Journal of Econometrics. Quasi-maximum likelihood estimation of volatility with high frequency data

Size: px
Start display at page:

Download "Journal of Econometrics. Quasi-maximum likelihood estimation of volatility with high frequency data"

Transcription

1 Joual of Ecoomecs Coes lss avalable a SceceDec Joual of Ecoomecs joual homeage: wwwelsevecom/locae/jecoom Quas-maxmum lkelhood esmao of volaly wh hgh fequecy daa Dacheg Xu Bedhem Cee fo Face, Pceo Uvesy, Pceo, NJ 85, USA acle fo absac Acle hsoy: Receved 8 Seembe 8 Receved evsed fom 8 Jue Acceed 7 July Avalable ole July JEL classfcao: C C C5 Keywods: Iegaed volaly Make mcosucue ose Quas-maxmum lkelhood esmao Realzed keels Sochasc volaly hs ae vesgaes he oees of he well-kow maxmum lkelhood esmao he esece of sochasc volaly ad make mcosucue ose, by exedg he classc asymoc esuls of quas-maxmum lkelhood esmao Whe yg o esmae he egaed volaly ad he vaace of ose, hs aamec aoach emas cosse, effce ad obus as a quas-esmao ude mssecfed assumos Moeove, shaes he model-fee feaue wh oaamec aleaves, fo sace ealzed keels, whle beg advaageous ove hem ems of fe samle efomace I lgh of quadac eeseao, hs esmao behaves lke a eave exoeal ealzed keel asymocally Comasos wh a vaey of mlemeaos of he ukey Hag keel ae ovded usg Moe Calo smulaos, ad a emcal sudy wh he Euo/US Dolla fuue llusaes s alcao acce Elseve BV All ghs eseved Ioduco he avalably of hgh fequecy daa s a double-edged swod fo he esmao of volaly O he oe had, faclaes ou emcal sudes of he asymoc oees of a aual esmao, Realzed Vaace RV, e he sum of squaed logeus; whle o he ohe had, alog wh he daa comes make mcosucue ose, whch dsus all he desable oees of he esmao Whou mcosucue ose, hs smle esmao s boh cosse ad effce I addo, ohe sudes have shed lgh o s ceal lm heoy, such as Jacod 99 ad Badoff-Nelse ad Shehad Howeve, he exsece of ose efees wh he esmao esecally whe he samlg fequecy aoaches zeo A commo acce s o samle sasely, say evey 5 m, dscadg a lage oo of he samle ad he fomao hee he oblem has eceved cosdeable aeo ecely Fo sace, Aï-Sahala e al 5 sugges samlg as fequely as ossble a he cos of modelg he ose hs ae assumes cosa volaly so ha ca efom he maxmum lkelhood esmao MLE I he seg of sochasc volaly, Zhag e al 5 bg fowad a oaamec esmao, wo-scale el: E-mal addess: dachegx@ceoedu Realzed Volaly SRV, whch s he fs cosse esmao he esece of ose, dese a elavely low covegece ae 6 Subsequely, Zhag 6 advocaes Mul-Scale Realzed Volaly MSRV, movg he covegece ae o, whch s he omal ae a model ca each as show by Gloe ad Jacod Recely, Badoff-Nelse e al 8 have desged vaous ealzed keels RKs ha ca be used o deal wh edogeous ose ad edogeously saced daa ad he covegece aes ae he same as ha of he MSRV I smulao, hese oaamec esmaos efom vey well wh omally seleced badwdh o keels; whle acce, esmaos wh badwdh based o ad hoc choces, o based o small samle efomace may behave bee, as llusaed Gaheal ad Oome 7 ad Bad ad Russell 8 Aohe gou of esmaos daes back o Zhou 996, who fs ales he auocovaace-based-aoach o cosa volaly cases Hase ad Lude 6 exed he Zhou esmao o he sochasc volaly models wh seally deede ose Howeve, he Zhou esmao ad s exesos ae cosse he movao ad sao of hs acle sem fom Gaheal ad Oome 7 ad Aï-Sahala ad Yu 9 Usg afcally smulaed zeo-ellgece daa, Gaheal ad Oome 7 comae a comehesve se of esmaos cludg hose lsed above ad he ad hoc modfcaos Accodg o he sudes, he MLE, hough ossbly mssecfed wh me vayg volaly, s amog he bes ems of effcecy ad obusess -76/$ see fo mae Elseve BV All ghs eseved do:6/jjecoom7

2 6 D Xu / Joual of Ecoomecs I addo, Aï-Sahala ad Yu 9 aly he MLE o aalyze he lqudy of NYSE socks he maxmum lkelhood esmaes wh he daa smulaed fom sochasc volaly models dcae good oees, alhough hs esmao s deved fom a cosa volaly assumo Relaed woks also clude Hase e al 8, whee he auhos sugges he cossecy of he MLE by examg movg aveage fles hese sudes movae us o cosde he MLE as a Quas-Maxmum Lkelhood Esmao QMLE ude mssecfed models, whch daes back o as ealy as Amemya 97, Whe 98, 98 I hese acles, he famewok of mssecfed esmao has bee bul ad s close coeco wh Kullback Leble Ifomao Ceo KLIC see Kullbacks ad Leble 95 has bee llusaed Sce hese semal woks, Domowz ad Whe 98 ad Baes ad Whe 985 have exeded he cossecy esuls of he QMLE wh d models o vaous cases cludg deede obsevaos, Quas-GMM-esmaos, ad Quas-M-esmaos Ou wok s heeby bul o he fuso of hgh fequecy daa ad mssecfed lkelhood esmao he coec model secfcao feaues sochasc volaly Howeve, hs model s eoally mssecfed o be oe of cosa volaly Ude hs assumo, we efom quas maxmum lkelhood esmao ad aalyze he esmao, whch s esseally he same as he MLE Aï-Sahala e al 5 Remakably, he coex of he coec model, he QMLE of volaly cossely esmaes he egaed volaly a he mos effce ae Also, he QMLE of ose vaace has he same asymoc dsbuo as befoe I ohe wods, he maxmum lkelhood esmaos ae obus o sochasc volaly I addo, hey ae sll obus o adom samlg evals ad o-gaussa make mcosucue oses he ae s ogazed as follows Seco evews he aamec lkelhood esmao ad ovdes he uo ad movao fo he QMLE Seco oules he classc asymoc heoy of he QMLE, ad deves a exeso o moe geeal segs Seco vesgaes he sascal oees of he QMLE, whee he cossecy, ceal lm heoy ad obusess of he QMLE ae esablshed Seco 5 comaes wh oaamec keel esmaos, ad Seco 6 uses Moe Calo smulaos o vefy he coclusos obaed fom he evous secos Seco 7 deals a emcal sudy wh he Euo/US Dolla fuue daa Seco 8 cocludes he Aedces ovdes all mahemacal oofs Revsg he MLE: he QMLE I hs seco, we ecaulae he aamec aoach ad oduce ou quas-esmao he adoal aamec mehodology ales o cases whee he ue value of he aamee of ees s a secal o he aamee sace heefoe, Aï-Sahala e al 5 have o make a assumo ha he volaly s ehe a sochasc ocess o a deemsc fuco, bu sead, a cosa ha s, he lae effce log ce ocess sasfes dx dw wh he obseved log asaco ce X coamaed by he mcosucue ose U a way such ha X X + U, whee s he obsevao me Fo smlcy, we assume ha he daa ae egulaly saced, sasfyg he obsevaos We gve he MLE a alas QMLE somemes ode o emhasze model mssecfcao ad kee he oao le wh he classc esuls of mssecfed models ae made wh [, ], whee s fxed, so ha he fll asymoc behavos ae deemed as goes o ad goes o smulaeously he sucue of he obseved log eu Y s feaues MA, whee Y X X W W + U U If we osulae ha he ose dsbuo s Gaussa, he ou log lkelhood fuco fo Ys s l, a log de log Y Y whee s gve Box I he MLE ˆ, â oves o be cosse a dffee aes fo s volaly a ad ose a eve f he ose dsbuo us ou o be ˆ A â a L N, 8a a + cum [U] whee cum [U] s he fouh cumula of he ue ose U Neveheless, amle evdece of sochasc volaly calls hs smlsc model o queso So, s aual o ask: wha s he mac of sochasc volaly o he MLE? Wll sochasc volaly chage he desed oees of he MLE, jus as mcosucue ose does o RV, o wll he MLE sll be cosse? Smulao sudes by Aï-Sahala ad Yu 9 ad Hase e al 8 seem o have suggesed ha he MLE may be a cosse esmao of he egaed volaly Iuvely, hs cojecue s lausble ha whe volaly becomes sochasc, he egaed volaly, he aamee of ees, haes o be he aveage of he volaly ocess, whch s execed o be a legmae caddae fo he esmae If cossecy s guaaeed, wha would be he covegece ae ad asymoc vaace? How would comae wh ohe aleave oaamec esmaos? Close scuy of he esmao he absece of mcosucue ose may yeld moe sghs Cosde a sochasc volaly model wh o ose: dx dw R he objecve s o esmae he egaed volaly d By desg, we msakely assume he so volaly o be cosa; heefoe, he quas-log lkelhood fuco s l, log log Y Y whee Y X X R ad Y Y, Y,,Y Aaely, he QMLE s ˆ X Y X X X Hee, he RV esmao ecus hough a dffee agume, ad s of couse he efec esmao ude he ue sochasc volaly model Howeve, he cossecy of he QMLE s o loge saghfowad he esece of ose, because hee may be o close fom avalable fo hs esmao Is asymoc vaace s fa moe comlcaed due o heeoskedascy ad auocoelao, as meoed by Hase e al 8 he emag ae goously vesgaes he asymoc behavo of he QMLE he seg of sochasc volaly ad mcosucue ose, whch leads us o he classc asymoc heoy of quas-esmaos

3 D Xu / Joual of Ecoomecs a a a + a a a + a C a A a + a Box I Asymoc heoy of quas-m-esmaos I hs seco, we a fs befly evew he cossecy of he QMLE ude mssecfed models, a heoy ally develoed Whe 98, 98 he aoale behd he heoy s elaed o Kullback Leble Ifomao Ceo KLIC Moe ecsely, suose ha we have a d adom samle, ad le gx be he ue ukow daa geeag desy, ad f X, ou ossbly mssecfed desy dexed by a aamee Whe 98 clams ha ude cea egula codos, he QMLE s cosse o whch mmzes KLIC: Ig : f, Elog[gX/f X, ] whee he execao s ake ude he ue model If he model s coecly secfed, ha s, hee exss, such ha gx f X,, he he KLIC aas s mmum a, hece hs esul s ageeme wh he cossecy of egula maxmum lkelhood esmaos Ohewse, he model s mssecfed, ad uvely, mmzes ou goace of he ue sucue I Domowz ad Whe 98, he auhos geealze hese esuls o clude deede obsevaos, ad show ha he QMLE ˆ, whch maxmzes Q,, sll coveges obably o, a maxmze of Q,, he execao of Q, ude he ue model I addo, Baes ad Whe 985 exed he heoy o geeal quas-m-esmaos, by oducg dsceacy fucos Now, we add moe adomess o he mssecfcao heoy, whch eables ou alcaos o sochasc volaly models, whee he aamee of ees self s adom he easog of he oof s smla o he egula oe gve Whe 98 ad Newey ad McFadde 99 heoem Le Q, ad Q, be wo adom fucos such ha fo each, a comac subse of R k, hey ae measuable fucos o ad, fo each, couous fucos o I addo, Q, s almos suely maxmzed a Fuhe, he followg wo codos ae sasfed as : Ufom covegece: su kq, Q, k P Idefably: fo evey >, hee exss a cosa >, such ha P Q, max Q, > 5 :k k he ay sequece of esmaos ˆ such ha Q, ˆ Q, + o coveges obably o, e, ˆ mgh deed o whe model s mssecfed su P Noe ha he defably codo hee s slghly dffee fom ha a commo seg, fo examle Va De Vaa ad Newey ad McFadde 99 I o oly eques some uqueess lke oey of he maxmze, bu also a oe omalzao such ha he maxmze ca be dsgushed asymocally Now we modfy he assumos o accommodae o he M-esmaos seg, case we eed dffee omalzaos fo dffee aamees heoem Le, ad, be adom veco-valued fucos Fo each, a comac subse of R k, hey ae measuable fuco o, ad fo each, couous fucos o I addo, hee exss a sequece of, sasfyg, almos suely, such ha as, Ufom covegece: su k,, k P 6 Idefably: Fo evey >, hee exss a cosa >, such ha, P m k, k > 7 :k k he ay sequece of esmaos ˆ such ha, ˆ o coveges obably o, e, ˆ P I he followg dscussos, we wll choose as he scoe fuco of a mssecfed model, u o a aoae omalzao, ad s caefully secfed coesodg o he ceal lm esul s gve by he ex heoem, whch s a exeso of heoem Domowz ad Whe 98 heoem Suose ha he codos of heoem ae sasfed I addo,, ad, ae couously dffeeable of ode o Also, hee exss a sequece of osve defe maces {V } such ha V, L N, Ik 8 If, s sochasc equcouous, ad, P,, ufomly fo all, he V, ˆ L N, Ik 9 he exesos of he classc asymoc heoy ave he way fo a hoough quy of asymoc oees of he QMLE Sascal oees of he QMLE hs seco shaes he seu wh mos volaly esmaos avalable he leaue Moe secfcally, we make he followg assumos

4 8 D Xu / Joual of Ecoomecs Assumo he udelyg lae log ce ocess sasfes dx dw wh he volaly ocess a osve ad locally bouded Iô semmagale Assumo he ose U s deedely ad decally dsbued, ad deede of ce ad volaly ocesses, wh mea, vaace a ad fe fouh mome he assumos of d ose ad s deedece wh ces ae o always cosse wh he emcal evdece, as oed ou Hase ad Lude 6 Aï-Sahala e al 5 have dscussed he way o modfy he log lkelhood fuco wh moe aamees, accodg o he assumed aamec sucue of he ose See also Gaheal ad Oome 7 fo a MA mlemeao As o he SRV, Aï-Sahala e al fohcomg have exeded o he case wh seally deede ose Kala ad Lo 8 have oosed a modfcao of he SRV wh edogeous ose ad heeoskedasc measueme eo Ideedely, Badoff-Nelse e al 8 have show he obusess of he ealzed keels wh esec o edogeous ose Howeve, seally deedece lus edogeous assumo ae sll usasfacoy, sce ealy he asaco ces ae ecoded wh oud-off eos L ad Myklad 7 have dscussed he obusess of he SRV wh esec o oudg eos whle Jacod e al 7 have ecely oosed a eaveagg aoach ha woks well fo a geeal class of eos cludg cea combao of oudg ad addve eos I hs ae, beg asmoous ad fo smlcy, we cosde he d whe ose ad ovde addo a heusc agume fo me-deede ose o valdae he alcaos of he QMLE acce Cossecy of he QMLE Cosde fs wha would hae he absece of ose Based o he dscusso Seco, he scoe fuco u o a oe omalzao ude mssecfed model s,, dl, Y Y d ad s oo s ˆ X Y he we choose, X X X Z R d heefoe, has a oo Because s a comac se, ad f we eque he aamee sace o be bouded away fom zeo, e, hee exs such ha >, he su,, X Y Z P Poof of he las se s well-kow, see eg Kaazas ad Sheve 99, heoem 58, hece ufom covegece obably heoem s show Idefably codo hs assumo accommodaes vually all couous me facal models See Jacod 8 hyohess L s fo moe deals vally holds, so he cossecy follows fom heoem he ae of covegece deeds o he ae of, whch s, as gve by Jacod 99 fo sace heefoe, ˆ X Z Y d O Aaely, we do o eed heoem a all hee I geeal, howeve, hs s cealy o he case Oe sao fom hs smle examle s egadg he seleco of If we add o leveage assumo, ha s o say, he volaly ocess s codoally deemsc, he s ohg bu he codoal execao of ude he ue model Whe he obseved daa ae osy, we have Y R + U U Besdes he cosa volaly assumo, we msakely assume ha he oses ae omally dsbued wh vaace a heefoe, he quas-log lkelhood fuco s exacly, hece he same lkelhood esmao ecus he fom of he QMLE ude mssecfed model Howeve, closedfom exessos ae o loge avalable, so we have o u o heoem fo cossecy Deoe, a ad we also assume ha he aamees say a comac se, whch s bouded away fom zeo As he o ose case, we choose he followg scoe fucos u o some @ logde @ logde @a Y ad + whee s as gve Box II Beg awae ha he covegece aes may be dffee fo he wo aamees, as show Seco, we choose dffee omalzaos accodgly hese omalzaos ae esseal o esue he defably codo Deoe j, ad j U j U j he dffeece bewee ad, fo sace s as gve Box III he fs em s a lea combao of magale dffeeces, whle he secod em s he sum of coss oducs ove dffee eces of egals he hd oe s a mxue of he magale a ad he ose a he es ae uely elaed o he ose I s clea ha he fs wo ems as a whole, he hd oe ad he es ae awse ucoelaed gve he oosed assumos Now we oceed wh vaace calculaos he lesso leaed fom he falue of RV dcaes ha he vaace of he ose may domae he ohes So fo he ose a, we accuaely comue he vaace of he devaves wh esec o dffee aamees Lemma Gve Assumos ad, we have 8 X < Z 9 Z dw : ; O

5 D Xu / Joual of Ecoomecs Z + a a a Z a + a a Z + a a a Z C A + a C X + Y : Z Box II {z } magale dffeece X Z X X j + 9 X X + ; j6,,+ j {z } due o he ose + a + j Z j j @ + + a {z } due o he ose Box III X X X Z j j6 X Z j j Z j O O j Lemma As o he ose a, we have a 6a + o 5 6 a + cum [U] + o 7 a 8 By vefyg he defably codo, solvg he equaos ad alyg he above lemmas, we ca ove heoem ad ae as gve, ˆ ˆ, â s as defed heoem Gve Assumos ad, he QMLE ˆ, â sasfes: â a P ad ˆ R P As execed, he cossecy of he volaly esmao sll holds, eve hough he volaly ocess becomes sochasc Ceal lm heoem of he QMLE Alhough we have show he cossecy of he QMLE, we have o exloed whehe he QMLE has he omal covegece aes, ehe have we show ayhg abou he magude of he asymoc vaace o ossble asymoc bas o aswe hese quesos, he followg lemma ad heoem ovde he ceal lm heoy Lemma Gve Assumos ad, we have he ceal lm heoem gve Box IV heoem 5 ad ae as gve, ˆ ˆ, â s as defed heoem Gve Assumos ad, he QMLE ˆ, â sasfes he ceal lm heoem gve Box V I he cosa volaly case, QMLE aas he omal effcecy, as ca be execed sce he QMLE s cosuced he same way as he MLE I geeal, QMLE acheves he omal covegece aes Howeve, hs QMLE mehod mgh o ovde a saghfowad esmao fo he egaed quacy, ha s, R d hus, as o he cosuco of cofdece evals, oe ca sead use he mehod gve by Jacod e al 7 Robusess of he QMLE Df Wha haes o ou coclusos f he udelyg X ocess has a df? Moe ecsely, suose X o be of he Iô ye, dx µ + dw whee µ s a locally bouded ad ogessvely measuable ocess Isead of aameezg he df em ad modfyg ou lkelhood esmao accodgly, we comleely goe he MN s a oao of mxed omal used Badoff-Nelse e al 8, ad L X meas X-sable covegece law

6 D Xu / Joual of Ecoomecs L X MN R R 5 6a 7 + a 8 5 a + a 6a 5 CC a + cum AA [U] a 8 Box IV Z ˆ â a C A L X MN 5a R R Box V + R a a + cum [U] CC AA esece of df, o ohe wods, mssecfy he model wh df I hs R case, he esmao s uchaged, ad we oly eed o elace R wh µ + R he oof Loosely seakg, hs wll o ale ou coclusos sce he hgh fequecy seg, he df s of ode d, whch s mahemacally eglgble wh esec o he dffusve comoe of ode d A aleave agume gve by Myklad ad Zhag 9 s o zeo ou he df by chagg obably measues Radom samlg evals Wha f he obsevaos ae made adomly? If he samlg evals bewee wo cosecuve obsevaos ae d, ad deede of he ce ocess, we may add oe moe mssecfcao ha he daa ae egulaly saced; hece we oba he same esmao as befoe Cossecy ca be esablshed decly by codog o he obsevao me I fac, hs esmao ca be vewed as he Peed Fxed Maxmum Lkelhood PFML esmao dscussed Aï-Sahala ad Myklad I lgh of hs, we may efom he Full Ifomao Maxmum Lkelhood FIML o Iegaed Ou Maxmum Lkelhood IOML esmao, ode o ulze he fomao egadg he dsbuo of he adom evals, f avalable Also, s ossble o exed he d samlg scheme o edogeous ad sochascally saced obsevaos, by cosdeg he me-chaged ocesses as Badoff-Nelse e al 8 o Myklad ad Zhag 6 he exeso s effoless lgh of he coeco bewee QMLE ad RKs selled ou Seco 5 No-Gaussa ad seal-deede mcosucue ose Wha abou he obusess wh esec o he ose? Fom Lemma ad heoem, we fd ha he dsbuo of he d ose does o affec he cossecy I ohe wods, whaeve he dsbuo of he ose s, s mssecfed o be Gaussa, ad he QMLE obaed by maxmzg gves he same esmao of a ad he same ode of covegece ae, hough he asymoc vaace may be dffee O he ohe had, f he oses ae fac me-deede, we ca combe he QMLE wh he subsamlg mehod Fo sace, f he ose self feaues a MAk sucue, we ca dvde he whole samle o k + dsjo as such ha he oses assocaed wh he adjace os each subsamle ae ucoelaed he, we ca smly aly he QMLE o each subsamle ad aggegae he esmaes by akg he aveage Jums If he ce ocess has jums, he ˆ wll covege o R + P X sead, whch cocdes wh he SRV esmao he oblem of seaag jums fom volaly hs seg s moe edous, ad may cobue lle, sce mos cases lage jums do o hae vey ofe wh a day If hey do occu, we may use he wavele mehod Fa ad Wag 7 o emove jums befoe esmao, o seaae he esmao eods by jums ad add u evey ece of egaed volaly ogehe, f he osos of jums ca be locaed accuaely 5 Comasos wh ealzed keels 5 Esmao mehods Realzed keels RKs clude a sees of oaamec esmaos desged fo volaly esmao he esece of ose Fla-o RKs ake o he followg fom X h K X X + k h X + h X 8 H whee h X h X X j X j X j X j h h 9 j s he hh samle auocovaace fuco ad he keel k s a wegh fuco he mos commo fe-lag fla-o keels ae of he modfed ukey Hag ye, defed by: k H x s x {alexale} I addo, hee ae fe-lag ealzed keels whch may assg ozeo wegh o all auocovaaces, such as he omal keel: k o x + xe x As execed, he sascal oees of RKs vay as dffee keels ad badwdhs ae seleced heefoe, eables us o make dffee choces fo secfc uoses Howeve, he choce of he badwdh may also become a bude acce, ha he omal badwdh gve by he heoy cao be esmaed vey accuaely, ad ha he ule-of-humb aoxmao of he badwdh may o efom as well as he oe seleced ad hoc ways hs oblem may become eve wose f he esmaes wee sesve o he choce of badwdh By coas, he QMLE s desged a aamec way, whch s fee of badwdh ad keel seleco, whle shag he desed model-fee feaue wh oaamec esmaos I s by aue a dffee esmao ad may o be egaded as oe of he RKs decly

7 D Xu / Joual of Ecoomecs AvaRK/AvaQMLE Omal ukeyhag Paze Cubc Asymoc Relave Effcecy Fg Rho Asymoc elave effcecy of he QMLE ad RKs 5 Asymoc behavo ad fe samle efomace I egad o asymoc effcecy, he keel ad badwdh ca be chose fo RKs o acheve he same omal covegece ae as he QMLE Neveheless, whe volaly s cosa, he asymoc vaace of fe-lag keels ca oly aoxmae he aamec vaace boud, whch, by coas, ca be obaed by he QMLE ad he omal keel wh fe lags Whe volaly s sochasc, he elave effcecy of he QMLE ad RKs deeds o he exe of heeoskedascy Moe ecsely, we have 5 AvaRK AvaQMLE 6 + k k q + k k / k + + q + k k / k 5 + whee R q du/ R u u du measues he vaably of he volaly ocess, ad k, k ad k ae cosas deemed by he seleced keel Fg los he elave effcecy of fou ycal RKs agas he QMLE Aaely, he QMLE eds o be moe favoable ha fe-lag keels as becomes lage, whle RKs ae bee whe s small he omal keel wh heoecally omal badwdh, howeve, fully domaes he QMLE asymocally exce fo he case, whe volaly s cosa Iuvely, he smalle s, he fuhe he mssecfed model devaes fom he uh A majo dawback of RKs s ha hey eque a umbe of ou-of-eod aday eus because of he cosuco of he auocovaace esmao h X Fo hs easo, felag keels, acula, ae o mlemeable emcally I acce, he feasble auocovaace esmao s mlemeed by ±h X X H X j X j X j X j h h jh+ Wh fe-lag fla-o keels, he cu-off ea he bouday s o a ssue asymocally, howeve may yeld a lage bas fe samle As he samle sze deceases, amely, H/ ceases, hs bas becomes moe evde he ex seco ales Moe Calo smulaos o demosae he edge effec 5 he asymoc vaace of he RK obaed hee eques a efeme of he edos as well as he omal badwdh 5 Quadac eeseao ad weghg maces A uve way o udesad he smlaes ad dffeeces bewee he QMLE ad RKs s o egad hem as quadac esmaos Moe secfcally, we have he followg quadac eave eeseao of he QMLE heoem 6 he QMLE ˆ, â sasfes he followg equaos: ˆ Y W ˆ, â Y â Y W ˆ, â Y he weghg maces sasfy: W, a W, a whee s gve by, ad C A Also, W, a ad W, a deed o a / ad a oly hough Smlaly, a feasble ealzed keel esmao ca be exessed as K X Y WY 5 whee W s deemed by he keel k ad badwdh H W, {+Haleale H} j W,j k {ale j aleh} {+Halejale H} H Fg los he weghg maces agas he ow ad colum dces fo he QMLE ad ukey Hag keel he lo dslays smla feaues such as he weghs o he dagoals ae ehe o vey close o wh seveal lags, ad decay o zeo beyod ha he followg heoem llusaes he mlc coeco bewee he wo esmaos ems of he asymoc behavo heoem 7 he QMLE s asymocally equvale o he omal R keel wh mlc badwdh H ˆ a / I ohe wods, fo ay K +, < <, ad ay K ale, j ale K, we have j W,,j, a k o heoem 7 also os ou ha he weghg max W, a s aoxmaely a symmec oelz max wh equal wegh alog he dagoal, bag he bouday effec he mlc badwdh deeds o he aamees of ees ad s subomal I heefoe exlas why he omal keel wh omal badwdh asymocally domaes he QMLE Fg Neveheless, he edg os of he dagoals have dffee aes, dcag dffee eames of he edge effec Aaely, he QMLE

8 D Xu / Joual of Ecoomecs QMLE of Volaly 5 6 QMLE of Nose Vaace he ose sasfes omal dsbuo wh sadad devao a 5% he jums follow a Posso ocess N deede of he ce ad volaly ocesses wh esy he jum sze volaly s J V exz, whee z N 5, he umbe of Moe Calo samle ahs s Fg shows he hsogams of he sadadzed esmaes comaed wh he asymoc dsbuos Moeove, able comaes he samle quale sascs wh he heoec bechmaks gve by he sadad Gaussa dsbuo All hese smulao esuls ecofm ou asymoc heoy ukeyhag Keel 5 6 Fg Weghg maces of he QMLE ad he RK Noe: We lo he wo weghg maces agas he colum ad ow dces he aamees ae, a 5, 65 ad /5 he badwdh H 57a / cools he weghs o he bouday a moe aual way, ad hece has bee fe samle efomace he quadac eeseao also sheds lgh o he dffeeces he ocedues of esmao he badwdh H of he RK s ehe esmaed as fs se o seleced a ad hoc way, whle he badwdh ˆ of he QMLE s auomacally udaed by he omzao algohm, o moe uvely, by eag ad he weghg max of he QMLE s heefoe moe adave I vew of hs, s aual o cosuc a oe-se aleave fo he QMLE, whch, sead of ug olea omzao, emloys a cosse lug- of ˆ fo ad hs oese volaly esmao cocdes wh oe of he ubased quadac esmaos gve by Su 6, whee s asymoc oees ae dscussed ude he cosa volaly assumo he quadac foms of ohe oaamec esmaos ae also cluded Adese e al fohcomg 6 Smulao sudes wh hgh fequecy daa 6 Asymoc behavo of he QMLE We a fs coduc Moe Calo smulaos o vefy he asymoc esuls gve heoem 5 We fx as day Wh [, ], he daa ae smulaed usg Eule scheme based o sochasc volaly models, fo sace he Heso Model wh jums volaly ocess dx µd + dw d ale + db + J V dn whee EdW db d he aval of asacos follows a Posso ocess ad he mea eaval me s s he ue values of aamees ae cosse wh hose chose by Aï- Sahala ad Yu 9 Secfcally, he df µ s, ad o add he leveage effec, s seleced o be 75 ale 5 ad he volaly of volaly s s samled fom he saoay dsbuo of he CIR ocess, ha s Gammaale /, /ale, so ha he ucodoal mea of volaly ocess s exacly, chose as 6 Comasos wh RKs Fo comaso uose, we mleme he ukey Hag keel whch s oe of he fla-o keels ha covege a he bes ae I s cosdeably effce comaed wh ohe keels Badoff-Nelse e al 8, ad does o eque oo may ou-of-eod daa As a bechmak, we fs mleme he ukey Hag keel wh he feasble badwdh, H c wh c gve by v u s u c k k + + k k k q whee a / R du, R q du/ R u u u du, k 9, k 7 ad k 7 Also, he edge effec s elmaed ha he calculao, usg fomula 9, cludes he ou-ofeod daa Hece, hs esmao would ovde us wh he bes ukey Hag keel esmae RK heoy Nex, we make he edge effec sad ou by aoxmag he auocovaaces usg, whle keeg he heoecal omal badwdh as befoe We deoe he secod esmao as RK Fally, o mmc he emcal alcaos of hese esmaos, we edo he exemes wh he smle ule-of-humb choce of he badwdh: q Ĥ 57â /RV X q whee â RV X/ ad RV X s he RV esmao based o m eus So we oba he coesodg RK ad RK he smulaos ae based o he same sochasc volaly model ad aamees as befoe wh a loge me wdow [, ] We egad [, ] as -samle eod, ad he es me eval s oly used fo feasble keels he esuls of esmao ae eoed able I s evde ha he QMLE domaes he fou mlemeaos of he ukey Hag keel ems of he RMSE As execed, RK s vey close o he QMLE, ad bee ha he ohe hee keels, ha emloys moe daa ad s equed wh heoecally omal badwdh Comag he fs wo keels, we ca also see ha he edge effec esuls a lage bas whe he samle sze s elavely small, ad become dameed as he samle sze goes u I addo, he dffeeces bewee he fs wo keels ad he las wo ae he losses due o he subomal choce of he badwdh I acce, oly RK ca be aled, whch may suffe fom lage bas ad vaace because of he wo souces of losses 7 Emcal wok wh he Euo/US Dolla fuue ces I hs seco, we ae eesed esmag he ealzed vaace of he Euo/US dolla fuues caed ou o he Chcago Mecale Exchage CME he yea 8 he coacs ae acvely aded o he -h clock, ad quoed ems of he u

9 D Xu / Joual of Ecoomecs able I hs able, we eo he fe samle quales of he feasble sadadzed QMLE, whch emloys he heoec asymoc vaace he bechmak quales ae hose fo he lm dsbuo N, No obs Mea Sdv 5% 5% 5% 95% 975% 995% a Iegaed Volaly 5 Nose Vaace Fg Hsogams of he sadadzed esmaes Noe: We lo he hsogams of he sadadzed esmaes he ue ose vaace s 5 he smulaos clude adom evals ad volaly jums he desy of he asymoc dsbuo s also loed able R hs able eos he esmaes fo ˆ, d whee ˆ s gve by he QMLE ad vaous mlemeaos of he ukey Hag keel esecvely Amog hem, RK ad RK have he heoecal badwdh, whle RK ad RK emloy ou-of-eod daa s 5 s s s s m m Bas QMLE Sdv RMSE RK Sdv Bas RMSE RK Sdv Bas RMSE RK Sdv Bas RMSE RK Sdv Bas RMSE value of he Euo as measued US dollas he hgh fequecy daa ae avalable fom ck Daa Ic he foeg exchage makes ae less acve dug he weekeds ad holdays; heefoe, we elmae he asacos o Saudays ad Sudays as well as US fedeal holdays I addo, we also exclude Jauay, he day afe haksgvg, Decembe 6, ad Decembe Fally, we make eveyday beg a 5 m Chcago me whe he elecoc adg sas so

10 D Xu / Joual of Ecoomecs Aualzed Daly Realzed Volaly of he EUR/USD FX Fuue 8 Realzed Keel QMLE Bea Seas Fe Sale Lehma Bohes Bakucy Fae Mae ad Fedde Mac Balou Iceladc Css 5 5 JAN FEB MAR APR MAY JUN JUL AUG SEP OC NOV DEC JAN Fg he EUR/USD fuue: a case sudy able Summay sascs of he log-eus of he Euo/US Dolla fuue Avg o of obs Avg feq Mea Sd e s lag d lag s 68e 9 6e as o elmae he oeal ce jums bewee he oe hou adg ga fom m o 5 m he summay sascs of he daa ae ovded able Evdely, he mcosucue ose dslays a MA sucue We aly boh he QMLE ad he ukey Hag keel wh he ule-of-humb badwdh ad lo he aualzed daly ealzed volaly Fg I s aae fom he lo ha he wo mehods gve almos decal esmaes ad he dffeeces ae sascally sgfca hs s ageeme wh ou Moe Calo smulao esul, whee he wo esmaes ae dsgushable whe he adg fequecy s as hgh as seveal secods Moeove, Fg ovdes amle evdece of sucue beaks o jums he volaly ocess sag fom Seembe 8, whch echoes he fac ha, sce he, he global facal css has eeed s mos ccal sage Fuhe, hese jums o lage movemes ae usually assocaed wh facal ews: fo sace, he goveme sezue of Fae Mae ad Fedde Mac, Lehma Bohes bakucy ealy Seembe, ad he collase of hee majo baks of Icelad ealy Ocobe, ec hese fdgs ae hade o oba by modelg lowe fequecy daa 8 Coclusos hs acle cobues o he esmao of egaed volaly by showg ha he oula MLE, as a ew quas-esmao, s cosse, effce ad obus wh esec o sochasc volaly Moeove, hs aamec esmao oly volves a omzao ocedue, whch s fee fom badwdh seleco, ad hece vey covee acce Moe eesgly, hs seemgly aoae esmao us ou o be asymocally equvale o he omal ealzed keel wh a mlcly secfed badwdh, ad doma ove aleave ealzed keels ems of he fe samle accuacy hs acle makes aohe cobuo o he aalyss of model secfcao by exedg he classc asymoc heoy of he QMLE o a sochasc aamee seg, ad by showg a examle of mssecfyg models o uose, whch gves se o facly ad feasbly esmao hs sudy may be alcable o moe cases whee sochasc volaly lagues he esmao, such as covaace esmao, o moe geeally, whee he objec of ees s sochasc Ackowledgemes I am debed o Yace Aï-Sahala fo ecouagg me o wok o hs oc ad helful suggesos heeafe I aecae he commes of he Co-Edo A Roald Galla ad wo efeees, whch led o cosdeable movemes of he ae I am also gaeful o Jaqg Fa, Pee Rehad Hase, Jea Jacod, Ygyg L, Ulch K Mülle, Pe Myklad, Ec Reaul, ad Roe Sca fo valuable commes, whch heled a lo movg he qualy of he ae Also, I would lke o hak odd Hes fo daa asssace, ad Ja L ad X Wa as well as he acas of he semas a Pceo Uvesy ad he Uvesy of Chcago, of he woksho o Facal Ecoomecs a Felds Isue, ad of he d Pceo-Humbold Face woksho fo fuful dscussos All eos ae me Aedx A Poof of heoem Fo each have 6 Q, ˆ >Q, Q, ˆ >Q, ˆ Q, > Q, >, wh obably aoachg wa, we 9 > due o ufom covegece >; hus, 8 >, wa, Q, ˆ > Q, Fuhe, fo ay >, le N : { : k k < } he fo each, N c \ s comac, ad max N c \ Q, Q, ale Q, Le : Q, We have, fo ay <, max Q, N c \ P Q, ˆ > max Q, N c \ P Q, ˆ > Q, P Q, ˆ > Q,, > P Q, ˆ > Q,, > heefoe, Pkˆ k < 6 he saemes hs oof ae well osed f ay desed measuably s guaaeed Howeve, hs measuably ssue ca be goed by edefg all coces ems of oue measue, see Newey ad McFadde 99,

11 D Xu / Joual of Ecoomecs Aedx B Poof of heoem X < Z 9 Z E he oof s smla o Va De Vaa s oof of cossecy of he M-esmaos I follows fom heoem, o alyg 8 : d ; o Q, k, k ad Q, k, k X < Z 9 Z ale M E dw : ; Aedx C Poof of heoem As he o ose case, we have Fo smlcy, s suffce o ove hs esul fo k 8 Because X < Z 9 Z,,, : d ; O 9 Plug ˆ ad mully boh sdes by V, he we have V, ˆ V, P Sce ˆ, s sochasc equcouous, ad P,,, ufomly fo all, follows fom a aalogous easog as heoem Domowz ad Whe 98 ha,, P, whch cocludes he oof Aedx D Poof of Lemma Le j ad X R We also defe M X hx, X R R he {X } aleale ad {M } aleale ae magales Also, we ca add oe moe codo ha B ale ale B, 8 [, ], ha a egula localzao ocedue always ales Fo coveece, we chage he vaables: + + a a he he vese chage of vaables s gve by, a + o a + a 6 a + + o a + 7 ad we have j j +j j+ j Oe ca check j I addo, s a cocave fuco of, ad clealy aas s maxmum a + ad mmum a he bouday I addo, by 6 ad 7 + a a + O + + O a a log + O a a + a + + O he, ca be easly show ha M : +, + O he magale oey, we have 8 X < Z Z : 9 A ; By Combg he wo, we oba 8 X < Z Z dw : O O O 9 A ; hus, we ove by Chebyshev s equaly o ove, oe ha j j, so follows fom he magale oey ha E X E X Z j j6 Z j j X X X X Z j kl Z k j> k l>k Z l Z j j k l 8 X < Z j X j Z 9 k E dw : kj j k ; ale B B j X E j Xj Z k kj k X Xj kj E j k kj ale B X X j k<j O k k Z k k Fally, we ove 5 Because fo ay j 6, j j, j+, ale, ad j > X X Z E j j j X X a j j j, j+, E j X ale a B, +, O Z So all he coclusos clamed he lemma hold

12 6 D Xu / Joual of Ecoomecs Aedx E Poof of Lemma Le,,, Deoe K, K,j, K,j,k, ad K,j,k,l he coesodg cumulas of { } he K, sce he mea of s zeo Accodg o McCullagh 987, Seco, we have X va j kl K,j,k,l + K K j,k,l + K,k K j,l whee V v,,j,k,l + K K k K j,l X j kl K,j,k,l + K,k K j,l,j,k,l X,j,k,l j kl cum, j, k, l + cov, k cov j, l : V, + V, X,j,k,l v j kl cum, j, k, l X V v, v j kl cov, k cov j, l,j,k,l NX NX a v j j, + j,+ j, j + j+, + j+,+ j+, j, + j,+ j, Recall ha Lemma of Aï-Sahala e al 5, cum, j, k, l 8 < cum [U], f j k l; s,j,k,l cum : [U], f max, j, k, l m, j, k, l + ;, ohewse whee s, j, k, l deoes he umbe of dces amog, j, k, l ha ae equal o m, j, k, l So usg hs lemma, we ca oba X X V v, cum [U] v + v,+ +,+ v +,+,+ + v, +,+ + v,+,+ o faclae calculao, we @ he dec comuao a { } 6 + o a + o a { + + }+o a + o a a 8a { } 5 + o 5a + o a a a + + O Combg, we + a 6a + o 5 a + a a V, a + V a + o a 8 I follows fom he smla calculao as above see also Aï-Sahala e al, 5, 96 V, @a, cum [U]+o a cum [U]+o cum [U]+o a cum [U]+o 8 cum [U]+o a cum [U]+o O cum [U] a 8 + o Hece, combg he above equales, we @ a 6a + o @ a + cum [U] + o a 8 hs cocludes he oof

13 D Xu / Joual of Ecoomecs Aedx F Poof of heoem We wa o ove by vefyg he codos of heoem I follows fom Lemmas ad ha o O + O + O + O O O + O O o + O + O So fa, we have show ha P, fo ay,, o ove ufom covegece obably, we eed sochasc equcouy of As agued Newey ad McFadde 99, see also Rockafella 97, heoem 8, he o-wse covegece of cocave fucos mles ufom covegece Noe ha ca be egaded as he dffeece of wo cocave fucos fom, he a slghly modfed oof of heoem 8 Rockafella 97, usg agle equales, sll gves se o ufom covegece heefoe, as Adese ad Gll 98, we ave a su k k P Nex, we show he defably @a J J + whee J J j whee J,, ad he ohe comoes of J s J a a a + J a + a 8a 6 + O 5 O he ohe had, choose K, ad le7 m K,K ad M +, + By 8, we have ale M + + m + e e a +O a 6 +O 5 6 e e a +O a +O 7 Hee ad subsequely, ca be elaced by + wh ay < < heefoe, fo ay K ale ale K, m +o; fo < K o > K, s domaed by m, ad he egao s ove a eval whch shks a he ae K/ /, so X Z Z m + o a Smlaly, fo K ale a + @a whle fo < K o > heefoe, a ad a + o + o + ae domaed by Z Z + o 6 + o 7 By calculao, we ca also oba R a + a a a 8a 5 a 8a + o 8 Hece, f we solve, we ge a a + a a + o a a R 9 We assume all hese lkelhood esmaes ae bouded almos suely, sce he aamee sace s self bouded Also, 6a + a 6 6a 5 8a R a a 6 a + + 8a 5 5a 6a 5 6a 7 + o

14 8 D Xu / Joual of Ecoomecs O he ohe had, le : be he dagam max wh he h eleme of he dagoal : R, he a + Z a a 6a d, a + > + q R + I J J a a + whe k k + a a > ad goes P m k k :k k a a X > P m k k > :k > P o < heefoe, follows ha he defably codo holds, hece by heoem, ˆ P, ad â a P hs cocludes he oof by 9 ad a X X a a 8a + a a 8a + O Z 8a d + o a a 8a + O whee he las equaly s gve by 7 Se, a, ha s, Z a O a a + O Aga, follows fom dec a + a R + 6a 8a + 6a 5 a + a R + d 6a 5 8a 6a + o As, m k k :k k Obseve ha, k k a P, so wh obably aoachg, + R m k k :k k a a + a 8a 5 a 8a + o Z + 8a d + o + a a 8a + O Aedx G Poof of Lemma he agume s vey smla o he oof of heoem gve he aedx of Badoff-Nelse e al 8, whch volves he coce of sable covegece law he deals abou have bee dscussed Jacod ad Shyaev ad Jacod 7 Assume fo coveece ha j, fo, j < o, j > Also, deoe M M M X X X X j< X j X j j X X j X X : Z j j E X whee X X X ad j j We beg wh M Pck K, ad cosde K ale ale K Whe j O +, j exoeally, fo ay < < Rewe as: M X whee f X, f x, x,,x K KX h h K x x h I follows fom heoem 7 Jacod 7 ha, M coveges sably law, ad s asymoc vaace ca be calculaed usg fomula 7 ad 7 Jacod 7: X Ava lm R, f

15 D Xu / Joual of Ecoomecs whee 8 < KX R, f : h U h h 9 h E U h ; h j ad U s ae d sadad Gaussa adom vaables Fo ay K ale ale K, dec calculao yelds j Kalej< 6 7 a By he same agume used he oof of heoem, we have Z M L X 5 MN, 6 7 a d Sce we have show Lemma ha M follows fom Lemma Badoff-Nelse e al 8 ha j, Z M + M L X 5 MN 6 7 a As o M, we a fs oce ha X j @ ad ha fo K ale ale K, j ad ha P @ 8 5 a become ufomly asymocally eglgble heefoe, by he sadad ceal lm heoem, codoal o he flao X, we have X X M P L N, a 8 5 a P Sce X R d, ad by Lemma ad Pooso 5 Badoff-Nelse e al 8, we ca coclude ha Z M L X a N, 8 5 a ad M + M ad M joly covege X- sably law Also, as o he ose a, codoal o he flao X, M L a N, 5 6a 5 he same easog as above yelds M + M + M + M Z L X 5 MN, 6 7 a d Z a + a 8 5 a + 6 6a 5 As o, sce he ose a domaes, mles ha L MN, a + cum [U] 7 a 8 Fally, he covaace of he ad ode O, hus he jo ceal lm heoem follows fom Pooso 5 Badoff-Nelse e al 8 Aedx H Poof of heoem 5 s of he Now we deve he ceal lm heoem of he esmaos Because, ˆ,, ˆ whee s bewee ˆ ad, ad by Lemma Badoff- Nelse e al 8, s clea ha we oly eed o vefy he assumos of heoem I fac, because of, we wll cosde Y Y ad ad he devaves mulled by oe omalzaos I s saghfowad o check ha hese fucos ae ehe covex o cocave, whch guaaees sochasc equcouy see he oof of heoem Combg heoem, we have he ceal lm esul gve Box VI Noe fom 9 ad ha because of he cossecy esuls R above, a a o, O, heefoe, he oeal asymoc bases ae elmaed, whch cocludes he oof Aedx I Poof of heoem 6 Fom he lkelhood fuco, we ca oba he scoe fucos: Y Y Y Y whee a I O he ohe had, he followg equales hold: a I a I a + As a esul, a Pluggg he scoe fucos gves he eeseao Clealy, he wo quadac foms ae o co-lea he secod clam s obvous I fac, we ca we, ad oly deeds o Pluggg o he eeseao s amou o elacg wh decly Aedx J Poof of heoem 7 Accodg o 8, we have ha fo K ale, j ale K,,j j j

16 5 D Xu / Joual of Ecoomecs ˆ â B 8a R 5a L X MN B, 8a C a A a 6 a + a R,,,, + a + cum [U] CC AA o o Box VI wh he dffeece exoeally small heefoe, we ca fuhe deduce ha + j,j X l lj l j a a j a + j,j a,j,j j a a j O he ohe had, dec calculaos deduce ha 5 + O a + O a a + O heefoe, follows fom he eeseao ha j W,,j + j j + e Refeeces j Aï-Sahala, Y, Myklad, P, he effec of adom ad dscee samlg whe esmag couous me dffusos Ecoomeca 7, 8 59 Aï-Sahala, Y, Myklad, P, Zhag, L, 5 How ofe o samle a couous-me ocess he esece of make mcosucue ose Revew of Facal Sudes 8, 5 6 Aï-Sahala, Y, Myklad, P, Zhag, L, 6 Ula hgh fequecy volaly esmao wh deede mcosucue ose Joual of Ecoomecs, fohcomg do:6/jjecoom8 Aï-Sahala, Y, Yu, J, 9 Hgh fequecy make mcosucue ose esmaes ad lqudy measues Aals of Aled Sascs, 57 Amemya,, 97 Regesso aalyss whe he deede vaable s ucaed omal Ecoomeca, Adese, G, Bolleslev,, Meddah, N, 9 Realzed volaly foecasg ad make mcosucue ose Joual of Ecoomecs, fohcomg do:6/jjecoom Adese, PK, Gll, RD, 98 Cox s egesso model fo coug ocesses: a lage samle sudy Aals of Sascs 9, Bad, M, Russell, J, 8 Make mcosucue ose, egaed vaace esmaos, ad he accuacy of asymoc aoxmaos Dscusso Pae Uvesy of Chcago, Gaduae School of Busess Badoff-Nelse, OE, Hase, PR, Lude, A, Shehad, N, 8 Desgg ealzed keels o measue he ex-os vaao of equy ces he esece of ose Ecoomeca 76, 8 56 Badoff-Nelse, OE, Shehad, N, Ecoomec aalyss of ealzed volaly ad s use esmag sochasc volaly models Joual of he Royal Sascal Socey Sees B 6, 5 8 Baes, C, Whe, H, 985 A ufed heoy of cosse esmao fo aamec models Ecoomec heoy, 5 78 Domowz, I, Whe, H, 98 Mssecfed models wh deede obsevaos Joual of Ecoomecs, 5 58 Fa, J, Wag, Y, 7 Mul-scale jum ad volaly aalyss fo hghfequecy facal daa Joual of he Ameca Sascal Assocao, 9 6 Gaheal, J, Oome, R, 7 Zeo-ellgece ealzed vaace esmao Wokg Pae Gloe, A, Jacod, J, Dffusos wh measueme eos: I Local asymoc omaly Euoea Sees Aled ad Idusal Mahemacs 5, 5 Hase, PR, Lage, J, Lude, A, 8 Movg aveage-based esmaos of egaed vaace Ecoomec Revews 7, 79 Hase, PR, Lude, A, 6 Realzed vaace ad make mcosucue ose Joual of Busess ad Ecoomc Sascs, 7 8 Jacod, J, 99 Lm of adom measues assocaed wh he cemes of a Bowa semmmagale Dscussg Pae Jacod, J, 7 Sascs ad hgh fequecy daa Uublshed Lecue Noes, Sémae Euoée de Sasque 7: Sascs fo Sochasc Dffeeal Equaos Models Jacod, J, 8 Asymoc oees of ealzed owe vaaos ad elaed fucoals of semmagales Sochasc Pocesses ad he Alcaos 8, Jacod, J, L, Y, Myklad, P, Podolskj, M, Vee, M, 7 Mcosucue ose he couous case: he e-aveagg aoach Wokg Pae Jacod, J, Shyaev, AN, Lm heoems fo Sochasc Pocesses, d ed Sge-Velag, New Yok Kala, I, Lo, O, 8 Esmag quadac vaao cossely he esece of edogeous ad dual measueme eo Joual of Ecoomecs 7, 7 59 Kaazas, I, Sheve, S, 99 Bowa Moo ad Sochasc Calculus, d ed I: GM, vol Sge-Velag Pess Kullbacks, S, Leble, R, 95 O fomao ad suffcecy Aals of Mahemacal Sascs, L, Y, Myklad, P, 7 Ae volaly esmaos obus wh esec o modelg assumos? Beoull, 6 6 McCullagh, P, 987 eso Mehods Sascs Chama ad Hall, Lodo, UK Myklad, P, Zhag, L, 6 ANOVA fo dffusos ad Iô ocesses Aals of Sascs, 9 96 Myklad, PA, Zhag, L, 9 Ifeece fo couous semmagales obseved a hgh fequecy Ecoomeca 77, 5 Newey, W, McFadde, D, 99 Lage samle esmao ad hyohess esg I: Hadbook of Ecoomecs, vol IV Chae 6 Rockafella,, 97 Covex Aalyss Pceo Uvesy Pess Su, Y, 6 Bes quadac ubased esmaos of egaed vaace he esece of make mcosucue ose Wokg Pae Va De Vaa, AW, Asymoc Sascs Cambdge Uvesy Pess Whe, H, 98 Nolea egesso o coss-seco daa Ecoomeca 8, 7 76 Whe, H, 98 Maxmum lkelhood esmao of mssecfed models Ecoomeca 5, 6 Zhag, L, 6 Effce esmao of sochasc volaly usg osy obsevaos: a mul-scale aoach Beoull, 9 Zhag, L, Myklad, PA, Aï-Sahala, Y, 5 A ale of wo me scales: deemg egaed volaly wh osy hgh fequecy daa Joual of he Ameca Sascal Assocao, 9 Zhou, B, 996 Hgh-fequecy daa ad volaly foeg-exchage aes Joual of Busess ad Ecoomc Sascs, 5 5

Suppose we have observed values t 1, t 2, t n of a random variable T.

Suppose we have observed values t 1, t 2, t n of a random variable T. Sppose we have obseved vales, 2, of a adom vaable T. The dsbo of T s ow o belog o a cea ype (e.g., expoeal, omal, ec.) b he veco θ ( θ, θ2, θp ) of ow paamees assocaed wh s ow (whee p s he mbe of ow paamees).

More information

The Stability of High Order Max-Type Difference Equation

The Stability of High Order Max-Type Difference Equation Aled ad Comuaoal Maemacs 6; 5(): 5-55 ://wwwsceceulsggoucom/j/acm do: 648/jacm653 ISSN: 38-565 (P); ISSN: 38-563 (Ole) Te Saly of g Ode Ma-Tye Dffeece Equao a Ca-og * L Lue Ta Xue Scool of Maemacs ad Sascs

More information

Fakultas Matematika dan Ilmu Pengetahuan Alam, Institut Teknologi Bandung, Bandung, 40132, Indonesia *

Fakultas Matematika dan Ilmu Pengetahuan Alam, Institut Teknologi Bandung, Bandung, 40132, Indonesia * MacWllams Equvalece Theoem fo he Lee Wegh ove Z 4 leams Baa * Fakulas Maemaka da Ilmu Pegeahua lam, Isu Tekolog Badug, Badug, 403, Idoesa * oesodg uho: baa@mahbacd BSTRT Fo codes ove felds, he MacWllams

More information

ANSWERS TO ODD NUMBERED EXERCISES IN CHAPTER 2

ANSWERS TO ODD NUMBERED EXERCISES IN CHAPTER 2 Joh Rley Novembe ANSWERS O ODD NUMBERED EXERCISES IN CHAPER Seo Eese -: asvy (a) Se y ad y z follows fom asvy ha z Ehe z o z We suppose he lae ad seek a oado he z Se y follows by asvy ha z y Bu hs oads

More information

Shrinkage Estimators for Reliability Function. Mohammad Qabaha

Shrinkage Estimators for Reliability Function. Mohammad Qabaha A - Najah Uv. J. es. (N. Sc.) Vol. 7 3 Shkage Esmaos fo elably Fuco مقدرات التقلص لدالة الفاعلية ohammad Qabaha محمد قبها Depame of ahemacs Faculy of Scece A-Najah Naoal Uvesy alese E-mal: mohqabha@mal.ajah.edu

More information

Comparison of the Bayesian and Maximum Likelihood Estimation for Weibull Distribution

Comparison of the Bayesian and Maximum Likelihood Estimation for Weibull Distribution Joural of Mahemacs ad Sascs 6 (2): 1-14, 21 ISSN 1549-3644 21 Scece Publcaos Comarso of he Bayesa ad Maxmum Lkelhood Esmao for Webull Dsrbuo Al Omar Mohammed Ahmed, Hadeel Salm Al-Kuub ad Noor Akma Ibrahm

More information

Exponential Synchronization of the Hopfield Neural Networks with New Chaotic Strange Attractor

Exponential Synchronization of the Hopfield Neural Networks with New Chaotic Strange Attractor ITM Web of Cofeeces, 0509 (07) DOI: 0.05/ mcof/070509 ITA 07 Expoeal Sychozao of he Hopfeld Neual Newos wh New Chaoc Sage Aaco Zha-J GUI, Ka-Hua WANG* Depame of Sofwae Egeeg, Haa College of Sofwae Techology,qogha,

More information

Parameter Estimation and Hypothesis Testing of Two Negative Binomial Distribution Population with Missing Data

Parameter Estimation and Hypothesis Testing of Two Negative Binomial Distribution Population with Missing Data Avlble ole wwwsceceeccom Physcs Poce 0 475 480 0 Ieol Cofeece o Mecl Physcs Bomecl ee Pmee smo Hyohess es of wo Neve Boml Dsbuo Poulo wh Mss D Zhwe Zho Collee of MhemcsJl Noml UvesyS Ch zhozhwe@6com Absc

More information

( ) ( ) Weibull Distribution: k ti. u u. Suppose t 1, t 2, t n are times to failure of a group of n mechanisms. The likelihood function is

( ) ( ) Weibull Distribution: k ti. u u. Suppose t 1, t 2, t n are times to failure of a group of n mechanisms. The likelihood function is Webll Dsbo: Des Bce Dep of Mechacal & Idsal Egeeg The Uvesy of Iowa pdf: f () exp Sppose, 2, ae mes o fale of a gop of mechasms. The lelhood fco s L ( ;, ) exp exp MLE: Webll 3//2002 page MLE: Webll 3//2002

More information

ON TOTAL TIME ON TEST TRANSFORM ORDER ABSTRACT

ON TOTAL TIME ON TEST TRANSFORM ORDER ABSTRACT V M Chacko E CONVE AND INCREASIN CONVE OAL IME ON ES RANSORM ORDER R&A # 4 9 Vol. Decembe ON OAL IME ON ES RANSORM ORDER V. M. Chacko Depame of Sascs S. homas Collee hss eala-68 Emal: chackovm@mal.com

More information

- 1 - Processing An Opinion Poll Using Fuzzy Techniques

- 1 - Processing An Opinion Poll Using Fuzzy Techniques - - Pocessg A Oo Poll Usg Fuzzy Techues by Da Peu Vaslu ABSTRACT: I hs ae we deal wh a mul cea akg oblem, based o fuzzy u daa : he uose s o comae he effec of dffee mecs defed o he sace of fuzzy umbes o

More information

8. Queueing systems lect08.ppt S Introduction to Teletraffic Theory - Fall

8. Queueing systems lect08.ppt S Introduction to Teletraffic Theory - Fall 8. Queueg sysems lec8. S-38.45 - Iroduco o Teleraffc Theory - Fall 8. Queueg sysems Coes Refresher: Smle eleraffc model M/M/ server wag laces M/M/ servers wag laces 8. Queueg sysems Smle eleraffc model

More information

Least Squares Fitting (LSQF) with a complicated function Theexampleswehavelookedatsofarhavebeenlinearintheparameters

Least Squares Fitting (LSQF) with a complicated function Theexampleswehavelookedatsofarhavebeenlinearintheparameters Leas Squares Fg LSQF wh a complcaed fuco Theeampleswehavelookedasofarhavebeelearheparameers ha we have bee rg o deerme e.g. slope, ercep. For he case where he fuco s lear he parameers we ca fd a aalc soluo

More information

Two kinds of B-basis of the algebraic hyperbolic space *

Two kinds of B-basis of the algebraic hyperbolic space * 75 L e al. / J Zhejag Uv SCI 25 6A(7):75-759 Joual of Zhejag Uvesy SCIECE ISS 9-395 h://www.zju.edu.c/jzus E-al: jzus@zju.edu.c Two ds of B-bass of he algebac hyebolc sace * LI Ya-jua ( 李亚娟 ) WAG Guo-zhao

More information

Technical Appendix for Inventory Management for an Assembly System with Product or Component Returns, DeCroix and Zipkin, Management Science 2005.

Technical Appendix for Inventory Management for an Assembly System with Product or Component Returns, DeCroix and Zipkin, Management Science 2005. Techc Appedx fo Iveoy geme fo Assemy Sysem wh Poduc o Compoe eus ecox d Zp geme Scece 2005 Lemm µ µ s c Poof If J d µ > µ he ˆ 0 µ µ µ µ µ µ µ µ Sm gumes essh he esu f µ ˆ > µ > µ > µ o K ˆ If J he so

More information

EMA5001 Lecture 3 Steady State & Nonsteady State Diffusion - Fick s 2 nd Law & Solutions

EMA5001 Lecture 3 Steady State & Nonsteady State Diffusion - Fick s 2 nd Law & Solutions EMA5 Lecue 3 Seady Sae & Noseady Sae ffuso - Fck s d Law & Soluos EMA 5 Physcal Popees of Maeals Zhe heg (6) 3 Noseady Sae ff Fck s d Law Seady-Sae ffuso Seady Sae Seady Sae = Equlbum? No! Smlay: Sae fuco

More information

Chapter 3: Maximum-Likelihood & Bayesian Parameter Estimation (part 1)

Chapter 3: Maximum-Likelihood & Bayesian Parameter Estimation (part 1) Aoucemes Reags o E-reserves Proec roosal ue oay Parameer Esmao Bomercs CSE 9-a Lecure 6 CSE9a Fall 6 CSE9a Fall 6 Paer Classfcao Chaer 3: Mamum-Lelhoo & Bayesa Parameer Esmao ar All maerals hese sles were

More information

Brownian Motion and Stochastic Calculus. Brownian Motion and Stochastic Calculus

Brownian Motion and Stochastic Calculus. Brownian Motion and Stochastic Calculus Browa Moo Sochasc Calculus Xogzh Che Uversy of Hawa a Maoa earme of Mahemacs Seember, 8 Absrac Ths oe s abou oob decomoso he bascs of Suare egrable margales Coes oob-meyer ecomoso Suare Iegrable Margales

More information

Professor Wei Zhu. 1. Sampling from the Normal Population

Professor Wei Zhu. 1. Sampling from the Normal Population AMS570 Pofesso We Zhu. Samplg fom the Nomal Populato *Example: We wsh to estmate the dstbuto of heghts of adult US male. It s beleved that the heght of adult US male follows a omal dstbuto N(, ) Def. Smple

More information

On the Quasi-Hyperbolic Kac-Moody Algebra QHA7 (2)

On the Quasi-Hyperbolic Kac-Moody Algebra QHA7 (2) Ieaoal Reeach Joual of Egeeg ad Techology (IRJET) e-issn: 9 - Volume: Iue: May- www.e.e -ISSN: 9-7 O he Qua-Hyebolc Kac-Moody lgeba QH7 () Uma Mahewa., Khave. S Deame of Mahemac Quad-E-Mllah Goveme College

More information

Leader-Following Consensus of Nonlinear Multi-Agent Systems Based on Parameterized Lyapunov Function

Leader-Following Consensus of Nonlinear Multi-Agent Systems Based on Parameterized Lyapunov Function ODRES JOURL OF ELECRICL EGIEERIG VOL 5 O 2 SUER 25 3 Leade-Followg Cosesus of olea ul-ge Sysems Based o Paameezed Lyauov Fuco Pegah aba Saad 2 ohammad ehd ada okha Shasadegh Behouz Safaeada bsac hs ae

More information

The Poisson Process Properties of the Poisson Process

The Poisson Process Properties of the Poisson Process Posso Processes Summary The Posso Process Properes of he Posso Process Ierarrval mes Memoryless propery ad he resdual lfeme paradox Superposo of Posso processes Radom seleco of Posso Pos Bulk Arrvals ad

More information

( m is the length of columns of A ) spanned by the columns of A : . Select those columns of B that contain a pivot; say those are Bi

( m is the length of columns of A ) spanned by the columns of A : . Select those columns of B that contain a pivot; say those are Bi Assgmet /MATH 47/Wte Due: Thusday Jauay The poblems to solve ae umbeed [] to [] below Fst some explaatoy otes Fdg a bass of the colum-space of a max ad povg that the colum ak (dmeso of the colum space)

More information

Fault-tolerant Output Feedback Control for a Class of Multiple Input Fuzzy Bilinear Systems

Fault-tolerant Output Feedback Control for a Class of Multiple Input Fuzzy Bilinear Systems Sesos & asduces Vol 7 Issue 6 Jue 04 pp 47-5 Sesos & asduces 04 by IFSA Publshg S L hp://wwwsesospoalco Faul-olea Oupu Feedbac Cool fo a Class of Mulple Ipu Fuzzy Blea Syses * YU Yag WAG We School of Eleccal

More information

Journal of Econometrics. Quasi-maximum likelihood estimation of volatility with high frequency data

Journal of Econometrics. Quasi-maximum likelihood estimation of volatility with high frequency data Joural of Ecoomercs 59 () 5 5 Coes lss avalable a SceceDrec Joural of Ecoomercs joural homepage: wwwelsevercom/locae/jecoom Quas-maxmum lkelhood esmao of volaly wh hgh frequecy daa Dacheg Xu Bedhem Ceer

More information

SECURITY EVALUATION FOR SNOW 2.0-LIKE STREAM CIPHERS AGAINST CORRELATION ATTACKS OVER EXTENSION FIELDS

SECURITY EVALUATION FOR SNOW 2.0-LIKE STREAM CIPHERS AGAINST CORRELATION ATTACKS OVER EXTENSION FIELDS SECURIY EVALUAION FOR SNOW.-LIKE SREAM CIPHERS AGAINS CORRELAION AACKS OVER EXENSION FIELDS A. N. Alekseychk * S. M. Koshok ** M. V. Poemsky *** Ise of Secal Commcao ad Ifomao Secy Naoal echcal Uvesy of

More information

Asymptotic Behavior of Solutions of Nonlinear Delay Differential Equations With Impulse

Asymptotic Behavior of Solutions of Nonlinear Delay Differential Equations With Impulse P a g e Vol Issue7Ver,oveber Global Joural of Scece Froer Research Asypoc Behavor of Soluos of olear Delay Dffereal Equaos Wh Ipulse Zhag xog GJSFR Classfcao - F FOR 3 Absrac Ths paper sudes he asypoc

More information

Generalisation on the Zeros of a Family of Complex Polynomials

Generalisation on the Zeros of a Family of Complex Polynomials Ieol Joul of hemcs esech. ISSN 976-584 Volume 6 Numbe 4. 93-97 Ieol esech Publco House h://www.house.com Geelso o he Zeos of Fmly of Comlex Polyomls Aee sgh Neh d S.K.Shu Deme of hemcs Lgys Uvesy Fdbd-

More information

k of the incident wave) will be greater t is too small to satisfy the required kinematics boundary condition, (19)

k of the incident wave) will be greater t is too small to satisfy the required kinematics boundary condition, (19) TOTAL INTRNAL RFLTION Kmacs pops Sc h vcos a coplaa, l s cosd h cd pla cocds wh h X pla; hc 0. y y y osd h cas whch h lgh s cd fom h mdum of hgh dx of faco >. Fo cd agls ga ha h ccal agl s 1 ( /, h hooal

More information

Solution set Stat 471/Spring 06. Homework 2

Solution set Stat 471/Spring 06. Homework 2 oluo se a 47/prg 06 Homework a Whe he upper ragular elemes are suppressed due o smmer b Le Y Y Y Y A weep o he frs colum o oba: A ˆ b chagg he oao eg ad ec YY weep o he secod colum o oba: Aˆ YY weep o

More information

The Eigenvalue Problem of the Symmetric Toeplitz Matrix

The Eigenvalue Problem of the Symmetric Toeplitz Matrix he Egevalue Poblem of he Smmec oelz Max bsac I hs assgme, he mehods ad algohms fo solvg he egevalue oblem of smmec oelz max ae suded. Fs he oelz ssem s oduced. he he mehods ha ca localze he egevalues of

More information

Available online at J. Math. Comput. Sci. 2 (2012), No. 4, ISSN:

Available online at   J. Math. Comput. Sci. 2 (2012), No. 4, ISSN: Available olie a h://scik.og J. Mah. Comu. Sci. 2 (22), No. 4, 83-835 ISSN: 927-537 UNBIASED ESTIMATION IN BURR DISTRIBUTION YASHBIR SINGH * Deame of Saisics, School of Mahemaics, Saisics ad Comuaioal

More information

(1) Cov(, ) E[( E( ))( E( ))]

(1) Cov(, ) E[( E( ))( E( ))] Impac of Auocorrelao o OLS Esmaes ECON 3033/Evas Cosder a smple bvarae me-seres model of he form: y 0 x The four key assumpos abou ε hs model are ) E(ε ) = E[ε x ]=0 ) Var(ε ) =Var(ε x ) = ) Cov(ε, ε )

More information

RECAPITULATION & CONDITIONAL PROBABILITY. Number of favourable events n E Total number of elementary events n S

RECAPITULATION & CONDITIONAL PROBABILITY. Number of favourable events n E Total number of elementary events n S Fomulae Fo u Pobablty By OP Gupta [Ida Awad We, +91-9650 350 480] Impotat Tems, Deftos & Fomulae 01 Bascs Of Pobablty: Let S ad E be the sample space ad a evet a expemet espectvely Numbe of favouable evets

More information

MULTI-OBJECTIVE GEOMETRIC PROGRAMMING PROBLEM AND ITS APPLICATIONS

MULTI-OBJECTIVE GEOMETRIC PROGRAMMING PROBLEM AND ITS APPLICATIONS Yugoslav Joual of Opeaos Reseach Volume (), Numbe, -7 DOI:.98/YJORI MULTI-OBJECTIVE GEOMETRIC PROGRAMMING PROBLEM AND ITS APPLICATIONS Sahdul ISLAM Depame of Mahemacs, Guskaa Mahavdyalaya, Guskaa, Budwa

More information

The Differential Approach to Superlative Index Number Theory

The Differential Approach to Superlative Index Number Theory The Dffeeal Aoach o Suelae Ide Numbe Theoy by Wllam A. Bae K-Hog Cho ad Taa M. Scla Decembe 8 Fohcomg he Ha Thel Memoal Secal ue of he Joual of Agculual ad Aled Ecoomc Wllam A. Bae ofeo of Ecoomc a Wahgo

More information

ESTIMATION OF PARAMETERS AND VERIFICATION OF STATISTICAL HYPOTHESES FOR GAUSSIAN MODELS OF STOCK PRICE

ESTIMATION OF PARAMETERS AND VERIFICATION OF STATISTICAL HYPOTHESES FOR GAUSSIAN MODELS OF STOCK PRICE Lhuaa Joual of Sascs Leuvos sasos daba 06, vol 55, o, pp 9 0 06, 55,, 9 0 p wwwsascsjouall ESTIMATIO OF PARAMETERS AD VERIFICATIO OF STATISTICAL YPOTESES FOR GAUSSIA MODELS OF STOCK PRICE Dmyo Maushevych,

More information

The Solutions of Initial Value Problems for Nonlinear Fourth-Order Impulsive Integro-Differential Equations in Banach Spaces

The Solutions of Initial Value Problems for Nonlinear Fourth-Order Impulsive Integro-Differential Equations in Banach Spaces WSEAS TRANSACTIONS o MATHEMATICS Zhag Lglg Y Jgy Lu Juguo The Soluos of Ial Value Pobles fo Nolea Fouh-Ode Ipulsve Iego-Dffeeal Equaos Baach Spaces Zhag Lglg Y Jgy Lu Juguo Depae of aheacs of Ta Yua Uvesy

More information

Continuous Time Markov Chains

Continuous Time Markov Chains Couous me Markov chas have seay sae probably soluos f a oly f hey are ergoc, us lke scree me Markov chas. Fg he seay sae probably vecor for a couous me Markov cha s o more ffcul ha s he scree me case,

More information

Learning of Graphical Models Parameter Estimation and Structure Learning

Learning of Graphical Models Parameter Estimation and Structure Learning Learg of Grahal Models Parameer Esmao ad Sruure Learg e Fukumzu he Isue of Sasal Mahemas Comuaoal Mehodology Sasal Iferee II Work wh Grahal Models Deermg sruure Sruure gve by modelg d e.g. Mxure model

More information

Density estimation III.

Density estimation III. Lecure 6 esy esmao III. Mlos Hausrec mlos@cs..eu 539 Seo Square Oule Oule: esy esmao: Bomal srbuo Mulomal srbuo ormal srbuo Eoeal famly aa: esy esmao {.. } a vecor of arbue values Objecve: ry o esmae e

More information

APPLICATION OF A Z-TRANSFORMS METHOD FOR INVESTIGATION OF MARKOV G-NETWORKS

APPLICATION OF A Z-TRANSFORMS METHOD FOR INVESTIGATION OF MARKOV G-NETWORKS Joa of Aed Mahema ad Comaoa Meha 4 3( 6-73 APPLCATON OF A Z-TRANSFORMS METHOD FOR NVESTGATON OF MARKOV G-NETWORKS Mha Maay Vo Nameo e of Mahema Ceohowa Uey of Tehoogy Cęohowa Poad Fay of Mahema ad Come

More information

Fairing of Parametric Quintic Splines

Fairing of Parametric Quintic Splines ISSN 46-69 Eglad UK Joual of Ifomato ad omputg Scece Vol No 6 pp -8 Fag of Paametc Qutc Sples Yuau Wag Shagha Isttute of Spots Shagha 48 ha School of Mathematcal Scece Fuda Uvesty Shagha 4 ha { P t )}

More information

Fundamentals of Speech Recognition Suggested Project The Hidden Markov Model

Fundamentals of Speech Recognition Suggested Project The Hidden Markov Model . Projec Iroduco Fudameals of Speech Recogo Suggesed Projec The Hdde Markov Model For hs projec, s proposed ha you desg ad mpleme a hdde Markov model (HMM) ha opmally maches he behavor of a se of rag sequeces

More information

Real-Time Systems. Example: scheduling using EDF. Feasibility analysis for EDF. Example: scheduling using EDF

Real-Time Systems. Example: scheduling using EDF. Feasibility analysis for EDF. Example: scheduling using EDF EDA/DIT6 Real-Tme Sysems, Chalmers/GU, 0/0 ecure # Updaed February, 0 Real-Tme Sysems Specfcao Problem: Assume a sysem wh asks accordg o he fgure below The mg properes of he asks are gve he able Ivesgae

More information

Maximum likelihood estimate of phylogeny. BIOL 495S/ CS 490B/ MATH 490B/ STAT 490B Introduction to Bioinformatics April 24, 2002

Maximum likelihood estimate of phylogeny. BIOL 495S/ CS 490B/ MATH 490B/ STAT 490B Introduction to Bioinformatics April 24, 2002 Mmm lkelhood eme of phylogey BIO 9S/ S 90B/ MH 90B/ S 90B Iodco o Bofomc pl 00 Ovevew of he pobblc ppoch o phylogey o k ee ccodg o he lkelhood d ee whee d e e of eqece d ee by ee wh leve fo he eqece. he

More information

Anouncements. Conjugate Gradients. Steepest Descent. Outline. Steepest Descent. Steepest Descent

Anouncements. Conjugate Gradients. Steepest Descent. Outline. Steepest Descent. Steepest Descent oucms Couga Gas Mchal Kazha (6.657) Ifomao abou h Sma (6.757) hav b pos ol: hp://www.cs.hu.u/~msha Tch Spcs: o M o Tusay afoo. o Two paps scuss ach w. o Vos fo w s caa paps u by Thusay vg. Oul Rvw of Sps

More information

Density estimation. Density estimations. CS 2750 Machine Learning. Lecture 5. Milos Hauskrecht 5329 Sennott Square

Density estimation. Density estimations. CS 2750 Machine Learning. Lecture 5. Milos Hauskrecht 5329 Sennott Square Lecure 5 esy esmao Mlos Hauskrec mlos@cs..edu 539 Seo Square esy esmaos ocs: esy esmao: Mamum lkelood ML Bayesa arameer esmaes M Beroull dsrbuo. Bomal dsrbuo Mulomal dsrbuo Normal dsrbuo Eoeal famly Noaramerc

More information

Chapter 8. Simple Linear Regression

Chapter 8. Simple Linear Regression Chaper 8. Smple Lear Regresso Regresso aalyss: regresso aalyss s a sascal mehodology o esmae he relaoshp of a respose varable o a se of predcor varable. whe here s jus oe predcor varable, we wll use smple

More information

Density estimation III.

Density estimation III. Lecure 4 esy esmao III. Mlos Hauskrec mlos@cs..edu 539 Seo Square Oule Oule: esy esmao: Mamum lkelood ML Bayesa arameer esmaes MP Beroull dsrbuo. Bomal dsrbuo Mulomal dsrbuo Normal dsrbuo Eoeal famly Eoeal

More information

On Probability Density Function of the Quotient of Generalized Order Statistics from the Weibull Distribution

On Probability Density Function of the Quotient of Generalized Order Statistics from the Weibull Distribution ISSN 684-843 Joua of Sac Voue 5 8 pp. 7-5 O Pobaby Dey Fuco of he Quoe of Geeaed Ode Sac fo he Webu Dbuo Abac The pobaby dey fuco of Muhaad Aee X k Y k Z whee k X ad Y k ae h ad h geeaed ode ac fo Webu

More information

Minimum Hyper-Wiener Index of Molecular Graph and Some Results on Szeged Related Index

Minimum Hyper-Wiener Index of Molecular Graph and Some Results on Szeged Related Index Joual of Multdscplay Egeeg Scece ad Techology (JMEST) ISSN: 359-0040 Vol Issue, Febuay - 05 Mmum Hype-Wee Idex of Molecula Gaph ad Some Results o eged Related Idex We Gao School of Ifomato Scece ad Techology,

More information

Upper Bound For Matrix Operators On Some Sequence Spaces

Upper Bound For Matrix Operators On Some Sequence Spaces Suama Uer Bou formar Oeraors Uer Bou For Mar Oeraors O Some Sequece Saces Suama Dearme of Mahemacs Gaah Maa Uersy Yogyaara 558 INDONESIA Emal: suama@ugmac masomo@yahoocom Isar D alam aer aa susa masalah

More information

The Central Limit Theorems for Sums of Powers of Function of Independent Random Variables

The Central Limit Theorems for Sums of Powers of Function of Independent Random Variables ScieceAsia 8 () : 55-6 The Ceal Limi Theoems fo Sums of Poes of Fucio of Idepede Radom Vaiables K Laipapo a ad K Neammaee b a Depame of Mahemaics Walailak Uivesiy Nakho Si Thammaa 86 Thailad b Depame of

More information

Some Probability Inequalities for Quadratic Forms of Negatively Dependent Subgaussian Random Variables

Some Probability Inequalities for Quadratic Forms of Negatively Dependent Subgaussian Random Variables Joural of Sceces Islamc epublc of Ira 6(: 63-67 (005 Uvers of ehra ISSN 06-04 hp://scecesuacr Some Probabl Iequales for Quadrac Forms of Negavel Depede Subgaussa adom Varables M Am A ozorga ad H Zare 3

More information

QR factorization. Let P 1, P 2, P n-1, be matrices such that Pn 1Pn 2... PPA

QR factorization. Let P 1, P 2, P n-1, be matrices such that Pn 1Pn 2... PPA QR facorzao Ay x real marx ca be wre as AQR, where Q s orhogoal ad R s upper ragular. To oba Q ad R, we use he Householder rasformao as follows: Le P, P, P -, be marces such ha P P... PPA ( R s upper ragular.

More information

_ J.. C C A 551NED. - n R ' ' t i :. t ; . b c c : : I I .., I AS IEC. r '2 5? 9

_ J.. C C A 551NED. - n R ' ' t i :. t ; . b c c : : I I .., I AS IEC. r '2 5? 9 C C A 55NED n R 5 0 9 b c c \ { s AS EC 2 5? 9 Con 0 \ 0265 o + s ^! 4 y!! {! w Y n < R > s s = ~ C c [ + * c n j R c C / e A / = + j ) d /! Y 6 ] s v * ^ / ) v } > { ± n S = S w c s y c C { ~! > R = n

More information

DUALITY IN MULTIPLE CRITERIA AND MULTIPLE CONSTRAINT LEVELS LINEAR PROGRAMMING WITH FUZZY PARAMETERS

DUALITY IN MULTIPLE CRITERIA AND MULTIPLE CONSTRAINT LEVELS LINEAR PROGRAMMING WITH FUZZY PARAMETERS Ida Joual of Fudameal ad ppled Lfe Sceces ISSN: 223 6345 (Ole) Ope ccess, Ole Ieaoal Joual valable a www.cbech.o/sp.ed/ls/205/0/ls.hm 205 Vol.5 (S), pp. 447-454/Noua e al. Reseach cle DULITY IN MULTIPLE

More information

Comparing Different Estimators for Parameters of Kumaraswamy Distribution

Comparing Different Estimators for Parameters of Kumaraswamy Distribution Compaig Diffee Esimaos fo Paamees of Kumaaswamy Disibuio ا.م.د نذير عباس ابراهيم الشمري جامعة النهرين/بغداد-العراق أ.م.د نشات جاسم محمد الجامعة التقنية الوسطى/بغداد- العراق Absac: This pape deals wih compaig

More information

Manifolds with Bakry-Emery Ricci Curvature Bounded Below

Manifolds with Bakry-Emery Ricci Curvature Bounded Below Advaces Pue Maemacs, 6, 6, 754-764 ://wwwscog/joual/am ISSN Ole: 6-384 ISSN P: 6-368 Maolds w Baky-Emey Rcc Cuvaue Bouded Below Issa Allassae Kaboye, Bazaaé Maama Faculé de Sceces e Tecques, Uvesé de Zde,

More information

Partial Molar Properties of solutions

Partial Molar Properties of solutions Paral Molar Properes of soluos A soluo s a homogeeous mxure; ha s, a soluo s a oephase sysem wh more ha oe compoe. A homogeeous mxures of wo or more compoes he gas, lqud or sold phase The properes of a

More information

Final Exam Applied Econometrics

Final Exam Applied Econometrics Fal Eam Appled Ecoomercs. 0 Sppose we have he followg regresso resl: Depede Varable: SAT Sample: 437 Iclded observaos: 437 Whe heeroskedasc-cosse sadard errors & covarace Varable Coeffce Sd. Error -Sasc

More information

= y and Normed Linear Spaces

= y and Normed Linear Spaces 304-50 LINER SYSTEMS Lectue 8: Solutos to = ad Nomed Lea Spaces 73 Fdg N To fd N, we eed to chaacteze all solutos to = 0 Recall that ow opeatos peseve N, so that = 0 = 0 We ca solve = 0 ecusvel backwads

More information

Applying Eyring s Model to Times to Breakdown of Insulating Fluid

Applying Eyring s Model to Times to Breakdown of Insulating Fluid Ieaoal Joual of Pefomably Egeeg, Vol. 8, No. 3, May 22, pp. 279-288. RAMS Cosulas Ped Ida Applyg Eyg s Model o Tmes o Beakdow of Isulag Flud DANIEL I. DE SOUZA JR. ad R. ROCHA Flumese Fed. Uvesy, Cvl Egeeg

More information

Lagrangian & Hamiltonian Mechanics:

Lagrangian & Hamiltonian Mechanics: XII AGRANGIAN & HAMITONIAN DYNAMICS Iouco Hamlo aaoal Pcple Geealze Cooaes Geealze Foces agaga s Euao Geealze Momea Foces of Cosa, agage Mulples Hamloa Fucos, Cosevao aws Hamloa Dyamcs: Hamlo s Euaos agaga

More information

A New Approach to Probabilistic Load Flow

A New Approach to Probabilistic Load Flow INDIAN INSTITUTE OF TECHNOLOGY, KHARAGPUR 73, DECEMBER 79, 837 A New Appoach o Pobablsc Load Flow T K Basu, R B Msa ad Puob Paoway Absac: Ths pape descbes a ew appoach o modellg of asmsso le uceaes usg

More information

Midterm Exam. Tuesday, September hour, 15 minutes

Midterm Exam. Tuesday, September hour, 15 minutes Ecoomcs of Growh, ECON560 Sa Fracsco Sae Uvers Mchael Bar Fall 203 Mderm Exam Tuesda, Sepember 24 hour, 5 mues Name: Isrucos. Ths s closed boo, closed oes exam. 2. No calculaors of a d are allowed. 3.

More information

COMPARISON OF ESTIMATORS OF PARAMETERS FOR THE RAYLEIGH DISTRIBUTION

COMPARISON OF ESTIMATORS OF PARAMETERS FOR THE RAYLEIGH DISTRIBUTION COMPARISON OF ESTIMATORS OF PARAMETERS FOR THE RAYLEIGH DISTRIBUTION Eldesoky E. Affy. Faculy of Eg. Shbee El kom Meoufa Uv. Key word : Raylegh dsrbuo, leas squares mehod, relave leas squares, leas absolue

More information

14. Poisson Processes

14. Poisson Processes 4. Posso Processes I Lecure 4 we roduced Posso arrvals as he lmg behavor of Bomal radom varables. Refer o Posso approxmao of Bomal radom varables. From he dscusso here see 4-6-4-8 Lecure 4 " arrvals occur

More information

Determination of Antoine Equation Parameters. December 4, 2012 PreFEED Corporation Yoshio Kumagae. Introduction

Determination of Antoine Equation Parameters. December 4, 2012 PreFEED Corporation Yoshio Kumagae. Introduction refeed Soluos for R&D o Desg Deermao of oe Equao arameers Soluos for R&D o Desg December 4, 0 refeed orporao Yosho Kumagae refeed Iroduco hyscal propery daa s exremely mpora for performg process desg ad

More information

REPLACEMENT CYCLES, INCOME DISTRIBUTION, AND DYNAMIC PRICE DISCRIMINATION

REPLACEMENT CYCLES, INCOME DISTRIBUTION, AND DYNAMIC PRICE DISCRIMINATION REPLACEMENT CYCLES, INCOME DISTRIBUTION, AND DYNAMIC PRICE DISCRIMINATION Eduado Coea de Souza IBMEC São Paulo. Rua Quaá 300, Vla Olíma, São Paulo SP Bazl 04546-042 (emal: eduadocs@s.edu.b) Joge Cham Basa

More information

Lecture 3 summary. C4 Lecture 3 - Jim Libby 1

Lecture 3 summary. C4 Lecture 3 - Jim Libby 1 Lecue su Fes of efeece Ivce ude sfoos oo of H wve fuco: d-fucos Eple: e e - µ µ - Agul oeu s oo geeo Eule gles Geec slos cosevo lws d Noehe s heoe C4 Lecue - Lbb Fes of efeece Cosde fe of efeece O whch

More information

5-1. We apply Newton s second law (specifically, Eq. 5-2). F = ma = ma sin 20.0 = 1.0 kg 2.00 m/s sin 20.0 = 0.684N. ( ) ( )

5-1. We apply Newton s second law (specifically, Eq. 5-2). F = ma = ma sin 20.0 = 1.0 kg 2.00 m/s sin 20.0 = 0.684N. ( ) ( ) 5-1. We apply Newon s second law (specfcally, Eq. 5-). (a) We fnd he componen of he foce s ( ) ( ) F = ma = ma cos 0.0 = 1.00kg.00m/s cos 0.0 = 1.88N. (b) The y componen of he foce s ( ) ( ) F = ma = ma

More information

L-MOMENTS EVALUATION FOR IDENTICALLY AND NONIDENTICALLY WEIBULL DISTRIBUTED RANDOM VARIABLES

L-MOMENTS EVALUATION FOR IDENTICALLY AND NONIDENTICALLY WEIBULL DISTRIBUTED RANDOM VARIABLES THE PUBLISHING HOUSE PROCEEDINGS OF THE ROMANIAN ACADEMY, Sees A, OF THE ROMANIAN ACADEMY Volume 8, Numbe 3/27,. - L-MOMENTS EVALUATION FOR IDENTICALLY AND NONIDENTICALLY WEIBULL DISTRIBUTED RANDOM VARIABLES

More information

D KL (P Q) := p i ln p i q i

D KL (P Q) := p i ln p i q i Cheroff-Bouds 1 The Geeral Boud Let P 1,, m ) ad Q q 1,, q m ) be two dstrbutos o m elemets, e,, q 0, for 1,, m, ad m 1 m 1 q 1 The Kullback-Lebler dvergece or relatve etroy of P ad Q s defed as m D KL

More information

θ = θ Π Π Parametric counting process models θ θ θ Log-likelihood: Consider counting processes: Score functions:

θ = θ Π Π Parametric counting process models θ θ θ Log-likelihood: Consider counting processes: Score functions: Paramerc coug process models Cosder coug processes: N,,..., ha cou he occurreces of a eve of eres for dvduals Iesy processes: Lelhood λ ( ;,,..., N { } λ < Log-lelhood: l( log L( Score fucos: U ( l( log

More information

Fault Tolerant Computing. Fault Tolerant Computing CS 530 Probabilistic methods: overview

Fault Tolerant Computing. Fault Tolerant Computing CS 530 Probabilistic methods: overview Probably 1/19/ CS 53 Probablsc mehods: overvew Yashwa K. Malaya Colorado Sae Uversy 1 Probablsc Mehods: Overvew Cocree umbers presece of uceray Probably Dsjo eves Sascal depedece Radom varables ad dsrbuos

More information

Computational Fluid Dynamics. Numerical Methods for Parabolic Equations. Numerical Methods for One-Dimensional Heat Equations

Computational Fluid Dynamics. Numerical Methods for Parabolic Equations. Numerical Methods for One-Dimensional Heat Equations Compuaoal Flud Dyamcs p://www.d.edu/~gyggva/cfd-couse/ Compuaoal Flud Dyamcs p://www.d.edu/~gyggva/cfd-couse/ Compuaoal Flud Dyamcs Numecal Meods o Paabolc Equaos Lecue Mac 6 7 Géa Tyggvaso Compuaoal Flud

More information

The Linear Regression Of Weighted Segments

The Linear Regression Of Weighted Segments The Lear Regresso Of Weghed Segmes George Dael Maeescu Absrac. We proposed a regresso model where he depede varable s made o up of pos bu segmes. Ths suao correspods o he markes hroughou he da are observed

More information

Quantitative Portfolio Theory & Performance Analysis

Quantitative Portfolio Theory & Performance Analysis 550.447 Quaave Porfolo heory & Performace Aalyss Week February 4 203 Coceps. Assgme For February 4 (hs Week) ead: A&L Chaper Iroduco & Chaper (PF Maageme Evrome) Chaper 2 ( Coceps) Seco (Basc eur Calculaos)

More information

are positive, and the pair A, B is controllable. The uncertainty in is introduced to model control failures.

are positive, and the pair A, B is controllable. The uncertainty in is introduced to model control failures. Lectue 4 8. MRAC Desg fo Affe--Cotol MIMO Systes I ths secto, we cosde MRAC desg fo a class of ult-ut-ult-outut (MIMO) olea systes, whose lat dyacs ae lealy aaetezed, the ucetates satsfy the so-called

More information

1 Constant Real Rate C 1

1 Constant Real Rate C 1 Consan Real Rae. Real Rae of Inees Suppose you ae equally happy wh uns of he consumpon good oday o 5 uns of he consumpon good n peod s me. C 5 Tha means you ll be pepaed o gve up uns oday n eun fo 5 uns

More information

dm dt = 1 V The number of moles in any volume is M = CV, where C = concentration in M/L V = liters. dcv v

dm dt = 1 V The number of moles in any volume is M = CV, where C = concentration in M/L V = liters. dcv v Mg: Pcess Aalyss: Reac ae s defed as whee eac ae elcy lue M les ( ccea) e. dm he ube f les ay lue s M, whee ccea M/L les. he he eac ae beces f a hgeeus eac, ( ) d Usually s csa aqueus eeal pcesses eac,

More information

ON THE EXTENSION OF WEAK ARMENDARIZ RINGS RELATIVE TO A MONOID

ON THE EXTENSION OF WEAK ARMENDARIZ RINGS RELATIVE TO A MONOID wwweo/voue/vo9iue/ijas_9 9f ON THE EXTENSION OF WEAK AENDAIZ INGS ELATIVE TO A ONOID Eye A & Ayou Eoy Dee of e Nowe No Uvey Lzou 77 C Dee of e Uvey of Kou Ou Su E-: eye76@o; you975@yooo ABSTACT Fo oo we

More information

Hydrodynamic Modeling: Hydrodynamic Equations. Prepared by Dragica Vasileska Professor Arizona State University

Hydrodynamic Modeling: Hydrodynamic Equations. Prepared by Dragica Vasileska Professor Arizona State University Hyoyamc Moelg: Hyoyamc Equaos Peae by Dagca Vasleska Poesso Azoa Sae Uesy D-Duso Aoaches Val he use aso omaes Exeso o DD Aoaches Valy Iouco o he el-eee mobly Velocy Sauao Imlcaos Velocy Sauao Dece Scalg

More information

University of Pavia, Pavia, Italy. North Andover MA 01845, USA

University of Pavia, Pavia, Italy. North Andover MA 01845, USA Iteatoal Joual of Optmzato: heoy, Method ad Applcato 27-5565(Pt) 27-6839(Ole) wwwgph/otma 29 Global Ifomato Publhe (HK) Co, Ltd 29, Vol, No 2, 55-59 η -Peudoleaty ad Effcecy Gogo Gog, Noma G Rueda 2 *

More information

MAT 516 Curve and Surface Methods for CAGD [Kaedah Lengkung dan Permukaan untuk RGBK]

MAT 516 Curve and Surface Methods for CAGD [Kaedah Lengkung dan Permukaan untuk RGBK] UNIVERSITI SAINS MALAYSIA Secod Semese Examao / Academc Sesso Jue MAT 56 Cuve ad Suface Mehods fo CAGD [Kaedah Legkug da Pemukaa uuk RGBK] Duao : hous [Masa : am] Please check ha hs examao ae cosss of

More information

Support Appendix The Logistics Impact of a Mixture of Order-Streams in a Manufacturer-Retailer System Ananth V Iyer and Apurva Jain

Support Appendix The Logistics Impact of a Mixture of Order-Streams in a Manufacturer-Retailer System Ananth V Iyer and Apurva Jain So Aedx Te og Ia o a Mxe o Ode-Sea a Maae-Reale Sye Aa V Iye ad Ava Ja Teoe 4: e ad q be e obably geeag o o e eady-ae be o ode ee e ye by a avg H ode ad a M ode eevely Te ad q Wee ad be e ee oo o e ollowg

More information

The Mean Residual Lifetime of (n k + 1)-out-of-n Systems in Discrete Setting

The Mean Residual Lifetime of (n k + 1)-out-of-n Systems in Discrete Setting Appled Mahemacs 4 5 466-477 Publshed Ole February 4 (hp//wwwscrporg/oural/am hp//dxdoorg/436/am45346 The Mea Resdual Lfeme of ( + -ou-of- Sysems Dscree Seg Maryam Torab Sahboom Deparme of Sascs Scece ad

More information

As evident from the full-sample-model, we continue to assume that individual errors are identically and

As evident from the full-sample-model, we continue to assume that individual errors are identically and Maxmum Lkelhood smao Greee Ch.4; App. R scrp modsa, modsb If we feel safe makg assumpos o he sascal dsrbuo of he error erm, Maxmum Lkelhood smao (ML) s a aracve alerave o Leas Squares for lear regresso

More information

Key words: Fractional difference equation, oscillatory solutions,

Key words: Fractional difference equation, oscillatory solutions, OSCILLATION PROPERTIES OF SOLUTIONS OF FRACTIONAL DIFFERENCE EQUATIONS Musafa BAYRAM * ad Ayd SECER * Deparme of Compuer Egeerg, Isabul Gelsm Uversy Deparme of Mahemacal Egeerg, Yldz Techcal Uversy * Correspodg

More information

Probability Bracket Notation and Probability Modeling. Xing M. Wang Sherman Visual Lab, Sunnyvale, CA 94087, USA. Abstract

Probability Bracket Notation and Probability Modeling. Xing M. Wang Sherman Visual Lab, Sunnyvale, CA 94087, USA. Abstract Probably Bracke Noao ad Probably Modelg Xg M. Wag Sherma Vsual Lab, Suyvale, CA 94087, USA Absrac Ispred by he Drac oao, a ew se of symbols, he Probably Bracke Noao (PBN) s proposed for probably modelg.

More information

FALL HOMEWORK NO. 6 - SOLUTION Problem 1.: Use the Storage-Indication Method to route the Input hydrograph tabulated below.

FALL HOMEWORK NO. 6 - SOLUTION Problem 1.: Use the Storage-Indication Method to route the Input hydrograph tabulated below. Jorge A. Ramírez HOMEWORK NO. 6 - SOLUTION Problem 1.: Use he Sorage-Idcao Mehod o roue he Ipu hydrograph abulaed below. Tme (h) Ipu Hydrograph (m 3 /s) Tme (h) Ipu Hydrograph (m 3 /s) 0 0 90 450 6 50

More information

The ray paths and travel times for multiple layers can be computed using ray-tracing, as demonstrated in Lab 3.

The ray paths and travel times for multiple layers can be computed using ray-tracing, as demonstrated in Lab 3. C. Trael me cures for mulple reflecors The ray pahs ad rael mes for mulple layers ca be compued usg ray-racg, as demosraed Lab. MATLAB scrp reflec_layers_.m performs smple ray racg. (m) ref(ms) ref(ms)

More information

4.1 Schrödinger Equation in Spherical Coordinates

4.1 Schrödinger Equation in Spherical Coordinates Phs 34 Quu Mehs D 9 9 Mo./ Wed./ Thus /3 F./4 Mo., /7 Tues. / Wed., /9 F., /3 4.. -. Shodge Sphe: Sepo & gu (Q9.) 4..-.3 Shodge Sphe: gu & d(q9.) Copuo: Sphe Shodge s 4. Hdoge o (Q9.) 4.3 gu Moeu 4.4.-.

More information

Minimizing spherical aberrations Exploiting the existence of conjugate points in spherical lenses

Minimizing spherical aberrations Exploiting the existence of conjugate points in spherical lenses Mmzg sphecal abeatos Explotg the exstece of cojugate pots sphecal leses Let s ecall that whe usg asphecal leses, abeato fee magg occus oly fo a couple of, so called, cojugate pots ( ad the fgue below)

More information

PORTFOLIO OPTIMIZATION PROBLEMS: A SURVEY Irina Bolshakova, Mikhail Kovalev Belarusian State University Minsk, Belarus

PORTFOLIO OPTIMIZATION PROBLEMS: A SURVEY Irina Bolshakova, Mikhail Kovalev Belarusian State University Minsk, Belarus PORFOLIO OPIMIZAION PROBLEMS: A SURVEY Ia Bolshakova, Mkhal Kovalev Belausa Sae Uvesy Msk, Belaus Ebehad Glch Faculy of Mahemacs Oo-vo-Guecke Uvesy Magdebug Posfach 40 3906 Magdebug Gemay Pep N 009 Fakulä

More information

CONTROL ROUTH ARRAY AND ITS APPLICATIONS

CONTROL ROUTH ARRAY AND ITS APPLICATIONS 3 Asa Joual of Cool, Vol 5, No, pp 3-4, Mach 3 CONTROL ROUTH ARRAY AND ITS APPLICATIONS Dazha Cheg ad TJTa Bef Pape ABSTRACT I hs pape he Rouh sably ceo [6] has bee developed o cool Rouh aay Soe foulas

More information

Chapter 3: Vectors and Two-Dimensional Motion

Chapter 3: Vectors and Two-Dimensional Motion Chape 3: Vecos and Two-Dmensonal Moon Vecos: magnude and decon Negae o a eco: eese s decon Mulplng o ddng a eco b a scala Vecos n he same decon (eaed lke numbes) Geneal Veco Addon: Tangle mehod o addon

More information

GLOBAL OPTIMIZATION FOR THE SYNTHESIS OF INTEGRATED WATER SYSTEMS IN CHEMICAL PROCESSES

GLOBAL OPTIMIZATION FOR THE SYNTHESIS OF INTEGRATED WATER SYSTEMS IN CHEMICAL PROCESSES GOBA OPTIMIZATIO O THE SYTHESIS O ITEGATED WATE SYSTEMS I HEMIA POESSES amuma Kauah ad Igaco E. Gossma* Deame o hemcal Egeeg aege Mello vesy Psbugh PA 523 Mach 2005 ABSTAT I hs ae we addess he oblem o

More information