ebook img

Predictable projections and predictable dual projections of a two parameter stochastic process PDF

2006·3.8 MB·English
by  GrayPeter
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Predictable projections and predictable dual projections of a two parameter stochastic process

PREDICTABLEPROJECTIONSANDPREDICTABLE DUALPROJECTIONSOFATWOPARAMETERSTOCHASTICPROCESS By PETERGRAY ADISSERTATIONPRESENTEDTOTHEGRADUATESCHOOL OFTHEUNIVERSITYOFFLORIDAINPARTIALFULFILLMENT OFTHEREQUIREMENTSFORTHEDEGREEOF DOCTOROFPHILOSOPHY UNIVERSITYOFFLORIDA 2006 ACKNOWLEDGMENTS Iamindebtedtomysupervisor,Dr.NicolaeDinculeanu,forhisinfinitepatience withmewhileIslowlylearnedthematerial.Also,Iamgratefultothemanyexcellent teachersthat I havehadduringmyjourneyattheUniversityofFlorida. Finally,for thefriendlybanterandheartylaughsthat Iholdinfondmemory,IthankJulia, Connie,andGretchen. II 1 TABLEOFCONTENTS ACKNOWLEDGMENTS ii ABSTRACT CHAPTER 1 THECROSSSECTIONTHEOREM 1 Introduction 1 Predictable^-Algebras 2 StoppingTimes 6 Projections 7t[A] g Sets fZs 1 TheCrossSectionTheorem 21 2 PREDICTABLEPROJECTIONS 37 Projections 37 TheUniquenessof 39 TheExistenceof /’X 43 3 PREDICTABLEDUALPROJECTIONS 50 StepFiltrations (^,) 50 PredictableDualProjections X/' 52 TheUniquenessof XF 54 PredictableDualProjectionsofMeasures 55 ProcessesAssociatedWithStochastic,K-ValuedMeasures 58 TheExistenceofX^ 64 4 VECTOR-VALUEDPREDICTABLEDUALPROJECTIONS 77 PredictableDualProjections W^ 77 TheUniquenessof W^ 81 ProcessesAssociatedWithStochastic,E-ValuedMeasures 82 TheExistenceof W^ 89 iii 5 ANEXTENSIONOFTHERADON-NIKODYMTHEOREM TOMEASURESWITHFINITESEMIVARIATION 94 6 SUMMARYANDCONCLUSIONS 106 REFERENCELIST 107 BIOGRAPHICALSKETCH 108 IV . AbstractofDissertationPresentedtotheGraduateSchool oftheUniversityofFloridainPartialFulfillmentofthe RequirementsfortheDegreeofDoctorofPhilosophy PREDICTABLEPROJECTIONSANDPREDICTABLE DUALPROJECTIONSOFATWOPARAMETERSTOCHASTICPROCESS By PETERGRAY August2006 Chair;NicolaeDinculeanu MajorDepartment:Mathematics Theframeworkofthisdissertationconsistsofaprobabilityspace (Q, afil- tration suchthatif isafiltrationsatisfying =3^s- forevery se thenwehave Qs = 3^s- foreverypredictablestoppingtime Sfor adouble filtration {3^s,i)s,ie^i^ suchthat 3^s,t = 3^s- forevery >0; andaBanachspace E. Westudyinitiallyareal-valued,twoparameterstochasticprocess X:Qx K, andthenweextendsomeofourresultstoavector-valuedprocess Y:Qx ^ In Chapter1 westartbydefiningthepredictable o-algebra p ofsubsetsof QX tobethe a-algebrageneratedbytheleftcontinuousprocesses X thatare adaptedtothedoublefiltration {3^s,t)s,K'SL^.Thenweprovethemainresultofthechap- ter,thecrosssectiontheoremforsetsin p InChapter2 wedefinethepredictableprojectionofameasurableprocess X:QXIR2 tobeapredictableprocess ^X; x oj suchthatforevery V stoppingtime Z wehave E[Xz1<z<oo>\3^z] = (^X)z1 almostsurely.Then, usingthecrosssectiontheorem,weshowthatthepredictableprojectionisunique uptoanevanescentset.Inaddition,wedemonstratethateverybounded,measurable process X hasapredictableprojection. in Chapter3 wedefinethepredictabledualprojectionofarightcontinuous,mea- surableprocess X:Qx integrablevariationtobearightcontinuous, predictableprocess X^:Qx k withintegrablevariationsuchthatforeach bounded,measurableprocess (p:ClxKl R wehave E[jPep dX] = E[|(p dX^]. Thenweshowthatthepredictabledualprojectionisuniqueuptoanevanescentset. Wealsoestablishthatthepredictabledualprojectionoftheprocess X existsifthe filtration isastepfiltration. In Chapters4and5 weturnourattentionfromreal-valuedprocesses X tovector- valuedprocesses Y. Inthissetting,ourformulationsarebasednotonfinitevariation andintegrablevariation,butonfinitesemivariationandintegrablesemivanation. VI CHAPTER 1 THECROSSSECTIONTHEOREM Introduction Thetheorysurroundingoneparameterstochasticprocesseshasapplicationsin manyfields.Infinanceforinstance,afiltration containsinformationthatis knownuptotime t about amarket; amartingale (X,),er^ for reflectsthe priceofstockoptions; apredictableprocess (H/),er^ housesthenumberofshares to beheldattime t\ andastoppingtime S for (^/),^r^ indicates whenstocks shouldbesoldforoptimalprofit. Noteadiscrete(orstep) filtrationsufficesforgood resultsinmanymarkets. Thepredictableprojection(^X,),eR^andthepredictabledualprojection (XO/eR+ fortheprocess (X,),eR^ bothplayaroleinoneparameterstochastictheory.Foran exampleofthisweretreadthefinancestagethatwassetabove.Therandomvariable PXs, whichisthepredictableprojection (^X,),er^ evaluatedata(predictable) stoppingtime S, mayberegardedasanupdatedversionoftheexpectedsellingprice E[Xs] ofstockoptions,giventhemarketinformation 3^s- Thegoalofthisdissertationistoextendthedefinitionand existence ofthepredic- tableprojectionandthepredictabledualprojectiontoatwoparameterprocess (Xj,/)j,,eR^. Thisextensionisdifficultbecause,whiletheset IR+ ofpositiverealnum- bersistotallyordered,theset oforderedpairsofpositiverealnumbersisnot. Inordertoreduceslightlythecomplexitythatweface,wewillretainaoneparameter 1 2 flavor: ourframeworkwillbebuilt around the doublefiltration = ^s-n where isarightcontinuous,completefiltration. Predictable a-Aigebras Themainresultofthischapterisacrosssectiontheoremforpredictablesubsets of QxIR2 relativetothedoublefiltration satisfying 3^s,t= 3^s- for s,t>0.Thecrosssectiontheoremwillbederivedwithinvaluablehelpfromthe MonotoneClasstheorem,whichmaybefoundinthetextProbabilitiesandPotential (Dellacherie&Meyer1975,p.13-1). Inthissectionweintroducethepredictable a-algebra p ofsubsetsofthespace QXK+. NotationandTerminology1.1 Thefollowingwillbeusedinthesequel. 1.1a(Q,^,P) isaprobabilityspace. 1.1bK+ isthesetofnon-negativerealnumbers. N isthesetofnaturalnumbers. Q+ isthesetofpositiverationalnumbers. 1R+ istheset IR+xK+, and Q+ istheset Q+xQ+. 1.1cR(lR+) istheBorel cr-algebrageneratedbytheintervals {s,t] of 0^+. istheBorel a-algebrageneratedbytherectangles (s,r]x(«,v] of Rl. I.ldAfunction X;Qx ^ iscalledatwoparameterprocess,andisdenoted (X.,). I.le^(g)B(K+) isthe <r-algebrageneratedbythesemiring B(K+). isthe (7-algebrageneratedbythesemiring J^xB(IK2). 3 1,1f(^/),€r, isafiltrationandthereforesatisfies • foreach r>0, isa <r-algebracontainedin 3^, and • 3^s 3^t s<t. Wewillwritesimply {3',),andwewillassumethatthefiltrationsatisfiestheusual conditions: • 3o containsallthenegligiblesets (thatis, (3,) iscomplete),and • 3, ^ Pi every r>0 (thatis, {3,) is rightcontinuous). S>t See 1.1j belowforanotherassumptionabout {3,). 1.1gAfunction S:Q [0,c»] isastoppingtimeforthefiltration(3,)if{S<r}g 3, forevery t>0. Let S,T betwostoppingtimes.Thestochasticinterval (S,T] istheset {(cT,r)GnXK+ I S(cj)<1<T(c7)}, while [S,T) istheset {(c7,r)eDXIR+ I S(cj)</<T(cj)}. Thestochasticintervals (S,T)and [S,T] aredefinedinasimilarfashion. 1.1hThe predictable cr-algebra V ofsubsetsofQxK+ isthe a-algebra generatedbythesets Ax(^,r] and Bx{0}, where Ag3^- and Bg^o. AstoppingtimeSispredictableifthestochasticinterval [S,oo) is apredictable set. 1.1iLetSbeastoppingtime. 3s denotesthe cr-algebra {AeJ^j An{S<r}e^, forevery r>0} while 3s- denotesthe cr-algebrageneratedbythesetsin 3o aswellassetsofthe form An{S>r}, wheret>0and Ag3,. isthefiltrationdefinedbythefollowingrules. , 4 • ^0= ^0 and • Qs = forevery 5’>0. Wewillwritesimply {Q,), andwewillassumethatthefiltration (^,) issuchthat Gs = J^s- foreverypredictablestoppingtimeSforthefiltration (Gs)- Forexample,this isthecasewhen forevery5e1R+, where [sj isthelargestintegerless thanorequalto5. 1.1ki3^s,t)s,eR+ isadoublefiltrationandthereforesatisfies •• f3o^rsjea<c=h3s^,u,tv>0i,f 3s^s,<tuisaancdr-algre<bvr.acontainedin 3^, and Wewillwritesimply andwewillassumethat = Gs forevery 5,/>0. 1.1ILet BcOx B(c7) istheset | (t(7,j:,r)eB}. ;r[B],calledtheprojection of B,istheset {t(7en I {fD,s,i)e B forsome s,teK+}. 1.1mThepoint (00,oo) willbedenotedby 00. Therefore,theinequality {s,t) < 00 meansthat 5 < qo and t < 00. I.lnLet g:n^[0 00]X[0,00] beafunction. [g] denotestheset {(c;,5,r)g x j g(gj)=(5^f) calledthegraphof g. Wenowcommenceourstudyofthereal-valuedtwoparameterprocess(Xsj). We beginbydefiningthepredictable<T-algebra. Definition1.2 Let X:Qx ^ g parameterprocess. 1.2a(X^,,) is leftcontinuous ifforevery ^oToeK+ and cjgQ, wehave

See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.