RESIDUAL EMPIRICAL PROCESSES AND WEIGHTED SUMS FOR TIME-VARYING PROCESSES WITH APPLICATIONS TO TESTING FOR HOMOSCEDASTICITY

Size: px
Start display at page:

Download "RESIDUAL EMPIRICAL PROCESSES AND WEIGHTED SUMS FOR TIME-VARYING PROCESSES WITH APPLICATIONS TO TESTING FOR HOMOSCEDASTICITY"

Transcription

1 JOURNAL OF TIME SERIES ANALYSIS J. Time. Ser. Aal. 38: 7 98 (7) Published olie 9 Jue 6 i Wiley Olie Library (wileyolielibrary.com) DOI:./jsa. ORIGINAL ARTICLE RESIDUAL EMPIRICAL PROCESSES AND WEIGHTED SUMS FOR TIME-VARYING PROCESSES WITH APPLICATIONS TO TESTING FOR HOMOSCEDASTICITY a b GABE CHANDLER a* AND WOLFGANG POLONI b Deparme of Mahemaics, Pomoa College, Claremo, CA, USA Deparme of Saisics, Uiversiy of Califoria, Davis, CA, USA I he coex of heeroscedasic ime-varyig auoregressive (AR)-process we sudy he esimaio of he error/iovaio disribuios. Our sudy reveals ha he o-parameric esimaio of he AR parameer fucios has a egligible asympoic effec o he esimaio of he empirical disribuio of he residuals eve hough he AR parameer fucios are esimaed o-paramerically. The derivaio of hese resuls ivolves he sudy of boh fucio-idexed sequeial residual empirical processes ad weighed sum processes. Expoeial iequaliies ad weak covergece resuls are derived. As a applicaio of our resuls we discuss esig for he cosacy of he variace fucio, which i special cases correspods o esig for saioariy. Received Augus 4; Acceped April 6 eywords: Cumulas; empirical process heory; expoeial iequaliy; locally saioary processes; o-saioary processes; ess of saioariy.. INTRODUCTION Cosider he followig ime-varyig auoregressive (AR) process ha saisfies he sysem of differece equaios p Y k Y k D ; D ;:::;; () kd where k are he AR parameer fucios, p is he order of he model, is a fucio corollig he volailiy ad.; / deoe he i.i.d. errors. Followig Dahlhaus (997), ime is rescaled o he ierval Œ; i order o make a large sample aalysis feasible. Observe ha his, i paricular, meas ha Y D Y ; saisfyig () i fac forms a riagular array. The cosideraio of o-saioary ime series models goes back o Priesley (965) who cosidered evoluioary specra, ha is, specra of ime series evolvig i ime. The ime-varyig AR process has always bee a impora special case, eiher i more mehodological ad heoreical cosideraios of o-saioary processes, or i applicaios such as sigal processig ad (fiacial) ecoomerics (e.g. Subba Rao, 97; Greier, 983; Hall e al., 983; Raja ad Rayer, 996; Giraul e al., 998; Eom, 999; Drees ad Sǎricǎ, ; Orbe e al., 5; Fryzlewicz e al., 6; Chadler ad Poloik, 6). Oe of our coribuios is he esimaio of he (average) disribuio fucio of he iovaios D. Suppose ha we have observed Y p ;Y p ;:::;Y,adleY D.Y ;:::;Y p / ;D ;:::;:Give a esimaor b of D. ;:::; p / ad correspodig residuals b D Y b Y, we cosider he sequeial empirical disribuio fucio of he residuals give by Correspodece o: Gabe Chadler, Deparme of Mahemaics, Pomoa College, Claremo, CA 97, USA. gabriel.chadler@pomoa.edu Copyrigh 6 Joh Wiley & Sos Ld

2 WEIGHTED SUMS AND RESIDUAL EMPIRICAL PROCESSES 73 bf. ; / D b c D ¹b º ; R; Œ; : The correspodig disribuio fucio of he rue iovaios is F. ; / D uder appropriae codiios (Theorem ), Œ;; R P b c D ¹ º: We show ha bf. ; / F. ; / D o P = ; () eve hough o-parameric esimaio of he parameer fucios k is ivolved. This meas ha by usig appropriae esimaors for he parameer fucios, he (average) disribuio fucio of he ca be esimaed jus as well as if he parameers were kow. I parameric siuaios, his pheomeo is o ew, ad i perhaps was firs observed i Boldi (98). We refer o oul (), Ch. 7, for more complex siuaios. No-parameric models usually do o allow for such a pheomeo. See, for isace, Akrias ad va eilegom (), oul (), Schick ad Wefelmeyer (), Cheg (5), Müller e al. (7, 9a, 9b, ) ad va eilegom e al. (8). I coras o his work, our model, eve hough o-parameric i aure, has he crucial srucure of beig liear i he lagged observaios, which also meas ha he Y have mea zero. For more o he role of a zero mea i his coex, see Wefelmeyer (994) ad Schick ad Wefelmeyer (). (Geeralized) Auoregressive Codiioal Heeroscedasic-ype processes are cosidered i Horváh e al. (), Sue (), oul (), oul ad Lig (6) ad Laïb e al. (8) (cf. Remark (a) give ex o Theorem 4 ). The proof of he resuls jus meioed ivolves he behaviour of wo ypes of sochasic processes, boh of which are ivesigaed i his aricle. Oe is he residual sequeial empirical process, ad he oher ype is a geeralized parial sum process or weighed sums process. Boh are of idepede ieres. The residual sequeial empirical process is defied as. ; ; g/ D p b c ² g Y C ³ F g Y C = ; (3) D where Œ; ; R; g W Œ;! R p ; g G p D ¹g D.g ;:::;g p / ;g i Gº wih G a appropriae fucio class such ha b G p wih probabiliy edig o (see i he succeedig exs) ad F. /deoes he disribuio fucio of he errors : Observe ha. ; ; / D p.f. ; / EF. ; //,where,here, D.;:::;/ deoes he p-vecor of ull fucios. The basic form of is sadard, ad residual empirical processes based o o-parameric models have bee cosidered i he lieraure as idicaed earlier. Our coribuio here is o sudy hese processes based o a o-saioary Y of he form (). The secod key ype of processes i his aricle is weighed sums of he form Z.h/ D p D h Y ; h H; where H is a appropriae fucio class. Such processes ca be cosidered as geeralizaios of parial sum processes. Expoeial iequaliies ad weak covergece resuls for such processes are derived laer. We use properies of Z.h/ for provig (), ad his moivaes is sudy here. We also sudy he empirical disribuio of he esimaed errors bf. ; / D P b c D.b /; where b D b b. / for a appropriae esimaor b of he variace fucio : We will see ha uder appropriae codiios, J. Time. Ser. Aal. 38: 7 98 (7) Copyrigh 6 Joh Wiley & Sos Ld wileyolielibrary.com/joural/jsa DOI:./jsa.

3 74 G. CHANDLER AND W. POLONI Œ;; R bf. ; / F. ; / f. / p d e D b p D o P = ; (4) where F. ; / D P b c D. / deoes he empirical disribuio of he rue errors ad f is heir desiy. The derivaio of his resul ivolves he sudy of a residual sequeial empirical process slighly more geeral ha (3) i order o accommodae for he esimaio of he variace. Noice ha, from (4), we ca see ha i coras o he esimaio of he AR parameer fucios, he esimaio of he variace fucio is o egligible, eve if he variace was o be assumed cosa. See Secio 3 for more o his. Dahlhaus (997) advaced he formal aalysis of ime-varyig processes by iroducig he oio of a locally saioary process. This is a ime-varyig process wih ime beig rescaled o Œ; ha saisfies cerai regulariy assumpios; see (4 6) preseed laer. We would like o poi ou, however, ha, i our aricle, local saioariy is oly used o calculae he asympoic covariace fucio i Theorem 5. All he oher resuls hold uder weaker assumpios. The oulie of he aricle is as follows. I Secios, 3 ad 4, we aalyze he large sample behaviour of he fucio-idexed residual empirical processes ad of fucio-idexed weighed sums, respecively, uder he ime-varyig model (), ad apply he obaied resuls o show () ad (4). As a applicaio of he heoreical resuls, we discuss i Secio 5 a mehod for esig for homoscedasiciy, which i special cases is equivale o esig for saioariy. Proofs are deferred o Secio 6. Remark o measurabiliy. Suprema of fucio-idexed processes will eer he heoreical resuls give i he succeedig exs. We assume hroughou he aricle ha such rema are measurable.. RESIDUAL EMPIRICAL PROCESSES UNDER TIME-VARYING AUTOREGRESSIVE MODELS I order o formulae oe of our mai resuls for he residual empirical process. ; ; g/ defied i (3) earlier, we firs iroduce some more oaios ad formulae he uderlyig assumpios. Le Hdeoe a class of fucios defied o Œ; ad le d deoe a meric o H.Foragiveı>;le N.ı;H;d/ deoe he miimal umber N of d-balls of radius ı ha are eeded o cover H. Thelog N.ı;H;d/is called he meric eropy of H wih respec o d: If he balls A k are replaced by brackes B k D¹h H W g k h g k º for pairs of fucios g k g k ;kd;:::;n wih d.g k ;g k / ı, he he miimal umber N D N B.ı; H;d/ of such brackes wih H S N kd B k is called a brackeig coverig umber, ad log N B.ı; H;d/ is called he meric eropy wih brackeig of H wih respec o d: For a fucio h W Œ;! R, we deoe khk WD uœ; jh.u/j ad khk WD P D h : We furher deoe by d he meric geeraed by kk ; ha is, d.h; g/ Dkhgk : Assumpios. (i) The process Y D Y ; has a Movig Average (MA)-ype represeaio Y ; D a ;.j / j ; (5) j D where i:i:d:.; /: The disribuio fucio F of has a sricly posiive Lipschiz coiuous Lebesgue desiy f: The fucio.u/ i () is of bouded variaio wih <m <.u/<m < for all u. (ii) The coefficies a ;./ i (5) saisfy ja ;.j /j ; j D ; ; ; : : : `.j / where, for j>, `.j / D j.log j/ C for some ; >. Forj D ; ; we le `.j / D : wileyolielibrary.com/joural/jsa Copyrigh 6 Joh Wiley & Sos Ld J. Time. Ser. Aal. 38: 7 98 (7) DOI:./jsa.

4 WEIGHTED SUMS AND RESIDUAL EMPIRICAL PROCESSES 75 (iii) We have max gg p g Y D OP./ as!: (6) (iv) xr jxj Cf.x/< for some >. Assumpios (i) ad (ii) have bee used i he lieraure o locally saioary processes before. I is show i Dahlhaus ad Poloik (5) (see also Dahlhaus ad Poloik, 9) by usig a proof similar o üsch (995) ha Assumpio (i) holds for ime-varyig AR processes () if he zeros of he correspodig AR polyomials are bouded away from he ui disk (uiformly i he rescaled ime u) ad he parameer fucios are of bouded j variaio. I case p D, he fucios a ;.j / are of he form a ;.j / D Q j `D ` ; where.u/ D./ for u<ad, similarly,.u/ D./ for u<. Assumpio (iii) is also o oo surprisig as i is similar o assumpios used i he aalysis of parameric residual empirical processes, [e.g. oul,, Thm..3, Ass (..8)]. Assumpio (iv) is eeded o corol he ails of f i he case whe he variable i he process. ; ; g/ is allowed o rage over he eire real lie. Recallig he defiiio of. ; ; g/ give i (3), we see ha he special case g D (he vecor of zero fucios) gives a process oly based o he iovaios, which are idepede bu o ecessarily ideically disribued. Our firs heorem says ha if he idex class G is o oo large, he. ; ; g/ ad. ; ;/ behave he same asympoically: Theorem. Suppose ha Assumpios (i) ad (ii) hold. Le G deoe a fucio class such ha, for some c>; Z c= The we have for ay <L< ad ı! as!ha q log N B u ; G;d du < 8 : (7) Œ;;.L;L/; gg p W P p kd kg k kı j. ; ; g/. ; ; /j Do P./: (8) If, i addiio, Assumpios (iii) ad (iv) hold, he (8) also holds wih L D. The proof of Theorem ress o he crucial echical Lemma, which is give i Secio 6. This lemma implies asympoic sochasic equicoiuiy of he residual empirical sequeial process. ; ; g/ of which he asserio of Theorem is a immediae cosequece. Saemes of ype (8) are of ypical aure for work o residual empirical processes [e.g. (8..3) i oul, ]. Observe, however, ha, here, we are dealig wih ime-varyig AR processes ad are cosiderig o-parameric idex classes. Also keep i mid ha we are cosiderig riagular arrays (recall ha Y D Y ; ). The fucio class G is geeric i he formulaio of Theorem. As idicaed i Secio, wha we have i mid is fucio classes G modellig he differeces b k k [cf. Assumpio (vii)]. A specific example of such a class is give i he discussio ex o he formulaio of Theorem. Noice ha Theorem is cosiderig he differece of wo processes where he righ cerig is used. I coras o ha, we ow cosider he differece bewee wo sequeial empirical disribuio fucios: oe of hem based o he residuals ad he oher o he iovaios. As already said earlier, uder appropriae assumpios, he differece of hese wo fucios is o P = p, so ha he o-parameric esimaio of he parameer fucios has a egligible asympoic effec o he esimaio of he (average) iovaio disribuio. J. Time. Ser. Aal. 38: 7 98 (7) Copyrigh 6 Joh Wiley & Sos Ld wileyolielibrary.com/joural/jsa DOI:./jsa.

5 76 G. CHANDLER AND W. POLONI Furher assumpios. (v) jcum k. /jkšc k for k D ;;:::for some C>. (vi) For k D ;:::;p;we have k b k k k D O P.m / wih m!as!: (vii) There exiss a class G wih b k./ k./ G; k D ;:::;p;wih probabiliy edig o as!such ha gg kgk < ; ad for some C;c > ; we have R c= log N B.u; G;d /du<c <8: Theorem. Assume Assumpios (i) pad (ii) ad ha G is such ha (7) holds. I addiio, assume ha Assumpios (v) (vii) hold, wih m saisfyig D o./ as m!,where is such ha max Y D O P. /. The we have for <L<ha bf. ; / F. ; / D o P.= p /: (9) Œ;;.L;L/ If furher Assumpios (iii) ad (iv) hold, he (9) also holds wih L D: Discussio of assumpios: Assumpio (v) holds, for isace, whe Ej j k C k for all k D ;;:::. This is obviously a srog assumpio. However, much of he lieraure o locally saioary processes is usig i, i paricular hose i which raes of covergece for he esimaors b ad/or b are derived. I our work, his assumpios eer he picure hrough Theorem 5, which is used i he proof of Theorem. I is worh poiig ou i his coex ha he assumpio o m is ied o he large-sample behaviour of he maximum of he Y ;D ;:::; T; ad he srog assumpio (v) eails weak codiios o he esimaors b: I fac, uder Assumpio (v), we ca choose D log, which follows as i he proof of Lem 5.9 of Dahlhaus ad Poloik (9). Weakeig codiio (v) would require a sroger codiio o ; he rae of covergece of he esimaor b : The assumpios o he eropy iegral [see Assumpios (vii) ad (7)] corol he complexiy of he class G. May classes G are kow o saisfy hese assumpios see below for a example. For more examples, R we refer o he lieraure o empirical process heory. A more sadard codiio o he coverig umbers is p c= log N.u;G;d /du < (or similarly wih brackeig). Compared wih ha, he eropy iegral i Assumpio p (iv) does o have a square roo, ad he laer is similar o codiio (7) where he iegrad is log NB.u ; G;d / (oice he u ). This makes boh our eropy codiios sroger ha he sadard assumpio. The reaso for his is ha he expoeial iequaliy uderlyig our derivaios is o of sub-gaussia ype (Lemma 3), which i ur is caused by he depedece srucure of our uderlyig ime-varyig process. A class of o-parameric esimaors saisfyig codiios (vi) ad (vii) is give by he wavele esimaors of Dahlhaus e al. (999). These esimaors lie i he Besov smoohess class Bp;q s.c / where he smoohess parameers saisfy he codiio s C >:The cosa C > is a uiform boud o he (Besov) max.;p/ orm of he fucios i he class. Dahlhaus e al. derive codiios uder which heir esimaors coverge a rae log s=.sc/ i he L -orm. For s ; he fucios i Bp;q s.c / have uiformly bouded oal variaio. Assumig ha he model parameer fucios also posses his propery, he rae of covergece i he d -disace is he same as he oe i L, because i his case, he error i approximaig he iegral by he average over equidisa pois is of order O. /: Cosequely, i his case, we have m D log s=.sc/. Iorder o verify he codiio o he brackeig coverig umbers from Assumpio (iii), we use Nickl ad Pöscher (7). Their Cor., applied wih s D ; p D q D ; implies ha he brackeig eropy wih respec o he L -orm ca be bouded by Cı =. (Whe applyig heir Corollary o our siuaio, choose, i heir oaio, D ; D U Œ; ; r D ad D, say). 3. ESTIMATING THE DISTRIBUTION FUNCTION OF THE ERRORS T Esimaig he error disribuio ca be reaed by usig similar echiques as used for he iovaio disribuio, alhough he esimaio of he variace is o egligible see Theorem 4 ad Remark (a) righ ex o he heorem. For oher work ivolvig he esimaio of he error disribuio ad he volailiy, see, for isace, Akrias ad va eilegom () ad Neumeyer ad va eilegom (). wileyolielibrary.com/joural/jsa Copyrigh 6 Joh Wiley & Sos Ld J. Time. Ser. Aal. 38: 7 98 (7) DOI:./jsa.

6 WEIGHTED SUMS AND RESIDUAL EMPIRICAL PROCESSES 77 The sequeial empirical disribuio fucio of heb ca be rewrie as bf. ; / D b c D ¹b º D b c Accordigly, we defie he modified residual sequeial empirical process as D. b / Y C b! μ C :. ; ; g;s/d b c p F D g ² ³ g Y C s C Y C s C = i ; () where we hik of s S wih he class S beig a model for he differece b S. The followig is he aalogue o Theorem. Theorem 3. Assume Assumpios (i) ad (ii) ad ha G ad S are classes of ses saisfyig he eropy codiio (7). The we have for ay <L<ad ı! as!ha Œ;;.L;L/ ¹.g;s/G p SW P p kd kgkkckskı;º j. ; ; g;s/. ; ; ;/jdo P./: () If furher Assumpios (iii) ad (iv) hold, he asserio () also holds wih L Dad ksk replaced by ksk : Remarks: (a) The fac ha, for L D; we have o replace ksk by ksk is agai remiisce of assumpios used i he aalysis of parameric error processes [e.g. oul ad Lig, 6, Ass (4.4)]. (b) The residual sequeial empirical process from (3) correspods o o esimaig he variace. Formally, his correspods o usig he esimae of he variace beig cosa equal o (i.e. o dividig he residuals by a esimae of he variace). Sice S models he differece b ; his resuls i he class S D¹º cosisig of oly oe fucio. Wih his choice, he process () reduces o he process (3). I fac, we have. ; ; g/ D. ; ; g; /: Furher assumpios. (viii) We have kb k D o P. = / as!: (ix) There exiss a fucio class S wih b S wih probabiliy edig o as!, ss ksk < ; ad for some C;c > ; we have R c= log N B.u; S;d /du<c < 8 : Theorem 4. Assume Assumpios (i), (ii), (v) (ix) ad ha G ad S are classes of ses saisfyig he eropy codiio (7). I addiio, assume p ha f is differeiable wih bouded derivaive ad ha he sequece ¹m º i Assumpio (vi) is saisfyig D o./ as m!,where is such ha max Y D O P. /.The we have for <L<ha Œ;;.L;L/ bf d e. ; / F. ; / f. / b p D o P.= p /: () If furher Assumpios (iii) ad (iv) hold ad if D kbk D O P./, he () also holds wih L D: J. Time. Ser. Aal. 38: 7 98 (7) Copyrigh 6 Joh Wiley & Sos Ld wileyolielibrary.com/joural/jsa DOI:./jsa.

7 78 G. CHANDLER AND W. POLONI Remarks. (a) The heorem shows ha, i coras o he esimaio of he AR parameers, he esimaio of he variace fucio is o egligible, ad his eve holds i he i.i.d. case whe he variace is cosa. Tha he esimaio of he AR parameers k is egligible is due o he echical fac ha, i he Taylor expasio of E F b F. ; /; he esimaes of he AR parameers appear i sums of he form P D.b k / Yk.Now,if. b k / P lies i a class of fucios G; he his sum ca be absoluely bouded by gg D g Yk, ad because he Y k have mea zero ad he fucios g have small orm, he laer sum is o P. = / uder appropriae assumpios. I coras o ha, he sum of he differeces.b / does o ivolve he zero mea Y s, so ha he same rick does o apply. P (b) The erm f. / d e b. p /. / D i () is he firs-order sochasic approximaio o he differece of he appropriae cerigs F.b /. / Y C b F. /[see (49) ad (5) show laer]. The facor f. / also appears i oher work ha ivolves he esimaio of variace fucios such as Horváh e al. () ad oul ad Lig (6). (c) The wavele-based esimaor of he variace fucio of Dahlhaus ad Neuma () saisfies he ecessary assumpios o b for <L<. (d) I order o be able o uilize he earlier resuls o develop saisical mehodology, fier kowledge abou he P asympoic behaviour of he erm p d e b. /. / D is eeded. I ca be expeced ha, uder appropriae assumpios, his erm becomes asympoically ormal for reasoable esimaors, bu o he bes of our. / kowledge, such a resul is o available i he lieraure for esimaors of he variace fucio uder local saioariy. While i migh be possible o derive such a resul for cerai esimaors uder appropriae assumpios, explorig his quesio goes beyod he scope of his work. I is also uclear, wheher he asympoic ormaliy, if i ca be derived, could be easily used for saisical purposes, for wha really is eeded is he kowledge abou he asympoic disribuio of he sum F. ; / C f. / P p d e b. /. / D : For his,. / P a sochasic expasio of he sum p d e b. /. / D. / would be ideal. Such a expasio will of course heavily deped o he specific esimaor cosidered. Moreover, oe he has o be able o esimae he resulig asympoic variace, which migh o be sraighforward eiher, because he asympoic variace migh have a complex form. We would like o poi ou, however, ha oe of he basic ideas uderlyig our esig approach discussed i Secio 5 is o avoid esimaig he variace fucio alogeher ad exrac iformaio of he shape of he variace fucios by oher meas. 4. WEIGHTED SUMS UNDER LOCAL STATIONARITY The secod ype of processes of imporace i our coex is weighed parial sums of locally saioary processes give by Z.h/ D p D h Y ; h H: (3) I he i.i.d. case, weighed sums have received some aeio i he lieraure. For fucioal ceral limi heorems ad expoeial iequaliies, see, for isace, Alexader ad Pyke (986) ad va de Geer () ad he refereces herei. We will show i he succeedig discussio ha, uder appropriae assumpios, Z.h/ coverges weakly o a Gaussia process. I order o calculae he covariace fucio of he limi, we assume ha he process Y is locally saioary as i Dahlhaus ad Poloik (9). Recallig Assumpio (i), we assume he exisece of fucios a.;j/w.;! R wih wileyolielibrary.com/joural/jsa Copyrigh 6 Joh Wiley & Sos Ld J. Time. Ser. Aal. 38: 7 98 (7) DOI:./jsa.

8 WEIGHTED SUMS AND RESIDUAL EMPIRICAL PROCESSES 79 j D ja.u; j /j u `.j / ; (4) a;.j / a ;j ; (5) TV.a.;j// `.j / ; (6) where for a fucio g W Œ;! R; we deoe by TV.g/he oal variaio of g o Œ; : Codiios (4) (6) hold if he zeros of he correspodig AR polyomials are bouded away from he ui disk (uiformly i he rescaled ime u) ad he parameer fucios are of bouded variaio (see Dahlhaus ad Poloik, 6). Furher, we defie he ime-varyig specral desiy as he fucio wih A.u; / WD f.u;/wd ja.u; /j j D a.u; j / exp.ij/; ad c.u;k/ WD Z f.u;/exp.ik/d D j D a.u; k C j/a.u;j/ is he ime-varyig covariace of lag k a rescaled ime u Œ; : We also deoe by cum k./, hekh-order cumula of a radom variable. Theorem 5. Le H deoe a class of uiformly bouded, real-valued fucios of bouded variaio defied o Œ;. Assume furher ha for some C;c > ; Z c= log N.u;H;d /du<c < 8 : (7) The we have uder Assumpios (i), (ii) ad (v) ha, as!, he process Z.h/; h H; coverges weakly o a igh, mea zero Gaussia process ¹G.h/;h Hº: If, i addiio, (4) (6) hold, he he variace covariace fucio of G.h/ ca be calculaed as C.h ;h / D R h.u/ h.u/ S.u/ du; where S.u/ D P kd c.u;k/: Remarks. (a) Here, weak covergece is mea i he sese of Hoffma Jørgese see va der Vaar ad Weller (996) for more deails. (b) Weighed parial sums of he form Z. ; h/ D p b c g Y J. Time. Ser. Aal. 38: 7 98 (7) Copyrigh 6 Joh Wiley & Sos Ld wileyolielibrary.com/joural/jsa DOI:./jsa. D

9 8 G. CHANDLER AND W. POLONI are i fac a special case of processes cosidered i he heorem. Here, h.u/ D h g;.u/ D Œ;.u/ g.u/. Noe ha if g G ad G saisfies he assumpios o he coverig iegral from he aforemeioed heorem, he so does he class ¹h g;.u/ W g G; Œ; º. I his case, he limi covariace ca he be wrie as C.h g; ; h g; / D R ^ g.u/ g.u/ S.u/ du: (c) Assumpios (4) (6) are oly used for calculaig he covariace fucio of he limi process. The mai igredies o he proof of Theorem 5 are preseed i he followig resuls. These resuls are of idepede ieres. Theorem 6. Le ¹Y ; D ;:::;º saisfy Assumpios (i), (ii) ad (v), ad le H D¹h W Œ;! Rº be oally bouded wih respec o d : Furher, le A D P D Y M,whereM>. There exis cosas c ;c ;c >such ha, for all >saisfyig <6M p (8) ad >c Z log N.u;H;d /du _ 8M p! ; (9) we have P " # ² jz.h/j >;A c exp ³ : hh; khk c 5. APPLICATIONS We cosider a applicaio ha moivaes he aforemeioed sudy of he wo ypes of processes. The applicaio cosiss of esig for he cosacy of he variace fucio, ha is, esig for homoscedasiciy. We oe ha, for our es saisics, his paricular applicaio does o require esimaio of he variace fucio.u/;raher, he quesio is abou he cosacy of his fucio. This allows us o use he resuls of Theorem, avoidig he eed o hadle he bias erm arisig i Theorem 4 [see Remark (i) followig his heorem]. As we are primarily cocered wih quesios of homoscedasiciy, as well as lackig mehodology dealig wih disribuioal ess regardig he error erms i his seig o which o compare, we do o explore quesios regardig F here. We oe ha such applicaios would be more difficul because of his addiioal erm. However, whe explorig quesios of homoscedasiciy ad relaed quesios abou saioariy, he curre mehodology is sigificaly easier o impleme ha he mehods o which we compare; see succeedig discussio. Addiioally, more geeral quesios of homoscedasiciy are o able o be hadled by hose mehodologies. I he case of a ime-varyig AR model, he homoscedasic ad heeroscedasic model boh live i he aleraive space for he ess of saioariy. A secod esig problem, closely relaed alhough more ivolved ha he firs, is a es for deermiig he modaliy of he variace fucio, deails of which ca be foud i Chadler ad Poloik (). 5.. A es of homoscedasiciy We are ieresed i esig he ull hypohesis H W.u/ D for all u Œ;. (Noe ha if we addiioally assume ha he AR parameers do o deped o ime, he uder mild codiios o he k, his is also a es for wileyolielibrary.com/joural/jsa Copyrigh 6 Joh Wiley & Sos Ld J. Time. Ser. Aal. 38: 7 98 (7) DOI:./jsa.

10 WEIGHTED SUMS AND RESIDUAL EMPIRICAL PROCESSES 8 weak saioariy.) To his ed, cosider he process bg ;. / D b c D b bq ; Œ; ; () where bq is he empirical quaile of he squared residuals. Noice ha bg ;. / cous he umber of large (squared) residuals wihi he firs. /% of he observaios Y ;:::;Y, where large is deermied by he choice of. If he variace fucio is cosa, he, sice oe ca expec o have a oal of bc large residuals, he expeced value of bg ;. / approximaely equals : This moivaes he form of our es saisic, T D r Œ;. / bg ;. / : Followig he discussio of our applicaio, we argue ha he large-sample behaviour of his es saisic uder he ull is ha of he remum of a Browia bridge (Secio 5.). While here seems o be a obvious robus compoe o his mehodology, i is o obvious ha i should be very powerful, ad perhaps, oo much iformaio is beig los i cosiderig he daa i his way. As we are uaware of ay compeig ess for cosacy of he variace fucio i a oherwise o-saioary seig, we aemp o allay fears abou a lack of power by cosiderig he followig es of saioariy via our mehodology. We cosider he uivariae locally saioary model used i Puchsei ad Preuß (6), ; D C Z ; () where he Z N.; / i.i.d. Thus, we ru our es o simply he squared observaios. As compared wih our es, mos ess for saioariy i he lieraure (e.g. vo Sachs ad Neuma, ; Paparodiis, 9, ; Preuß e al., 3; Puchsei ad Preuß, 6) work i he specral domai, comparig a esimae of he ime-varyig specrum wih he saioary esimae. We compare he power behaviour of our es based o T o hree of hese es. To his ed, we use resuls preseed i Puchsei ad Preuß (6), which compare he mehods of Paparodiis () ad Preuß e al. (3) wih heir proposed mehod. This laes mehod has he beefi of o requirig a widow size o be chose, while he oher wo mehods were ru a wo differe widow sizes o opimize he procedures. For our proposed ime domai mehod, we use a D :8, alhough his does o prove o be oo criical. For example, wih D 56 ad uder Model (), he power was.87,.885 ad.86 for D.:7; :8; :9/. The proposed model is compeiive (Table I) ad much simpler boh cocepually ad implemeaio-wise. Accordig o our heory, he esimaio of he parameer fucios should o have a effec o he es, asympoically. To see how far his also holds for fiie samples, we simulaed from a var() of he form ; D :8 si ; C Z Table I. Simulaio resuls for model () T Paparodiis () Preuß e al. (3) Puchsei ad Preuß (6) J. Time. Ser. Aal. 38: 7 98 (7) Copyrigh 6 Joh Wiley & Sos Ld wileyolielibrary.com/joural/jsa DOI:./jsa.

11 8 G. CHANDLER AND W. POLONI q Figure. T D q.q/ jj OG Gjj (horizoal) vs p jjfo F jj (verical) ad he esimaed he parameer fucio.u/ D :8 si.u/ usig local leas squares followed by a kerel smooh. Noe ha his paricular form of he esimaor does o saisfy he assumpios uderlyig our heory, bu he discussio laer demosraes he robusess of he resul o hese assumpios. We geeraed daa ses accordig o his model wih D 56, ad he es saisic regardig he residuals was compued for each. We he geeraed a sample from a uiform disribuio ad compued he olmogorov Smirov saisic, properly ormalized for D as a sample from he limiig ull disribuio. A QQ plo of he wo es saisics is provided i Figure. We he simulaed from he model ; D :8 si ; C C Z ; which has he same volailiy fucio as he modulaed whie oise process (). Wih D 56, 884 of ime series from his model resuled i a rejecio of he ull of cosa variace usig a D :5. We oe ha his is very close o he esimaed power (.878) foud for Model (), as expeced. Based o he resuls comparig his idea wih muliple mehods i he beer sudied quesio abou saioariy, we are led o believe ha hese resuls are promisig, despie lackig a secod mehod o which o compare. 5.. Asympoic disribuio of a es saisic T Noice ha bg ;. / is closely relaed o he sequeial residual empirical process, ad as ca be see from he proof of Theorem 7, weighed empirical processes eer he aalysis of bg ;. / hrough hadlig he esimaio of q. Theorem 7 give i he succeedig exs provides a approximaio of he es saisic T by idepede (bu o ecessarily ideically disribued) radom variables. This resul crucially eers he proofs i Chadler ad Poloik (), where a similar es saisic is used o es for he modaliy of he variace fucio. I paricular, i implies ha he large-sample behaviour of he es saisic T uder he ull hypohesis is o iflueced by he, i geeral, o-parameric esimaio of he parameer fucios, as log as he rae of covergece of hese esimaors is sufficiely fas. Firs, we iroduce some addiioal oaio. Le f u deoe he pdf of.u/,hais,f u. / D f,.u/.u/ ad G ;. / D b c q ; where q is defied by meas of. / D Z F D du.u/ Z F du; ; ().u/ wileyolielibrary.com/joural/jsa Copyrigh 6 Joh Wiley & Sos Ld J. Time. Ser. Aal. 38: 7 98 (7) DOI:./jsa.

12 WEIGHTED SUMS AND RESIDUAL EMPIRICAL PROCESSES 83 as he soluio o he equaio.q / D : (3) Noice ha his soluio is uique sice we have assumed F o be sricly moooic, ad if.u/ D is cosa for all u Œa; b, heq equals he upper -quaile of he squared iovaios D : The approximaio resul ha follows does o assume ha he variace is cosa, however. Theorem 7. Le Œ; ad pose ha a<bare o-radom. The, uder Assumpios (i) (v), wih = log D o./; we have as!ha m where p bg ;. / G ;. / C c. /.G ;./ EG ;.// D op./; (4) Œ; c. / D R Œ f u.q / C f u.q / du R Œf u.q / C f u.q / du : Uder he ull hypohesis.u/ >for u Œ; ; we have c. / D : Moreover, i case he AR parameer i Model () is cosa ad p -cosise esimaors are used, he he mome assumpios o he iovaios ca be sigificaly relaxed o E < : Uder he ull hypohesis,.u/ D for all u Œ;, he iovaios are i.i.d. Usig ha i his case c. / D ad EG ;. / D ; we see ha he aforemeioed resul implies ha, i his case,.. // =p.bg ;. / / coverges weakly o a sadard Browia Bridge (cf. Chadler ad Poloik, ). 6. PROOFS Throughou he proofs, we use he oaio F u. / D P..u/ /. 6.. A crucial maximal iequaliy The proofs of Theorems ad 3 res o he followig lemma, which is of idepede ieres. I is modelled afer a similar resul for empirical processes (va de Geer,, Thm 5.). Le H L D Œ;.L; L/ Gp S deoe he idex space of he process defied i (), where <L. Defie a meric o H L as d ;W.h ;h / Dj jcjw. / W. /jc p kg ;k g ;k k Cks s k ; (5) where W.x/ D R x w.y/dy wih w.y/ > iegrable such ha, wih >from Assumpio (iv), we have lim jyj! y Cw.y/ D. Observe ha W is a sricly icreasig, posiive, bouded fucio o R ad he ails of f are o heavier ha he oes of w, assumig ha Assumpio (iv) holds. Lemma. Assume Assumpios (i) ad (ii). Furher assume ha boh G ad S are oally bouded wih respec o d.lea D P sdpc Y s C \ gg jg ± Y j <C : Defie D ss ksk C m C L C p pc : Suppose ha C ;; >are such ha kf k m J. Time. Ser. Aal. 38: 7 98 (7) Copyrigh 6 Joh Wiley & Sos Ld wileyolielibrary.com/joural/jsa DOI:./jsa. kd

13 84 G. CHANDLER AND W. POLONI 6 p ; (6) p. ^ /; (7) C Z = 8 p The for <L< ad C 6 p, we have wih C D q log N B u ; H L ;d ;W du _ : (8) 6. 6 C/ C C ha " # P.h /.h /j ; A C exp h ;h H L W d;w.h;h/ 6. 6 C / : If, i addiio, Assumpio (iv) holds ad, for some c >;we have if ss if uœ; Œ.s C /.u/ >c ; he he asserio also holds for L Dwih a modified (see Proof). Remarks: (a) The asserio also holds (wih a slighly modified ) o he simplified se A D P sdpc Y s C if we, i addiio, assume ha gg kgk < : (b) This lemma of course also applies o H L D Œ;.L; L/ G p replacig H L [wih he appropriaely modified (simplified) d ;W ]. Sice his case formally correspods o puig S D¹º, all he assumpios ivolvig S are rivially saisfied. Moreover, i his case, he assumpios gg kgk < ad xr jxj Cf.x/< for some >ca boh be dropped (see also Proof). (c) The assumpio if ss if uœ; Œ.s C /.u/ >c for some c >is fulfilled if ss ksk is arbirarily small (recall ha we assumed o be bouded away from zero). I our applicaio, we hik of s modellig he differece b, so ha (uiform) cosisecy of he esimaor b will esure ha, wih high probabiliy, b will lie i S; saisfyig he assumpio. Proof We oly prese a oulie of he proof. Le h D. ; ; g;s/ H L. Firs oice ha.h/ D. ; ; g;s/ is a sum of bouded marigale differeces. To see his, le ;g;s D g Y C s C ad ;g;s Q D ;g;s E ;g;s jf,wheref D. ; ;:::/ deoes he -algebra geeraed by ¹ ; ;:::º,he. ; ; g;s/d p D Q ;g;s : Obviously, also. ; ; g ;s /. ; ; g ;s / are sums of marigale differeces. The proof of he lemma is based o he basic chaiig device ha is well kow i empirical process heory, uilizig he followig expoeial iequaliy for sums of bouded marigale differeces from Freedma (975). Lemma (Freedma 975). Le Z ;:::;Z deoe marigale differeces wih P respec o a filraio ¹F ; D ;:::; º wih jz j C for all D ;:::;: Le furher S D p D Z ad V D V.S / D wileyolielibrary.com/joural/jsa Copyrigh 6 Joh Wiley & Sos Ld J. Time. Ser. Aal. 38: 7 98 (7) DOI:./jsa.

14 WEIGHTED SUMS AND RESIDUAL EMPIRICAL PROCESSES 85 P D E Z j F : The we have for all ; >ha P S ; V exp C C p! : (9) I order o be able o apply (9) o our problem, i is crucial o corol he quadraic variaio V.Weow idicae how o do his. Le ; ;g;s D ;g;s Q : We have for h D. ; ; g ;s /; h D. ; ; g ;s / H wih d ;W.h ;h / ha V DV..h /.h // " D E e ;s ; ;g D D j jc C D E ;g ;s For he las sum, we obai by elescopig C C D ;g E ;s F D F D F D e ; ;g;s E ;g;s F Q ;g ;s F ;g;s F : C F# D hq ;g E ;s ;g;s Q F i g Y C.s C / F g Y C.s C / (3) g Y C.s C / F g Y C.s C / (3) g Y C.s C / F g Y C.s C / (3) We esimae he hree erms o he righ separaely. To corol (3), le V.y/ D F Y C.s C / W.y/,whereW is iroduced i (5) earlier. The g.3/ D D V.W. // V.W. // D f g Y C.s C / W./ w W./ W. / W. /.s C / D (33) wih bewee W. / ad W. /, ad hus, by moooiciy of W; we have W./.L; L/ (sice ;.L; L/). I fac, for L<; we ca choose w.x/ D for x.l; L/ ad w.x/ D else, ad he erm (33) J. Time. Ser. Aal. 38: 7 98 (7) Copyrigh 6 Joh Wiley & Sos Ld wileyolielibrary.com/joural/jsa DOI:./jsa.

15 86 G. CHANDLER AND W. POLONI (i.e. (3)) ca herefore be bouded by kf k. ss ksk C m / m : For L D ; we also eed o cosider cases i which W./ is large. We fix L large eough so ha, for W./ > L,wehaveoA ha g Y C.s C / W./ > jw./=j ad ha jxj>l jxj Cw.x/ > =. The laer ca be achieved by our assumpio o w (give righ before he formulaio of he lemma). To see he former, assume ha (a) s C >c ad (b) gg jg Y j <C. The, if j j C c, he i is easy o see ha j. /j j j ; which is he asserio. (a) holds by assumpio ad (b) holds o A. Usig hese properies, we obai f g Y C.s C / W./ M w W./ j g Y C.s C / W./ /Cw W./ C.m / W./ C C.m / M: w.w.// Thus, for L D; we have he boud kf k.3/ max m ; C.m / xcf.x/ ksk C m : xr ss Nex, we cosider (3). By a applicaio of a oe-erm Taylor expasio, here exiss lyig bewee s ad s such ha, wih he shorhad oaio. / D g Y C C,wehaveha F g Y C.s C/ F g Y C.s C/ D f. /.s s / : Cosequely, for L<,.3/ D D f.s. / s / kf k L: For L D, we argue as follows. Sice lies bewee s ad s, we have from our assumpio ha C >c : As meioed earlier i he esimaio of (33), i ow follows ha if j j C c,he j. /j j j ; adwehaveforj j C c ha.3/ D D D f.s. / s / f m xr jxf.x/j m m xr jxf.x/j m :. /.s s. / /.s s / D m. /. / j j j. /j wileyolielibrary.com/joural/jsa Copyrigh 6 Joh Wiley & Sos Ld J. Time. Ser. Aal. 38: 7 98 (7) DOI:./jsa. (34)

16 WEIGHTED SUMS AND RESIDUAL EMPIRICAL PROCESSES 87 Cosequely, we have uiformly over R.3/ m C max kf k c ; m xr jxf.x/j : As for (3), we have o A (ad oe ha we do o eed he eve gg jg ± Y j <C o hold here).3/ u;x f u.x/ kf k m.g g / D v p u kg k g k k p kd Y Dp Y p pc kf k m : (35) Puig everyhig ogeher, we obai ha, for h ;h H L wih d ;W.h ;h / ad,wehaveo A for <L< ha V..h /.h // ; (36) wih D kf k m ss ksk C m C L C p pc ; as defied i he formulaio of he lemma. If L D; he he formula jus show shows ha (36) holds where he correspodig ca immediaely be obaied from he aforemeioed esimaes. Proofs of saemes i Remarks give afer he lemma (a) If we modify A by droppig he eve gg jg ± Y j <C gg kgk <, he we ca esimae he las average i (34) as follows: D ad isead assume ha v.s s / u g Y.s s / g Y / D D v u g Y / D v u kgk p Y gg DpC C = kgk p! : gg This he leads o a slighly modified cosa, which ca easily be deermied. (b) If H L is replaced by H L he, as has bee discussed earlier, his formally correspods o seig S D¹ º, which is a class cosiss of jus oe fucio, implyig ha, i he aforemeioed proof, s s D.Ioher words, all he assumpios used o boud (3) become obsolee. J. Time. Ser. Aal. 38: 7 98 (7) Copyrigh 6 Joh Wiley & Sos Ld wileyolielibrary.com/joural/jsa DOI:./jsa.

17 88 G. CHANDLER AND W. POLONI The jus show corol of he quadraic variaio i cojucio wih Freedma s expoeial boud for marigales ow eables us o apply he (resriced) chaiig argume i a way similar o he proof of Thm 5. i va de Geer (). Deails are omied. 6.. Proofs of Theorems ad 3 We oly prese he proof of Theorem 3. The proof of Theorem is exacly he same. Recall ha he meric d ;W.h ;h / o H L is defied i (5), ad observe ha, for h D. ; ; g;s/ ad h D. ; ; ;/; we have d ;W.h ;h / D P p kd kg k k Cksk. We hus obai for >ad <L<ad C>ha B C j. ; ; g;s/. ; ; ;/ja Œ;;.L;L/;gG Pp p kd kg k kckskı P A { C P d ;W.h ;h /ı h ;h H L C j.h /.h /j; A A ; where as i Lemma, A D P sdpc Y s C \ gg jg ± Y j <C for some C ;C >: A applicaio of Lemma gives he asserio oce we have show ha P A { ca be made arbirarily small q log N B u ; H L ;d ;W du <. (for sufficiely large ) adha R c To see his, oice ha, by our assumpios, we have boh R c p log NB.u ; G;d /du < ad R q R q c log N B u ; S;d du < : This implies c log N B u ; H L ;d ;W du <, because for hi D. i; i; g i ;s i /; i D ; wih j j < ; jw. / W. /j < ; kg 4 4 k g k k < ;k D ;:::;p ad 4p ksk <, we obviously have d 4 ;W.h ;h /<;ad hus, by usig sadard argumes, i is o difficul o see ha log N B ; H L ;d ;W C log 4 C p log N B ; G;d 4p C log N B ; S;d 4 (37) for some C >. I remais o show ha P.A { / ca be made arbirarily P small (for large) by choosig C ;C sufficiely large. Because of Assumpio (iii), his follows from D E Y < ; whichiurfollows easily from EY D E 4 j D 3 a ;. j/ j 5 D E for some C>. Here, we are usig Assumpio (ii) Proof of Theorem 5 D j D kd j D a ;. j/a ;. k/ j k a ;. j/ j D <C <; `.j / Firs, we formulae ad prove a lemma ha is eeded i he proof of Theorem 5. For radom variables ;:::; k, we deoe by cum. ;:::; k / heir joi cumula, ad if i D for all i D ;:::;k, he cum. ;:::; k / D cum.;:::;/d cum k./,hekh-order cumula of. wileyolielibrary.com/joural/jsa Copyrigh 6 Joh Wiley & Sos Ld J. Time. Ser. Aal. 38: 7 98 (7) DOI:./jsa.

18 WEIGHTED SUMS AND RESIDUAL EMPIRICAL PROCESSES 89 Lemma. Le ¹Y ;D;:::;º saisfy Assumpios (i) ad (ii). For j D ;;:::,leh j be fucios defied o Œa; b wih kh j k < : The here exiss a cosa < such ha for all k such ha cum.z.h /;:::;Z.h k // k cum k. / If, i addiio, kh j k M<; jd ;:::;k, he for k 3; ky kh j k : j D cum.z.h /;:::;Z.h k //. / k M k cum k. / k : Proof P We have cum.z.h /;:::;Z.h k // D k h k ; ;:::; kd :::hk cum.y ;:::;Y k / by uilizig mulilieariy of cumulas. I order o esimae cum.y ;:::;Y k /; we uilize he special srucure of he Y - variables. Sice he j are idepede ad have mea zero, we have ha cum. j ;:::; jk / D uless all he j`; `D ;:::;k are equal, ad by agai usig mulilieariy of he cumulas, we obai cum.y ;:::;Y k / D cum k. / mi¹ ;:::; kº j D a ;. j/:::a k;. k j/: (38) Thus, cum.y ;:::;Y k / cum k. / P cum.z.h /;:::;Z.h k // k D k j D Q k id cum k. / cum k. / `.j i ; ad cosequely, j j/ ky 4 hi j D id i D Y 4 hi j D id i D ky 4 i hi id3 i D i i `.j i j j/ 3 5 `.j i j j/ 3 5 `.j i j j/ 3 5 : Uilizig he Cauchy Schwarz iequaliy, we have for he las produc 3 v ky 4 i hi ky u 5 i h i `.j i j j/ id3 i D id3 i D v k Y u k kh i k id3 k k v uu `.j i j j/ D ky kh i k ; id3 i D `.jj/ (39) J. Time. Ser. Aal. 38: 7 98 (7) Copyrigh 6 Joh Wiley & Sos Ld wileyolielibrary.com/joural/jsa DOI:./jsa.

19 9 G. CHANDLER AND W. POLONI r P P where we used he fac ha D `.jj/ D boud (39) does o deped o he idex j aymore, so ha cum.z.h /;:::;Z.h k // k By usig P j D `.j j j/ `.jj/ ky kh i k cum k. / id3 for some < : Noice ha he j D id 3 Y 4 i hi 5 : `.j i j j/ i D D P `.j j j/ j D P `.j j j/ `.j. /C j j/ `.j j/ j D for some `.j j/ >ad Cauchy Schwarz iequaliy, we obai 3 Y 4 i hi 5 D h `.j i j j/ j D id i D D D h h `.j j/ D v D v u u h h `.j j/ D kh k kh k : D D h D j D `.j j/ `.j j j/ `.j j j/ `.j j j/ This complees he proof of he firs par of he lemma. The secod par follows similar o he aforemeioed by observig ha if kh i k <Mfor all i D ;:::;k, he, isead of he esimae (39), we have wih D P D ha `.jj/ 3 ky 4 i hi Y k 5 M k `.j i j j/ id3 i D id3 i D `.j i j j/.m / k : Now, we coiue wih he proof of Theorem 5. Showig weak covergece of Z.h/ meas provig asympoic ighess ad covergece of he fiie dimesioal disribuio (e.g. va der Vaar ad Weller, 996). Tighess follows from Theorem 6. I remais o show he covergece of he fiie dimesioal disribuios. To his ed, we will uilize he Cramér Wold device i cojucio wih he mehod of cumulas. I follows from Lemma ha all he cumulas of Z.h/ of order k 3 coverge o zero as!: Usig he lieariy of he cumulas, he same holds for ay liear combiaio of Z.h i /; i D ;:::;:The mea of all he Z.h/ equals zero. I remais o show ha covergece of he covariaces cov.z.h /; Z.h //: The rage of he summaio idices show ex are such ha he idices of he Y -variables are bewee ad. For ease of oaio, we achieve his by formally seig h i.u/ D for u ad u>;id ; : We have cov.z.h /; Z.h // D D h D jkj p D sd h h s cov.ys ;Y / cov.y ;Y k / C R ; h k (4) wileyolielibrary.com/joural/jsa Copyrigh 6 Joh Wiley & Sos Ld J. Time. Ser. Aal. 38: 7 98 (7) DOI:./jsa.

20 WEIGHTED SUMS AND RESIDUAL EMPIRICAL PROCESSES 9 where for sufficiely large, jr j h D jkj> p k h cov.y ;Y k /: From Proposiio 5.4 of Dahlhaus ad Poloik (9), we obai ha jcov.y ;Y k /j for some cosa : Sice boh h ad h are bouded ad P kd < ; we ca coclude ha R `.k/ D o./: The mai erm i (4) ca be approximaed as where jr ; j D jkj p h D jkj p h `.k/ k h c ;k C R ; (4) k h cov.y ;Y k / c : ;k Proposiio 5.4 of Dahlhaus ad Poloik (9) also gives us ha, for jkj p,wehave P c ;k C jkj for some >. Usig his, we obai jr ; j p h D kd p p kd p D h k cov.y ;Y k / c cov.y ;Y k / c ;k as! : Nex, we replace h. k / i he mai erm of (4) by h bouded by h D jkj p h k h ;k p kd p D cov.y ;Y k / C jkj D o./ `.jkj/. The approximaio error ca be `.jkj/ D o./: Here, we are usig he fac ha u jc.u;k/j (see Prop 5.4 i Dahlhaus ad Poloik, 9) ogeher `.jkj/ wih he assumed (uiform) coiuiy of h, he boudedess of h ad he boudedess of P kd.we `.jkj/ have see ha cov.z.h /; Z.h //D h h c k p ;k C o./: D Sice S.u/ D P kd c ;k < ; we also have cov.z.h /; Z.h //D kd h h c ;k C o./: D J. Time. Ser. Aal. 38: 7 98 (7) Copyrigh 6 Joh Wiley & Sos Ld wileyolielibrary.com/joural/jsa DOI:./jsa.

21 9 G. CHANDLER AND W. POLONI Fially, we uilize he fac ha TV.c.;k//, which is aoher resul from Prop 5.4 of Dahlhaus ad `.jkj/ Poloik (9). This resul, ogeher wih he assumed bouded variaio of boh h ad h ; allows us o replace he average over by he iegral A crucial expoeial iequaliy The followig expoeial iequaliy is a crucial igredie o he proof of Theorem 6. I relies o he corol of he cumulas, which is provided i Lemma show earlier. Lemma 3. Le ¹Y ;D;:::;º saisfy Assumpios (i) ad (ii). Le h be a fucio wih khk <. Assume ha here exiss a cosa C>such ha, for all k D ;;:::;we have jcum k. /jkšc k : The here exis cosas c ;c >such ha, for ay >;we have ² P Œ jz.h/j > c exp c khk Proof Usig Lemma, we have he assumpios o he cumulas implyig ha Z.h/./ D log Ee Z.h/.C khk / k ; kd assumig ha >is such ha he ifiie sum exiss ad is fiie. We obai Choosig D P ŒjZ.h/j >e E e Z.h/ exp C ³ : (4) μ.c khk / k : C khk gives he asserio wih c D e = ad c D C : The fac ha cum j. / jšc j ; j D ;;::: (43) kd holds if Ej j k C k ;kd;;:::;ca be see by iducio. Deails are omied Proof of Theorem 6 Usig Lemma 3, we ca mimic he proof of Lem 3. from va de Geer (). As compared wih ha of va de ± Geer, our expoeial boud is of he form c exp c khk raher ha c exp ²c khk ³.Iiswellkow ha his ype of iequaliy leads o he coverig iegral beig he iegral of he meric eropy raher ha he square roo of he meric eropy. (See, for isace, Thm..4 i va der Vaar ad Weller, 996.) This idicaes he ecessary modificaios o he proof i va de Geer. Deails are omied Proofs of Theorems ad 4 Proof of Theorem 4 Recall ha bf. ; / D P b c D.b / Y C.b / ± C ad ha, by wileyolielibrary.com/joural/jsa Copyrigh 6 Joh Wiley & Sos Ld J. Time. Ser. Aal. 38: 7 98 (7) DOI:./jsa.

22 WEIGHTED SUMS AND RESIDUAL EMPIRICAL PROCESSES 93 assumpio, P b ; b G p S! as!.le F. ; ; g;s/d b c D ² g Y C.s C / ³ : (44) The we have wih F. ; ; ;/ D F. ; ;b ;b /. Furher, le P b c D ¹ º ha F. ; / D F. ; ; ;/ ad bf. ; / D E. ; ; g;s/d b c D F g Y C.s C / (45) deoe he codiioal expecaio of F. ; ; g;s/ give F, wheref D.Y ;Y ;:::/. The purpose of iroducig E is o make F E a marigale differece (see i he followig exs). We ow have bf. ; / F. ; / D F. ; ; b ;b / F. ; ; ;/ i D h.f E /. ; ; b ;b /.F E i /. ; ; ;/ C he. ; ; b ;b / E. ; ; ;/ (46) DW T. ; ;b ;b /C T. ; ;b ;b /: Usig his decomposiio, we will show he asserio by provig he followig wo properies: p T. ; ;b ;b / D op./ (47) Œ;;.L;L/ " p T. ; ;b ;b / f. / Œ;;.L;L/ b # D o P./: (48) D Verificaio of (47). Le. ; ; g;s/ D p F E. ; ; g;s/deoe he error sequeial empirical process. By akig io accou ha, by assumpio, P p kd kb k k k D O P.m / ad kb k D O P. =4 /; we have ha, for ay >;we ca fid a C>such ha, wih probabiliy a leas for large eough ; p T. ; ;b ;b / Œ;;.L;L/ Œ;;.L;L/; gg p ; P p kd kg k kcm ss; kskc =4 j p T. ; ; g;s/j: Tha he righ-had side is o P./ is a immediae applicaio of Theorem 3, ad (47) is verified. Verificaio of (48). We have by a simple oe-erm expasio ha J. Time. Ser. Aal. 38: 7 98 (7) Copyrigh 6 Joh Wiley & Sos Ld wileyolielibrary.com/joural/jsa DOI:./jsa.

23 94 G. CHANDLER AND W. POLONI " p T. ; ;b ;b / D p d e F D f. / D d e p C p d e D. b / Y C b! # F. / ". b / Y C b # (49) ". b / f.. // Y C b # (5) D wih. / bewee ad.b/. / Y C b. /. The erm (5) is a remaider erm ha will be reaed as show. /. / laer. The sum i (49) ca be wrie as a sum of wo erms f. / d e p D.b / Y C f. / d e p D b : (5) The firs erms are (fiie sum of) weighed sums of he Y s, ad we will ow use ± Theorem 6 o show ha his erm is o P./. By recallig ha, by assumpio, he probabiliy of b G p eds o as!, we ca assume ha.b / k G for all k D ;:::;p.forh H D¹h ;g D.u/ g.u/ I Œ; ; g Gº wih he.u/ shorhad oaio.u/ D.u /, le Z k;.h/ D p D The we ca wrie he firs erm i (5) as f. / P p ± eve b G p )ha h Y k ; k D ;:::;p: (5) kd Z k;.b/ k : Wih his oaio, we have (o he d e f. / p D.b / Y f.x/ x p kd hh jz k;.h/j: I hus remais o show ha hh jz k;.h/j D o P./ for all k D ;:::;p: To his ed, we will apply Theorem 6. Clearly, hh khk m gg kgk m m D o./: Sice EZ k;.h/ D for all h H, a applicaio of Theorem 6 ow gives he asserio oce we have verified ha he class H saisfies he codiio o he coverig umbers. This, however, follows easily because by assumpio, G has a fiie coverig iegral wih respec o d ; is jus a sigle fucio ha is bouded above ad below ad he class of fucios.u/ ¹.u/; Œ; º is a Vapik-Chervoekis (VC)-subgraph class (ad hus has a fiie coverig iegral as well). Now, we show ha he erm i (5) is o P./. Wehave.5/ p C p D D ". b / # f.. // Y (53) " b # f.. // ; (54) wileyolielibrary.com/joural/jsa Copyrigh 6 Joh Wiley & Sos Ld J. Time. Ser. Aal. 38: 7 98 (7) DOI:./jsa.

24 WEIGHTED SUMS AND RESIDUAL EMPIRICAL PROCESSES 95 ad uder our assumpios, boh (53) ad (54) are o P./ uiformly over. I fac, for (53), we have p D ". b / f.. // x f.x/ p m D # Y.b / Y ; where for he las sum we have (recall ha is such ha max Y D O P / p D.b / Y p D b / k kd p max Y p OP p m p Y j j D p D p m A O P./ D o P./; where he las equaliy uses our assumpios. The fac ha (54) i o P./ follows immediaely oce we have show ha f.. // D O P./: (55) R To see ha (55) holds, firs observe ha, by assumpio, we have lim jxj! jxf.x/j D; ad a applicaio of l Hospial s Rule gives lim jxj! jx f.x/j D: Sice f is coiuous, we coclude ha f is bouded, ad hus, i is clear ha (55) holds for j j <L<: For L D, he argume is as follows. The previous argumes show ha xr jf.x/x j < : Sice max R see ha i is sufficie o show ha max R jf.. // j D R max jf.. //. /jj. /j, we. / D O P./: (56) This follows by a argume alog he lies used o boud (33) i he proof of Lemma 3. Furher deails are omied. Treaig residual empirical process, i.e. provig Theorem, is very similar. We omi he deails. Proof of Theorem 7 Firs recall ha bf. ; / D we ca wrie P b c D.b / ad F. ; / D p b G ;. / G ;. / P b c D : Wih his oaio, D p h i bf. ;bq / C bf. ; bq /. F. ; q / C F. ; q // h p D F. ;bq / bf. ;bq / p i F. ; bq / bf. ; bq / p. F. ;bq / F. ; q / / p. F. ; bq / F. ; q / / DW I. / II. / (57) J. Time. Ser. Aal. 38: 7 98 (7) Copyrigh 6 Joh Wiley & Sos Ld wileyolielibrary.com/joural/jsa DOI:./jsa.

25 96 G. CHANDLER AND W. POLONI wih I. / ad II. /; respecively, deoig he wo quaiies iside he wo Œ-brackes. The asserio of Theorem 7 follows from ji. /j Do P./ ad (58) Œ; jii. / c. /.G ;./ EG ;.// jdo P./: (59) Œ; Clearly, Œ; ji. /j Œ;; R jbf. ; / F. ; /j; ad hus, propery (58) follows from Theorem. Propery (59) will ow be show by usig empirical process heory based o idepede, bu o ideically disribued, radom variables. Proof of (59) Defie We ca wrie F. ; / WD EF. ; / D b c D F! : II. / D p.f F /. ;bq /.F F /. ; q / (6) C p.f F /. ; bq /.F F /. ; q / (6) C p.f. ;bq / F. ; q // p.f. ; bq / F. ; q //: (6) The process. ; ; / D p.f F /. ; / is a sequeial empirical process, or a iefer Müller process, based o idepede, bu o ecessarily ideically disribued, radom variables. This process is asympoically sochasically equicoiuous, uiformly i wih respec o.v; w/ D F.; v/ F ;.; w/. Tha is, for every >;here exiss a >wih lim P! " # j. ; ; /. ; ; /j > Œ;;. ; / I fac, wih. ; /;. ; / / Dj jc. ; /; we have Œ; j. ; ; /. ; ; /j ; ;R;. ; / ; Œ;; ; R.. ; /;. ; // D : (63) j. ; ; /. ; ; /j: Thus, (63) follows from asympoic sochasic d -equicoiuiy of. ; ; /. This i ur follows from a proof similar o, bu simpler ha, he proof of Lemma. I fac, i ca be see from (35) ha, for g D g D ; we simply ca use he meric.. ; /;. ; / / Dj j C. ; / i he esimaio of he quadraic variaio, which i he simple case of g D g D amous o he esimaio of he variace, because he radomess oly comes i hrough he. Wih his modificaio, he proof of he -equicoiuiy of. ; ; / follows he proof of Lemma. Thus, if bq is cosise for q wih respec o, he i follows ha boh (6) ad (6) are o P./. The proof of he cosisecy is omied here. (I is available from he auhors.) This complees he proof. wileyolielibrary.com/joural/jsa Copyrigh 6 Joh Wiley & Sos Ld J. Time. Ser. Aal. 38: 7 98 (7) DOI:./jsa.

BEST LINEAR FORECASTS VS. BEST POSSIBLE FORECASTS

BEST LINEAR FORECASTS VS. BEST POSSIBLE FORECASTS BEST LINEAR FORECASTS VS. BEST POSSIBLE FORECASTS Opimal ear Forecasig Alhough we have o meioed hem explicily so far i he course, here are geeral saisical priciples for derivig he bes liear forecas, ad

More information

Lecture 15 First Properties of the Brownian Motion

Lecture 15 First Properties of the Brownian Motion Lecure 15: Firs Properies 1 of 8 Course: Theory of Probabiliy II Term: Sprig 2015 Isrucor: Gorda Zikovic Lecure 15 Firs Properies of he Browia Moio This lecure deals wih some of he more immediae properies

More information

1 Notes on Little s Law (l = λw)

1 Notes on Little s Law (l = λw) Copyrigh c 26 by Karl Sigma Noes o Lile s Law (l λw) We cosider here a famous ad very useful law i queueig heory called Lile s Law, also kow as l λw, which assers ha he ime average umber of cusomers i

More information

STK4080/9080 Survival and event history analysis

STK4080/9080 Survival and event history analysis STK48/98 Survival ad eve hisory aalysis Marigales i discree ime Cosider a sochasic process The process M is a marigale if Lecure 3: Marigales ad oher sochasic processes i discree ime (recap) where (formally

More information

Extremal graph theory II: K t and K t,t

Extremal graph theory II: K t and K t,t Exremal graph heory II: K ad K, Lecure Graph Theory 06 EPFL Frak de Zeeuw I his lecure, we geeralize he wo mai heorems from he las lecure, from riagles K 3 o complee graphs K, ad from squares K, o complee

More information

Math 6710, Fall 2016 Final Exam Solutions

Math 6710, Fall 2016 Final Exam Solutions Mah 67, Fall 6 Fial Exam Soluios. Firs, a sude poied ou a suble hig: if P (X i p >, he X + + X (X + + X / ( evaluaes o / wih probabiliy p >. This is roublesome because a radom variable is supposed o be

More information

Ideal Amplifier/Attenuator. Memoryless. where k is some real constant. Integrator. System with memory

Ideal Amplifier/Attenuator. Memoryless. where k is some real constant. Integrator. System with memory Liear Time-Ivaria Sysems (LTI Sysems) Oulie Basic Sysem Properies Memoryless ad sysems wih memory (saic or dyamic) Causal ad o-causal sysems (Causaliy) Liear ad o-liear sysems (Lieariy) Sable ad o-sable

More information

A Note on Prediction with Misspecified Models

A Note on Prediction with Misspecified Models ITB J. Sci., Vol. 44 A, No. 3,, 7-9 7 A Noe o Predicio wih Misspecified Models Khresha Syuhada Saisics Research Divisio, Faculy of Mahemaics ad Naural Scieces, Isiu Tekologi Badug, Jala Gaesa Badug, Jawa

More information

ODEs II, Supplement to Lectures 6 & 7: The Jordan Normal Form: Solving Autonomous, Homogeneous Linear Systems. April 2, 2003

ODEs II, Supplement to Lectures 6 & 7: The Jordan Normal Form: Solving Autonomous, Homogeneous Linear Systems. April 2, 2003 ODEs II, Suppleme o Lecures 6 & 7: The Jorda Normal Form: Solvig Auoomous, Homogeeous Liear Sysems April 2, 23 I his oe, we describe he Jorda ormal form of a marix ad use i o solve a geeral homogeeous

More information

Lecture 9: Polynomial Approximations

Lecture 9: Polynomial Approximations CS 70: Complexiy Theory /6/009 Lecure 9: Polyomial Approximaios Isrucor: Dieer va Melkebeek Scribe: Phil Rydzewski & Piramaayagam Arumuga Naiar Las ime, we proved ha o cosa deph circui ca evaluae he pariy

More information

The Central Limit Theorem

The Central Limit Theorem The Ceral Limi Theorem The ceral i heorem is oe of he mos impora heorems i probabiliy heory. While here a variey of forms of he ceral i heorem, he mos geeral form saes ha give a sufficiely large umber,

More information

Exercise 3 Stochastic Models of Manufacturing Systems 4T400, 6 May

Exercise 3 Stochastic Models of Manufacturing Systems 4T400, 6 May Exercise 3 Sochasic Models of Maufacurig Sysems 4T4, 6 May. Each week a very popular loery i Adorra pris 4 ickes. Each ickes has wo 4-digi umbers o i, oe visible ad he oher covered. The umbers are radomly

More information

10.3 Autocorrelation Function of Ergodic RP 10.4 Power Spectral Density of Ergodic RP 10.5 Normal RP (Gaussian RP)

10.3 Autocorrelation Function of Ergodic RP 10.4 Power Spectral Density of Ergodic RP 10.5 Normal RP (Gaussian RP) ENGG450 Probabiliy ad Saisics for Egieers Iroducio 3 Probabiliy 4 Probabiliy disribuios 5 Probabiliy Desiies Orgaizaio ad descripio of daa 6 Samplig disribuios 7 Ifereces cocerig a mea 8 Comparig wo reames

More information

David Randall. ( )e ikx. k = u x,t. u( x,t)e ikx dx L. x L /2. Recall that the proof of (1) and (2) involves use of the orthogonality condition.

David Randall. ( )e ikx. k = u x,t. u( x,t)e ikx dx L. x L /2. Recall that the proof of (1) and (2) involves use of the orthogonality condition. ! Revised April 21, 2010 1:27 P! 1 Fourier Series David Radall Assume ha u( x,) is real ad iegrable If he domai is periodic, wih period L, we ca express u( x,) exacly by a Fourier series expasio: ( ) =

More information

Comparison between Fourier and Corrected Fourier Series Methods

Comparison between Fourier and Corrected Fourier Series Methods Malaysia Joural of Mahemaical Scieces 7(): 73-8 (13) MALAYSIAN JOURNAL OF MATHEMATICAL SCIENCES Joural homepage: hp://eispem.upm.edu.my/oural Compariso bewee Fourier ad Correced Fourier Series Mehods 1

More information

Supplement for SADAGRAD: Strongly Adaptive Stochastic Gradient Methods"

Supplement for SADAGRAD: Strongly Adaptive Stochastic Gradient Methods Suppleme for SADAGRAD: Srogly Adapive Sochasic Gradie Mehods" Zaiyi Che * 1 Yi Xu * Ehog Che 1 iabao Yag 1. Proof of Proposiio 1 Proposiio 1. Le ɛ > 0 be fixed, H 0 γi, γ g, EF (w 1 ) F (w ) ɛ 0 ad ieraio

More information

FIXED FUZZY POINT THEOREMS IN FUZZY METRIC SPACE

FIXED FUZZY POINT THEOREMS IN FUZZY METRIC SPACE Mohia & Samaa, Vol. 1, No. II, December, 016, pp 34-49. ORIGINAL RESEARCH ARTICLE OPEN ACCESS FIED FUZZY POINT THEOREMS IN FUZZY METRIC SPACE 1 Mohia S. *, Samaa T. K. 1 Deparme of Mahemaics, Sudhir Memorial

More information

Solution. 1 Solutions of Homework 6. Sangchul Lee. April 28, Problem 1.1 [Dur10, Exercise ]

Solution. 1 Solutions of Homework 6. Sangchul Lee. April 28, Problem 1.1 [Dur10, Exercise ] Soluio Sagchul Lee April 28, 28 Soluios of Homework 6 Problem. [Dur, Exercise 2.3.2] Le A be a sequece of idepede eves wih PA < for all. Show ha P A = implies PA i.o. =. Proof. Noice ha = P A c = P A c

More information

Big O Notation for Time Complexity of Algorithms

Big O Notation for Time Complexity of Algorithms BRONX COMMUNITY COLLEGE of he Ciy Uiversiy of New York DEPARTMENT OF MATHEMATICS AND COMPUTER SCIENCE CSI 33 Secio E01 Hadou 1 Fall 2014 Sepember 3, 2014 Big O Noaio for Time Complexiy of Algorihms Time

More information

COS 522: Complexity Theory : Boaz Barak Handout 10: Parallel Repetition Lemma

COS 522: Complexity Theory : Boaz Barak Handout 10: Parallel Repetition Lemma COS 522: Complexiy Theory : Boaz Barak Hadou 0: Parallel Repeiio Lemma Readig: () A Parallel Repeiio Theorem / Ra Raz (available o his websie) (2) Parallel Repeiio: Simplificaios ad he No-Sigallig Case

More information

An interesting result about subset sums. Nitu Kitchloo. Lior Pachter. November 27, Abstract

An interesting result about subset sums. Nitu Kitchloo. Lior Pachter. November 27, Abstract A ieresig resul abou subse sums Niu Kichloo Lior Pacher November 27, 1993 Absrac We cosider he problem of deermiig he umber of subses B f1; 2; : : :; g such ha P b2b b k mod, where k is a residue class

More information

Notes 03 largely plagiarized by %khc

Notes 03 largely plagiarized by %khc 1 1 Discree-Time Covoluio Noes 03 largely plagiarized by %khc Le s begi our discussio of covoluio i discree-ime, sice life is somewha easier i ha domai. We sar wih a sigal x[] ha will be he ipu io our

More information

Comparisons Between RV, ARV and WRV

Comparisons Between RV, ARV and WRV Comparisos Bewee RV, ARV ad WRV Cao Gag,Guo Migyua School of Maageme ad Ecoomics, Tiaji Uiversiy, Tiaji,30007 Absrac: Realized Volailiy (RV) have bee widely used sice i was pu forward by Aderso ad Bollerslev

More information

Dynamic h-index: the Hirsch index in function of time

Dynamic h-index: the Hirsch index in function of time Dyamic h-idex: he Hirsch idex i fucio of ime by L. Egghe Uiversiei Hassel (UHassel), Campus Diepebeek, Agoralaa, B-3590 Diepebeek, Belgium ad Uiversiei Awerpe (UA), Campus Drie Eike, Uiversieisplei, B-260

More information

Moment Generating Function

Moment Generating Function 1 Mome Geeraig Fucio m h mome m m m E[ ] x f ( x) dx m h ceral mome m m m E[( ) ] ( ) ( x ) f ( x) dx Mome Geeraig Fucio For a real, M () E[ e ] e k x k e p ( x ) discree x k e f ( x) dx coiuous Example

More information

Lecture 8 April 18, 2018

Lecture 8 April 18, 2018 Sas 300C: Theory of Saisics Sprig 2018 Lecure 8 April 18, 2018 Prof Emmauel Cades Scribe: Emmauel Cades Oulie Ageda: Muliple Tesig Problems 1 Empirical Process Viewpoi of BHq 2 Empirical Process Viewpoi

More information

Calculus Limits. Limit of a function.. 1. One-Sided Limits...1. Infinite limits 2. Vertical Asymptotes...3. Calculating Limits Using the Limit Laws.

Calculus Limits. Limit of a function.. 1. One-Sided Limits...1. Infinite limits 2. Vertical Asymptotes...3. Calculating Limits Using the Limit Laws. Limi of a fucio.. Oe-Sided..... Ifiie limis Verical Asympoes... Calculaig Usig he Limi Laws.5 The Squeeze Theorem.6 The Precise Defiiio of a Limi......7 Coiuiy.8 Iermediae Value Theorem..9 Refereces..

More information

Additional Tables of Simulation Results

Additional Tables of Simulation Results Saisica Siica: Suppleme REGULARIZING LASSO: A CONSISTENT VARIABLE SELECTION METHOD Quefeg Li ad Ju Shao Uiversiy of Wiscosi, Madiso, Eas Chia Normal Uiversiy ad Uiversiy of Wiscosi, Madiso Supplemeary

More information

OLS bias for econometric models with errors-in-variables. The Lucas-critique Supplementary note to Lecture 17

OLS bias for econometric models with errors-in-variables. The Lucas-critique Supplementary note to Lecture 17 OLS bias for ecoomeric models wih errors-i-variables. The Lucas-criique Supplemeary oe o Lecure 7 RNy May 6, 03 Properies of OLS i RE models I Lecure 7 we discussed he followig example of a raioal expecaios

More information

CLOSED FORM EVALUATION OF RESTRICTED SUMS CONTAINING SQUARES OF FIBONOMIAL COEFFICIENTS

CLOSED FORM EVALUATION OF RESTRICTED SUMS CONTAINING SQUARES OF FIBONOMIAL COEFFICIENTS PB Sci Bull, Series A, Vol 78, Iss 4, 2016 ISSN 1223-7027 CLOSED FORM EVALATION OF RESTRICTED SMS CONTAINING SQARES OF FIBONOMIAL COEFFICIENTS Emrah Kılıc 1, Helmu Prodiger 2 We give a sysemaic approach

More information

λiv Av = 0 or ( λi Av ) = 0. In order for a vector v to be an eigenvector, it must be in the kernel of λi

λiv Av = 0 or ( λi Av ) = 0. In order for a vector v to be an eigenvector, it must be in the kernel of λi Liear lgebra Lecure #9 Noes This week s lecure focuses o wha migh be called he srucural aalysis of liear rasformaios Wha are he irisic properies of a liear rasformaio? re here ay fixed direcios? The discussio

More information

The Moment Approximation of the First Passage Time for the Birth Death Diffusion Process with Immigraton to a Moving Linear Barrier

The Moment Approximation of the First Passage Time for the Birth Death Diffusion Process with Immigraton to a Moving Linear Barrier America Joural of Applied Mahemaics ad Saisics, 015, Vol. 3, No. 5, 184-189 Available olie a hp://pubs.sciepub.com/ajams/3/5/ Sciece ad Educaio Publishig DOI:10.1691/ajams-3-5- The Mome Approximaio of

More information

A TAUBERIAN THEOREM FOR THE WEIGHTED MEAN METHOD OF SUMMABILITY

A TAUBERIAN THEOREM FOR THE WEIGHTED MEAN METHOD OF SUMMABILITY U.P.B. Sci. Bull., Series A, Vol. 78, Iss. 2, 206 ISSN 223-7027 A TAUBERIAN THEOREM FOR THE WEIGHTED MEAN METHOD OF SUMMABILITY İbrahim Çaak I his paper we obai a Tauberia codiio i erms of he weighed classical

More information

Mathematical Statistics. 1 Introduction to the materials to be covered in this course

Mathematical Statistics. 1 Introduction to the materials to be covered in this course Mahemaical Saisics Iroducio o he maerials o be covered i his course. Uivariae & Mulivariae r.v s 2. Borl-Caelli Lemma Large Deviaios. e.g. X,, X are iid r.v s, P ( X + + X where I(A) is a umber depedig

More information

Research Article A Generalized Nonlinear Sum-Difference Inequality of Product Form

Research Article A Generalized Nonlinear Sum-Difference Inequality of Product Form Joural of Applied Mahemaics Volume 03, Aricle ID 47585, 7 pages hp://dx.doi.org/0.55/03/47585 Research Aricle A Geeralized Noliear Sum-Differece Iequaliy of Produc Form YogZhou Qi ad Wu-Sheg Wag School

More information

B. Maddah INDE 504 Simulation 09/02/17

B. Maddah INDE 504 Simulation 09/02/17 B. Maddah INDE 54 Simulaio 9/2/7 Queueig Primer Wha is a queueig sysem? A queueig sysem cosiss of servers (resources) ha provide service o cusomers (eiies). A Cusomer requesig service will sar service

More information

Review Exercises for Chapter 9

Review Exercises for Chapter 9 0_090R.qd //0 : PM Page 88 88 CHAPTER 9 Ifiie Series I Eercises ad, wrie a epressio for he h erm of he sequece..,., 5, 0,,,, 0,... 7,... I Eercises, mach he sequece wih is graph. [The graphs are labeled

More information

Actuarial Society of India

Actuarial Society of India Acuarial Sociey of Idia EXAMINAIONS Jue 5 C4 (3) Models oal Marks - 5 Idicaive Soluio Q. (i) a) Le U deoe he process described by 3 ad V deoe he process described by 4. he 5 e 5 PU [ ] PV [ ] ( e ).538!

More information

F D D D D F. smoothed value of the data including Y t the most recent data.

F D D D D F. smoothed value of the data including Y t the most recent data. Module 2 Forecasig 1. Wha is forecasig? Forecasig is defied as esimaig he fuure value ha a parameer will ake. Mos scieific forecasig mehods forecas he fuure value usig pas daa. I Operaios Maageme forecasig

More information

If boundary values are necessary, they are called mixed initial-boundary value problems. Again, the simplest prototypes of these IV problems are:

If boundary values are necessary, they are called mixed initial-boundary value problems. Again, the simplest prototypes of these IV problems are: 3. Iiial value problems: umerical soluio Fiie differeces - Trucaio errors, cosisecy, sabiliy ad covergece Crieria for compuaioal sabiliy Explici ad implici ime schemes Table of ime schemes Hyperbolic ad

More information

SUMMATION OF INFINITE SERIES REVISITED

SUMMATION OF INFINITE SERIES REVISITED SUMMATION OF INFINITE SERIES REVISITED I several aricles over he las decade o his web page we have show how o sum cerai iiie series icludig he geomeric series. We wa here o eed his discussio o he geeral

More information

Calculus BC 2015 Scoring Guidelines

Calculus BC 2015 Scoring Guidelines AP Calculus BC 5 Scorig Guidelies 5 The College Board. College Board, Advaced Placeme Program, AP, AP Ceral, ad he acor logo are regisered rademarks of he College Board. AP Ceral is he official olie home

More information

Convergence theorems. Chapter Sampling

Convergence theorems. Chapter Sampling Chaper Covergece heorems We ve already discussed he difficuly i defiig he probabiliy measure i erms of a experimeal frequecy measureme. The hear of he problem lies i he defiiio of he limi, ad his was se

More information

TAKA KUSANO. laculty of Science Hrosh tlnlersty 1982) (n-l) + + Pn(t)x 0, (n-l) + + Pn(t)Y f(t,y), XR R are continuous functions.

TAKA KUSANO. laculty of Science Hrosh tlnlersty 1982) (n-l) + + Pn(t)x 0, (n-l) + + Pn(t)Y f(t,y), XR R are continuous functions. Iera. J. Mah. & Mah. Si. Vol. 6 No. 3 (1983) 559-566 559 ASYMPTOTIC RELATIOHIPS BETWEEN TWO HIGHER ORDER ORDINARY DIFFERENTIAL EQUATIONS TAKA KUSANO laculy of Sciece Hrosh llersy 1982) ABSTRACT. Some asympoic

More information

A note on deviation inequalities on {0, 1} n. by Julio Bernués*

A note on deviation inequalities on {0, 1} n. by Julio Bernués* A oe o deviaio iequaliies o {0, 1}. by Julio Berués* Deparameo de Maemáicas. Faculad de Ciecias Uiversidad de Zaragoza 50009-Zaragoza (Spai) I. Iroducio. Le f: (Ω, Σ, ) IR be a radom variable. Roughly

More information

CSE 241 Algorithms and Data Structures 10/14/2015. Skip Lists

CSE 241 Algorithms and Data Structures 10/14/2015. Skip Lists CSE 41 Algorihms ad Daa Srucures 10/14/015 Skip Liss This hadou gives he skip lis mehods ha we discussed i class. A skip lis is a ordered, doublyliked lis wih some exra poiers ha allow us o jump over muliple

More information

Approximating Solutions for Ginzburg Landau Equation by HPM and ADM

Approximating Solutions for Ginzburg Landau Equation by HPM and ADM Available a hp://pvamu.edu/aam Appl. Appl. Mah. ISSN: 193-9466 Vol. 5, No. Issue (December 1), pp. 575 584 (Previously, Vol. 5, Issue 1, pp. 167 1681) Applicaios ad Applied Mahemaics: A Ieraioal Joural

More information

1. Solve by the method of undetermined coefficients and by the method of variation of parameters. (4)

1. Solve by the method of undetermined coefficients and by the method of variation of parameters. (4) 7 Differeial equaios Review Solve by he mehod of udeermied coefficies ad by he mehod of variaio of parameers (4) y y = si Soluio; we firs solve he homogeeous equaio (4) y y = 4 The correspodig characerisic

More information

Mean Square Convergent Finite Difference Scheme for Stochastic Parabolic PDEs

Mean Square Convergent Finite Difference Scheme for Stochastic Parabolic PDEs America Joural of Compuaioal Mahemaics, 04, 4, 80-88 Published Olie Sepember 04 i SciRes. hp://www.scirp.org/joural/ajcm hp://dx.doi.org/0.436/ajcm.04.4404 Mea Square Coverge Fiie Differece Scheme for

More information

th m m m m central moment : E[( X X) ] ( X X) ( x X) f ( x)

th m m m m central moment : E[( X X) ] ( X X) ( x X) f ( x) 1 Trasform Techiques h m m m m mome : E[ ] x f ( x) dx h m m m m ceral mome : E[( ) ] ( ) ( x) f ( x) dx A coveie wa of fidig he momes of a radom variable is he mome geeraig fucio (MGF). Oher rasform echiques

More information

L-functions and Class Numbers

L-functions and Class Numbers L-fucios ad Class Numbers Sude Number Theory Semiar S. M.-C. 4 Sepember 05 We follow Romyar Sharifi s Noes o Iwasawa Theory, wih some help from Neukirch s Algebraic Number Theory. L-fucios of Dirichle

More information

A Note on Random k-sat for Moderately Growing k

A Note on Random k-sat for Moderately Growing k A Noe o Radom k-sat for Moderaely Growig k Ju Liu LMIB ad School of Mahemaics ad Sysems Sciece, Beihag Uiversiy, Beijig, 100191, P.R. Chia juliu@smss.buaa.edu.c Zogsheg Gao LMIB ad School of Mahemaics

More information

On the Validity of the Pairs Bootstrap for Lasso Estimators

On the Validity of the Pairs Bootstrap for Lasso Estimators O he Validiy of he Pairs Boosrap for Lasso Esimaors Lorezo Campoovo Uiversiy of S.Galle Ocober 2014 Absrac We sudy he validiy of he pairs boosrap for Lasso esimaors i liear regressio models wih radom covariaes

More information

Inference of the Second Order Autoregressive. Model with Unit Roots

Inference of the Second Order Autoregressive. Model with Unit Roots Ieraioal Mahemaical Forum Vol. 6 0 o. 5 595-604 Iferece of he Secod Order Auoregressive Model wih Ui Roos Ahmed H. Youssef Professor of Applied Saisics ad Ecoomerics Isiue of Saisical Sudies ad Research

More information

K3 p K2 p Kp 0 p 2 p 3 p

K3 p K2 p Kp 0 p 2 p 3 p Mah 80-00 Mo Ar 0 Chaer 9 Fourier Series ad alicaios o differeial equaios (ad arial differeial equaios) 9.-9. Fourier series defiiio ad covergece. The idea of Fourier series is relaed o he liear algebra

More information

Samuel Sindayigaya 1, Nyongesa L. Kennedy 2, Adu A.M. Wasike 3

Samuel Sindayigaya 1, Nyongesa L. Kennedy 2, Adu A.M. Wasike 3 Ieraioal Joural of Saisics ad Aalysis. ISSN 48-9959 Volume 6, Number (6, pp. -8 Research Idia Publicaios hp://www.ripublicaio.com The Populaio Mea ad is Variace i he Presece of Geocide for a Simple Birh-Deah-

More information

Online Supplement to Reactive Tabu Search in a Team-Learning Problem

Online Supplement to Reactive Tabu Search in a Team-Learning Problem Olie Suppleme o Reacive abu Search i a eam-learig Problem Yueli She School of Ieraioal Busiess Admiisraio, Shaghai Uiversiy of Fiace ad Ecoomics, Shaghai 00433, People s Republic of Chia, she.yueli@mail.shufe.edu.c

More information

Section 8 Convolution and Deconvolution

Section 8 Convolution and Deconvolution APPLICATIONS IN SIGNAL PROCESSING Secio 8 Covoluio ad Decovoluio This docume illusraes several echiques for carryig ou covoluio ad decovoluio i Mahcad. There are several operaors available for hese fucios:

More information

N! AND THE GAMMA FUNCTION

N! AND THE GAMMA FUNCTION N! AND THE GAMMA FUNCTION Cosider he produc of he firs posiive iegers- 3 4 5 6 (-) =! Oe calls his produc he facorial ad has ha produc of he firs five iegers equals 5!=0. Direcly relaed o he discree! fucio

More information

Order Determination for Multivariate Autoregressive Processes Using Resampling Methods

Order Determination for Multivariate Autoregressive Processes Using Resampling Methods joural of mulivariae aalysis 57, 175190 (1996) aricle o. 0028 Order Deermiaio for Mulivariae Auoregressive Processes Usig Resamplig Mehods Chaghua Che ad Richard A. Davis* Colorado Sae Uiversiy ad Peer

More information

MASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.265/15.070J Fall 2013 Lecture 4 9/16/2013. Applications of the large deviation technique

MASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.265/15.070J Fall 2013 Lecture 4 9/16/2013. Applications of the large deviation technique MASSACHUSETTS ISTITUTE OF TECHOLOGY 6.265/5.070J Fall 203 Lecure 4 9/6/203 Applicaios of he large deviaio echique Coe.. Isurace problem 2. Queueig problem 3. Buffer overflow probabiliy Safey capial for

More information

EXISTENCE THEORY OF RANDOM DIFFERENTIAL EQUATIONS D. S. Palimkar

EXISTENCE THEORY OF RANDOM DIFFERENTIAL EQUATIONS D. S. Palimkar Ieraioal Joural of Scieific ad Research Publicaios, Volue 2, Issue 7, July 22 ISSN 225-353 EXISTENCE THEORY OF RANDOM DIFFERENTIAL EQUATIONS D S Palikar Depare of Maheaics, Vasarao Naik College, Naded

More information

xp (X = x) = P (X = 1) = θ. Hence, the method of moments estimator of θ is

xp (X = x) = P (X = 1) = θ. Hence, the method of moments estimator of θ is Exercise 7 / page 356 Noe ha X i are ii from Beroulli(θ where 0 θ a Meho of momes: Sice here is oly oe parameer o be esimae we ee oly oe equaio where we equae he rs sample mome wih he rs populaio mome,

More information

Using Linnik's Identity to Approximate the Prime Counting Function with the Logarithmic Integral

Using Linnik's Identity to Approximate the Prime Counting Function with the Logarithmic Integral Usig Lii's Ideiy o Approimae he Prime Couig Fucio wih he Logarihmic Iegral Naha McKezie /26/2 aha@icecreambreafas.com Summary:This paper will show ha summig Lii's ideiy from 2 o ad arragig erms i a cerai

More information

The analysis of the method on the one variable function s limit Ke Wu

The analysis of the method on the one variable function s limit Ke Wu Ieraioal Coferece o Advaces i Mechaical Egieerig ad Idusrial Iformaics (AMEII 5) The aalysis of he mehod o he oe variable fucio s i Ke Wu Deparme of Mahemaics ad Saisics Zaozhuag Uiversiy Zaozhuag 776

More information

12 Getting Started With Fourier Analysis

12 Getting Started With Fourier Analysis Commuicaios Egieerig MSc - Prelimiary Readig Geig Sared Wih Fourier Aalysis Fourier aalysis is cocered wih he represeaio of sigals i erms of he sums of sie, cosie or complex oscillaio waveforms. We ll

More information

Asymptotic statistics for multilayer perceptron with ReLu hidden units

Asymptotic statistics for multilayer perceptron with ReLu hidden units ESANN 8 proceedigs, Europea Symposium o Arificial Neural Neworks, Compuaioal Ielligece ad Machie Learig. Bruges (Belgium), 5-7 April 8, i6doc.com publ., ISBN 978-8758747-6. Available from hp://www.i6doc.com/e/.

More information

Department of Mathematical and Statistical Sciences University of Alberta

Department of Mathematical and Statistical Sciences University of Alberta MATH 4 (R) Wier 008 Iermediae Calculus I Soluios o Problem Se # Due: Friday Jauary 8, 008 Deparme of Mahemaical ad Saisical Scieces Uiversiy of Albera Quesio. [Sec.., #] Fid a formula for he geeral erm

More information

Extended Laguerre Polynomials

Extended Laguerre Polynomials I J Coemp Mah Scieces, Vol 7, 1, o, 189 194 Exeded Laguerre Polyomials Ada Kha Naioal College of Busiess Admiisraio ad Ecoomics Gulberg-III, Lahore, Pakisa adakhaariq@gmailcom G M Habibullah Naioal College

More information

The Eigen Function of Linear Systems

The Eigen Function of Linear Systems 1/25/211 The Eige Fucio of Liear Sysems.doc 1/7 The Eige Fucio of Liear Sysems Recall ha ha we ca express (expad) a ime-limied sigal wih a weighed summaio of basis fucios: v ( ) a ψ ( ) = where v ( ) =

More information

Approximately Quasi Inner Generalized Dynamics on Modules. { } t t R

Approximately Quasi Inner Generalized Dynamics on Modules. { } t t R Joural of Scieces, Islamic epublic of Ira 23(3): 245-25 (22) Uiversiy of Tehra, ISSN 6-4 hp://jscieces.u.ac.ir Approximaely Quasi Ier Geeralized Dyamics o Modules M. Mosadeq, M. Hassai, ad A. Nikam Deparme

More information

Some Properties of Semi-E-Convex Function and Semi-E-Convex Programming*

Some Properties of Semi-E-Convex Function and Semi-E-Convex Programming* The Eighh Ieraioal Symposium o Operaios esearch ad Is Applicaios (ISOA 9) Zhagjiajie Chia Sepember 2 22 29 Copyrigh 29 OSC & APOC pp 33 39 Some Properies of Semi-E-Covex Fucio ad Semi-E-Covex Programmig*

More information

Outline. simplest HMM (1) simple HMMs? simplest HMM (2) Parameter estimation for discrete hidden Markov models

Outline. simplest HMM (1) simple HMMs? simplest HMM (2) Parameter estimation for discrete hidden Markov models Oulie Parameer esimaio for discree idde Markov models Juko Murakami () ad Tomas Taylor (2). Vicoria Uiversiy of Welligo 2. Arizoa Sae Uiversiy Descripio of simple idde Markov models Maximum likeliood esimae

More information

INVESTMENT PROJECT EFFICIENCY EVALUATION

INVESTMENT PROJECT EFFICIENCY EVALUATION 368 Miljeko Crjac Domiika Crjac INVESTMENT PROJECT EFFICIENCY EVALUATION Miljeko Crjac Professor Faculy of Ecoomics Drsc Domiika Crjac Faculy of Elecrical Egieerig Osijek Summary Fiacial efficiecy of ivesme

More information

MATH 507a ASSIGNMENT 4 SOLUTIONS FALL 2018 Prof. Alexander. g (x) dx = g(b) g(0) = g(b),

MATH 507a ASSIGNMENT 4 SOLUTIONS FALL 2018 Prof. Alexander. g (x) dx = g(b) g(0) = g(b), MATH 57a ASSIGNMENT 4 SOLUTIONS FALL 28 Prof. Alexader (2.3.8)(a) Le g(x) = x/( + x) for x. The g (x) = /( + x) 2 is decreasig, so for a, b, g(a + b) g(a) = a+b a g (x) dx b so g(a + b) g(a) + g(b). Sice

More information

Available online at J. Math. Comput. Sci. 4 (2014), No. 4, ISSN:

Available online at   J. Math. Comput. Sci. 4 (2014), No. 4, ISSN: Available olie a hp://sci.org J. Mah. Compu. Sci. 4 (2014), No. 4, 716-727 ISSN: 1927-5307 ON ITERATIVE TECHNIQUES FOR NUMERICAL SOLUTIONS OF LINEAR AND NONLINEAR DIFFERENTIAL EQUATIONS S.O. EDEKI *, A.A.

More information

A Study On (H, 1)(E, q) Product Summability Of Fourier Series And Its Conjugate Series

A Study On (H, 1)(E, q) Product Summability Of Fourier Series And Its Conjugate Series Mahemaical Theory ad Modelig ISSN 4-584 (Paper) ISSN 5-5 (Olie) Vol.7, No.5, 7 A Sudy O (H, )(E, q) Produc Summabiliy Of Fourier Series Ad Is Cojugae Series Sheela Verma, Kalpaa Saxea * Research Scholar

More information

BE.430 Tutorial: Linear Operator Theory and Eigenfunction Expansion

BE.430 Tutorial: Linear Operator Theory and Eigenfunction Expansion BE.43 Tuorial: Liear Operaor Theory ad Eigefucio Expasio (adaped fro Douglas Lauffeburger) 9//4 Moivaig proble I class, we ecouered parial differeial equaios describig rasie syses wih cheical diffusio.

More information

Affine term structure models

Affine term structure models /5/07 Affie erm srucure models A. Iro o Gaussia affie erm srucure models B. Esimaio by miimum chi square (Hamilo ad Wu) C. Esimaio by OLS (Adria, Moech, ad Crump) D. Dyamic Nelso-Siegel model (Chrisese,

More information

Stochastic Processes Adopted From p Chapter 9 Probability, Random Variables and Stochastic Processes, 4th Edition A. Papoulis and S.

Stochastic Processes Adopted From p Chapter 9 Probability, Random Variables and Stochastic Processes, 4th Edition A. Papoulis and S. Sochasic Processes Adoped From p Chaper 9 Probabiliy, adom Variables ad Sochasic Processes, 4h Ediio A. Papoulis ad S. Pillai 9. Sochasic Processes Iroducio Le deoe he radom oucome of a experime. To every

More information

ECE-314 Fall 2012 Review Questions

ECE-314 Fall 2012 Review Questions ECE-34 Fall 0 Review Quesios. A liear ime-ivaria sysem has he ipu-oupu characerisics show i he firs row of he diagram below. Deermie he oupu for he ipu show o he secod row of he diagram. Jusify your aswer.

More information

Page 1. Before-After Control-Impact (BACI) Power Analysis For Several Related Populations. Richard A. Hinrichsen. March 3, 2010

Page 1. Before-After Control-Impact (BACI) Power Analysis For Several Related Populations. Richard A. Hinrichsen. March 3, 2010 Page Before-Afer Corol-Impac BACI Power Aalysis For Several Relaed Populaios Richard A. Hirichse March 3, Cavea: This eperimeal desig ool is for a idealized power aalysis buil upo several simplifyig assumpios

More information

BRIDGE ESTIMATOR AS AN ALTERNATIVE TO DICKEY- PANTULA UNIT ROOT TEST

BRIDGE ESTIMATOR AS AN ALTERNATIVE TO DICKEY- PANTULA UNIT ROOT TEST The 0 h Ieraioal Days of Saisics ad Ecoomics Prague Sepember 8-0 06 BRIDGE ESTIMATOR AS AN ALTERNATIVE TO DICKEY- PANTULA UNIT ROOT TEST Hüseyi Güler Yeliz Yalҫi Çiğdem Koşar Absrac Ecoomic series may

More information

2 f(x) dx = 1, 0. 2f(x 1) dx d) 1 4t t6 t. t 2 dt i)

2 f(x) dx = 1, 0. 2f(x 1) dx d) 1 4t t6 t. t 2 dt i) Mah PracTes Be sure o review Lab (ad all labs) There are los of good quesios o i a) Sae he Mea Value Theorem ad draw a graph ha illusraes b) Name a impora heorem where he Mea Value Theorem was used i he

More information

NEWTON METHOD FOR DETERMINING THE OPTIMAL REPLENISHMENT POLICY FOR EPQ MODEL WITH PRESENT VALUE

NEWTON METHOD FOR DETERMINING THE OPTIMAL REPLENISHMENT POLICY FOR EPQ MODEL WITH PRESENT VALUE Yugoslav Joural of Operaios Research 8 (2008, Number, 53-6 DOI: 02298/YUJOR080053W NEWTON METHOD FOR DETERMINING THE OPTIMAL REPLENISHMENT POLICY FOR EPQ MODEL WITH PRESENT VALUE Jeff Kuo-Jug WU, Hsui-Li

More information

Semiparametric and Nonparametric Methods in Political Science Lecture 1: Semiparametric Estimation

Semiparametric and Nonparametric Methods in Political Science Lecture 1: Semiparametric Estimation Semiparameric ad Noparameric Mehods i Poliical Sciece Lecure : Semiparameric Esimaio Michael Peress, Uiversiy of Rocheser ad Yale Uiversiy Lecure : Semiparameric Mehods Page 2 Overview of Semi ad Noparameric

More information

C(p, ) 13 N. Nuclear reactions generate energy create new isotopes and elements. Notation for stellar rates: p 12

C(p, ) 13 N. Nuclear reactions generate energy create new isotopes and elements. Notation for stellar rates: p 12 Iroducio o sellar reacio raes Nuclear reacios geerae eergy creae ew isoopes ad elemes Noaio for sellar raes: p C 3 N C(p,) 3 N The heavier arge ucleus (Lab: arge) he ligher icomig projecile (Lab: beam)

More information

ASSESSING GOODNESS OF FIT

ASSESSING GOODNESS OF FIT ASSESSING GOODNESS OF FIT 1. Iroducio Ofe imes we have some daa ad wa o es if a paricular model (or model class) is a good fi. For isace, i is commo o make ormaliy assumpios for simpliciy, bu ofe i is

More information

6/10/2014. Definition. Time series Data. Time series Graph. Components of time series. Time series Seasonal. Time series Trend

6/10/2014. Definition. Time series Data. Time series Graph. Components of time series. Time series Seasonal. Time series Trend 6//4 Defiiio Time series Daa A ime series Measures he same pheomeo a equal iervals of ime Time series Graph Compoes of ime series 5 5 5-5 7 Q 7 Q 7 Q 3 7 Q 4 8 Q 8 Q 8 Q 3 8 Q 4 9 Q 9 Q 9 Q 3 9 Q 4 Q Q

More information

Paper 3A3 The Equations of Fluid Flow and Their Numerical Solution Handout 1

Paper 3A3 The Equations of Fluid Flow and Their Numerical Solution Handout 1 Paper 3A3 The Equaios of Fluid Flow ad Their Numerical Soluio Hadou Iroducio A grea ma fluid flow problems are ow solved b use of Compuaioal Fluid Damics (CFD) packages. Oe of he major obsacles o he good

More information

Skewness of Gaussian Mixture Absolute Value GARCH(1, 1) Model

Skewness of Gaussian Mixture Absolute Value GARCH(1, 1) Model Commuicaios for Saisical Applicaios ad Mehods 203, Vol. 20, No. 5, 395 404 DOI: hp://dx.doi.org/0.535/csam.203.20.5.395 Skewess of Gaussia Mixure Absolue Value GARCH(, Model Taewook Lee,a a Deparme of

More information

11. Adaptive Control in the Presence of Bounded Disturbances Consider MIMO systems in the form,

11. Adaptive Control in the Presence of Bounded Disturbances Consider MIMO systems in the form, Lecure 6. Adapive Corol i he Presece of Bouded Disurbaces Cosider MIMO sysems i he form, x Aref xbu x Bref ycmd (.) y Cref x operaig i he presece of a bouded ime-depede disurbace R. All he assumpios ad

More information

Procedia - Social and Behavioral Sciences 230 ( 2016 ) Joint Probability Distribution and the Minimum of a Set of Normalized Random Variables

Procedia - Social and Behavioral Sciences 230 ( 2016 ) Joint Probability Distribution and the Minimum of a Set of Normalized Random Variables Available olie a wwwsciecedireccom ScieceDirec Procedia - Social ad Behavioral Scieces 30 ( 016 ) 35 39 3 rd Ieraioal Coferece o New Challeges i Maageme ad Orgaizaio: Orgaizaio ad Leadership, May 016,

More information

GAUSSIAN CHAOS AND SAMPLE PATH PROPERTIES OF ADDITIVE FUNCTIONALS OF SYMMETRIC MARKOV PROCESSES

GAUSSIAN CHAOS AND SAMPLE PATH PROPERTIES OF ADDITIVE FUNCTIONALS OF SYMMETRIC MARKOV PROCESSES The Aals of Probabiliy 996, Vol, No 3, 3077 GAUSSIAN CAOS AND SAMPLE PAT PROPERTIES OF ADDITIVE FUNCTIONALS OF SYMMETRIC MARKOV PROCESSES BY MICAEL B MARCUS AND JAY ROSEN Ciy College of CUNY ad College

More information

The Connection between the Basel Problem and a Special Integral

The Connection between the Basel Problem and a Special Integral Applied Mahemaics 4 5 57-584 Published Olie Sepember 4 i SciRes hp://wwwscirporg/joural/am hp://ddoiorg/436/am45646 The Coecio bewee he Basel Problem ad a Special Iegral Haifeg Xu Jiuru Zhou School of

More information

Economics 8723 Macroeconomic Theory Problem Set 2 Professor Sanjay Chugh Spring 2017

Economics 8723 Macroeconomic Theory Problem Set 2 Professor Sanjay Chugh Spring 2017 Deparme of Ecoomics The Ohio Sae Uiversiy Ecoomics 8723 Macroecoomic Theory Problem Se 2 Professor Sajay Chugh Sprig 207 Labor Icome Taxes, Nash-Bargaied Wages, ad Proporioally-Bargaied Wages. I a ecoomy

More information

Clock Skew and Signal Representation

Clock Skew and Signal Representation Clock Skew ad Sigal Represeaio Ch. 7 IBM Power 4 Chip 0/7/004 08 frequecy domai Program Iroducio ad moivaio Sequeial circuis, clock imig, Basic ools for frequecy domai aalysis Fourier series sigal represeaio

More information

The Solution of the One Species Lotka-Volterra Equation Using Variational Iteration Method ABSTRACT INTRODUCTION

The Solution of the One Species Lotka-Volterra Equation Using Variational Iteration Method ABSTRACT INTRODUCTION Malaysia Joural of Mahemaical Scieces 2(2): 55-6 (28) The Soluio of he Oe Species Loka-Volerra Equaio Usig Variaioal Ieraio Mehod B. Baiha, M.S.M. Noorai, I. Hashim School of Mahemaical Scieces, Uiversii

More information

On The Eneström-Kakeya Theorem

On The Eneström-Kakeya Theorem Applied Mahemaics,, 3, 555-56 doi:436/am673 Published Olie December (hp://wwwscirporg/oural/am) O The Eesröm-Kakeya Theorem Absrac Gulsha Sigh, Wali Mohammad Shah Bharahiar Uiversiy, Coimbaore, Idia Deparme

More information

Fresnel Dragging Explained

Fresnel Dragging Explained Fresel Draggig Explaied 07/05/008 Decla Traill Decla@espace.e.au The Fresel Draggig Coefficie required o explai he resul of he Fizeau experime ca be easily explaied by usig he priciples of Eergy Field

More information