12
I AD-RI60 942 MEAN CONVERGENCE OF THE VARIATIONS ITERATED INTEURALS 1/1 1 AND DOLERfiS-DODE E.. (U) NORTH CAOLINA UNIV AT CHAPEL HILL CENTER FOR STOCHASTIC PROC.. F AVRAN MAR 6 UNCLSSIFIED T-135 AFOSR-TR-86-0327 F4962-85-C-S144 F/G 2/1 NL

THE VARIATIONS ITERATED INTEURALS 1 AND DOLERfiS-DODE … · 2014. 9. 27. · i ad-ri60 942 mean convergence of the variations iterated inteurals 1/1 1 and dolerfis-dode e.. (u) north

  • Upload
    others

  • View
    2

  • Download
    0

Embed Size (px)

Citation preview

  • I AD-RI60 942 MEAN CONVERGENCE OF THE VARIATIONS ITERATED INTEURALS 1/11 AND DOLERfiS-DODE E.. (U) NORTH CAOLINA UNIV AT CHAPEL

    HILL CENTER FOR STOCHASTIC PROC.. F AVRAN MAR 6

    UNCLSSIFIED T-135 AFOSR-TR-86-0327 F4962-85-C-S144 F/G 2/1 NL

  • 12.8

    L 1 3 2 .111:6'IJill

    1ii.5*11

  • I~OCMENTATION PAGEIn. RESTRICTIVE MARKCINGS

    2& SE '- :.ASS.r CA;ON A RbILI 4n6 ISTRISUTIONAVAILAGIL OF REPOR'

    %au 06S86o ~ pproved "or :jblic Releasei Distributicn21. 0ECLAS~ FCATON, DOWNG S~FUEUni imi tea

    4 PERPORllMING ORGANIZATION R~EPORT NUMBdw 5. VIONITORING ORGANIZATION REPORT NI.JMSERIS)

    lechi-_-iI RportV".AFOSR Tt. S -0 32 76&. NAME OF OERFORMING ORGANIZATION b. OFFICE SYMBO0L 7s. NAME OF MONITORING ORGANIZATION

    .n~vers4-.:v o" North Carclina AFOSR/NM

    Oc. ADDRESS i(Y, 5St# and ZIP Cod., 7b. ADDRESS (City. Stat and ZIP Code)Center for Stochastic Processes, Statistics Bldg. 410Department, Phillips Hall 039-A, Bolling AFB, DC 20332-6448Chapel Hill, NC 27514 ____________________________

    G& NAME OF PUNOING/SPONSORING 8b. OFPICE SYMBOL 9. PROCUREMENT INSTRUMENT IDENTIFICATION NUMBER

    QAFOSRAINIapicbu F49620 85 C 0144

    &C. ADDRESS City, State aind ZIP Code.) 10. SOURCE OF FUNOiNG NOS.______

    Bd.40PROGRAM PROJECT TASK WORK UNITBalin roDELE MENT NO. NO. NO. NO.

    6.1102F 230411I. TITLE ticgau4a Secunty, CLawf~cation, aWeak- convergence of the variations, iterated integrals, n Doleans-Dide e.ponent aso

    * 12. PERSONAL AUTHORIS) sequences of semimartingales"Avram - F,

    134L TYPE OF REPORT 13b., TIME COVERED 14. GATE OF REPORT (Yr-. M.. Day) iS PAGE COUNT

    * Tehnicl IFROM ** L.. TO.fL.. M1arch 1986 616, SUPPLEMENTARY P1OTATION

    17 COSATI COOES IS& SUBJECT TERMS ICOanune on muerm if neesery and identfy~ by Ndock Nnmitbri

    * IELD GROUP SUB seGx. -Keywords:

    It. ABSTRACT 'cont"InU O1 '"ePse I f neceDIWy and identify by blocil number,

    If X is a sequence of semimartingales, converging to a senunmirtingale(n) (n)

    X, and such that [X, X converges to [X,XI, then all higher order variat ions(n)

    and all the iterated integrals of X converge jointly to the respective

    functionals of X.

    20. GISTRIBUTION/AVAILASILITY OP ABSTRACT 21. ABSTRACT SECURITY CLASSIFICATION

    UNCLAS31PIEG/UNLIMAITEC ) SAME AS PIPT OTZ ocusERS 7 UNCLASSIFIED221L NAME OF RESPONSIBLE INDIVIDUAL 22.TELEPM.ONE NUMBER I22c OFFICE SYMBOL

    I / (2~ '( 4,~yNvA4{)c)) ~ '>s C AFOSR/NM0D FORM 1473, 83 APR eoriON OP I IA4-13 IS OBSOLETE. UNCLASS I FIED

  • AYOSR-TR. 8 6~O2 - -..

    CENTER FOR STOCHASTIC PROCESSES

    Department of StatisticsUniversity of North CarolinaChapel Hill, North Carolina

    WEAK CONVERGENCE OF THE VARIATIONS, 1ITERATED INTEGRALS,

    AND DOLEANS-DADE EXPONENTIALS OF S1.QUNCES OF SEMIMARTINGALES

    by

    Florin Avram

    Technical Report No. 135

    March 1980

    t r6 0249

  • - -o m h 1i

    WEAK CONVERGENCE OF THE VARIATIONS, ITERATED INTEGRALS,

    AND DOLEANS-DADE EXPONENTIALS OF SEQUENCES OF SEMIMARTINGALES

    by

    Florin Avram

    University of North Carolina at Chapel Hill

    Abstract

    (n)If X is a sequence of semimartingales, converging to a semimartingale

    (n)(n)X, and such that [ X, X ] converges to [X,X], then all higher order variations

    (n)and all the iterated integrals of X converge jointly to the respective

    functionals of X.wI.,

    AT?' Y. .m'1

    . o . +: J:- 2.

    AMS 1980 Subject Classifications: Primary, 60F17; Secondary, 60105.

    Keywords and Phrases: Semimartingales, weak J1 Skoronod topology, variations,multiple integrals, Dolians-Dade exponential.

    This research supported by the Air Force Office of Scientific Research

    Contract No. F49620 85C 0144.

    I,-,

    S

    . . ,.:.,','- ' .,. ' +-,.%'%L,',-+.+., -,'-.+, ' +.+.,.. '"'" .,.- .""'''"."-". -" . .,', ,'" - +," .' .- -.- ''''. . . . "."",'."." ".."-"- .- ".

    .. . .. . . . . ,. , I Ii. + , +L + " + - ' + - + -! ' + + ' +

    ° t ! ' ' '

  • 1. Introduction

    (n)A. Let X t be a sequence of semimartingales, with tE [0,1], such that

    (n)(I.1) x x,

    where X is a semimartingale, and WO denotes weak convergence on D[OI]

    with respect to the Jj-Skorohod topology.

    -We investigate thelconvergence of the variations, iterated integrals

    (n)and Dolans Dade exponentials.of X , which are defined as follows: For Y

    a semimartingale, /

    r Y for k= 1t

    (1.2) V k(Y) = [YY]t = t+ 2 (AY s) for k= 2k t tt( ~ (AY) for k >t3

    sst

    Y for k-t

    (1.3) k (Y)t trI (Y)s dY for k 2

    (4E ) Ik-l s(1.4) E(XY) exp[XY - RY t:(XAYs

    t t T ts_

  • - - . .W - .-. w.- - P - ._ .-... X.,- - .- RJ

    2

    (n)

    Ik ( t 1 ,nn tIk( )t

  • I.&

    r.

    Corollary: Ifl>(n) w(JJ)

    (1. 9) x >x

    and the condition of Jacod (1983) holds:

    (1.10) lim sup P{Var(B hn)1 >b} = 0b-o n-o

    h~n(where h is a truncation function and (Bh'n) is the previsible projection

    tn

    of the truncated semimartingale X), then (1.5), (1.6), (1.7) and (1.8) hold.

    Proof: cf. Jacod (1983), Theorem 5.1.1, (1.9) and (1.10) imply (1.5).

    2. Proofs

    Introduce the following notation: For any real number x,

    '" >a,'-'"x := x'l{x 1>a}

    x X.1

    We establish now the following:(n)

    Lemma 1: a) Suppose X are semimartingales such that

    (n)(n)

    (2.1) lim lim PQ X, X ]>b} 0,b-~ n-o

    2and let f(x) be any real function such that f(x)=o(x2), as x-0. Then,

    for all c,

    (2.2) lim lim P{ f jf(AX-a) c = 0.a- 0 n- s

  • 4

    (n) (n) 2 o

    Proof: a) Note first that Estf(A Xs)jX t (X) " ts Xs-

  • Since the Do1e'ans-Dade exponential

    EXt=exp{x t - 2 [X X]I + f[/X T(X) ts tt

    it remains only to note that the functional:

    X a:D 2)[0,1] -~ D ()[0,1]

    Is continuous a.s., if both spaces are endowed with the respective J1 topo-

    *logies. Letting then a -0, as in the proof of Lemma 1, one gets:

    * ((n), (n)(n) f(n)< ~ £( (n)>, X X t 9 x( X-), R 'eAX>))

    s(J) [1x'x f(AX R II(AX >)t 5 5s~t s!- t

    2 2*since enl+ x) -x + =o(x ), and since (1.5) implies (2.1). Finally,

    * applying the continuous functional

    (4)

    P D [0,1] D[0,1]'

    O(Z19 Z ,Z , exp[Z -Z + Z Izl2394 1 2 2 3 4

    * we get that(n) ()

    E(X X ) -4')E(XX).

    *Since (1.6) is equivalent to (1.7) (by the use of the polynomial mapping),

    * and (1.6) trivially implies (1.5), Theorem I is proved.

  • References

    I.Avram, .,Taqqu, M.S.: Symetric Polynomials of Random VariablesAttracted to an Infinitely Divisible Law. To appear in Z. fUr Wahr.

    Z. Billingsley, ?.: Convergence of probability -measures. Wiley: New York,(1968).

    . Dnk-'n. 7.3. . Mandelbaun, A.: Symmetr~c szatistilcs, Poisson jDoint pro-cesses and miltioie Wiener integ-rals. Ann. of Statistics 11, 739-745,

    -. Denker, M. , Grillenberger, Chr. , Keller, 0.: Anote on invarianceorincfioles for von Mises' statistics. Metrika (to appear 1985).

    * ~ . cinilv -2 .. :. : Soecta' unctions, orobabilizy semigrouvs and 1-amiltonSlos. oo. iot-es in Mlath. , 696. Springer Verlag: New York (197S).

    licoDc 2. hecr~rneST L~ic;e ?our Les Prccessus. E-cole d'Eta de?rcoao:.. de esc .aint-F-lour Y=--l9SS. Laot. Noteas --n Mt

    -. >ln acc,,,Ms : nvariance 7rfnciola --:r svnetriz sta:is-

    Mjev.er...... n cours sur !os intoo4rales st-ochastfocues. Sen. Probl X,c:. Notzes '-.a'1-. ill. 2_53-00. :_-rincer -erlac.- N ew Yrk ~l7 6,

    iuL .....ae .. : Avooi distribution a:sv~erc statist-.cs.__________5 17:-0 (19 30) .

    -71.

  • Oo.

    ".5

    "" . .. . . ." - -" - ." . " " ' ' '. . ., ; . , , , . . , " . ." ' -' - ." ." - , ' 5" ' . .' ." . . ..-, -, -

    - " " -'U ' ' ' ' " ' " " " " .. ' "" ' '' " '"'*" " " "" " ' ,