ¬ ¥ � { W � z�{�g�~ rangeweredetermined.
i2.3 FOCUS 38
2.3 FOCUS
2.3.1 Intr oduction
Sincehadronswith quarksfrom the ��ÒHÓ and ��ÔÕÓ family exist only rarely in nature,
we needa way to producethem copiously to study them effectively. The methodof
photoproductionin the experiment E831 [30] that took place in the Fermi National
AcceleratorLaboratory (Fermilab) near Chicago, Illinois was used to createparticles
with the charmquark. The E831 experiment,or FOCUS(“FotoproductionOf Charm
with an Upgraded Spectrometer”), was an upgradeversion of the E687 experiment
(Section2.2.3) and was a heavy–flavor (producesparticleswith � ÒHÓ family, or higher,
quarks)photoproductionexperimentlocatedat theWide BandPhotonArea.
Figure2.2: ProtonsynchrotronandbeamlinesatFermilab. TheFOCUSexperimentislocatedat theWide BandPhotonAreain theEastProtonArea.
i2.3 FOCUS 39
2.3.2 The FOCUSPhotonBeam
To photoproduceparticleswhich containthecharmquarkit is necessaryto generate
photonswith a high energy (about300 ����� ). At Fermilabthe initial beamis a 800 �ª���Tevatronprotonbeam[31, p.10].
Figure2.3: TheE831PhotonBeam(FOCUSCollaborationFigure).
As shown in Figure2.3, a photonbeamwasproducedafter threedifferentstages.
A neutralbeamproducedfrom 800 ����� protonsincident on a liquid deuteriumtarget
includeda variety of neutralparticlesalong with high energy photonsfrom � a decays.
Chargedsecondariesanduninteractedprotonswere taken out of the beamandabsorbed
by theberyllium targetdump(not shown). All theneutralsecondarytrajectoriesproduced
i2.3 FOCUS 40
weretransmitedwithin anapertureof 1.0mr horizontallyan0.7mr verticallywith respect
to the incidentprotonbeamtrajectories.To absorba smallerportion of thesesecondary
photonsa Deuteriumtargetmaterialwasused.
Thephotonsthatwereleft produceda � e � g pair dueto electromagneticinteractions
in passingthrougha 60%radiationlengthleadconvertor. Thepairswereseparatedfrom
theneutralsusingmomentumselectingdipolesandthenput togetherfocusingthemonto
anothertargetThis momentumselectionwasnecessaryto purify the � g and � e beamsby
reducingtheir charged hadronicbackgroundconsiderably. A neutraldumpwasusedto
removehadronicneutralparticles.
Finally, photonsthat camefrom bremsstrahlungwere producedfrom the focused
� e � g beamwhich was passedthrougha 20% radiation length lead radiator. Recoiled
electronsweretakenoutof thephotonbeamandabsorbedby anelectrondump.
2.3.3 Inside the FOCUSExperiment
FOCUSutilized a forward large aperturefixed-target multi-particle spectrometer,
to be explained in Figure 2.4, to measurethe interactionsof high energy photonson
a segmented Ö±��× target. This spectrometerhad excellent particle identification and
vertexing for chargedhadronsandleptons.
The experiment collected data during the z ¬�¬�� ¥ ¬�¤ fixed target run and is
investigatingseveral topics in charmphysicsincluding high precisionstudiesof charm
semileptonicdecays,Quantum Chromodynamicsstudiesusing double charm events,
i2.3 FOCUS 41
a measurementof the absolutebranching fraction for the D a meson, a systematic
investigationof charm baryonsand their lifetimes, and searchesfor D a mixing, CP
violation, rareandforbiddendecays,andfully leptonicdecaysof the ¿ e .
Ø�Ù1Ú ÛÝÜlÞ/Ù1ß1àáDâ ã�â äDâ å æç ÛÝèKéê�ë ÜìÛ1í1Ú ë ßîà
ã�â ï§â äDâ å æ Ø�Ù1Ú ÛðÜñ�ò Û1í\ÚFÜìßîé�è1óîàîÛ1Ú ë íä è ò ßÌÜ ë é�Û*ÚFÛ1Ü
ä ÛðÜFÛðà*ô*ß*õä ßÌÙÌàYÚ�Û*Ü æö Ü ë ó*ó1ÛÝÜ÷ ßîøÌß æ í1ßÌùÌÛ ã�â ïúâ ä�â
ã�â ïúâ ä�âä ÛÝÜ�Ûðà1ô*ß*õä ßYÙYàYÚ�ÛûÜÞ_è*óÌàÌÛ*Ú
ü�ý*þÝÿ��������ðÿ�� �
Þ_è*óÌàÌÛ*Ú
� ë�ò�ë í ß*àÂÞ ë íðÜ�ß æ Ú�Ü ë ù æö Ü ë óûóûÛðÜ ä ßûÙûàûÚ�Û Ü æö è*ÜìóÌÛ1Ú æ � à*à1ÛðÜñ�ò Û1í1ÚFÜ ßîé�è1óîàîÛ\Ú ë íä è ò ßîÜ ë é�Û1ÚFÛ*Ü÷ èûø Ü ß àä è ò ßÌÜ ë é�Û1ÚFÛ*Ü
ö Ü ë ó1ó1ÛðÜ÷ ßÌøÌß æ í*ßîùÌÛ
Þ/Ù1ß1à÷ ßÌøÌß æ í*ßÌùÌÛ
ÞfÙ*ß*à� ë ò Ú�ÛðÜç ÛÝèKéä è ò ßÌÜ ë é�Û1ÚFÛ*Ü�ÝÚ Ü è��ö Ù��YÛ æ
ö è*ÜìóYÛ*Ú� ë�ò ë í ßÕà ������������������� ���ç ÛKèÝéêDë ÜìÛ1í1Ú ë ßîà! "$#&%!�"'#&%!�"'#&%
Figure2.4: TheFOCUSSpectrometer(FOCUSCollaborationFigure).
In theFOCUSspectrometeraBeOtargetwasusedto produceprimaryandsecondary
particlescomingfrom interactionsbetweentheTevatronphotonbeamandprotonswithin
the target ( ¢ Å p b ( Å p). The targetwassegmentedin four separatepartsin orderto
maximizethepercentageof charmdecayswhichoccuredoutsidethetargetmaterial.
Two trigger countersandtwo silicon trackingsystemswere in the target region as
shown in the Figure2.4. Chargedparticleswhich emergedfrom the target weretracked
by two systemsof silicon microvertex detectors(SSD).Thefirst systemwasmingledwith
i2.3 FOCUS 42
theexperimentaltargetwhile thesecondwasjust downstreamof the targetandconsisted
of twelve planesof microstripsarrangedin threeviews. Altogetherthesetwo detectors
allowedto differentiateprimaryandsecondaryverticeswith high resolution.Two magnets
(M1 andM2) of oppositepolarity with five stationsof multiwire proportionalchambers
(PWC’s) wereusedto measurethemomentumof chargedparticlesusingthedeflectionof
the particleswhenpassingthroughthe magnets.TherewerethreePWCsbeforeM1 and
two betweenM1 andM2.
Threethresholdmulticell Cerenkov counterswereusedto identify electrons,pions,
kaonsandprotons[31, p.16].Two typesof calorimeterwereusedto determinetheparticle
energy. Thefirst calorimetersweretwo electromageticcalorimeters.Theinnercalorimeter,
a leadglassblockarray, coveredthecentralsolidangleanddetectedparticleswhichpassed
throughthe aperturesof both magnets.The outercalorimetercoveredthe outerangular
annulusdescribedby particlesthatpassthroughthefirst magnetbut not thesecond.The
secondcalorimetertypewasahadroniccalorimeterconsistingof iron andscintillatingtiles
whichwasusedprimarily in theexperimenttriggerbut wasalsousedto reconstructneutral
hadrons.
To identify muons,FOCUShada detaileddetectingsystem.Muonswereidentified
in eithera fine grainedscintilatorhodoscopewith aniron filter (coveringtheinnerregion)
or in an outersystemthat usedresistive platechambersandthe iron yoke of the second
magnetasa filter. Themuonsystemwasconstitutedby two sub-systems:theInnerMuon
Systemwhich wascomposedof 9 planesof scintillatormaterial,in which threewereused
astriggersof dataacquisition,andthreeiron filters [32,33], andtheOuterMuon System
i2.3 FOCUS 43
composedof resistiveplatechamberswhichdetectedmuonswith ananglegreaterthan125
milliradians.
In orderto analyzetheinteractionsin thespectrometer, theanalogsignalsfrom each
of thedetectorshadto bedigitalizedandrecorded.Thiswasthetaskof thedataacquisition
system.The FOCUSdataacquisitionsystemhadto dealwith input datain a numberof
differentdataformats,merge all this datainto onestream,andput the outputon 8 mm
magnetictapes.Informationfrom thedifferentdetectorswasreadout andthenthesedata
wereput onadataacquisitionsystembus.
Data for each event was taken off the data adquisition systembus and placed
into a temporarymemorysystem;datacould be written to and read from this system
simultaneously. Later on, the datawasbufferedon disk andwritten to tape. During the
run,FOCUScollectedmorethan z�{*) eventsandreconstructedmorethan zH{ � .The FOCUSCollaborationhaspublishedfour papersstudyingdifferent topics in
charmphysics[34–39].
Chapter 3
Objectives
As adeparturepoint for this thesis,thefollowing objectiveswereconsidered:
1. Thedeterminationof theprobabilityof misidentifyinga mesonasa muon(MisID)
in FOCUSasa functionof momentum.
2. The developmentof software for the searchfor the raredecayD a b d e d g using
blind analysismethods.
3. Thepreselection(from thelargeFOCUSdataset)of asmallandflexible setof event
candidatesfor D a b d e d g aswell asfor theD a b K g � e normalizingmode.
4. The determinationof the relative efficiency for the detectionof D a b d e d g and
D acb K g � e in FOCUS using Monte Carlo simulation to generateand analyze
appropriatesamplesof thedecays.
44
i3 Objectives 45
5. Thedeterminationof thebackgroundlevel in theD acb�d$e´d�g analysisby usingan
independentdatasetandthemisidmeasurements.
6. Thedeterminationof the90%confidenceregion for thebranchingratioof thedecay
D acb�d$e´dhg . In using the high statisticsFOCUSdataset and in optimizing the
analysiscuts, the objective is to improve existing measurementsandobtaintighter
(moreprecise)confidencelimits.
Chapter 4
Procedure
Thischapterwill describehow thesearchfor theraredecayD a b d e d g wascarried
out. Initially, will be discussedthe FOCUSdataformat and the reconstructionprocess.
Thenwill bedefinedeachoneof thevariablesusedin this study. Next, thevariousstages
of datareductionwill be discussed.The useof a normalizationmodewill be explained.
Basically the methodusedin this work is known asblind analysiswhich consistsin the
optimizationof theselectioncriteriaby minimizing thebackgroundin thedatasidebands.
Theuseof MonteCarlosimulationto determinedetectionefficienciesaswell asto study
backgroundswill bediscussed.Themethodologyleadsto thedeterminationof confidence
intervalsfor thevalueof theD ab dle´dhg branchingratio.
46
i4.1 DataFormat 47
4.1 Data Format
Wewill begin bydiscussingtheformof thedataproducedin theexperiment.For this,
it is advisableto befamiliarwith theconceptof an“event”. This refersto oneinstanceof a
photoninteractingin thetargetproducingvariousparticleswhichtravel throughsomeor all
of thedetectors.Someof theparticlesproduceddecayin or nearthetarget into secondary
decayproductswhich aretheonesactuallydetected.Theexperimentalideal is to usethe
informationfrom thedetectorsin orderto measurethecharacteristicsof thedecay. To this
end,it wouldbeeasierif thesetof detectorsignalscame+�¸ ·-, from thedecayof interestand
its associatedproductionpartners.In practice,this is never perfectlyachieved. The“raw”
dataconsistsof setsof signalswhich approachthe ideal of coming from one and only
onephysicalevent. Eachof thesesetsis calleda “raw dataevent”. Typically it contains
informationaboutthephysicaleventbut alsodetectornoiseandinformationaboutparticle
trajectoriesunrelatedto the physicalevent. Given the experimentaldifficulty, many raw
dataeventscontainno interestingphysicalevents.
Each raw data event is an independentunit of data which is “reconstructed”
separately. Reconstructionis the processwhereby the physical characteristicsof the
underlyingparticletrajectoriesareextractedfrom theraw detectordata.Theraw dataevent
is eitherdiscardedif it is not at all interestingor convertedto a “reconstructeddataevent”
which containsthevaluesof thephysicalvariablessuchasthenumberof trajectoriesthat
wereregisteredby thePWCchambers,themomentumof theparticles,etc. In FOCUS,the
reconstructionprocessis calledPass1(seeSection4.3.2).All data(raw andreconstructed)
is storedonmagnetictape.
i4.2 ReconstructionProcess 48
Reconstructedoutputtapesfrom Pass1containall kindsof physicalevents,instances
of amyriadof interestingprocessesall in thesametape.Therearethousandsof tapes.
The next stepis to group eventsof the samekind with the goal of having all the
eventsof theprocessunderstudybeingcontainedin a small datafile. Themostefficient
wayto do this is to do it in stepscalled“skims”. At eachskimlevel, strictereventselection
criteriaareappliedandthedatais separatedinto a largernumberof groupseachof which
containsasmalleramountof data.
4.2 ReconstructionProcess
To have a morepreciseideaof how the dataselectionprocessis donein FOCUS,
it is necessaryto be familiar with the trajectoryreconstructionprocess.A greatdealof
the information depositedin the FOCUSdetectorsare presentedashits which we have
to interrelateto be able to deteminewhich hits in differentdetectorscorrespondto one
particulartrajectory.
Betweenthetargetandthefirst magnet(M1) wefind theSiliconMicrostripDetector
(SSD) whichhasfour stations.After M1 thetrackingis doneby fivemultiwire proportional
chambers(PWCs). Eachoneof these,calledP0,P1,P2,P3andP4,hasfour planes.
Outsideof themagnetsthe trajectorieswill bestraightline segmentswhich will be
reconstructedin theSSDsandPWCs.Thetrajectorieswill bendin themagneticfieldsso
that the straightline segmentswill be different in the threeregions: (1) beforeM1, (2)
i4.2 ReconstructionProcess 49
betweenM1 andM2 and(3) afterM2. Thesewill bereconstructedby (1) SSD,(2) P0,P1,
P2and,(3) P3andP4.
Theprocessof SSDtrackreconstructionconsistsof findingtrajectorieswith theSSD
detector. Thestepsaregroupinghit stripsinto clustersof hits,findingeachoneof thethree
measurementdirectionsto projecttheclusters,andthen,combiningtheseinto trajectories.
ThenPWCtrajectoriesarefound.Theremustbehits in at leastthreechambers.Thosewith
hits in all five PWCsarereferredto as“tracks”. Theoneswith hits in only thefirst three
chambersarereferredto as“stubs”. TheSSDandthePWCtrajectoriesmustbeassociated
with oneanotherto beableto get thenecessaryinformationfor the reconstructionof the
charmdecay. To identify a “link” theSSDandthePWCtrajectoriesmustbeextrapolated
to thecenterof M1 andmusthaveconsistentslopesandintercepts.
µ
µ
µ+
−
−
µ+
Direction
Magnets
Beam
Muon Filters
PhotonBeam
Muon HodoscopeR.P.C. ’sOuter Muon
P.W.C. s
Target
3
2
1
4
Figure4.1: TrajectoriesthroughtheFOCUSspectrometer. A view of sometrajectorieswithin theFOCUSspectrometer. Thefirst andsecondtrajectoriesaredetectedby theOuter Muon System. Trayectoriestwo, threeand four aredetectedusing the InnerMuonSystem.
i4.3 DataSelection 50
Figure4.1 shows sometrajectoriespassingthroughthe spectrometer. Trajectory1
correspondsto a three-chambertrackwhile thethird andfourth trajectories(3 and4 in the
figure)correspondto five-chambertracks.Thesecondtrajectory(2) hashits in all thePWC
chambersexceptin P3. This trajectorycanstill bereconstructedin theregion afterM2 by
combiningtheinformationfrom P0,P1,P2andP4.
4.3 Data Selection
4.3.1 SelectionVariables
In thissectionwewill discussthedifferentvariablesusedfor thiswork. They will be
dividedinto two types:vertexing andparticleidentification.
a) Vertexing . UsingtheFOCUSmicrostripdetectorswe cantake advantageof the
factthatcharmedparticlestravel shortdistanceswithin thespectrometerbeforethey decay
while non-charmedparticlestendto eitherdecayimmediatelyat theproductionpointor to
passthroughthewholespectrometerwithoutdecaying.Thismeansthatfor charmeddecays
we canreconstructtwo distinctverticesin theevent: theprimaryor productionvertex and
thesecondaryor decayvertex. A diagramof aD a b�d$e´dhg decayandits verticesis shown
in Figure4.2. By reconstructingthe two vertices,we obtainseveral quantitieson which
we cando “cuts” to reducenon-charmedbackgrounds.A cut refersto aneventselection
criteriawhichis implementedby requiringacertainrangeof valuesfor aparticularvariable.
A vertexing algorithm[40] is usedto reconstructa D a b d e d g decaycandidate.It
i4.3 DataSelection 51
µ
µ
π
π
µ −
D+
0
+
−
γ
Σ +
0D
Primary
Vertex
Vertex
Secondary
−
Figure4.2: Productionanddecayverticesfor theD a r t e t g process.Figureshowstheschematizationof theproductionanddecayverticesfor theD asrut e t g process.
startswith two distinct trajectoriesthat meetthe requirementsto be calleda d e d g pair.
Thevertexing algorithmcalculatesa confidencelevel (CLS) for thehypothesisthat these
trajectoriescomefrom a commonpoint andfinds that point for the caseswhich have a
��.0/21 z43 .
Let ussaythatausefulvertex is found.TheputativeD a trajectorycanbecalculated
by addingthemuonmomentumvectors.Thenthealgorithmtriesto find theprimaryvertex
usingothertrajectoriesin theevent in combinationwith theD a candidate.Thealgorithm
beginsby takingall of thetwo-trajectorycombinations(oneof which mustbetheD a and
theothermustnotbeadaughtermuon)andfits themto avertex hypothesisagainrequiring
a minimumCL of 1% [41]. If thetrajectoriesareaccepted,thealgorithmtriesto put more
andmoretrajectoriesin thatvertex until theconfidencelevel fallsbelow 1%. Theresulting
primaryhasamaximumnumberof daughterswith aCL (CLP) greaterthan1% [41].
i4.3 DataSelection 52
Usinginformationcomingfrom thevertexing algorithmwecandefinethefollowing
variablesto beusedin theanalysisfor theraredecayD a b d e d g :
1. L/ 5 . Short lived hadronicbackgroundis removed by requiring a minimum
separationbetweentheprimaryandsecondaryvertices.Thedetachmentcut will requirea
minimum .s¯ ° where . is thedistancebetweenthetwo verticesand ° is theerroron that
measurement.Figure4.2 illustratesthis situationin which the errorson vertex positions
areshown asellipses.
2. Multiplicity of the Primary Vertex. This refersto the numberof trajectories
producedin the interactionthat occurswithin the target. Higher multiplicity primary
verticeshave lessbackground.
3. Vertex ConfidenceLevels. Eachvertex is formedby a reconstructionalgorithm
that hasan associatedconfidencelevel. Due to measurementerrorsthe trajectorieswill
not all intersectat the samepoint. However we can defineerror ellipsesfor eachpair
of trajectories(Figure4.2). We thencalculatetheprobability that,within their errors,all
candidatetracksareconsistentwith intersecting.Confidencelevelsabove1% arerequired
for minimal cutson thesevalues. Sometimesa larger cut is usedto rejectbackgrounds.
CLP standsfor theprimaryvertex confidencelevel andCLSfor thesecondary.
4. Primary and SecondaryVertex z Position. The FOCUStargetsare thin flat
wafersin thexy plane.Thustheprimaryvertex z position(ZVPRIM) becomesthecritical
variablein identifying goodeventswheretheprimaryvertex occurswithin the targetand
backgroundwhenit occursoutsidethe target. For certainlong livedcharmedparticles,a
large percentageof the decays(secondaryvertex) will occuroutsidethe target segments
i4.3 DataSelection 53
wheretherewill be lessbackgrounddue to interactionsof secondaryparticleswith the
target material. The secondaryvertex z position (ZVSEC) canbe usedto maximizethe
signalto noiseratio.
Dueto experimentalresolution,thesevariablesarenot useddirectly but insteadare
usedin a ratio with the experimentaluncertainty. For this purposewe will definethe
variable °76 which is the differencebetweenthe z vertex position and the z target edge
dividedby theuncertainty. Dueto thefour-parttargetsegmentationweneedto checkif the
vertex is occurringbetweenthesegmentsor within these.For °76x� { thevertex is inside
thetargetmaterialandfor °76 1£{ thevertex is outside.For theprimaryvertex we use °7698andfor thesecondary°76 P .
5. Vertex Isolation. The primary vertex isolation, ISOP, givesa large confidence
level whenoneof the trajectoriesin thesecondaryvertex is alsoconsistentwith beingin
the primary. Therefore,the lower the ISOPvalue, therewill be an unlikely chancethat
any charmcandidatetrajectorieswill be in fact associatedwith the primary vertex. The
secondaryvertex isolation,ISOS, is theconfidencelevel for a trajectorythat is not in the
primaryvertex to bein thesecondaryvertex togetherwith thepresumeddecaydaughters.
This variablecomesin handyfor rejectingbackgroundfrom highermultiplicity decaysby
requiringlow valuesof ISOS.
b) Partic le Identification . This analysisconcentrateson identifying two typesof
particles:muonsandhadrons.
1. Muon Identification . Muon identification is possiblebecausemuonsare the
only chargedparticleswhich canpenetratelarge amountsof material. The methodused
i4.3 DataSelection 54
to detectmuonsis to placechargedparticledetectorsbehinda large amountof shielding
material(typically steel).Thereareapairof differentkindsof detectionsystemsfor muons
in FOCUS.Theseare:theInnerMuondetectorandtheOuterMuondetector. Thefirst one
usescommonscintillatordetectorelementsandthesecondusesresistiveplatechambersto
detectpassingmuons.
For the inner region the systemfor muon detectionusesthreemuon scintillating
hodoscopestations:MH1, MH2 andMH3. The first two stationsregisterhits in x andy
whereasthethird stationregistershits in auvplanewhich is oriented¦ � {�: with respectto
thexyplane.
Thealgorithmof muonidentification[33] wasdevelopedat theUniversityof Puerto
Rico, MayaguezCampus,with thepurposeof calculatingfor eachtrajectorya confidence
level to decidethedegreeof reliability in theclassificationof thattrajectoryasarealmuon.
Thealgorithmis basedon the following considerations:themultiple Coulombscattering
and the geometryof the scintillatorsthat make up eachone of the planesof the muon
identificationsystem.It is describedin detailby Ramirez[33] andWiss[42,43].
Therearetwo waysamuonidentificationalgorithmcanfail. Oneis to fail to identify
a muonasa muon.Thischaracteristicis measuredasthealgorithm’sefficiency. Theother
is to identify anon-muonasamuon.Thatis measuredby themisidentification.Thesetwo
characteristicsarecorrelated.A goodalgorithmachievesabalancebetweenthem.
In themuonidentificationweareconsideringthefollowing variables:
i. Kaon Consistencyfor Muons. We canusethe kaonconsistency variable(see
i4.3 DataSelection 55
hadronidentification)to separatekaonsfrom muonsalso,sincepionsandmuonsbehave
almostidenticallyin theCerenkov system.Thenamegivenfor this variableis �¡¯�d .
ii. Inner Muons. For the Inner Muon System(IMU) the algorithm begins by
verifying if thecandidatetrajectoriesto bemuonsleave signalin thefive PWCchambers.
The trajectoriesare projectedinto the IMU planesand associatedhits are identified by
lookingwithin acirclearoundtheprojection.
In the IMU systemthere are six MH planes. There must be four hits in these
planes.We canproceedto calculate] ^ which givesus the trajectorycompatibility with
theMH hits. Therearetwo otherthingsthataretakeninto considerationwhenwe do this
calculation.Thesearethemultiple Coulombscatteringthatoccursinsidethe steelfilters
andthegranularityof thescintillatorpaddles.
Lastly, anisolationvariableis calculatedwhichcorrespondsto thelargestCL for any
othertrajectoryto havecausedthesameMH hits.
Becausetherearesteelfilters in theIMU thatcanstopmuonswith low momentum,
the numberof requiredplanesis reducedto two from the original numberof four for
candidateswith a momentumlessthan10 ����� . This will increasethe efficiency at low
momentum.
The efficiency for the muon identificationalgorithm is over 98% for momentum
above 10 ����� while misidentificationis of theorderof 0.5%. In-flight � b d q4; decays
candidateto themisidentificationrate.For thisdecaythepionandmuondirectionof flight
areverycloselymatched.
i4.3 DataSelection 56
Informationobtainedfrom theIMU is containedin thefollowing variables:
A) Number of Missed Muon Planes. The number of missed muon planes
correspondsto the numberof MH planeswithout signal. In this analysisthis variable
is calledMISSPL.
B) Inner Muon ConfidenceLevel. The muon identificationalgorithmcalculates
a ] ^ value for eachtrajectorythat passesa minimal requirementin the numberof MH
planeswith signals.Trajectoriesthatdo not correspondto muonstendto have higher ]_^becausethey havedifferentbehavior comparedwith muons[32,p.11].TheinnermuonCL
(IMUCL) is calculatedfrom the ] ^ .iii. Outer Muons. Muonsareidentifiedin theOuterMuonSystem(OMU) almostthe
sameway asin theIMU. Thereis aninternalmagneticfield in theiron shield(M2) which
cancausesomeproblemswhenit comesto OMU identificationbecauseit deflectsmuons.
Muons must be tracedthroughthe magneticfield in the M2 steel. Multiple Coulomb
scatteringsmearingmustbeappliedto thetracedposition.
TheOMU reconstructionalgorithmis efficient for muonsdown to Ï � ����� , the
rangeof muonsin theM2.
A) Outer Muon ConfidenceLevel. In thesamewayasis calculatedtheIMUCL, the
OMUCL is determinedcomingfrom themeasurementof the ]_^ valuefor eachtrajectory
thatpassedaminimal requirementwithin theOMU.
2. Hadron Identification . To identify hadronssuchas � µ , � µ , p ¯ p, FOCUS
developed an algorithm called CITADL (Cerenkov Identification Through A Digital
i4.3 DataSelection 57
Likelihood)[44] whichcalculatestheprobabilitiesfor severalhypothesisabouttheidentity
of a trajectory. The identitiesconsideredare electrons,pions, kaonsand protons(and
their antiparticles). The d$µ hypothesisis not consideredseparatelyfrom the �$µ since
the momentumrangeover which the two hypothesescanbe separatedis limited. In the
Cerenkov systemthe behavior of thesetwo particles,muonsandpions,is similar dueto
their closemassvalues.For eachof thefour hypothesis,CITADL calculatestheconedue
to the Cerenkov effect andthe averagenumberof photonsemitted. Poissonstatisticsare
usedto calculatetheprobabilitythattheobservedsizeof thesignalsin theCerenkov cells
is consistentwith theidentity hypothesis.
From the probability < , the algorithm calculates the negative log-likelihood
k =?>A@ U © WCB ¥ �EDGF <cà for eachparticlehypothesis© .Variablesusedin hadronidentificationare:
i. Kaonicity . It is definedasthedifferencebetweenthe k ÀûÁ y for thekaonandpion
hypothesis.
� k U � K WCB k =?>A@ U � W ¥ k =?>A@ U K W (4.1)
Statistically, kaonicityis largerfor realkaonsthatfor realpions.
ii. Pion Consistency. A variablecalledpion consistency or �7H =?I is usedto identify
pions. This variable is relatedto the probability that the pion hypothesisis better in
i4.3 DataSelection 58
comparisonto otherkindsof hypothesis.�7H =?I is definedas
�7H =?I B k =?>A@ U\p � n J Ñ , �7+ J Ñ?� n © n�W ¥ k =?>A@ U � W (4.2)
wherethe k =?>A@ U1p � n J Ñ , �$+ J Ñ?� n © n�W correspondsto the minimal valueof the likelihoodfor
all thehypothesisconsidered.Typically weuseacut of �7H =?I 1�{ for pion identification.
Othervariablesusedin theanalysisare:
3. Momentum. To measurethe momentumof a particle,FOCUSdeterminedthe
deflectionanglesin eachoneof the magneticfieldsproducedby M1 andM2. For three-
chambertrajectoriesthemomentumis measuredby M1 andfor five-chambertrajectories
themomentumis measuredby M2.
Themomentumresolutionfor trajectoriesmeasuredin M1 is approximately
° K� � { � { ����� �z�{�{[����� zÅ L z ¤ �ª���� M ^ (4.3)
andtheresolutionfor trajectoriesmeasuredin M2 is
° K� � { � {�z ��� �z�{�{[����� zÅ L ��� �ª���� M ^ (4.4)
At high momentum,theresolutionis limited by thepositionresolutionof thePWC
system;at low momentumit is dominatedby multipleCoulombscattering.
4. Dimuonic Invariant Mass. Themassof primaryparticlesis calculatedfrom the
i4.3 DataSelection 59
informationobtainedfrom theparticlesinto which thesedecay(Section1.6). In our case
this is thetwo muontrajectoriesinto which theD a decays.
4.3.2 FOCUSData SelectionStages
Thereare6.5 billion photoninteractioneventson about60008 mm magnetictapes
which constitutethe raw dataset of FOCUS.Obviously, this is way too much datafor
an individual researcherto handle. Therefore,the analysiswasdivided into threesteps.
Figure4.3 shows a summaryof the FOCUSanalysisprocess.The reasonfor doing the
selectionin stagesis to avoid having to handlethetotality of thedataagainin thecasethat
someproblemis later found in theselectionprocess.In a multistageprocedure,oneonly
needbegin at thestagepreviousto theproblem.
a) Pass1 . This is thebasiceventreconstructionstageincludingtrackreconstruction
and particle identification. It lastedfrom January, 1998 to October, 1998. The output
consistedof “reconstructedevents”madeup of thevaluesfor thephysicalvariablesof the
event. It wasputontoanothersetof 60008 mmmagnetictapes.
b) Skim1 . Here,the reconstructeddatawasdivided into six “Superstreams”.This
wasdonein orderto make the sizeof the dataeasierto handle. Approximatelyonehalf
of the eventsfrom Pass1met the requirementsfor Skim1 anddependingon the physical
topics that were consideredthey were put into differentSuperstreams(seeTable4.1).
EachSuperstreamconsistedof 200–5008 mm magnetictapes.Skim1wascarriedout at
theUniversitiesof ColoradoandVanderbiltfrom October, 1998to February, 1999.
i4.3 DataSelection 60
Skim2
Reconstruction Overview
Skimmed
3% Raw Data
3% Recon Data
15% of Charm
10 days of Skim1
StorageMass
DAQ
Pass1
Skim1
Skim2
FNAL
Disk
Mainline Tape Path
Expressline Network Path
Express
Analysis
MonitorCharm Yield
Shift Monitor
3% Recon
Institutions
DataRaw
DataRecon
Data
Figure4.3: Analysisoverview of FOCUS.The solid lines show the pathof the datawritten to 8 mm magnetictapes. The dottedlines illustrate the distribution of largeamountsof datavia theInternet,whichwasusedto helpcollaboratorsgetdataquicklyfor studiesandpreliminaryanalyses(E. Vaandering’s ThesisFigure).
c) Skim2 . Thesix Skim1Superstreamsweredividedinto many otherSubstreams.
Thoseeventsthatdid not passthe morespecificcutswerenot includedbut not all of the
skimshadadditionalcuts.
i4.3 DataSelection 61
Table 4.1: Descriptionsof Superstreams.There are about 30 different datasubsetsgroupedinto six Superstreamsbasedon physicstopicsandthe typesof informationpresent.
Super Physics Skim2Stream Topics Institution
1 Semileptonic PuertoRico2 Topologicalvertexing andK a y Illinois3 Calibration CBPF, Brazil4 Baryons Fermilab5 Diffractive (light quarkstates) California,Davis6 Hadronicmesondecays California,Davis
Table4.1showsthefiveinstitutionsinvolvedin Skim2.Therewere5–12Substreams
writtenfromeachof theSuperstreams.Calibrationdatasamplesandmany topicsin physics
werebroughttogetherby theseSubstreams.SimilarcomputingmodelsbetweentheSkim1
andtheSkim2wereusedbut they variedby institutions.Skim2beganin January, 1999and
finishedby June,1999.
Table4.2refersto theoutputof theSuperstream1 reconstructedandanalyzedby the
HEPGroupof theUniversityof PuertoRico at MayaguezCampusduringJanuaryto June
of 1999. More than500tapesof semileptoniceventswereskimmedanddividedinto five
differentssubsamples.
d) Skim3 . It is in this stagewherewe really apply strongselectioncriteria which
leadus to obtaina reducedoutputyet still leavesus with a certainamountof flexibility
in the detailedanalysisandoptimization. It is very difficult to have to work with dozens
of tapesin the detailedanalysis.Skim3 reducedthis numberto onetape. Justthe same
asin earlierstages,the ideawith Skim3wasto obtaina morecompactsamplethatwould
containthegreatmajorityof acertainkind of event.
i4.3 DataSelection 62
Table4.2: Descriptionsof Superstream1. This Superstreamwasdevelopedbythe University of PuertoRico at MayaguezCampus.Therearefive differentskimsgroupedaccordingto physicstopics.
Sub Physics OutputStream Topics Tapes
1 Semimuonic 262 Dileptonic/PPbar 453 Semielectronicwith mesons 374 Semielectronicwith baryons 275 Normalization 58
The datain Substream2 of Superstream1 wasusedasthe startingpoint in Skim3
becauseit provided the highestefficiency for the detectionof D a b d e d g . The main
idea of Skim3 was to require the existenceof two linked muon candidatetrajectories
with oppositechargecomingfrom a parentthatwasidentifiedwith a cut in thedimuonic
invariant massaroundthe massof the D a (between1.7 and 2.1 �ª��� ). In addition to
this, certainvertex identificationconditionswererequiredconsistingof �N.POQ1 z43 and
��.0/R1 z43 . Furthermore,the productionvertex wasrequiredto be separatedfrom the
decayvertex by at leastthreetimestheseparationerror( .s¯ ° 1 � ). With this we achieved
a reductionin thedatasizefrom 45 tapesof 4.5 Gbyteseachto one3 Gbytetape. The3
Gbyteswereeasilykepton disk for greaterconvenience.
At the endof Skim3, we continuedto reducethe databy a processwhich we call
Skim4.Theselectioncriteriarequiredfor Skim4werethefollowing:
Thesecutswerebasedon theexperienceacquiredin previousFOCUSdataanalysis;
thesewereknown to havehighefficiency.
Theoutputfrom Skim4wasamatrixof S � ¸ asaneventfile wherem is thenumber
i4.4 NormalizingMode 63
b Vertex isolationcuts � 1% b Verticeszposition, ¥�z�{ � V �3�b °76T8VU@� b S K Ô Ã�ËXW �b Daughtermomentum1 7 ����� b WerequiredaMISSPL Y �b IMUCL 1 z43 b OMUCL 1£z43
of variableswith informationto bestoredandn is thenumberof pair trajectoriesaccepted
asmuoncandidates.Thetotal datasizedid not exceed500Mbytes. This matrix is called
an ntuple, which is analysedusing software developedby CERN called PAW (Physics
AnalysisWorkstation)[45] which letsuseasilymanipulatetheinformationstored.It uses
histograms,vectorsand other familiar objectsto provide the statisticalor mathematical
analysisinteractively. An ntupleallows usto try differentanalysiscriteria in a shorttime
interval.
4.4 Normalizing Mode
Theobjectof this studyis to determineat what frequency do D a b d e d g occurin
nature. This meansthat we will indirectly measurehow many timesthis processoccurs
with respectto a largesamplethatcontainsmany differentprocesses.
Thebranchingratio (BR) is definedasthefractionof decaysto aparticularchannel.
Mathematically
Ö[Z]\D¸_^�Ñ.©*¸'`Nab\ J ©�+�� `dc$S p �eZ�+*f D a b d e d g �7Z*+ X c'^�� Xg + J \ · ¸_c$S p �eZh+*f D al�7Z*+ X c'^�� X (4.5)
i4.4 NormalizingMode 64
If weconsiderawell known processwherethefrequency atwhich it occursis known
with somecertaintythen we can usethat processto measurethe total numberof D a ’sproduced.g + J \ · ¸_c$S p �eZh+*f D a �7Z*+ X c'^�� X � `ic'S p �eZj+]f ¿ a blk � n �mZ]+ X cn^�� XÖ[a U1¿ a bok W (4.6)
Theprocess¿ a b k is calledthenormalizingmode. In FOCUSa processcoming
from D a asprimary particle is D ab K g �_e . It is known asa GoldenMode due to its
largeBR value.In theanalysisfor thesearchof theprocessD a b d e d g we canconsider
D a b K g � e as the normalizingmode. Then the BR of the decayD a b�d e d g canbe
calculatedas:
Öpa U D a b d e d g W � ` ;q; U �7Z*+ X�W`Nr�s U �7Z]+ X�W Ö[a U D a b K g � e W (4.7)
Thenumberof producedparticlesis calculatedasthenumberof observedparticles
dividedby theefficiency ( t ) of detectioncalculatedby MonteCarlosimulations.Theneq.
(4.7) turnsinto
Öpa U D acb�d e d g W � `«ÀûÁ y U D a b�d e d g W`«À*Á y U D a b K g � e WtAr�st ;u; Öpa U D acb K g � e W (4.8)
where the ratio tAr�s�¯]t ;q; correspondsto the relative efficiency to be calculatedmaking
parallelMonte Carlo simulationsfor the two processes,D ab d$efdhg andD acb K g �_e .
BR(D a b K g � e ) hasbeenpreviouslymeasuredbyotherexperimentsas ����¨�� ¦�{ � { ¬$� z�{ g ^[4, p.36].
The dataeventsfor the normalizingmodewere selectedusing the sameselection
i4.5 MonteCarloSimulation 65
criteria (except for particle identification)usedto selectthe dimuonic dataevents. The
normalizing mode data was obtainedfrom Substream5 of the Superstream1. This
Substreamcontainedhadronicdecaysselectedwith thesamebasicconditionsastheother
Superstream1 eventswith thespecificpurposeof servingasa normalizationsample.
4.5 Monte Carlo Simulation
Becauseof thecomplexity of High Energy Physicsexperimentsandtheimpossibility
of calculatingeveryeffectanalytically, simulationmethodsarecommonlyused.Computers
have pseudo-randomnumbergeneratorswhich are usedto simulateapproximationsof
expectedresultscomingfrom the theoreticalpredictions.We call this simulationmethod
MonteCarlo. In FOCUS,generatedMonteCarloeventswereusedasa measurementtool
for theefficienciesandalsoasaninformativetool abouttheform of thebackgroundwithin
our dimuonicsample.
Two mathematicians,the AmericanJ. Neyman and the HungarianS. Ulam gave
birth to this method in Monaco in 1949 as an alternative solution to probabilistic as
well as deterministicproblems. Using a computer, the Monte Carlo methodperforms
statisticalsamplingexperimentsby simulating randomquantities,therebyproviding us
with approximatecalculationsin High Energy Physicsproblems[46].
FOCUSusedthePythia6.127[47] generatorasa Monte Carlo simulatorto model
charmproductionby photoninteractionswith the target material. Pythiagenerateda cc
pair whereit waspossibleto specifythetypeof charm( ¿ a , ¿ e , v ew , etc.) generatedand
i4.5 MonteCarloSimulation 66
a specificdecaypath.Typically this wasdonefor oneof thecharmquarkswhile theother
wasleft freeto hadronizeanddecayaccordingto theknown crosssectionsandbranching
ratios. To simulateall theprocesseswithin theFOCUSspectrometer, analgorithmcalled
ROGUEwasdeveloped[48].
D xzy {}|0{�~ Events . To calculatedimuonic efficiencies it was necessaryto
generateMonte Carlo data which were required to passthrough the sameselection
criteria that were applied to real data, in other words, the requirementsset by Pass1,
Skim1, Skim2, Skim3 and Skim4. 800,000 events were generatedfor the process�z����� � D � �o���_���n��� c representedin Figure4.4.
µ
µ
c−
−
+D 0
γ
Figure4.4: MonteCarlogeneratedprocessschematizationfor thedecayD �C�R� � � � .
Thenumberof eventsprovideduswith enoughsamplingstatisticsto make accurate
determinationsof thedetectionefficiencies.
D �z� K ���j� Events . 800,000D � � K �7� � eventsweregenerated.The same
selectioncriteriawereappliedaswith thenormalizingdata.
The decaychannelD � � K � � � is very importantin this work becausethis is the
�4.6 AnalysisMethodology 67
normalizingmodeusedin the BR(D � �o� � � � ) measurement.In addition to that, the
samedecaychannelis presentaspossiblebackgroundin ourdimuonicsample.
D � � � � � � and D � � K � K � Events . The processesD � � � � � � and
D � � K � K � were also consideredin the backgroundstudy. For the first process,1.5
million eventsweresimulatedand4.3million eventsfor thesecond.
4.6 AnalysisMethodology
4.6.1 Blind Analysis
Our Blind analysisexcludesall the eventswithin a definedsymmetricareaaround
thevalueof theD � massin theprocessof cut selection.This areais calledthesignalarea.
In otherwords,cut optimizationis basedon reducingbackgroundandnot on maximizing
thesignalunderstudy. Thepointof this is notto introducebiasdueto thepossibleapparent
absenceor presenceof a potentialsignalafterhaving applieda specificgroupof selection
cuts. Oncethe optimizationstageendedwe openedthe “hidden” areaof the signaland
calculatedtheconfidenceinterval.
4.6.2 Rolkeand LopezLimits
Thestatisticalmethodproposedby Rolke andLopez[49] wasusedto calculatethe
sensitivities and the confidenceintervals. They suggesta techniqueto place limits on
�4.6 AnalysisMethodology 68
signalsthat are small when backgroundnoiseis present. This techniqueis basedon a
combinationof a two dimensionalconfidenceregion anda largesampleapproximationto
thelikelihood-ratioteststatistic.Automaticallyit quotesupperlimits for smallsignalsand
alsotwo-sidedconfidenceintervalsfor largesamples.
The techniqueof Feldmanand Cousins [25] does not consider the amount of
uncertaintyin the background. The new Rolke and Lopez techniquedealswith the
backgrounduncertaintyasa statisticalerror. The techniqueperformsvery well. It has
goodpower andhascorrectcoverage.This methodcanbeusedfor two situations:when
sidebandsdatagiveanestimatedbackgroundrateandwhenit comesfrom MonteCarlo. It
canalsobeusedif thereis asecondbackgroundsourcein thesignalregion.
To understandtheresultsof thistechniquewewill needthefollowing notation:x will
representtheobservedeventsin our signalregion, y will be theeventsin thebackground
region and � will be theratio of thesidebandsbackgroundto thesignalbackground.The
expectedbackgroundrate is ���Q����� and the SensitivityNumber, �N� � D � �o���_���n� , is
definedastheaverage90%upperlimit for anensembleof experimentshaving anaverage
of y eventsin thesidebandsandno truesignal.Table4.3givesthesensitivity numbersfor
differentvaluesof theconfidencelevel, � andy. During thevariousstagesof this analysis
computerroutinesprovidedby RolkeandLopezwereusedto calculatesensitivity numbers
andconfidencelimits. Thevalueof � for eachsetof cutswasdeterminedasexplainedin
Section4.7.Thesensitivity numberswereusedto selecttheoptimumsetof cuts.Thenthe
confidencelimits werecomputedfrom thevaluesof x, y and � for thatsetof cuts.Examples
of confidencelimits aregivenin Table4.4
�4.6 AnalysisMethodology 69
Table 4.3: Sensitivity number as per Rolke and Lopez. The number ofbackgroundeventsin the sidebandsis y. The estimatedbackgroundrate is��������� . �� ¡
C.L.
����¡C.L.
y �¢�¤£ �¢�¦¥ �¢�§£ �¢�¦¥0 2.21 2.21 4.52 4.521 3.31 2.82 6.22 5.352 3.97 3.17 7.12 6.143 4.62 3.49 8.17 6.454 5.29 4.03 8.99 7.125 5.90 4.23 9.77 7.436 6.37 4.60 10.52 8.007 6.86 4.79 11.21 8.308 7.23 5.14 11.85 8.869 7.64 5.32 12.51 9.0210 8.01 5.62 13.04 9.5211 8.40 5.78 13.55 9.7612 8.71 6.08 14.15 10.1413 9.09 6.19 14.63 10.3714 9.39 6.45 15.14 10.7615 9.71 6.59 15.64 10.96
4.6.3 Determination of Sensitivity and ConfidenceLimits
We definethe experimentalsensitivity asthe averageupperlimit for the branching
ratio thatwouldbeobtainedby anensembleof experimentswith theexpectedbackground
andno truesignal.Therefore,theexperimentalsensitivity is calculatedin thesamewayas
thebranchingratio exceptthat thesensitivity numberis usedfor thenumberof observed
signalevents.For eachparticularsetof cuts, � is determinedasexplainedin Section4.7.3.����� is the predictedlevel of background,b. Having this in mind, we candeterminethe
�4.6 AnalysisMethodology 70
Table4.4: An exampleof Rolke andLopez90% ConfidenceLimits. y is thenumberof eventsobserved in thebackgroundregion andandx is thenumberof eventsobservedin thesignalregion. A valueof ����¥ hasbeenassumed.
yx 0 1 2 3 40 0,2.21 0,2.27 0,1.58 0,1.28 0,0.91 0,3.65 0,3.22 0,3.44 0,3.06 0,2.872 0.42,5.3 0,4.87 0,4.43 0,4 0,4.413 0.97,6.81 0.08,6.37 0,5.93 0,5.49 0,5.054 1.54,8.25 0.74,7.8 0.03,7.36 0,6.91 0,6.475 2.16,9.63 1.42,9.18 0.74,8.73 0.07,8.28 0,7.836 2.82,10.98 2.12,10.53 1.45,10.08 0.8,9.62 0.17,9.177 3.5,12.3 2.83,11.84 2.18,11.39 1.54,10.94 0.91,10.488 4.2,13.59 3.55,13.14 2.91,12.68 2.29,12.23 1.67,11.779 4.92,14.87 4.29,14.42 3.66,13.96 3.04,13.5 2.43,13.0510 5.66,16.14 5.03,15.68 4.42,15.22 3.81,14.77 3.21,14.31
experimentalsensitivity as¨ � D � �o� � � � � � �N� � D � �o���_���$��N©�ª¬« � D � � K � � � �®°¯�±�²q²Q³[´ � D � � K � � � � (4.9)
where, �N� � D � � �µ�_���n� is the sensitivity number(Section4.6.2) which is determined
usingtheRolke andLopezroutinesfrom thenumberof eventsobservedin thesidebands,
y, andthebackgroundratio, � .
In simpleterms,thesensitivity is mainlydependenton theratioof �[�¶� �²q² . Wewish
to minimizethis ratiowhichmeanshaving low backgroundwith highefficiency.
Oncetheoptimumsetof cutshadbeendetermined,weproceededto openthehidden
signal region and determinedthe correspondingconfidencelimits for the D � � ���'���
�4.6 AnalysisMethodology 71
branchingratio by usingeq. (4.9) andsubstitutingtheappropiateRolke andLopezlimits
in placeof �[� .
4.6.4 SidebandsBackground and Signal Ar ea
In this analysis,the signal areaas well as the sidebandswere definedfrom the
invariantmassdistributionof MonteCarlogenerateddata(Figure4.5a).Theability of the
FOCUSMonte Carlo to predictsignalwidth hasbeenverified with othercopiouscharm
signals.
0.016 GeVσ =·σ·|¸
Figure 4.5: D � � � � � � Invariant Mass distributions. (a) distribution forgeneratedMonteCarlodatawith a gaussianfit. Thegaussianwidth is equalto 0.016¹Vº¼»
. (b) Representationof thesidebandsandthesignalareafor theFOCUSproduceddatashowing the signalandsidebandareas. Skim3 cutshave beenappliedand thesignalareahasbeenmasked.
Thewidth of theinvariantmassdistributionfor D � � ���n��� eventsis approximately
equalto ½¾� �¿À £ÂÁÄÃbÅuÆ . The signalareawastaken as ÇN¥*½ arounda D � massof £ ¿ÀÈ Á�É
�4.7 BackgroundStudy 72ÃbÅuÆ [4, p.36]. Eachoneof thesidebandshada width equalto thesignalareawidth but
thesidebandswerenot locatedimmediatelyafter thesignalarea.Betweeneachsideband
andthesignalarea,abuffer areawasleft with awidth alsoequalto ¥*½ . In thisway theleft
sidebandwasdefinedas £ ¿ËÊ]Ì�ʢͦΠ²q² Í £ ¿ÀÈ] £ andright sideband£ ¿Ï� ¥ �ÐÍÑÎ ²q² Í £ ¿À���]Ìasshown in Figure4.5b.
4.7 Background Study
4.7.1 Intr oduction
Themostimportantaspectof theanalysisis thechoiceof a setof selectioncriteria
which optimize the probability of observingthe signal if it is present. Generallythis
choicewill not beonewhich eliminatesall backgroundbecausesuchcutsusuallyreduce
the efficiency of detectionof the signal to unacceptablysmall values. It is therefore
importantto be able to predict the backgroundlevel accuratelyin order to measurethe
significanceof thecandidateevents.Dueto theuncertaintiesin simulatingthebackground,
our analysismethodologyreliesmainly on the real data(in the sidebands)to predict the
backgroundin thesignal.However, this methodologyrequiresa determinationof theratio
of sidebandto signalbackground,� . This, in turn, requiresadeterminationof theshapeof
thebackground.As is explainedin this section,this shapeis alsoobtainedfrom realdata.
Fortunately, sincethey arenot strong-interacting,muonsarenotproducedcopiously
andit is evenrarerthata pair of muonsis producedin eithera productionor decayvertex.
In addition, eventswherethe pair of muonscomefrom a non-D � two-body decayare
�4.7 BackgroundStudy 73
eliminatedby the invariantmasscut while thosewherethe muon pair is formed at the
productionvertex (e.g. thosefrom the Bethe-Heitlerprocess)areeliminatedby the ÒÓ�*½cut. In fact,themainsourceof backgroundin ourfinal sampleareeventswhereoneor two
mesonshavebeenmisidentifiedasmuons.Thereforetheunderstandingof thebackground
startedwith a studyof this misidentificationusingreal data. Theseresultswereusedto
validatethecapabilityof theMonteCarloto simulatethiseffect. TheMonteCarlowasthen
usedto studysomeknown decayprocesseswhichcouldcontributeto thebackgroundif the
decayproductsweremisidentifiedasmuons.Finally, thebackgroundratiowasdetermined
fromanindependentrealdatasampleof eventswith nomuonsin thefinal stateby assuming
that the backgroundconsistedof sucheventsexcept that both decayproductshad been
misidentified.
4.7.2 MisID Study
Thegoalof thisstudywasto determinethemisidentificationprobabilityasafunction
of momentum. Samplesof real and simulatedK � « � � � � � decayswere usedfor this
purpose.
Therealdatawasobtainedfrom 140raw datatapes.K � « decayswereidentifiedwith
thefollowing requirements:vertex conditionsfor a pair of oppositechargedparticles,the
two pion trajectoriesmusthave left a signal in the five PWC chambers,andthe dipionic
massin a rangearoundtheK � « mass(
�¿ÏÔ Ê�ÊÐͦΠ±u± ÍÑ Õ¿ ÉÕ£ Ê ), beingtheK � « massequalto �¿ÖÔ ��È Ã×ÅqÆ [4, p.32].Furthermore,decaycandidateshadto havelinkedtrajectoriesbetween
the silicon detectorsand the PWCs. Eventswhich passtheserequirementsare called
�4.7 BackgroundStudy 74
Type-9K � « [32, p.27]. By determininghow many of thesetrajectorieswereidentifiedas
muonsonecoulddeterminethemisidentificationprobabilityasa functionof momentum.
A sampleof simulatedType-9K � « � � � � � decayswasgeneratedwith MonteCarlo
by artificially lowering the K � « lifetime thusforcing the K � « to decaynearthe target and
obtaininga largesampleof pionscomingfrom thetarget.Thissamplewasanalyzedin the
samewayastherealType-9K � « dataobtainingconsistentresultsbetweenthetwo samples
(SeeSection5.1).
4.7.3 Background Ratio Study
In orderto havea roughideaof thenatureof thebackground,MonteCarlowasused
to generatetwo-bodymesonicD � decayswhich wereexpectedto contributesignificantly
to the background. The first decaysimulatedwas D � � � � � � . Figure 4.6 shows the
comparisonof thedimuonicinvariantmassdistributionsfor D � � � � � � andD � � � � � � .
TheD � � � � � � peakis slightly shiftedfrom theD � massdueto thedifferencebetween
the pion and muon mass. Sincethat differenceis small, the shift in the peakis small
andthe invariantmasscut will not helpus to eliminatethis background.Fortunately, the
D � � � � � � decayis Cabibbo-SuppressedandtheFOCUSmuonsystemis very goodat
rejectingpions(low misidentificationprobabilities).
The other backgrounddecaygeneratedwas D � � K � � � . Figure 4.7 shows the
dimuonic invariant massdistribution for D � � K �7� � events. Due to the large mass
differencebetweenthe kaon and the muon, the peak is much more shifted but it does
fall within theleft sideband.D � � K � � � is aCabibbo-Favoreddecay. Fortunately, it lies
�4.7 BackgroundStudy 75
Figure 4.6: Dimuonic Invariant Mass comparison of Monte Carlo data fromD �Ó�ÙØ � Ø � andD �Ó�Ú� � � � .
outsideour signalareaandcanbe almostentirely eliminatedby the invariantmasscut.
This distributionalsoshowsanadditionalcontinuousbackgroundinsidethesignalregion.
This backgroundis from theothercharmparticlein theevent. (Charmis alwaysproduced
in pairs). It is madeup of semimuoniceventswheretheneutrinois not observedor from
partially-reconstructedhadronicdecays.It is verydifficult to simulateall suchbackground
componentsandto know theirrelativeabundance.Thisis whywedonotrely onsimulation
to obtainthebackgroundsidebandto signalratio.
As analternative to simulationdata,a sampleof realdatafrom Superstream2 (also
�4.7 BackgroundStudy 76
Figure4.7: Dimuonicinvariantmassof D � � K � Ø � generatedby MonteCarlo.
calledGlobalVertex) wasusedto determinethebackgroundratio. This superstreamwas
selectedwith a minimum of requirementsandcontainsall backgroundsto our signal. In
particular, no particleidentificationwasusedin makingSS2. Our procedureconsistedin
submittingthis sampleto thesameskimmingprocessastherealdataexceptthatno muon
requirementwasused.It wasonly necessaryto work with Û 10%of the total SS2datain
orderto obtainsufficientstatisticalpower for our purposes.
As an example of the dimuonic invariant massdistributions obtainedwith this
procedure,Figure 4.8 shows the resultswhenone appliestight cuts to the SS2sample.
�4.7 BackgroundStudy 77
The D � � K �$� � is the dominantfeaturein the distribution but a D � � � � � � peakis
alsoevidentaswell asotherbackground.TheD � � � � � � peakis muchsmallerthanthe
D � � K � � � peakbecauseit is Cabibbo-Suppressed.The peakfrom D � � K � K � lies
completelyoutsideany regionof interest.As thecutsarevaried,theratioof thebackground
in thesidebandsto thatin thesignal( � ) variesbut thenumberof eventsin thehistogramsis
sufficient to allow a precisedeterminationof this ratio which is calculatedusingweighted
histogramsasexplainedbelow.
Figure4.8: BackgroundcomparisonbetweenrealandMonteCarlodata.Distributionfrom SS2datawith tight cutsbut no muonidentification.Theseeventsconstitutethemainsourceof backgroundwhenthedecayproductsaremisidentifiedasmuons.Thereareevidentpeaksfrom theprocessesD � � K � Ø � , D � �RØ � Ø � andD � � K � K �alongwith otherbackground.
�4.8 Optimizationof theSelectionCriteria 78
In order to take into considerationthe probability of the decay productsbeing
misidentifiedasmuons,the backgroundratio wasdeterminedfrom weightedhistograms.
Eachentryin thedimuonicinvariantmasshistogramsof theSS2data(suchasFigure4.8)
wasweightedwith theproductof themomentum-dependentMisID probabilitiesfor each
of the two decayproducts.Eventswith two muonswereeliminatedfrom the samplebut
eventswith one muon were allowed in order to include semimuonicbackground. The
weightwascalculatedasfollows: Ü ��Ý Þ�ß}Ý4à (4.10)
where Ýi�ᣠ¿À for muons,and Ýi� Îãâ?äeå&æfor non-muons.Then,wehave that
çèÜ � Îãâ?äeå&æ £éß Î�âêäeåÕæ ¥ if neitherdecayproductis identifiedasamuon.çèÜ � Îãâ?äeå&æif oneof thetwo decayproductsis identifiedasamuon.
Resultsfrom this procedurearepresentedin Chapter5. A valuefor thebackground
ratio ( � ) wasobtainedfor eachof thesetsof cutsconsidered.
4.8 Optimization of the SelectionCriteria
The optimization procedureof our selectioncriteria was the longest and most
tiresomephaseof this effort. It establishedthe final criteria for the selectionof muon
candidateevents. The aspectthat wasconstantlykept in mind was to eliminatea large
numberof backgroundeventsmaintaininghighefficiency values.
�4.8 Optimizationof theSelectionCriteria 79
For this procedurewe startedwith the applicationof fixed (constant)cuts which
correspondto the cuts applied in the selection processincluding Skim4. Due to
correlations,the optimizationof the cuts requireda procedurewhich considereda large
numberof cut sets.This implied a selectionof the appropiatecut rangesfor eachoneof
thesevariables.Onceeachcut rangehadbeendeterminedwe proceededto optimizeour
selectioncriteriawith the ideaof obtainingthebestsetof cutsto beappliedin our search
of dimuonicevents.
4.8.1 Fixed Cuts and Variable Cut Ranges
In additionto the cutsappliedin the earlierstages,in this analysisthe Skim4 cuts
wereusedasfixedor constantrequirements.Table4.5presentsthefixedcutsappliedto the
invariantmassvariablewhichdefinethesidebandandsignalregions.
Table4.5: DimuonicInvariantMassCutsfor SidebandsandSignalRegions.
Variable CutName Applied
Sidebands ë7ì-ímî_í�ïñðóò òôïQë$ìöõ7÷¶ë (left)ë7ìöø$ù'øôïñðóò òôïQë$ìöø$ø$î (right)SignalArea ë7ìöõ$î'îôïñðóò òôïQë$ìöõ$øní
Therangeof thevariablecutsincluding thevariablesusedfor hadronidentification
were chosenon the basisof the experiencewith previous FOCUSanalysis. Theseare
presentedin Table4.6.
�4.8 Optimizationof theSelectionCriteria 80
Table4.6: Cut valuesfor eachvariableconsideredin theoptimizationprocess.
VariableName Cut Applied� Vertex IDÒ0��½ úñûµü ínü*øCLS úý÷�ìö÷¶ë7ü�÷¶ìG÷'ûµü4÷¶ì�ëISOP ïþëÿ÷ ��� ü&ë&÷ � � üÕëÿ÷ � �ISOS ïþëÿ÷ ��� ü&ë&÷ � � üÕëÿ÷ � �½���� � ÷�ü*ù� Muon IDK/ � ïñûµü]õMISSPL ïþë7ü]ùµü*îIMUCL úý÷�ìö÷¶ë7ü�÷¶ìG÷'îµü4÷¶ìG÷'û� MesonID� � � K � � ë7ü� ¶ü í������� ��� ë7ü � îµü � û
4.8.2 Optimization Basedon Sensitivity
All possiblecut setsthat could be formed from the cut valuesin Table 4.6 were
considered.The total numberof cut setswas26244. The sensitivity wascalculatedfor
eachcut setaspresentedin eq.(4.9).Theoptimumcut setwasdefinedastheonewith the
lowestsensitivity.
Chapter 5
Resultsand Conclusions
5.1 MisID
The measurementof the percentageof pions that were misidentified as muons
was carried out using a large sampleof K � « � � � � � decaysusing Type-9 K � « data
events.Resultsfrom this studyarepresentedin Figure5.1 for which thefollowing muon
identificationcutswereapplied:
ç InnerMuonCL �¦£ ¡ .ç MISSPL
Í ¥ .Wecanseethatfor low momentumtheprobabilityof pionmisidentificationis greater
thanfor high momentumfor realdataaswell asfor MonteCarlosimulateddata. This is
81
�5.2 BackgroundRatio 82
Figure5.1: MisID asa functionof momentum.
dueto the fact thatseveraleffectsthatcontribute to pion MisID, suchaspion decaysand
multiple Coulombscattering,becomelargerat low momentum.Theresultsfrom dataand
MonteCarloareconsistentwithin errors.
5.2 Background Ratio
Figure5.2showsa typicalweightedinvariantmassdistribution for backgroundratio
of a specificsetof cutsselectionusingGlobalVertex sample.Thebackgroundratio � was
�5.3 SkimmingResults 83
calculatedby integratingthis histogramnumericallyin thesidebandsaswell asthesignal
region.
Figure 5.2: Typical WeightedInvariant Mass Distribution Used in CalculatingtheBackgroundRatio.
5.3 Skimming Results
Invariantmassdistributionswith Skim3selectioncriteriaareshown in Figure5.3for
dimuonicaswell asnormalizingsamples.
�5.3 SkimmingResults 84
Figure5.3: InvariantMassDistributionsfor Skim3Cuts.(a)Dimuonrealdatasample.(b) D �P�Ú� � � � Monte Carlo sample. (c) Normalization real data sample. (d)NormalizationMonteCarlosample.Thesignalregionhasbeenmaskedfor thedimuonrealdata.Therealdatasampleshave largebackgroundlevelswith theseloosecuts.
Following theblind analysismethodology, only sidebandeventsarepresentedfor the
dimuonrealdata. For the realdatadistributionswe canseea higherlevel of background
(Figure 5.3.a and Figure 5.3.c). Figure 5.4 shows the samedistributions displayedin
�5.3 SkimmingResults 85
Figure5.3 with thedifferencethat thesedistributionscorrespondto eventsselectedusing
Skim4criteria.
Figure5.4: InvariantMassDistributionsfor Skim4Cuts. A considerablereductioninthelevel of backgroundis achievedwith theapplicationof thesestrongercuts.
Although the backgroundin the dimuonreal data(Figure5.4.a)hasbeenreduced
considerablywith the Skim4 cuts, it is necessaryto reduceit even further to obtain the
�5.4 CutOptimizationResults 86
bestmeasurement.Thedeterminationof theoptimumcut setto achieve this reductionis
discussedin thenext Section.
5.4 Cut Optimization Results
Using the cut rangesshown in Table4.6, the valuesof � that hadbeenpreviously
determinedand the Rolke and Lopezroutines,the sensitivity was determinedfor each
of the 26244differentcombinationsof cuts. The lowestsensitivity wasobtainedfor the
following cut set.
Table5.1: BestCut Combination.����� �7.0���! #" $1%���! %� $1%& �'� �1%�)(+* ,1- ���!�."�� $3�/- 0 & � �
1% #- 0 & � �1%1 ��2 $8.03 4 576 1 8 91.06;:=<?> 9-5.0
�5.4 CutOptimizationResults 87
Figure5.5: Massdistributionsfor thebestcut combination.
For these cuts the dimuon mass distributions for the sidebandsdata and the
D � �o� � � � Monte Carlo arepresentedin Figure5.5.aandFigure5.5.b. Therearefive
eventsin thesidebands.Theefficiency for detectingD � �o� � � � is 2.8%.
The distributionsfor the normalizingmodeareFigure5.5.candFigure5.5.d. For
therealdatatheyield correspondsto thenumberof eventsin theGaussiancurve. For the
�5.4 CutOptimizationResults 88
MonteCarlogenerateddata,theefficiency A¯�± is calculatedasthenumberof eventsin the
Gaussiancurvedividedby thetotal numberof generatedevents.
Our reconstructionprocessfollowedby theanalysisalgorithmlet usobserve135432
D � � K �7� � eventswith a detectionefficiency of 3.9%.Table5.2summarizestheresults
for the bestcut combination. For the backgoundratio study the resultsusing the final
selectioncriteriaareshown in Figure5.6.
Table5.2: Resultsof Cut Optimization.
@BADCFEHGJILKJM!N!O PQASR�TUK 5WV 85.72"XI�YZO!T[C\R]YZO @BA^C_EHGJI�KJM!N!O`T[N �aTUGJNbADc0.9�dYZN!e]TfR�ThgiThRkj l`M!m noYZIXp[l`qar2.96
D ÷ts 2vuw2 � Yyx CzTUYZN!C\j 5?{ 82.8
D ÷ s K� 6 u|Yyx C}TUY~N!C�j 5�{ 8
3.9�dYZN!e]TfR�ThgiThRkj 5}���Z� �d� 81.1
Using the Rolke andLopezalgorithm, the sensitivity numberis determinedto be¥ ¿Ï� Á . Thentheexperimentalsensitivity is calculatedas¨ � D � � � � � � � � ¥ ¿À� Á£ Ì É Ô Ì ¥Ì�¿Ï��� £ � ॠ¿ÏÈ�� £ � à ÌÕ¿ÀÈ�È�� £ � à �§£ ¿ £4É � £ ��� (5.1)
This result is encouragingsinceit is muchlower thanthe existing 90% confidence
limit of
Ôm¿ £ � £ ��� . This meansthatour measurementwill bemoreprecisethanprevious
measurements.
�5.5 ConfidenceLimits 89
Figure 5.6: Invariant Mass Distribution for BackgroundRatio using the best cutcombination.
5.5 ConfidenceLimits
Onceour selectioncriteriawereoptimized,we proceededto openthemaskedarea.
Applying thebestcut combinationto dimuoniceventswithin our signalareawe have two
eventsin that region. Thus,we have five eventsin sidebandsarea( �i� É ) andtwo events
in thesignalarea( � �¦¥ ).UsingtheRolke andLopezroutines,for � �§É ¿ÀÊ ¥ , thebackgroundrate( �j�¦�m�4� ) is �¿Ï�
and,the90%confidencelevel lower limit is 0 andthe90%confidencelevel upperlimit
�5.5 ConfidenceLimits 90
Figure5.7: Eventcandidatesfor theD �Ó�Ú� � � � process.Figuredisplaytheonly twoeventswithin thesignalareawhichsurvivestheoptimumselectioncriteriain thestudyof theraredecayD �Ó�Ú� � � � .
is 5.34.The90%upperconfidencelimit for thebranchingratio is calculatedas
³p´ � D � �o� � � � � � É ¿ÏÌ]Ô£ Ì É Ô�Ì ¥ÌÕ¿À��� £ � ॠ¿ÀÈ�� £ � à Ì�¿ÏÈ�È�� £ � à �¦¥ ¿ £ � £ ��� (5.2)
�5.6 Conclusions 91
No definitive evidenceof the rare decayD � �o� � � � was found in this analysis.
However, ourwork allowsusto seta lower90%confidencelevel upperlimit for thisdecay.
5.6 Conclusions
Monte Carlo simulation of the level of meson misidentification in FOCUS
was confirmedby experimentalmeasurementsusing real data. The misidentification
measurementswereusedas the basisof a methodto determinethe ratio of background
in theinvariantmasssidebandsversusbackgroundin thesignalregion. This methodused
realdataandavoidedsystematicuncertaintiesin theuseof simulation.
It waspossibleto reducethe backgroundin this analysissignificantly usingwell-
known selectionvariablesbasedon the characteristicsof the signal being sought. The
blind analysismethodologyprovided an excellentway of optimizing the event selection
criteriawith aminimumof bias.
Themainconclusionobtainedfrom thiswork wasthereductionof theupperlimit in
thebranchingratioatahalf of thevaluepreviouslypublishedby PDG.No evidencefor the
raredecayD � � � � � � wasfound.The90%CL branchingratio limit was ¥ ¿ £ � £ ��� .
List of References
[1] M. Herrero,TheStandard Model. hep-ph/9812242(1998).
[2] B.P. Roe, Particle Physicsat the New Millennium. Springer-Verlag, Inc. p.246.
New York, USA (1996).
[3] C. Quigg, Gauge Theoriesof the Strong, Weak, and Electromagnetic Interactions.
Addison-Wesley Lognman,Inc. p.106.Reading,USA (1997).
[4] ParticleDataGroup(D.E. Groomet al.). TheEuropeanPhysicalJournal.Review of
Particle Physics. Vol.3,Number 1-4 (2000).
[5] N. Cabibbo,Unitary SymmetryandLeptonicDecays. Phys.Rev. Lett.10, 531(1963).
[6] M. KobayashiandT. Maskawa,CP Violation in theRenormalizableTheoryof Weak
Interactions. Prog.Theor. Phys.49, 652(1973).
[7] L.-L. Chau and W.-Y. Keung, Comments on the Parametrization of the
Kobayashi-MaskawaMatrix. Phys.Rev. Lett. 53, 1802(1984).
[8] H. HarariandM. Leurer, Recommendinga Standard Choiceof CabibboAnglesand
KM Phasesfor AnyNumberof Generations. Phys.Lett. B181, 123(1986).
92
�Bibliography 93
[9] H. Fritzschand J. Plankl, The Mixing of Quarks Flavors. Phys.Rev. D35, 1732
(1987).
[10] F.J. Botella and L.-L. Chao,Anticipating the Higher Generations of Quarksfrom
RephasingInvarianceof theMixing Matrix. Phys.Lett. B168, 97 (1986).
[11] L. Wolfenstein, Parametrization of the Kobayashi-Maskawa Matrix.
Phys.Rev. Lett. 51, 1945(1983).
[12] S.L. Glashow, J. Iliopoulos andL. Maiani, Weak Interactionswith Lepton-Hadron
Symmetry. Phys.Rev. D 2, 1285(1970).
[13] B.R. Poe,Probability and Statisticsin ExperimentalPhysics. Springer-Verlag, Inc.
p.102.New York, USA (1992).
[14] S. Pakvasa,Symposiumon Flavor ChangingNeutral Currents: Presentand Future
Studies(FCNC97).UH-511-871-97, p.9.SantaMonica,USA (1997).
[15] BEATRICE Collaboration(M. Adamovich et al.). Search for theFlavour-Changing
Neutral-CurrentDecayD � � � � � � . Phys.Lett. B 408, 469(1997).
[16] T. Sjostrand,TheLundMonteCarlo for JetFragmentationand � � � � Physics.JETSET
VERSION6.2. Comput.Phys.Commun.39, 347(1986).
[17] T. SjostrandandM. Bengtsson,TheLund MonteCarlo for Jet Fragmentationand
� � � � Physics.JETSETVERSION6.3. Comput.Phys.Commun.43, 367(1987).
[18] H.-U. BengtssonandT. Sjostrand,TheLund MonteCarlo for Hadronic Processes:
PYTHIAVERSION4.8. Comput.Phys.Commun.46, 43 (1987).
�Bibliography 94
[19] T. Sjostrand, High-Energy-Physics Event Generation with PYTHIA 5.7 and
JETSET7.4. Comput.Phys.Commun.82, 74 (1994).
[20] K. Lassila, CERN Computing and Networks Division rep. CN/91/13. Geneva,
Switzerland(1991).
[21] A. Fasso, A. Ferrari,J.Ranft,P.R.Sala,G.R.StevensonandJ.M.Zazula.Proceeding
Workshopon SimulatingAccelerator Radiation Enviroments, Los Alamos Report
LA-12835-C, p. 134-144.SantaFe,USA (1993).
[22] FermilabE791 Collaboration(E.M. Aitala et al.). Search for Rare and Forbidden
DileptonicDecaysof the
æ � ,
æ �« and
æ � CharmedMesons. Phys.Lett. B 462, 401
(1999).
[23] H.-U. Bengtssonand T. Sjostrand, High-Energy-PhysicsEvent Generation with
PYTHIA5.7andJETSET7.4. Comput.Phys.Commun.82, 74 (1994).
[24] T. Sjostrand,Phytia 5.7 and Jetset7.4 Physicsand Manual, CERN-TH 7112/93
Geneva,Switzerland(1995).
[25] G.J.FeldmanandR.D.Cousins,UnifiedApproachto theClassicalStatisticalAnalysis
of SmallSignals. Phys.Rev. D 57, 3873(1998).
[26] R.D. Cousinsand V.L. Highland, Incorporating SystematicsUncertaintiesinto an
UpperLimit. Nuc. Inst.andMeth. in PhysicsResearch.A320, 331(1992).
[27] FermilabE687Collaboration(P.L. Frabettiet al.). Search for Rare and Forbidden
SemileptonicDecaysof theCharmedMeson
æ � . Phys.Lett. B 398, 239(1997).
�Bibliography 95
[28] Fermilab FOCUS Collaboration(J.M. Link et al.). Description and Performance
of theFermilabFOCUSSpectrometer. http://www-focus.fnal.gov/nim/
focus-nim/nim.ps, (1999).ActiveJuly, 2001.
[29] H.-U. Bengtssonand T. Sjostrand, High-Energy-PhysicsEvent Generation with
PYTHIA5.7andJETSET7.4. Comput.Phys.Commun.82, 74 (1994).
[30] E.W. Vaandering,Mass and Width Measurementsof �_� Baryons. Ph.D. Thesis.
Universityof Colorado,p.15(2000).
[31] C. Rivera,NuevosAspectosdel Analisis de los Datosde Cerenkov del Experimento
831deFermilab. M.S. Thesis.UniversidaddePuertoRico, RecintoUniversitariode
Mayaguez(1998).
[32] A. Mirles, Identificacion de Muonesen el ExperimentoE831 de Fermilab. M.S.
Thesis.UniversidaddePuertoRico,RecintoUniversitariodeMayaguez(1997).
[33] E. Ramirez,Algoritmo de Identidicacion de Muones. M.S. Thesis.Universidadde
PuertoRico,RecintoUniversitariodeMayaguez.p.66(1997).
[34] FermilabFOCUSCollaboration(J.M. Link et al.). A Measurementof the Lifetime
Differencesin theNeutral D-mesonSystem. Phys.Lett. B 485, 62 (2000).
[35] FermilabFOCUSCollaboration(J.M. Link et al.). Measurementof the � �� and � �7��MassSplittings. Phys.Lett. B 488, 218(2000).
[36] FermilabFOCUSCollaboration(J.M.Link etal.). Search for CPViolation in
æ � andæ � Decays. Phys.Lett. B 491, 232(2000).
�Bibliography 96
[37] FermilabFOCUSCollaboration(J.M.Link etal.). A Studyof theDecayD � � K �'� � .
Phys.Rev. Lett. 86, 2955(2001).
[38] Fermilab FOCUS Collaboration(J.M. Link et al.). A measurementof branching
ratios of
æ � and
æ �« hadronic decaysto four-body final statescontaininga ��� .hep-ex/0105031, (2001).(Submittedto PhysicalReview Letters).
[39] Fermilab FOCUS Collaboration(J.M. Link et al.). Measurementof the relative
branchingratio BR(� �� � � � K � � � )/BR(� �� � � � � � � � ). hep-ex/0102040, (2001).
[40] R. Culbertson,Vertexing in the90’s. InternalFOCUSmemo,(1991).
[41] J.A. Quinones,NuevosAspectosen el Estudiode la Partıcula
æen el Experimento
FOCUSdeFermilab. M.S.Thesis.UniversidaddePuertoRico,RecintoUniversitario
deMayaguez.p.44-46(2001).
[42] J. Wiss, Thoughtson Muon Identification Algorithms for E831. FOCUS memo,
http://web.hep.uiuc.edu/e687/muon/mu_id_thoughts.ps, (1994).
ActiveJuly, 2001.
[43] C. Cawlfield, M. Ruesnink and J. Wiss, Muon Identification � à Confidence
Levels. FOCUSmemo,http://web.hep.uiuc.edu/e687/muon/chisq.
ps, (1994).ActiveJuly, 2001.
[44] J. Wiss, CITADL: A Proposalfor a New Cerenkov Algorithm for FOCUS. FOCUS
memo, http://web.hep.uiuc.edu/e831/citadl/citadl.ps, (1997).
ActiveJuly, 2001.
[45] R. Brun,O. Couet,C. VandoniandP. Zanarini.PAW - PhysicsAnalysisWorkstation.
V 2.10/09.http://paw.web.cern.ch/paw/, (2000).ActiveJuly, 2001.
�Bibliography 97
[46] N. Metropolis and S. Ulam, The Monte Carlo Method. J. Am. Stat.
Assoc.44, 335(1949).
[47] T. Sjostrand,P. Eden,C. Friberg, L. Lonnblad,G. Miu, S. MrennaandE. Norrbin,
High-Energy-PhysicsEventGeneration with PYTHIA6.1. hep-ph/0010017(2000).
[48] Fermilab FOCUS Collaboration (J.M. Link et al.). MCFocus Home Page,
http://www-focus.fnal.gov/internal/mcfocus.html. Active July,
2001.
[49] W. A. Rolke andA. M. Lopez,ConfidenceIntervalsand Upper Boundsfor Small
Signals in the Presenceof Background Noise. Nucl. Inst. and Meth. in Physics
Research.A458, 745(2001).