1. Introduction. Markovian stochastic control problems on a given horizon of time T can typically be written as. β(xu α,α u,e)ñ(du, de).

Size: px
Start display at page:

Download "1. Introduction. Markovian stochastic control problems on a given horizon of time T can typically be written as. β(xu α,α u,e)ñ(du, de)."

Transcription

1 SIAM J. CONROL OPIM. Vol. 54, No. 4, pp c 2016 Sociey for Idurial ad Applied Mahemaic A WEAK DYNAMIC PROGRAMMING PRINCIPLE FOR COMBINED OPIMAL SOPPING/SOCHASIC CONROL WIH E f -EXPECAIONS ROXANA DUMIRESCU, MARIE-CLAIRE QUENEZ, AND AGNÈS SULEM Abrac. We udy a combied opimal corol/oppig problem uder a oliear expecaio E f iduced by a BSDE wih jump, i a Markovia framework. he ermial reward fucio i oly uppoed o be Borelia. he value fucio u aociaed wih hi problem i geerally irregular. We fir eablih a ub- rep., uper- opimaliy priciple of dyamic programmig ivolvig i upper- rep., lower- emicoiuou evelope u rep., u. hi reul, called he weak dyamic programmig priciple DPP, exed ha obaied i [Bouchard ad ouzi, SIAM J. Corol Opim., , pp i he cae of a claical expecaio o he cae of a E f -expecaio ad Borelia ermial reward fucio. Uig hi weak DPP, we he prove ha u rep., u iavicoiy ub- rep., uper- oluio of a oliear Hamilo Jacobi Bellma variaioal iequaliy. Key word. Markovia ochaic corol, mixed opimal corol/oppig, oliear expecaio, backward ochaic differeial equaio, weak dyamic programmig priciple, Hamilo Jacobi Bellma variaioal iequaliy, vicoiy oluio, E f -expecaio AMS ubjec claificaio. 93E20, 60J60, 47N10 DOI /15M Iroducio. Markovia ochaic corol problem o a give horizo of ime ca ypically be wrie a [ 1.1 u0,x=upe α A fα,x α d + gx α, 0 where A i a e of admiible corol procee α,adx α i a corolled proce of he form X α = x + bxu α,α u du + σxu α,α u dw u βxu α,α u,eñdu, de. R he radom variable gx α may repree a ermial reward ad fα,x αai- aaeou reward proce. Formally, for all iiial ime i [0, ad iiial ae y, he aociaed value fucio i defied by [ 1.2 u, y =upe α A fα,x α d + gx α Xα = y. Received by he edior Jue 19, 2015; acceped for publicaio i revied form May 17, 2016; publihed elecroically Augu 24, hp:// Fudig: he reearch of he fir auhor leadig o hee reul ha received fudig from he Regio Ile-de-Frace. Iiu für Mahemaik, Humbold-Uiveriä zu Berli, Uer de Lide 6, Berli, Germay roxaa@ceremade.dauphie.fr. LPMA, Uiverié Pari 7 Dei Didero, Boie courrier 7012, Pari cedex 05, Frace queez@mah.uiv-pari-didero.fr. INRIA Pari, 3 rue Simoe Iff, CS 42112, Pari Cedex 12, Frace, ad Uiverié Pari-E, Pari, Frace age.ulem@iria.fr. 2090

2 WEAK DPP WIH E f -EXPECAIONS 2091 he dyamic programmig priciple DPP ca formally be aed a [ 1.3 u0,x=upe fα,x α d + u, Xα for i [0,. α A 0 hi priciple i claically eablihed uder aumpio which eure ha he value fucio u aifie ome regulariy properie. From hi priciple, i ca be derived ha he value fucio i a vicoiy oluio of he aociaed Hamilo Jacobi Bellma HJB equaio. Similar reul are obaied for opimal oppig ad mixed opimal oppig/corol problem. he cae of a dicoiuou value fucio ad i lik wih vicoiy oluio ha bee udied for deermiiic corol i he eighie. Barle ad Perhame udy i [3 a deermiiic opimal oppig problem wih a reward map g oly uppoed o be Borelia. o hi purpoe, hey iroduce a oio of vicoiy oluio which exed he claical oe o he dicoiuou cae: a fucio v i aid o be a weak vicoiy oluio of he HJB equaio if i upper emicoiuou u..c. evelope, deoed by v,iavicoiy uboluio of hi PDE, ad if i he lower emicoiuou l..c. evelope, deoed by v,iavicoiy uperoluio of hi equaio. he, by he claical DPP provided i he previou lieraure, hey ge ha he u..c. evelope u of he value fucio aifie a ubopimaliy priciple i he ee of Lio ad Sougaidi i [19. Uig hi ubopimaliy priciple, hey he how ha u i a vicoiy uboluio of he HJB equaio. Moreover, uig he fac ha he l..c. evelop u of u i he value fucio of a relaxed problem, hey how ha u i a vicoiy uperoluio ad hu ge ha u i a weak vicoiy oluio of he HJB equaio. hey re ha i geeral, he weak vicoiy oluio of hi PDE i o uique. However, uder a regulariy aumpio o he reward g, by uig he corol formulae, hey obai ha he u..c. evelope u of he value fucio i he uique u..c. vicoiy oluio of he HJB equaio. See Remark 23 for addiioal referece ad comme. More recely, i a ochaic framework, Bouchard ad ouzi [8 have proved a weak DPP whe he ermial map g i irregular: hey prove ha he value fucio u aifie a ubopimaliy priciple of dyamic programmig ivolvig i u..c. evelope u, ad uder a addiioal regulariy lower emicoiuiy aumpio of he reward g, hey obai a uperopimaliy priciple ivolvig he l..c. evelope u. he, uig he ubopimaliy priciple, hey derive ha u i a vicoiy uboluio of he aociaed HJB equaio. Moreover, whe g i l..c., uig he uperopimaliy priciple, hey how ha u i a vicoiy uperoluio ad hu ge ha u i a weak vicoiy oluio of hi PDE i he ame ee a above or [3. A weak DPP ha bee furher eablihed, whe g i l..c. for problem wih ae corai by Bouchard ad Nuz i [7, ad, whe g i coiuou, for zero-um ochaic game by Bayrakar ad Yao i [4. I hi paper we are iereed i geeralizig hee reul o he cae whe g i oly Borelia ad whe he liear expecaio E i replaced by a oliear expecaio iduced by a backward ochaic differeial equaio BSDE wih jump. ypically, uch problem i he Markovia cae ca be formulaed a 1.4 up E0, α [gx α, α A where E α i he oliear expecaio aociaed wih a BSDE wih jump wih corolled driver fα,x α,y,z,k. Noe ha problem 1.1 i a paricular cae of

3 2092 R. DUMIRESCU, M.-C. QUENEZ, AND A. SULEM 1.4 whe he driver f doe o deped o he oluio of he BSDE, ha i, whe fα,x α,y,z,k fα,x α. We fir provide a weak DPP ivolvig he u..c. ad l..c. evelope of he value fucio. For hi purpoe, we prove ome prelimiary reul, i paricular ome meaurabiliy ad pliig properie. No regulariy codiio o g i required o obai he ub- ad uper-opimaliy priciple, which i o he cae i he previou lieraure i he ochaic cae, eve wih a claical expecaio ee [8, [7, ad [4. Uig hi weak DPP, we he how ha he value fucio, which i geerally eiher u..c. or l..c., i a weak vicoiy oluio i he ee of [3 of a aociaed oliear HJB equaio. Moreover, i hi paper, we coider he combied problem whe here i a addiioal corol i he form of a oppig ime. We hu coider mixed geeralized opimal corol/oppig problem of he form 1.5 up up E0,τ α [ hτ,x τ α, α A τ where deoe he e of oppig ime wih value i [0,, ad h i a irregular reward fucio. Noe ha i he lieraure o BSDE, ome paper ee, e.g., Peg [20, Li ad Peg [18, Buckdah ad Li [9, ad Buckdah ad Nie [10 udy ochaic corol problem wih oliear E-expecaio i he coiuou cae wihou opimal oppig. heir approach i differe from our ad relie o he coiuiy aumpio of he reward fucio. he paper i orgaized a follow: i ecio 2, we formulae our geeralized mixed corol-opimal oppig problem. Uig reul o refleced BSDE RBSDE, we expre hi problem a a opimal corol problem for RBSDE. I ecio 3, we prove a weak DPP for our mixed problem wih E f -expecaio. hi require ome pecific echique of ochaic aalyi ad BSDE o hadle meaurabiliy ad oher iue due o he olieariy of he expecaio ad he lack of regulariy of he ermial reward. Uig he DPP ad properie of RBSDE, we prove i ecio 4 ha he value fucio of our mixed problem i a weak vicoiy oluio of a oliear HJB variaioal iequaliy. I he appedix, we give everal fie meaurabiliy properie which are ued i he paper. 2. Formulaio of he mixed oppig/corol problem. We coider he produc pace Ω := Ω W Ω N,whereΩ W := C[0, i he Wieer pace, ha i, he e of coiuou fucio ω 1 from [0,ioR p uch ha ω 1 0 = 0, ad Ω N := D[0, i he Skorohod pace of righ-coiuou wih lef limi RCLL fucio ω 2 from [0,ioR d, uch ha ω 2 0 = 0. Recall ha Ω i a Polih pace for he Skorohod meric. Here p, d 1, bu, for oaioal impliciy, we hall coider oly R-valued fucio, ha i, he cae p = d =1. Le B =B 1,B 2 be he caoical proce defied for each [0,adeach ω =ω 1,ω 2 bybω i =Bω i i :=ω i for i =1, 2. Le u deoe he fir coordiae proce B 1 by W.LeP W be he probabiliy meaure o Ω W, BΩ W uch ha W i a Browia moio. Here BΩ W deoe he Borelia σ-algebra o Ω W. Se E := R \{0} equipped wih i Borelia σ-algebra BE, where 1. We defie he jump radom meaure N a follow: for each >0adeachB BE, 2.1 N., [0, B := 1 {ΔB 2 B}. 0<

4 WEAK DPP WIH E f -EXPECAIONS 2093 he meaurable e E, BE i equipped wih a σ-fiie poiive meaure ν uch ha E 1 e νde <. Le P N be he probabiliy meaure o Ω N, BΩ N uch ha N i a Poio radom meaure wih compeaor νded ad uch ha B 2 = 0< ΔB2 a.. Noe ha he um of jump i well defied up o a P N -ull e. We e Ñdr, de :=Ndr, de νded. he pace Ω i equipped wih he σ-algebra BΩ ad he probabiliy meaure P := P W P N. Le F := F 0 be he filraio geeraed by W ad N compleedwihrepecobω ad P, defied a follow ee [17, p. 3 or [13, IV: le F be he compleio σ-algebra of BΩ wih repec o P 1 For each [0,, F i he σ-algebra geeraed by W,N, ad he P -ull e. Noe ha F = F ad F 0 i he σ-algebra geeraed by he P -ull e. Le P be he predicable σ-algebra o Ω [0, aociaed wih he filraio F. Le >0befixed. LeH 2 deoed alo by H2 be he e of real-valued predicable procee Z uch ha E 0 Z2 d < ad le S 2 be he e of real-valued RCLL adaped procee ϕ wih E[up 0 ϕ 2 <. Le L 2 ν be he e of meaurable fucio l :E, BE R, BR uch ha l 2 ν := E l2 eνde <. he e L 2 ν i a Hilber pace equipped wih he calar produc l, l ν := E lel eνde for all l, l L 2 ν L2 ν. Le H2 ν deoe he e of predicable real-valued procee k wih E 0 k 2 L d <. 2 ν Le A be he e of corol, defied a he e of predicable procee α valued i a compac ube A of R p,wherep N.Foreachα Aad each iiial codiio x i R, le X α,x 0 be he uique R-valued oluio i S 2 of he ochaic differeial equaio SDE: 2.2 X α,x = x + bxr α,x,α r dr + σxr α,x,α r dw r + βx α,x r,α r,eñdr, de, 0 0 where b, σ : R A R, are Lipchiz coiuou wih repec o x ad α, ad β : R A E R i a bouded meaurable fucio uch ha for ome coa C 0, ad for all e E βx, α, e C Ψe, x R,α A, where Ψ L 2 ν, βx, α, e βx,α,e C x x + α α Ψe, x,x R,α,α A. he crierio of our mixed corol problem, depedig o α, i defied via a BSDE wih driver fucio f aifyig he followig hypohei. Aumpio 1. f : A [0, R 3 L 2 ν R, BR i BA B[0, BR3 BL 2 ν -meaurable ad aifie i fα,, x, 0, 0, 0 C1 + x p for all α A, [0,,x R, wherep N, ii fα,, x, y, z, k fα,,x,y,z,k C α α + x x + y y + z z + k k L 2 ν for all [0,, x, x,y,y,z,z R, k, k L 2 ν,α,α A, iii fα,, x, y, z, k 2 fα,, x, y, z, k 1 γα,, x, y, z, k 1,k 2,k 2 k 1 ν for all, x, y, z, k 1,k 2,α, where γ : A [0, R 3 L 2 ν 2 L 2 ν, BL 2 ν i BA B[0, BR 3 BL 2 ν 2 - meaurable, aifyig γ.e 1ad γ.e Ψe, dνe-a.., where Ψ L 2 ν. For all x R ad all corol α A,lef α,x be he driver defied by f α,x r, ω, y, z, k :=fα r ω,r,xr α,x ω,y,z,k. 1 For he defiiio of he compleio of a σ-algebra ad he oe of P -ull e, ee, e.g., Lemma E

5 2094 R. DUMIRESCU, M.-C. QUENEZ, AND A. SULEM We iroduce he oliear expecaio E f α,x deoed more imply by E α,x aociaed wih f α,x, defied for each oppig ime τ ad for each η L 2 F τ a Er,τ α,x [η :=Xr α,x, 0 r τ, where Xr α,x i he oluio i S 2 of he BSDE aociaed wih driver f α,x,ermial ime τ, ad ermial codiio η, ha i aifyig dx α,x r = fα r,r,x α,x r Zr α,x dw r, Xr α,x E,Z α,x r,kr α,x dr Kr α,x α,x eñdr, de; XS = η, ad Z α,x, K α,x are he aociaed procee, which belog repecively o H 2 ad H 2 ν. Codiio iii eure he odecreaig propery of he Efα,x -expecaio ee [21. For all x R ad all corol α A, we defie he reward by h, X α,x for 0 < ad gx α,x for =,where g : R R i Borelia, h :[0, R R i a fucio which i Lipchiz coiuou wih repec o x uiformly i, ad coiuou wih repec o o [0,, h, x + gx C1 + x p for all [0,,x R, wih p N. Le be he e of oppig ime wih value i [0,. Suppoe he iiial ime i equal o 0. Noe ha E α,,x 0,τ [ hτ,xτ α,,x ca be ake a coa. 2 For each iiial codiio x R, we coider he mixed opimal corol/oppig problem: 2.3 u0,x:=upup α A τ E α,x α,x 0,τ [ hτ,xτ, where h, x :=h, x1 < + gx1 =. Noe ha h i Borelia bu o ecearily regular i, x. We ow make he problem dyamic. We defie for [0,adeachω Ωhe -ralaed pah ω =ω := ω ω. Noe ha ω 1, := ω 1 ω 1 correpod o he realizaio of he ralaed Browia moio W := W W ad ha he ralaed Poio radom meaure N := N,,. ca be expreed i erm of ω 2, := ω 2 ω 2 imilarly o 2.1. Le F =F be he filraio geeraed by W ad N compleed wih repec o BΩ ad P.Noeha for each [,, F i he σ-algebra geeraed by W r, N r, r, adf 0. Recall alo ha we have a marigale repreeaio heorem for F -marigale a ochaic iegral wih repec o W ad Ñ. Le u deoe by he e of oppig ime wih repec o F wih value i [,. Le P be he predicable σ-algebra o Ω [, equipped wih he filraio F. We ow iroduce he followig pace of procee. Le [0,. Le H 2 be he P -meaurable procee Z o Ω [, uch ha Z H 2 := E[ Zudu 2 <. We defie H 2,ν a he e of P -meaurable procee K o Ω [, uch ha K H 2,ν := E[ K u 2 ν du <. We deoe by S2 he e of real-valued RCLL procee ϕ o Ω [,, F -adaped, wih E[up ϕ 2 <. 2 Ideed, he oluio of a BSDE wih a Lipchiz driver i uique up o a P -ull e. I iiial value may hu be ake coa for all ω, modulo a chage of i value o a P -ull e, becaue F 0 i he σ-algebra geeraed by he P -ull e.

6 WEAK DPP WIH E f -EXPECAIONS 2095 Le A be he e of corol α : Ω [, A, which are P -meaurable. We coider he oluio deoed by X α,,x i S 2 of he followig SDE drive by he ralaed Browia moio W ad he ralaed Poio radom meaure N wih filraio F : 2.4 X α,,x = x + + E bx α,,x r,α r dr + βx α,,x r,α r,eñ dr, de. σxr α,,x,α r dwr For all, x [0, R ad all corol α A,lef α,,x be he driver defied by f α,,x r, ω, y, z, k :=fα r ω,r,xr α,,x ω,y,z,k. Le E.,τ α,,x [ hτ,xτ α,,x deoed alo by X α,,x be he oluio i S 2 of he BSDE wih driver f α,,x α,,x, ermial ime τ, ad ermial codiio hτ,xτ, drive by W ad N, which i olved o [, Ω wih repec o he filraio F : dxr α,,x = fα r,r,xr α,,x, Xr α,,x,zr α,,x,kr α,,x dr 2.5 Zr α,,x dwr E Kα,,x r eñ dr, de Xτ α,,x α,,x = hτ,xτ, where Z α,,x, K α,,x are he aociaed procee, which belog, repecively, o H 2 ad H 2,ν. Noe ha Eα,,x,τ [ hτ,xτ α,,x ca be ake deermiiic modulo a chage of i value o a P -ull e. 3 For each iiial ime ad each iiial codiio x, we defie he value fucio a 2.6 u, x := up α A up τ E α,,x,τ [ hτ,xτ α,,x, which i a deermiiic fucio of ad x. For each α A, we iroduce he fucio u α defied a We hu ge u α, x := up τ E α,,x,τ [ hτ,xτ α,,x. 2.7 u, x = up u α, x. α A For each α, u α, x h, x, ad hece u, x h, x. Moreover, u α,x = u,x=gx. By [22, heorem 3.2, for each α, he value fucio u α i relaed o a refleced BSDE. More preciely, le Y α,,x,z α,,x,k α,,x S 2 H 2 H 2 ν, be he oluio of he refleced BSDE aociaed wih driver f α,,x := fα,,x α,,x,y,z,k, RCLL 3 Ideed, he oluio E.,τ α,,x [ hτ, Xτ α,,x = X. α,,x of he BSDE 2.5 i uique up o a P -ull e. Moreover, i value a ime i F -meaurable, ad F i equal o he σ-algebra geeraed by he P -ull e ha i, F 0. he ame propery hold for he oluio of he refleced BSDE 2.8. See alo he addiioal Remark 6 ad 8.

7 2096 R. DUMIRESCU, M.-C. QUENEZ, AND A. SULEM obacle proce ξ α,,x filraio F,hai, := h, X α,,x, ermial codiio gx α,,x, ad wih 2.8 dyr α,,x = fα r,r,xr α,,x,yr α,,x,zr α,,x,kr α,,x dr + da α,,x Zr α,,x dwr E Kα,,x r, eñ dr, de, Y α,,x = gx α,,x ady α,,x ξ α,,x = h, X α,,x, 0 < a.., A α,,x i a RCLL odecreaig P -meaurable proce wih = 0 ad uch ha 0 Y α,,x ξ α,,x da α,,x,c = 0 a.. ad ΔA α,,x,d A α,,x = ΔA α,,x 1 {Y α,,x =ξα,,x } a.. Here A α,,x,c deoe he coiuou par of A ad A α,,x,d i dicoiuou par. I he paricular cae whe h,x gx, he obacle ξ α,,x aifie for all F - predicable oppig ime τ, ξ τ ξ τ a.., which implie he coiuiy of he proce A α,,x ee [22. I he followig, for each α A, Y α,,x will be alo deoed by Y α,,x, Noe ha i value a ime ca be ake a deermiiic modulo a chage of i value o a P -ull e. Uig [22, heorem 3.2, we ge ha for each α A, 2.9 u α, x =Y α,,x = Y α,,x, [gx α,,x. [gx α,,x. By uig hee equaliie, we ca reduce our mixed opimal oppig/corol problem 2.6 o a opimal corol problem for refleced BSDE. heorem 2 characerizaio of he value fucio. For each, x [0, R, he value fucio u, x of he mixed opimal oppig/ corol problem 2.6 aifie 2.10 u, x = up u α, x = up α A α A Y α,,x, [gx α,,x. hi key propery will be ued o olve our mixed problem. We poi ou ha i he claical cae of liear expecaio, hi approach allow u o provide aleraive proof of he DPP o hoe give i he previou lieraure. Remark 3. Some mixed opimal corol/oppig problem wih oliear expecaio have bee udied i [5, 22. I hee paper, he reward proce doe o deped o he corol, which yield he characerizaio of he value fucio a he oluio of a RBSDE. hi i o he cae here. 3. Weak DPP. I hi ecio, we prove a weak DPP for our mixed opimal corol/oppig problem 2.6. o hi purpoe, we fir provide ome pliig properie for he forward-backward yem We he how ome meaurabiliy properie of he fucio u α, x, defied by 2.9, wih repec o boh he ae variable x ad he corol α. Uig hee reul, we how he exiece of ε-opimal corol aifyig ome appropriae meaurabiliy properie. Moreover, we eablih a Faou lemma for RBSDE, where he limi ivolve boh ermial codiio ad ermial ime. Uig hee reul, we he prove a ub- rep., uper- opimaliy priciple of dyamic programmig, ivolvig he u..c. rep., l..c. evelope of he value fucio.

8 WEAK DPP WIH E f -EXPECAIONS Spliig properie. Le [0,. For each ω, le ω := ω r 0 r ad ω := ω r ω r. We hall ideify he pah ω wih ω, ω, which mea ha a pah ca be pli io wo par: he pah before ime ad he -ralaed pah afer ime. Le α be a give corol i A. We how below he followig: a ime, forfixed pa pah ω := ω, he proce α ω,. which oly deped o he fuure pah ω i a -admiible corol, ha i, α ω,. A ;furhermore,he crierium Y α,0,x ω,. from ime coicide wih he oluio of he refleced BSDE drive by W ad Ñ, corolled by α ω,. ad aociaed wih iiial ime ad iiial ae codiio X α,0,x ω. We iroduce he followig radom variable defied o Ω by S : ω ω ; : ω ω. Noe ha hey are idepede. For each ω Ω, we have ω = S ω+ ω1,, or equivalely ω r = ω r + ωr1, r, for all r [0,. For all pah ω, ω Ω, ω, ω deoe he pah uch ha he pa rajecory before i ha of ω, adhe-ralaed rajecory afer i ha of ω. hi ca alo be wrie a ω, ω := ω + ω 1,. Noe ha for each ω Ω, we have ω, ω = ω. Lemma 4. Le [0,. Le Z H 2. here exi a P -ull e N uch ha for each ω i he compleme N c of N, eig ω := ω = ω., he proce Z ω, deoed alo by Z ω,. defied by Z ω, :Ω [, R ;ω,r Z r ω, ω belog o H 2. Moreover, if Z A,heZ ω, A. hi propery alo hold for all iiial ime [0,. More preciely, le [,. Le Z H 2 rep., A. For a.e. ω Ω, he proce Z ω,. =Z r ω, r belog o H 2 rep., A. Proof. Claically, we have E[ Z2 r dr =E[E[ Z2 r dr F < +. Uig he idepedece of wih repec o F ad he meaurabiliy of S wih repec o F,wederiveha [ [ E Zr 2 dr F = E Z r S, 2 dr F = F S < +, P-a.., where F ω :=E[ Z r ω, 2 dr. Le u ow prove ha he proce Z ω, : ω,r Z r ω, ω i P - meaurable. here exi a proce idiiguihable of Z r, ill deoed by Z r, which i meaurable wih repec o he predicable σ-algebra aociaed wih he filraio geeraed by W ad N ee [13, IV, ecio 79. We ca hu uppoe i hi proof wihou lo of geeraliy ha F rep., F i he filraio geeraed by W ad N rep., W ad N, ad P rep., P i i aociaed predicable σ-algebra. Suppoe we have how ha he map ψ :Ω [, Ω [0,; ω,r ω, ω,rip, P-meaurable. Now, we have Z ω, ω,r=z ψω,r for each ω,r Ω [,. Sice Z i P-meaurable, by compoiio, we derive ha Z ω, ip -meaurable. I remai o how he P, P-meaurabiliy of ψ. Recall ha he σ-algebra P i geeraed by he e H v,, where v [0,[adH i of he form H =

9 2098 R. DUMIRESCU, M.-C. QUENEZ, AND A. SULEM {B i A i, 1 i }, where A i BR 2 ad 1 < 2 < v. I i hu ufficie o how ha ψ 1 H v, P. Noe ha ψ 1 H v, = H v,, where H = {ω Ω, ω, ω H}. Ifhereexii uch ha i ad ω i A i,he H =. Oherwie, we have H = {ω i ω A i for all i uch ha i >}. Hece H Fv, which implie ha ψ 1 H v, P. he proof i hu complee. Le Z H 2. Le u give a iermediary ime [0, ad a fixed pa pah ω. Noe ha he Lebegue iegral u Z rdr ω,. i equal a.. o he iegral u Z r ω,.dr. We ow how ha he ochaic iegral u Z rdw r ω,. coicide wih he ochaic iegral of he proce Z ω,. wih repec o he ralaed Browia moio W,hai, u Z r ω,.dwr. Lemma 5 pliig properie for ochaic iegral. Le [0,. Le Z H 2 ad K H 2 ν.hereexiap-ull e N which deped o uch ha for each ω N c,ad ω := ω, we have Z r ω, r H 2 ad K r ω, r H 2,ν, ad u u 3.1 Z r dw r ω, = Z r ω, dwr, P-a.., u u K r eñdr, de ω, = K r ω,,eñ dr, de, P-a.. E Remark 6. I he lieraure, he -ralaed Browia moio i ofe defied by W v := W +v W = W+v,0 v. ForeachZ H2 ad for each u, we have u Z rdwr = u Z 0 +r dw r a.. he ue of W hu allow u o avoid a chage of ime. he ame remark hold for he Poio radom meaure. Noe ha equaliy 3.1 i equivale o u Z rdw r ω, ω = u Z r ω, dwr ω forp-almo every ω Ω. he ame remark hold for he ecod equaliy. Proof. We hall oly prove he fir equaliy wih he Browia moio. he ecod oe wih he Poio radom meaure ca be how by imilar argume. Le u fir how ha equaliy 3.1 hold for a imple proce. Le a< ad le H L 2 F a. For each ω ω, ω =S ω, ω Ω, we have u H1 a, dw r ω, ω =H ω, ω ωu ωa u u = H ω, 1 a, dwr ω. Le ow Z H 2.LeuhowhaZaifie equaliy 3.1. he idea i o approximae Z by a appropriae equece of imple procee Z N o ha he equece Z N coverge i H 2 o Z ad ha, for almo every pa pah ω, he equece Z ω, N coverge o Z ω, ih 2.Foreach N, defie 1 i Z r := i=1 i 1 E Z u du 1 i, i+1 By iequaliy A.2 i he appedix, we have u Z r ω2 dr u Z rω 2 dr, adfor each ω Ωad u, u Z r ω Z rω 2 dr 0. Sice u Z2 r dr L1 Ω, i follow, by he Lebegue heorem for he codiioal expecaio, ha [ u 3.2 E Zr Z r 2 dr F 0 r.

10 WEAK DPP WIH E f -EXPECAIONS 2099 excep o a P -ull e N. Sice S i F -meaurable ad i idepeda of F, here exi a P -ull e icluded i he previou oe, uch ha for each ω N c, eig ω = ω,wehave [ u [ u E Zr Z r 2 dr F ω =E Zr ω, Z r ω, 2 dr [ u u = E Zr ω, dwr Z r ω, dwr. he ecod equaliy follow by he claical iomery propery. Now, for each quare iegrable marigale M, M 2 M i a marigale. Hece, for each ω N c,where N i a P -ull e icluded i he previou oe, eig ω = ω,wehave 3.4 [ u E Zr Z r 2 dr F ω [ u u 2 = E Zr dw r Z r dw r F ω [ u u 2 = E Zr dw r ω, Z r dw r ω,. For each N,iceZ i a imple proce, i aifie equaliy 3.1 everywhere, ha i, u Z r dw r ω, = u Z r ω, dwr. By he covergece propery 3.2, equaliie 3.3 ad 3.4, ad he uiquee propery of he limi i L 2,wederiveha for each ω N c, eig ω = ω, equaliy 3.1 hold. he proof i hu complee. Uig he above lemma, we ow how ha for each, for almo every ω Ω, eig ω = ω, he proce Y α,,x ω, coicide wih he oluio of he refleced BSDE o Ω [,, aociaed wih driver f α ω,,,η ω wih obacle hr, X α ω,,,x α ω,,x ω r ad filraio F, ad drive by W ad Ñ. o implify oaio, will be replaced by i he followig. I paricular Y α,,x ω, will be imply deoed by Y α,,x ω,.. heorem 7 pliig properie for he forward-backward yem. Le [0,, α A,ad [,. here exi a P -ull e N which deped o ad uch ha for each ω N c, eig ω = ω, he followig properie hold: here exi a uique oluio Xr α ω,,,η ω r i S 2 of he followig SDE: 3.5 X α ω,,,η ω r = η ω+ + + r r r bxv α ω,,,η ω,α v ω, dv σx α ω,,,η ω v E,α v ω, dw v βx α ω,,,η ω v,α v ω,,eñ dv, de, where η ω :=X α ω,,x ω. We alo have Xr α,,x ω,. =Xr α ω,,,η ω, r P-a.. here exi a uique oluio Yr α ω,,,η ω,zr α ω,,,η ω,kr α ω,,,η ω i S 2 H 2 H 2,ν of he refleced BSDE o Ω [, drive by W ad Ñ r

11 2100 R. DUMIRESCU, M.-C. QUENEZ, AND A. SULEM ad aociaed wih filraio F, driver f α ω,.,,η ω, ad obacle hr, Xr α ω,,,η ω. We have he followig: Yr α,,x ω,. =Yr α ω,.,,η ω, r, P-a.., ω,. =Z α ω,,,η ω Z α,,x r Y α,,x r ad K α,,x r ω,. = Kr α ω,,,η ω, r, dp dr-a.., ω,. =Y α ω,,,η ω = u α ω,, η ω, P-a.. Proof. Recall ha by Lemma 4, here exi a P -ull e N uch ha for each ω N c, he proce α ω, :=α r ω, r belog o A. Le u how he fir aerio. o implify he expoiio, we uppoe ha here i o Poio radom meaure. here exi a P -ull e, ill deoed by N, icluded i he above oe uch ha for each ω N c, eig ω = ω, Xr α,,x ω,. =η ω+ r r bxv α,,x ω,.,α v ω,.dv + σxv α,,x,α v dw v ω,., o [,, P -a.. Now, by he fir equaliy i Lemma 5, here exi a P -ull e N uch ha for each ω N c, eig ω = ω,wehave r σx α,,x v,α v dw v ω,. = r σx α ω,,,η ω v,α v ω, dw v, P-a.., which implie ha he proce Xr α,,x ω, r [, i a oluio of SDE 3.5, ad he, by uiquee of he oluio of hi SDE, we have Xr α,,x ω,. =Xr α ω,.,,η ω, r, P-a.. Le u how he ecod aerio. Fir, oe ha ice he filraio F i he compleed filraio of he aural filraio of W ad Ñ wih repec o he iiial σ-algebra BΩ, we have a marigale repreeaio heorem for F - marigale wih repec o W ad Ñ. Hece, here exi a uique oluio Yr α ω,,,η ω,zr α ω,,,η ω,kr α ω,,,η ω r i S 2 H 2 H 2,ν of he refleced BSDE o Ω [, drivebyw ad Ñ ad aociaed wih filraio F ad wih obacle hr, Xr α ω,,,η ω. Equaliie 3.6 he follow from imilar argume a above ogeher wih he uiquee of he oluio of a Lipchiz RBSDE. Equaliy 3.7 i obaied by akig r = i equaliy 3.6 ad by uig he defiiio of u α ω,.. Remark 8. Iheaboveproof,wehavereaedheP-ull e iue carefully. We re ha all he filraio are compleed wih repec o BΩ ad P. he uderlyig probabiliy pace i hu alway he compleio of he iiial probabiliy pace Ω, BΩ,P ha i, Ω, F,P. Noe ha he P -ull e remai alway he ame, which i paricularly impora for ochaic iegral ee Lemma 5 ad alo for BSDE becaue he oluio of a BSDE i uique up o a P -ull e. Moreover, i he proof of Lemma 5, he choice of he equece of ep fucio approximaig he proce Z i appropriae o hadle he iue of P -ull e. Noe ha heorem 7 applied o he impler cae whe α A eure ha he oluio of 2.8 wih replaced by coicide o [, Ω wih he oluio i S 2 H 2 H 2 ν of he refleced BSDE imilar o 2.8 bu drive by W ad Ñ iead of W ad Ñ, ad aociaed wih F.

12 WEAK DPP WIH E f -EXPECAIONS Meaurabiliy properie ad ε-opimal corol. We eed o how a meaurabiliy propery of he fucio u α, x wih repec o corol α ad iiial codiio x. o hi purpoe, we fir provide a prelimiary reul, which will allow u o hadle he olieariy of he expecaio. Propoiio 9. Le Ω, F,P be a probabiliy pace. For each q 0, wedeoe by L q he e L q Ω, F,P. Suppoe ha he Hilber pace L 2 equipped wih he uual calar produc i eparable. Le g : R R be a Borelia fucio uch ha gx C1 + x p for each real x wih p 0. hemapϕ g defied by ϕ g : L 2p L 2 L 2 ; ξ g ξ =gξ i he B L 2p L 2 /BL 2 -meaurable, where BL 2 i he Borelia σ-algebra o L 2, ad B L 2p L 2 i he σ-algebra iduced by BL 2 o L 2p L 2. he proof of hi propoiio i popoed uil he appedix. Uig hi reul, we ow how he followig meaurabiliy propery. heorem 10. Le [0,. he map α, x u α, x; A R R, i B A BR/BR-meaurable, where B A deoe he σ-algebra iduced by BH2 o A. Proof. Recall ha u α, x =Y α,,x, [gx α,,x i alo deoed by Y α,,x, [ h., X. α,,x. Le x 1,x 2 R, adα 1,α 2 A. By claical eimae o diffuio procee ad he aumpio made o he coefficie, we ge [ 3.8 E up X α1,,x 1 r X α2,,x 2 r 2 C α 1 α 2 2 H + 2 x1 x 2 2. r We iroduce he map Φ : A R S2 L2 S2 ;α, x, ζ,ξ Y α,,x, [η,ξ, where Y α,,x, [ζ,ξ deoe here he oluio a ime of he refleced BSDE aociaed wih driver f α,,x := fα r,r,xr α,,x,.1 r, obacle η <, ad ermial codiio ξ. By he eimae o RBSDE ee he appedix i [14, uig he Lipchiz propery of f w.r.. x, α ad eimae 3.8, for all x 1,x 2 R, α 1,α 2 A, η1,η2 S 2 ad ξ 1,ξ 2 L 2,wehave Y α1,,x 1, [η 1,ξ 1 Y α2,,x 2, [η 2,ξ 2 2 C α 1 α 2 2 H 2 + x1 x η 1 η 2 2 S 2 + ξ1 ξ 2 2 L 2. he map Φ i hu Lipchiz-coiuou wih repec o he orm. 2 H S L. 2 Recall ha by aumpio, h, x C1 + x p, ad ha h i Lipchiz coiuou wih repec o x uiformly i. Oe ca derive ha he map S 2p S 2 S2, η h., η i Lipchiz-coiuou for he orm. 2 S ad hu Borelia, S 2 2 beig equipped wih he Borelia σ-algebra BS 2 ad i ubpace S2p S 2 wih he σ-algebra iduced by BS 2. Moreover, by Lemma 25, he Hilber pace L 2 i eparable. We ca hu apply Propoiio 9 ad ge ha he map L 2p L 2 L 2, ξ gξ i Borelia. We hu derive ha he map α, x α, x, h., X. α,,x,gx α,,x defied o A R ad valued i A R S2 L2 i B A BR/B A BR BS2 BL2 - meaurable. By compoiio, i follow ha he map α, x Y α,,x, [h., X. α,,x, = u α, x i meaurable. gx α,,x

13 2102 R. DUMIRESCU, M.-C. QUENEZ, AND A. SULEM For each, wih, we iroduce he e A of rericio o [, ofhe corol i A. hey ca alo be ideified o he corol α i A which are equal o 0o[,. Le η L 2 F. Sice η i F -meaurable, up o a P -ull e, i ca be wrie a a meaurable map, ill deoed by η, of he pa rajecory ω. See he argume ued i he proof of Lemma 25 for deail. For each ω Ω, by uig he defiiio of he fucio u, wehave 3.9 u, η ω = up u α, η ω. α A By heorem 10 ogeher wih a meaurable elecio heorem, we how he exiece of early opimal corol for 3.9 aifyig ome pecific meaurabiliy properie. heorem 11 exiece of ε-opimal corol. Le [0,, [, [, ad η L 2 F. Le ε>0. here exi αε A uch ha, for almo every ω Ω, α ε ω, i ε-opimal for Problem 3.9, i he ee ha u, η ω u αε ω,, η ω + ε. Proof. Wihou lo of geeraliy, we may aume ha = 0. We iroduce he pace Ω:={ω r 0 r ; ω Ω}, equipped wih i Borelia σ-algebra deoed by B Ω, ad he probabiliy meaure P, which correpod o he image of P by S :Ω Ω; ω ω r r. he Hilber pace H 2 of quare-iegrable predicable procee o Ω [,, equipped wih he orm H 2, i eparable ee Lemma 25. Moreover, A i a cloed ube of H2. Alo, he pace Ω of pah RCLL before i Polih for he Skorohod meric. Now, a ee above, ice η i F -meaurable, up o a P -ull e, we ca uppoe ha i i of he form η S,whereη i B Ω- meaurable. Moreover, by heorem 10, he map ω, α u α, η ω i B Ω BA -meaurable wih repec o x, α. We ca hu apply [6, Propoiio 7.50 o he problem 3.9. Hece, here exi a map α ε : Ω A ; ω α ε ω,, which i uiverally meaurable, ha i U Ω/BA -meaurable, ad uch ha u, η ω u αε ω,, η ω + ε ω Ω. Here, U Ω deoe he uiveral σ-algebra o Ω. Le u ow apply Lemma 26 o X = Ω, o E = H 2, ad o probabiliy Q = P. By defiiio of U Ω ee, e.g., [6, we have U Ω B Q Ω, where B Q Ω deoe he compleio of B Ω wih repec o Q. Hece, here exi a map ˆα ε : Ω A ; ω ˆα ε ω, whichi Borelia, ha i B Ω/BA -meaurable, ad uch ha ˆα ε ω, =α ε ω, for P -almo every ω Ω. Sice H 2 i a eparable Hilber pace, for each ω, wehaveˆα ε u ω, ω = i βi,ε ωe i uω dp ω du-a.., where β i,ε ω =< ˆα ε ω,,e i > H 2 ad {e i,i N} i a couable orhoormal bai of H 2.Noehaβ i,ε i Borelia, ha i, B Ω/BR-meaurable. Le ᾱ ε : Ω A ; ω ᾱε ω, = i βi,ε ωe i. I i Borelia, ha i, B Ω/BA - meaurable. We ow defie a proce α ε o [0, Ωbyα ε r ω := i βi,ε S ωe i ω. I remai o prove ha i i P-meaurable. Noe ha β i,ε S i F -meaurable by compoiio. Sice he proce e i u u i P -meaurable, he proce β i,ε S e i u

14 WEAK DPP WIH E f -EXPECAIONS 2103 i P-meaurable. Ideed, if we ake e i of he form e i u = H1 r, u wih r ad H a radom variable Fr -meaurable, he he radom variable β i,ε S H i F r - meaurable ad hece he proce β i,ε S H1 r, i P-meaurable. he proce α ε i hu P-meaurable. Noe alo ha α ε ω, ω = i βi,ε ωe i ω, ω. Now, we have e i ω, ω = e i ω becaue e i ω deped o ω oly hrough ω. Hece, α ε ω, ω = ᾱ ε ω, ω, which complee he proof A Faou lemma for refleced BSDE. We eablih a Faou lemma for refleced BSDE, where he limi ivolve boh ermial codiio ad ermial ime. hi reul will be ued o prove a uper- rep., ub- opimaliy priciple ivolvig he l..c. rep., u..c. evelope of he value fucio u ee heorem 17. We fir iroduce ome oaio. A fucio f i aid o be a Lipchiz driver if f : [0, Ω R 2 L 2 ν R ω,, y, z, k fω,, y, z, k i P BR 2 BL 2 ν -meaurable, uiformly Lipchiz wih repec o y, z, k, ad uch ha f., 0, 0, 0 H 2. A Lipchiz driver f i aid o aify Aumpio 12 if he followig hold. Aumpio 12. Aume ha dp d-a. for each y, z, k 1,k 2 R 2 L 2 ν 2, f, y, z, k 1 f, y, z, k 2 γ y,z,k1,k2,k 1 k 2 ν wih γ :[0, Ω R 2 L 2 ν 2 L 2 ν ;ω,, y, z, k 1,k 2 γ y,z,k1,k2 ω,., uppoed o be P BR 2 BL 2 ν 2 -meaurable, uiformly bouded i L 2 ν, ad aifyig dp ω d dνe-a.., for each y, z, k 1,k 2 R 2 L 2 ν 2, he iequaliy γ y,z,k1,k2 ω, e 1. hi aumpio eure he compario heorem for BSDE wih jump ee [21, heorem 4.2. Le η be a give RCLL obacle proce i S 2 ad le f be a give Lipchiz driver. I he followig, we will coider he cae whe he ermial ime i a oppig ime ad he ermial codiio i a radom variable ξ i L 2 F. I hi cae, he oluio, deoed Y., ξ,z., ξ,k., ξ, of he refleced BSDE aociaed wih ermial oppig ime, driverf,obacleη <, ad ermial codiio ξ i defied a he uique oluio i S 2 H 2 H 2 ν of he refleced BSDE wih ermial ime, driver f, y, z, k1 { }, ermial codiio ξ, ad obacle η 1 < + ξ1.noeha Y, ξ =ξ,z, ξ =0,k, ξ =0for. We fir prove a coiuiy propery for refleced BSDE where he limi ivolve boh ermial codiio ad ermial ime. Propoiio 13 a coiuiy propery for refleced BSDE. Le > 0. Le η be a RCLL proce i S 2. Le f be a give Lipchiz driver. Le N be a o icreaig equece of oppig ime i, covergig a.. o a ed o. Leξ N be a equece of radom variable uch ha E[up ξ 2 < +, ad for each, ξ i F -meaurable. Suppoe ha ξ coverge a.. o a F -meaurable radom variable ξ a ed o. Suppoe ha 3.10 η ξ a.. Le Y., ξ ; Y., ξ be he oluio of he refleced BSDE aociaed wih driver f, obacle η < rep.. η <, ermial ime rep.,, ad ermial codiio ξ rep., ξ. We have Y 0, ξ = lim Y 0, + ξ a.. Whe for each, = a.., he reul ill hold wihou aumpio 3.10.

15 2104 R. DUMIRESCU, M.-C. QUENEZ, AND A. SULEM By imilar argume a i he Browia cae ee, e.g., [16, oe ca prove he followig eimae o refleced BSDE, which will be ued i he proof of he above propoiio. Lemma 14. Le ξ 1,ξ 2 L 2 F ad η 1, η2 S2. Le f 1,f 2 be Lipchiz driver wih Lipchiz coa C>0. Fori =1, 2, ley i,z i,k i,a i be he oluio of he refleced BSDE wih driver f i,ermialime, obacle η i,adermial codiio ξ i. For [0,, ley := Y 1 Y 2, η := η 1 η2, ξ := ξ1 ξ 2 ad f :=f 1, Y 2,Z 2,k 2 f, Y 2,Z 2,k. 2 he, we have [ 3.11 Y 2 S E[ξ K 2 +E f 2 d + φ 2 up η 0 < 0 L 2, where he coa K i uiveral, ha i, deped oly o he Lipchiz coa C ad, ad where he coa φ deped oly o C,, η i S 2, ξ i L 2, ad f i, 0, 0, 0 H 2, i =1, 2. Proof of Propoiio 13. Le N. We apply 3.11 wih f 1 = f1, f 2 = f1, ξ 1 = ξ, ξ 2 = ξ, η 1 = η 1 < + ξ 1 <,adη 2 = η 1 < + η 1 < + ξ1 <. We have Y 1 = Y., ξ a.. Moreover, ice by aumpio η ξ a.., we have Y 2 = Y., ξ a.. NoehaY 2,Z2,k2 =ξ,0, 0 a.. o { }. Wehu obai [ Y 0, ξ Y 0, ξ 2 K E[ξ ξ 2 +E f 2, ξ, 0, 0d φ up η η < L 2, where he coa K deped oly o he Lipchiz coa C of f ad he ermial ime, ad where he coa φ deped oly o C,, η S 2,up ξ L 2,ad f, 0, 0, 0 H 2. Sice he obacle η i righ-coiuou ad a.., we have lim + up η η L 2 =0. he righ member of 3.12 hu ed o 0 a ed o +. he reul follow. Remark 15. Compared wih he cae of orefleced BSDE ee [21, Propoiio A.6, here i a exra difficuly due o he preece of he obacle ad he variaio of he ermial ime. he addiioal aumpio 3.10 o he obacle i here required o obai he reul. Uig Propoiio 13, we derive a Faou lemma i he refleced cae, where he limi ivolve boh ermial codiio ad ermial ime. Propoiio 16 a Faou lemma for refleced BSDE. Le > 0. Le η be a RCLL proce i S 2. Le f be a Lipchiz driver aifyig Aumpio 12. Le N be a oicreaig equece of oppig ime i, covergig a.. o a ed o. Le ξ N be a equece of radom variable uch ha E[up ξ 2 < +, adforeach, ξ i F -meaurable. Le Y., ξ ; Y., lim if + ξ ad Y., lim up + ξ be he oluio of he refleced BSDE aociaed wih driver f, obacle η < rep., η <, ermial ime rep.,, ad ermial codiio ξ rep., lim if + ξ ad lim up + ξ.

16 3.13 Suppoe ha WEAK DPP WIH E f -EXPECAIONS 2105 lim if + ξ η he Y 0, lim if + ξ rep., lim if + Y 0, ξ rep., Y 0, lim up ξ η + lim up + a.. ξ lim up Y 0, ξ. + Whe for each, = a.., he reul ill hold wihou aumpio Proof. We pree oly he proof of he fir iequaliy, ice he ecod oe i obaied by imilar argume. For all, we have by he moooiciy of refleced BSDE wih repec o ermial codiio, Y 0, if p ξ p Y 0, ξ. We derive ha lim if Y 0, + ξ lim if Y 0, + if p ξp = Y 0, lim if + ξ, where he la equaliy follow from aumpio 3.13 ogeher wih Propoiio A weak DPP. We will ow provide a weak DPP ha i boh a weak ub- ad uper-opimaliy priciple of dyamic programmig, ivolvig, repecively, he u..c. evelope u ad he l..c. evelope u of he value fucio u, defied by u, x := lim up u,x ; u, x := lim if,x,x,x,x u,x, x [0, R. We ow defie he map ū ad ū for each, x [0, R by ū, x :=u, x1 < + gx1 = ; ū, x :=u, x1 < + gx1 =. Noe ha he fucio ū ad ū are Borelia. We have ū u ū ad ū,.= u,.=ū,. =g.. Noe ha ū rep., ū i o ecearily upper rep., lower emicoiuou o [0, R, ice he ermial reward g i oly Borelia. o prove he weak DPP, we will ue he pliig properie heorem 7, he exiece of ε-opimal corol heorem 11, ad he Faou lemma for RBSDE Propoiio 16. heorem 17 a weak DPP. he value fucio u aifie he followig weak ubopimaliy priciple of dyamic programmig: for each [0, ad for each oppig ime, we have 3.14 u, x up up E α,,x [, τ hτ,x α,,x τ 1 τ< +ū, X α,,x 1 τ. α A τ Moreover, he followig weak uperopimaliy priciple of dyamic programmig hold: for each [0, ad for each oppig ime, we have 3.15 u, x up α A up τ E α,,x, τ [ hτ,x α,,x τ 1 τ< +ū, X α,,x 1 τ. Remark 18. he proof give below alo how ha hi weak DPP ill hold wih replaced by α i iequaliie 3.14 ad 3.15, give a family of oppig ime idexed by corol { α,α A }.

17 2106 R. DUMIRESCU, M.-C. QUENEZ, AND A. SULEM Noe ha o regulariy codiio i required o g o eure hi weak DPP, eve hi i o he cae i he lieraure eve for claical expecaio ee [8, [7, [4. Moreover, our DPP are roger ha hoe give i hee paper, where iequaliy 3.14 rep., 3.15 i eablihed wih u rep., u iead of ū rep., ū. Now, ū u ad ū u. Before givig he proof, we iroduce he followig oaio. For each ad each ξ i L 2 F, we deoe by Y α,,x., ξ,z α,,x., ξ,k α,,x., ξ he uique oluio i S 2 H 2 H 2 ν of he refleced BSDE wih driver f α,,x 1 { }, ermial ime,ermial codiio ξ, ad obacle hr, Xr α,,x 1 r< + ξ1 r. Proof. By eimae for refleced BSDE ee [14, Propoiio 5.1, he fucio u ha a mo polyomial growh a ifiiy. Hece, he radom variable ū, X α,,x ad ū, X α,,x are quare iegrable. Wihou lo of geeraliy, o implify oaio, we uppoe ha =0. We fir how he ecod aerio which i he mo difficul, or equivalely: [ 3.16 up Y α,0,x 0, ū, X α,0,x u0,x. α A Le. For each N, we defie 3.17 := 2 1 k=0 k 1 Ak + 1 =, where k := k+1 2 ad A k := { k 2 < k+1 2 }.Noeha ad. O { = } we have = for each. Wehugeū,X α,0,x =ū, X α,0,x for each o { = }. Moreover, o { <}, he lower emicoiuiy of ū o [0,[ R ogeher wih he righ coiuiy of he proce X α,0,x implie ha ū, X α,0,x lim if ū,x α,0,x + a.. Hece, by he compario heorem for refleced BSDE, we ge O { <}, wehave [ Y α,0,x 0, ū, X α,0,x Y α,0,x 0, [ lim if ū,x α,0,x + lim if ū,x α,0,x lim if h,x α,0,x = lim h,x α,0,x =h, X α,0,x a.. by he regulariy properie of h o [0,[ R. O { = }, = ad ū,x α,0,x =ū,x α,0,x =gx α,0,x = h,x α,0,x. Hece, we have lim if + ū,x α,0,x h, X α,0,x a.. Codiio 3.13 i hu aified wih ξ =ū,x α,0,x adξ = h, X α,0,x. We ca hu apply he Faou lemma for refleced BSDE Propoiio 16. We hu ge [ [ Y α,0,x 0, ū, X α,0,x Y α,0,x 0, lim if ū,x α,0,x [ lim if Y α,0,x 0, ū,x α,0,x..

18 WEAK DPP WIH E f -EXPECAIONS 2107 Le ε>0. Fix N. For each k<2 1, le A k be he e of he rericio o [ k,ofhecorolα i A. By heorem 11, here exi a P -ull e N which deped o ad ε uch ha for each k<2 1, here exi a ε-opimal corol corol α,ε,k i A 0 k = A k for he corol problem a ime k wih iiial codiio η = X α,0,x k ha i aifyig he iequaliy 3.19 u k,x α,0,x k k ω u α,ε,k k ω, k,x α,0,x k ω + ε for each ω N c. Uig he defiiio of he map u α,ε,k k ω, ogeher wih he pliig propery for refleced BSDE 3.7, we derive ha here exi a P -ull e N which coai he above oe uch ha for each ω N c ad for each k<2 1, we have u α,ε,k k ω, k,x α,0,x k k ω = Y α,ε,k k ω,, k,x α,0,x k, k k k ω = Y α,ε,k, k,x α,0,x k k, k ω. Here, Y α,ε,k, k,x α,0,x k., =Y f α,ε,k,k,x α,0,x k., [ hr, X α,ε,k, k,x α,0,x k r deoe he oluio of he refleced BSDE aociaed wih ermial ime,obacle hr, X α,ε,k, k,x α,0,x k r k r, ad driver f α,ε,k, k,x α,0,x k r, y, z, k :=fα,ε,k r,r,x α, k,x α,0,x k r,y,z,k. Se α,ε := k<2 1 α,ε,k 1 Ak + α 1 {= }. Sice for each k, A k F k,here exi a P -ull e N uch ha, o N c,foreachk<2 1, we have he followig equaliie: Y α,ε,k, k,x α,0,x k k, 1 Ak = Y f α k,,ε,k,k,x α,0,x k 1Ak = Y f α,ε,,x k, α,0,x 1 Ak [ h [ h r, X α,ε,k, k,x α,0,x k r 1 Ak r, X α,ε,,x α,0,x r 1 Ak, where, for a give driver f, Y f1a k deoe he oluio of he refleced BSDE aociaed wih f1 Ak.WehugeY α,ε,k, k,x α,0,x k k, 1 Ak = Y α,ε,,x α,0,x, 1 Ak o N c.uig iequaliie 3.19, we ge ū,x α,0,x = u k,x α,0,x k 1 Ak + gx α,,x 1 {= } 0 k<2 1 Y α,ε,,x α,0,x, + ε o N c. We e α,ε := α 1 < + α,ε 1. Noe ha α,ε A. Uig he compario heorem ogeher wih he eimae o refleced BSDE ee [14, we obai [ Y α,0,x 0, [ū,x α,0,x Y α,0,x 0, Y α,ε,,x α,0,x, + Kε = Y α,ε,0,x 0, + Kε, where he la equaliy follow from he flow propery. Sice Y α,ε,0,x 0, u0,x, uig 3.18, we ge Y α,0,x 0, [ū, X α,0,x Y α,0,x 0, [ū,x α,0,x u0,x+kε. akig he upremum o α Aad leig ε ed o 0, we obai iequaliy I remai o how he fir aerio. I i ufficie o how ha for each, [ 3.20 u0,x up Y α,0,x 0, ū, X α,0,x. α A

19 2108 R. DUMIRESCU, M.-C. QUENEZ, AND A. SULEM Le. Le α A. A above, we approximae by he equece of oppig ime N defied above. Le N. By applyig he flow propery for refleced BSDE, we ge Y α,0,x 0, = Y α,0,x 0, [Y α,,x α,0,x,. By imilar argume a i he proof of he uperopimaliy priciple bu wihou uig he exiece of ε-opimal corol, we derive ha Y α,,x α,0,x, ū,x α,0,x a.. By he compario heorem for refleced BSDE, i follow ha [ Y α,0,x 0, = Y α,0,x 0, Y α,,x α,0,x, Y α,0,x 0, [ū,x α,0,x. Uig he Faou lemma for refleced BSDE Propoiio 16, we ge Y α,0,x 0, lim up Y α,0,x 0, [ū,x α,0,x Y α,0,x 0, [ lim up ū,x α,0,x. Uig he upper emicoiuiy propery of ū o [0,[ R ad ū,x=gx, we obai [ Y α,0,x 0, Y α,0,x 0, lim up ū,x α,0,x Y α,0,x 0, [ū, X α,0,x. Sice α Ai arbirary, we ge iequaliy 3.20, which complee he proof. 4. Noliear HJB variaioal iequaliie Some exeio of compario heorem for BSDE ad refleced BSDE. We provide wo reul which will be ued o prove ha he value fucio u, defied by 2.6, i a weak vicoiy oluio of ome oliear HJB variaioal iequaliie ee heorem 22. We fir how a ligh exeio of he compario heorem for BSDE give i [21, from which we derive a compario reul bewee a BSDE ad a refleced BSDE. Lemma 19. Le 0 [0, ad le 0. Le ξ 1 ad ξ 2 L 2 F. Le f 1 be a driver. Le f 2 be a Lipchiz driver wih Lipchiz coa C>0, aifyig Aumpio 12. For i =1, 2, lex i,πi,li be a oluio i S2 H 2 H 2 ν of he BSDE aociaed wih driver f i,ermialime, ad ermial codiio ξ i.suppoe ha f 1, X 1,π 1,l 1 f 2, X 1,π 1,l 1 0, d dp a.., ad ξ 1 ξ 2 + ε a.., where ε i a real coa. he, for each [ 0,, we have X 1 X 2 + εe C a.. Proof. From iequaliy 4.22 i he proof of he compario heorem i [21, we derive ha X 1 0 X 2 0 e C E [H 0, ε F 0 a.., where C i he Lipchiz coa of f 2, ad H 0, [0, i he oegaive marigale aifyig dh 0, = H [β 0, dw + E γuñd, du wih H 0, 0 =1,β beig a predicable proce bouded by C. he reul follow. Propoiio 20 a compario reul bewee a BSDE ad a refleced BSDE. Le 0 [0, ad le 0.Leξ 1 L 2 F ad le f 1 be a driver. Le X 1,π 1,l 1 be a oluio of he BSDE aociaed wih f 1,ermialime, ad ermial codiio ξ 1.Leξ 2 S2 ad le f 2 be a Lipchiz driver wih Lipchiz coa C>0which aifie Aumpio 12. Le Y 2 be he oluio of he refleced BSDE aociaed

20 WEAK DPP WIH E f -EXPECAIONS 2109 wih f 2,ermialime, ad obacle ξ 2. Suppoe ha 4.1 f 1, X 1,π 1,l 1 f 2, X 1,π 1,l 1, 0, d dp -a.. ad X 1 ξ 2 + ε, 0 a.. he, we have X 1 Y 2 + εe C, 0 a.. Proof. Le [ 0,. By he characerizaio of he oluio of he RBSDE a he value fucio of a opimal oppig problem ee heorem 3.2 i [21, Y 2 = e up τ [, E f 2,τ ξτ 2. By Lemma 19, for each τ [,, X 1 E f 2,τ ξτ 2 +e C ε. akig he upremum over τ [,, he reul follow Lik bewee he mixed corol problem ad HJB equaio. We iroduce he followig HJB variaioal iequaliy HJBVI: 4.2 miu, x h, x, if α A u, x Lα u, x f u,x=gx,x R, α,, x, u, x, σ u, x,b α u, x =0,, x [0, R, where L α := A α + K α, ad for φ C 2 R, A α φx := 1 2 σ2 x, α 2 φ x+bx, α φ 2 x adbα φx :=φx + βx, α, φx, K α φx := φ E φx + βx, α, e φx xβx, α, eνde. Defiiio 21. Afuciou i aid o be a vicoiy uboluio of 4.2 if i i u..c. o [0, R, ad if for ay poi 0,x 0 [0,[ R ad for ay φ C 1,2 [0, R uch ha φ 0,x 0 =u 0,x 0 ad φ u aai i miimum a 0,x 0, we have 4.3 miu 0,x 0 h 0,x 0, φ if α A f 0,x 0 L α φ 0,x 0 α, 0,x 0,u 0,x 0, σ φ 0,x 0,B α φ 0,x 0 0. I oher word, if u 0,x 0 >h 0,x 0,he if α A φ 0,x 0 L α φ 0,x 0 f α, 0,x 0,u 0,x 0, σ φ 0,x 0,B α φ 0,x 0 0. A fucio u i aid o be a vicoiy uperoluio of 4.2 if i i l..c. o [0, R, ad if for ay poi 0,x 0 [0,[ R ad ay φ C 1,2 [0, R uch

21 2110 R. DUMIRESCU, M.-C. QUENEZ, AND A. SULEM ha φ 0,x 0 =u 0,x 0 ad φ u aai i maximum a 0,x 0, we have miu 0,x 0 h 0,x 0, if α A φ 0,x 0 L α φ 0,x 0 f α, 0,x 0,u 0,x 0, σ φ I oher word, we have boh u 0,x 0 h 0,x 0 ad 4.4 if α A φ 0,x 0 L α φ 0,x 0 f α, 0,x 0,u 0,x 0, σ φ 0,x 0,B α φ 0,x ,x 0,B α φ 0,x 0 0. Uig he weak DPP give i heorem 17 ad Propoiio 20, we ow prove ha he value fucio of our problem i a weak vicoiy oluio of he above HJBVI. heorem 22. he value fucio u, defied by 2.6, i a weak vicoiy oluio of he HJBVI 4.2, i he ee ha i u..c. evelope u i a vicoiy uboluio of 4.2 ad i l..c. evelope u i a vicoiy uperoluio of 4.2 wih ermial codiio u,x=gx. Proof. We fir prove ha u i a uboluio of 4.2. Le 0,x 0 [0,[ R ad φ C 1,2 [0, R be uch ha φ 0,x 0 =u 0,x 0 adφ, x u, x for all, x [0, R. Wihou lo of geeraliy, we ca uppoe ha he miimum of u φ aaied a 0,x 0 i ric. Suppoe for coradicio ha u 0,x 0 >h 0,x 0 ad ha if α A φ 0,x 0 L α φ 0,x 0 f α, 0,x 0,φ 0,x 0, σ φ 0,x 0,B α φ 0,x 0 > 0. By uiform coiuiy of K α φ ad B α φ :[0, R L 2 ν wih repec o α, weca uppoe ha here exi ɛ>0,η ɛ > 0 uch ha for all, x uch ha η ɛ < ad x x 0 η ɛ,wehaveφ, x h, x+ɛ ad 4.5 φ, x Lα φ, x f α,, x, φ, x, σ φ, x,b α φ, x ɛ α A. We deoe by B ηε 0,x 0 he ball of radiu η ε ad ceer 0,x 0. By defiiio of u, here exi a equece,x i B ηε 0,x 0, uch ha,x,u,x 0,x 0,u 0,x 0. Fix N. Leα be a arbirary corol of A ad X α,,x he aociaed ae proce. We defie he oppig ime α, a α, := 0 + η ɛ if{, X α,,x x 0 η ɛ }.

22 WEAK DPP WIH E f -EXPECAIONS 2111 Le ψ α, x := φ, x +Lα φ, x. Applyig Iô lemma o φ, X α,,x, we derive ha φ, X α,,x, σ φ, X α,,x,b α φ, X α,,x ; [, α, i he oluio of he BSDE aociaed wih he driver proce ψ α, X α,,x, ermial ime α,, ad ermial value φ α,,x α,,x. By 4.5 ad by defiiio α, of α,,wege 4.6 ψ α, X α,,x f α,,x α,,x,φ, X α,,x, σ φ, X α,,x,bφ, X α,,x + ɛ for each [, α,. hi iequaliy give a relaio bewee he driver ψ α, X α,,x adfα, of wo BSDE. Now, ice he miimum 0,x 0 i ric, here exi γ ɛ uch ha 4.7 u, x φ, x γ ɛ o [0, R \ B ηɛ 0,x 0. We have φ α,, X α,,x α, =φ, Xα,,x 1 < α, + φ α,,x α,,x α, 1 α,, a.. o implify oaio, e δ ε := miɛ, γ ɛ. Uig 4.7 ogeher wih he defiiio of α,,wege φ, X α,,x h, X α,,x +δ ε 1 < α, +u α,,x α,,x α, +δ ε1 = α,, α, a.. hi, ogeher wih iequaliy 4.6 o he driver ad he above compario heorem bewee a BSDE ad a refleced BSDE ee Propoiio 20 lead o φ,x Y α,,x, α, [h, X α,,x 1 < α, + u α,,x α,,x 1 α, = α+δ ε K, where K i a poiive coa which oly deped o ad he Lipchiz coa of f. Now, recall ha,x,u,x 0,x 0,u 0,x 0 ad φ i coiuou wih φ 0,x 0 = u 0,x 0. We ca hu aume ha i ufficiely large o ha φ,x u,x δ ε K/2. Hece, u,x Y α,,x, α, [h, Xα,,x 1 < α, + u α,,x α,,x 1 α, = α+δ εk/2. A hi iequaliy hold for all α A ad ice u ū, we ge a coradicio of he ubopimaliy priciple of DPP 3.14 ee alo Remark 18. We ow prove ha u i a vicoiy uperoluio of 4.2. Le 0,x 0 [0,[ R ad φ C 1,2 [0, R be uch ha φ 0,x 0 =u 0,x 0 adφ, x u, x for all, x [0, R. Wihou lo of geeraliy, we ca uppoe ha he maximum i ric i 0,x 0. Sice he oluio Y α,0,x0 ay above he obacle, for each α A,wehaveu 0,x 0 h 0,x 0. Our aim i o how ha iequaliy 4.4 hold.

23 2112 R. DUMIRESCU, M.-C. QUENEZ, AND A. SULEM Suppoe for coradicio ha hi iequaliy doe o hold. By coiuiy, we ca uppoe ha here exi α A, ɛ>0, ad η ɛ > 0uch ha for all, x wih η ɛ < ad x x 0 η ɛ,wehave 4.8 φ, x Lα φ, x f α,, x, φ, x, σ φ, x,b α φ, x ɛ. We deoe by B ηε 0,x 0 he ball of radiu η ε ad ceer 0,x 0. Le,x be a equece i B ηε 0,x 0 uch ha,x,u,x 0,x 0,u 0,x 0. We iroduce he ae proce X α,,x aociaed wih he above coa corol α ad defie he oppig ime a := 0 + η ɛ if{, X α,,x x 0 η ɛ }. By Iô formula, he proce φ, X α,,x, σ φ, Xα,,x,B α φ, X α,,x ; [, i he oluio of he BSDE aociaed wih ermial ime, ermial value φ,x α,,x, ad driver ψ α, X α,,x. he defiiio of he oppig ime ad iequaliy 4.8 lead o 4.9 ψ α, X α,,x f α,, X α,,x,φ, X α,,x, σ φ, X α,,x,b α φ, X α,,x for d dp -a.. Now, ice he maximum 0,x 0 i ric, here exi γ ɛ which deped o η ɛ uch ha u, x φ, x +γ ɛ o [0, R \ B ηɛ 0,x 0 which implie φ,x α,,x u,x α,,x γ ɛ. Hece, uig iequaliy 4.9 o he driver, ogeher wih he compario heorem for BSDE, we derive ha φ,x =E ψα, [φ,x α,,x E α,,x, Eα,,x, [u,x α,,x γ ɛk, [u,x α,,x γ ɛ where he ecod iequaliy follow from a exeio of he compario heorem Lemma 19. We ca aume ha i ufficie large o ha φ,x u,x δ ε K/2. We hu ge 4.10 u,x E α,,x, [u,x α,,x γ ɛk/2. Sice u aifie he uperopimaliy DPP heorem 17, we have u,x E α,,x, [ū,x α,,x. Sice ū u, hi iequaliy wih 4.10 lead o a coradicio. Remark 23. Whe g i oly Borelia, he weak oluio of he HJB equaio 4.2 i geerally o uique, eve i he deermiiic cae a reed i [2, [1, [3. Noe ha whe g i l..c., he value fucio u of our problem ca be how o be he miimal l..c. vicoiy uperoluio of he HJB equaio 4.2, wih ermial value greaer ha g by uig imilar argume a i he proof of [15, heorem 6.5.

24 WEAK DPP WIH E f -EXPECAIONS 2113 Noe alo ha he paper [2 ee alo [1 provide a characerizaio of he u..c. evelope u of he value fucio of he deermiiic corol problem u, x := up α A gx,x,α, which correpod o our problem wih σ = f = 0 ad o oppig ime corol. More preciely, whe g i u..c., he map u i characerized a he uique u..c. vicoiy oluio of he HJB equaio i.e., aifie u,x=gx ad he aalogou of 4.3 bu wih a equaliy. he proof i baed o PDE argume ad deermiiic corol heory. A iereig furher developme of our paper ad of [8 would be o udy aalogou properie i he ochaic cae. Appedix A. Appedix. We give here ome meaurabiliy reul which are ued i ecio 3.2. We ar by he proof of Propoiio 9. o hi purpoe, we fir provide he followig lemma. Lemma 24. Le Ω, F, P be a probabiliy pace. Suppoe ha he Hilber pace L 2 := L 2 Ω, F,P equipped wih he uual calar produc i eparable. Le F L 2. Coider a equece of fucio g N uch ha for each, g : R R i Borelia wih g x C1 + x p. Suppoe ha equece g N coverge poiwie. Le g be he limi, defied for each x R by gx := lim + g x. Suppoe alo ha for each N, hemapψ g,f deoed alo by ψ g defied by ψ g : L 2p L 2 R; ξ E[g ξ F i Borelia, L 2p L 2 beig equipped wih he σ-algebra iduced by BL 2. he, he map ψ g deoed alo by ψ g,f defied by A.1 ψ g : L 2p L 2 R; ξ E[gξ F i Borelia. Proof. By he Lebegue heorem, for each ξ L 2p L 2, we have ψ g ξ = E[gξ F = lim + E[g ξ F = lim + ψ g ξ. Sice he poiwie limi of a equece of R-valued meaurable map i meaurable, we derive ha he map ψ g i Borelia. Proof of Propoiio 9. Sice by aumpio he Hilber pace L 2 i eparable, here exi a couable orhoormal bai {e i,i N} of L 2. For each ξ L 2p L 2, we have ϕ g ξ =gξ = i ψg,i ξ e i i L 2,whereψ g,i ξ :=E[gξ e i foreachi N. Hece, i order o how he meaurabiliy of he map ψ, i i ufficie o how he meaurabiliy of he map ψ g,f, F L 2. For hi purpoe, we iroduce he e H of bouded Borelia fucio g : R R uch ha for each F L 2,hemapψ g,f i Borelia. Noe ha H i a vecor pace. Suppoe we have how ha for all real umber a, b wih a<b, 1 a,b[ H. he, by Lemma 24 ogeher wih a moooe cla heorem, we derive ha H i equal o he whole e of bouded Borelia fucio. Whe g i o bouded, he reul follow by approximaig g by a equece of bouded Borelia fucio, ad by uig Lemma 24. I remai o how ha for all a, b R wih a<b,wehave1 a,b[ H. Sice 1 a,b[ i l..c., i follow ha here exi a odecreaig equece g N of Lipchiz coiuou fucio akig heir value i [0, 1 uch ha for each x R, 1 a,b[ x := lim + g x. For each, iceg i Lipchiz coiuou, by uig he Cauchy Schwarz iequaliy, oe ca derive ha he map ψ g : L 2 R; ξ E[g ξ F i Lipchiz coiuou for he orm L 2, ad hece Borelia. he reul he follow from Lemma 24.

Présentée pour obtenir le grade de. Docteur en Science **************TITRE**************

Présentée pour obtenir le grade de. Docteur en Science **************TITRE************** UNIVRSITÉ MOHAMD KHIDR FACULTÉ DS SCINCS XACTS T SCINC D LA NATUR T D LA VI BISKRA *************************** THÈS Préeée pour obeir le grade de Doceur e Sciece Spécialié: Probabilié **************TITR**************

More information

Minimal Supersolutions of Convex BSDEs

Minimal Supersolutions of Convex BSDEs Miimal Superoluio of Covex BSDE Samuel Drapeau a,1,, Gregor Heye a,2,, Michael Kupper a,3, Jauary 22, 2013 ABSTRACT We udy he oliear operaor of mappig he ermial value ξ o he correpodig miimal uperoluio

More information

Conditional distributions, exchangeable particle systems, and stochastic partial differential equations

Conditional distributions, exchangeable particle systems, and stochastic partial differential equations Codiioal diribuio, exchageable paricle yem, ad ochaic parial differeial equaio Da Cria, Thoma G. Kurz, Yoojug Lee 23 July 2 Abrac Sochaic parial differeial equaio whoe oluio are probabiliy-meaurevalued

More information

arxiv:math/ v1 [math.fa] 1 Feb 1994

arxiv:math/ v1 [math.fa] 1 Feb 1994 arxiv:mah/944v [mah.fa] Feb 994 ON THE EMBEDDING OF -CONCAVE ORLICZ SPACES INTO L Care Schü Abrac. I [K S ] i wa how ha Ave ( i a π(i) ) π i equivale o a Orlicz orm whoe Orlicz fucio i -cocave. Here we

More information

Hadamard matrices from the Multiplication Table of the Finite Fields

Hadamard matrices from the Multiplication Table of the Finite Fields adamard marice from he Muliplicaio Table of he Fiie Field 신민호 송홍엽 노종선 * Iroducio adamard mari biary m-equece New Corucio Coe Theorem. Corucio wih caoical bai Theorem. Corucio wih ay bai Remark adamard

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

Weak Solutions of Mean Field Game Master Equations

Weak Solutions of Mean Field Game Master Equations Weak Soluio of Mea Field Game Maer Equaio Cheche Mou ad Jiafeg Zhag March 5, 19 Abrac I hi paper we udy maer equaio ariig from mea field game problem, uder he Lary-Lio moooiciy codiio. Claical oluio of

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

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

MODERN CONTROL SYSTEMS

MODERN CONTROL SYSTEMS MODERN CONTROL SYSTEMS Lecure 9, Sae Space Repreeaio Emam Fahy Deparme of Elecrical ad Corol Egieerig email: emfmz@aa.edu hp://www.aa.edu/cv.php?dip_ui=346&er=6855 Trafer Fucio Limiaio TF = O/P I/P ZIC

More information

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

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

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

Ruled surfaces are one of the most important topics of differential geometry. The

Ruled surfaces are one of the most important topics of differential geometry. The CONSTANT ANGLE RULED SURFACES IN EUCLIDEAN SPACES Yuuf YAYLI Ere ZIPLAR Deparme of Mahemaic Faculy of Sciece Uieriy of Aara Tadoğa Aara Turey yayli@cieceaaraedur Deparme of Mahemaic Faculy of Sciece Uieriy

More information

Heat Equation Derivative Formulas for Vector Bundles

Heat Equation Derivative Formulas for Vector Bundles Joural of Fucioal Aalyi 183, 4218 (21) doi:1.16jfa.21.3746, available olie a hp:www.idealibrary.com o Hea Equaio Derivaive Formula for Vecor Budle Bruce K. Driver 1 Deparme of Mahemaic-112, Uiveriy of

More information

Economics 8723 Macroeconomic Theory Problem Set 3 Sketch of Solutions Professor Sanjay Chugh Spring 2017

Economics 8723 Macroeconomic Theory Problem Set 3 Sketch of Solutions Professor Sanjay Chugh Spring 2017 Deparme of Ecoomic The Ohio Sae Uiveriy Ecoomic 8723 Macroecoomic Theory Problem Se 3 Skech of Soluio Profeor Sajay Chugh Sprig 27 Taylor Saggered Nomial Price-Seig Model There are wo group of moopoliically-compeiive

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

ME 321 Kinematics and Dynamics of Machines S. Lambert Winter 2002

ME 321 Kinematics and Dynamics of Machines S. Lambert Winter 2002 ME 31 Kiemaic ad Dyamic o Machie S. Lamber Wier 6.. Forced Vibraio wih Dampig Coider ow he cae o orced vibraio wih dampig. Recall ha he goverig diereial equaio i: m && c& k F() ad ha we will aume ha he

More information

Types Ideals on IS-Algebras

Types Ideals on IS-Algebras Ieraioal Joural of Maheaical Aalyi Vol. 07 o. 3 635-646 IARI Ld www.-hikari.co hp://doi.org/0.988/ija.07.7466 Type Ideal o IS-Algebra Sudu Najah Jabir Faculy of Educaio ufa Uiveriy Iraq Copyrigh 07 Sudu

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

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

TIME RESPONSE Introduction

TIME RESPONSE Introduction TIME RESPONSE Iroducio Time repoe of a corol yem i a udy o how he oupu variable chage whe a ypical e ipu igal i give o he yem. The commoly e ipu igal are hoe of ep fucio, impule fucio, ramp fucio ad iuoidal

More information

e x x s 1 dx ( 1) n n!(n + s) + e s n n n=1 n!n s Γ(s) = lim

e x x s 1 dx ( 1) n n!(n + s) + e s n n n=1 n!n s Γ(s) = lim Lecure 3 Impora Special FucioMATH-GA 45. Complex Variable The Euler gamma fucio The Euler gamma fucio i ofe ju called he gamma fucio. I i oe of he mo impora ad ubiquiou pecial fucio i mahemaic, wih applicaio

More information

Basic Results in Functional Analysis

Basic Results in Functional Analysis Preared by: F.. ewis Udaed: Suday, Augus 7, 4 Basic Resuls i Fucioal Aalysis f ( ): X Y is coiuous o X if X, (, ) z f( z) f( ) f ( ): X Y is uiformly coiuous o X if i is coiuous ad ( ) does o deed o. f

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

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

The Inverse of Power Series and the Partial Bell Polynomials

The Inverse of Power Series and the Partial Bell Polynomials 1 2 3 47 6 23 11 Joural of Ieger Sequece Vol 15 2012 Aricle 1237 The Ivere of Power Serie ad he Parial Bell Polyomial Miloud Mihoubi 1 ad Rachida Mahdid 1 Faculy of Mahemaic Uiveriy of Sciece ad Techology

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

State-Space Model. In general, the dynamic equations of a lumped-parameter continuous system may be represented by

State-Space Model. In general, the dynamic equations of a lumped-parameter continuous system may be represented by Sae-Space Model I geeral, he dyaic equaio of a luped-paraeer coiuou ye ay be repreeed by x & f x, u, y g x, u, ae equaio oupu equaio where f ad g are oliear vecor-valued fucio Uig a liearized echique,

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

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

Exercise: Show that. Remarks: (i) Fc(l) is not continuous at l=c. (ii) In general, we have. yn ¾¾. Solution:

Exercise: Show that. Remarks: (i) Fc(l) is not continuous at l=c. (ii) In general, we have. yn ¾¾. Solution: Exercie: Show ha Soluio: y ¾ y ¾¾ L c Þ y ¾¾ p c. ¾ L c Þ F y (l Fc (l I[c,(l "l¹c Þ P( y c

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

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

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

( ) ( ) ( ) ( ) ( ) ( ) ( ) (2)

( ) ( ) ( ) ( ) ( ) ( ) ( ) (2) UD 5 The Geeralized Riema' hypohei SV aya Khmelyy, Uraie Summary: The aricle pree he proo o he validiy o he geeralized Riema' hypohei o he bai o adjume ad correcio o he proo o he Riema' hypohei i he wor

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

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

SLOW INCREASING FUNCTIONS AND THEIR APPLICATIONS TO SOME PROBLEMS IN NUMBER THEORY

SLOW INCREASING FUNCTIONS AND THEIR APPLICATIONS TO SOME PROBLEMS IN NUMBER THEORY VOL. 8, NO. 7, JULY 03 ISSN 89-6608 ARPN Jourl of Egieerig d Applied Sciece 006-03 Ai Reerch Publihig Nework (ARPN). All righ reerved. www.rpjourl.com SLOW INCREASING FUNCTIONS AND THEIR APPLICATIONS TO

More information

Chapter 14 Necessary and Sufficient Conditions for Infinite Horizon Control Problems

Chapter 14 Necessary and Sufficient Conditions for Infinite Horizon Control Problems Chaper 14 Neceary ad Sufficie Codiio for Ifiie Horizo Corol Problem I may opimal corol problem i ecoomic he plaig horizo i aumed o be of ifiie legh. Thi mea ha he pero who olve uch a opimal corol problem

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

t = s D Overview of Tests Two-Sample t-test: Independent Samples Independent Samples t-test Difference between Means in a Two-sample Experiment

t = s D Overview of Tests Two-Sample t-test: Independent Samples Independent Samples t-test Difference between Means in a Two-sample Experiment Overview of Te Two-Sample -Te: Idepede Sample Chaper 4 z-te Oe Sample -Te Relaed Sample -Te Idepede Sample -Te Compare oe ample o a populaio Compare wo ample Differece bewee Mea i a Two-ample Experime

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

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

Review - Week 10. There are two types of errors one can make when performing significance tests:

Review - Week 10. There are two types of errors one can make when performing significance tests: Review - Week Read: Chaper -3 Review: There are wo ype of error oe ca make whe performig igificace e: Type I error The ull hypohei i rue, bu we miakely rejec i (Fale poiive) Type II error The ull hypohei

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

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

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

Generalized Orlicz Spaces and Wasserstein Distances for Convex-Concave Scale Functions

Generalized Orlicz Spaces and Wasserstein Distances for Convex-Concave Scale Functions Generalized Orlicz Space and Waerein Diance for Convex-Concave Scale Funcion Karl-Theodor Surm Abrac Given a ricly increaing, coninuou funcion ϑ : R + R +, baed on he co funcional ϑ (d(x, y dq(x, y, we

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

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

arxiv: v1 [math.nt] 13 Dec 2010

arxiv: v1 [math.nt] 13 Dec 2010 WZ-PROOFS OF DIVERGENT RAMANUJAN-TYPE SERIES arxiv:0.68v [mah.nt] Dec 00 JESÚS GUILLERA Abrac. We prove ome diverge Ramauja-ype erie for /π /π applyig a Bare-iegral raegy of he WZ-mehod.. Wilf-Zeilberger

More information

arxiv: v1 [math.pr] 16 Dec 2018

arxiv: v1 [math.pr] 16 Dec 2018 218, 1 17 () arxiv:1812.7383v1 [mah.pr] 16 Dec 218 Refleced BSDEs wih wo compleely separaed barriers ad regulaed rajecories i geeral filraio. Baadi Brahim ad Oukie Youssef Ib Tofaïl Uiversiy, Deparme of

More information

Variational Iteration Method for Solving Differential Equations with Piecewise Constant Arguments

Variational Iteration Method for Solving Differential Equations with Piecewise Constant Arguments I.J. Egieerig ad Maufacurig, 1,, 36-43 Publihed Olie April 1 i MECS (hp://www.mec-pre.e) DOI: 1.5815/ijem.1..6 Available olie a hp://www.mec-pre.e/ijem Variaioal Ieraio Mehod for Solvig Differeial Equaio

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

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

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

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

Computable Analysis of the Solution of the Nonlinear Kawahara Equation

Computable Analysis of the Solution of the Nonlinear Kawahara Equation Diache Lu e al IJCSE April Vol Iue 49-64 Compuale Aalyi of he Soluio of he Noliear Kawahara Equaio Diache Lu Jiai Guo Noliear Scieific eearch Ceer Faculy of Sciece Jiagu Uiveri Zhejiag Jiagu 3 Chia dclu@uj.edu.c

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

S n. = n. Sum of first n terms of an A. P is

S n. = n. Sum of first n terms of an A. P is PROGREION I his secio we discuss hree impora series amely ) Arihmeic Progressio (A.P), ) Geomeric Progressio (G.P), ad 3) Harmoic Progressio (H.P) Which are very widely used i biological scieces ad humaiies.

More information

CHAPTER 2 Quadratic diophantine equations with two unknowns

CHAPTER 2 Quadratic diophantine equations with two unknowns CHAPTER - QUADRATIC DIOPHANTINE EQUATIONS WITH TWO UNKNOWNS 3 CHAPTER Quadraic diophaie equaio wih wo ukow Thi chaper coi of hree ecio. I ecio (A), o rivial iegral oluio of he biar quadraic diophaie equaio

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

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

MASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.265/15.070J Fall 2013 Lecture 16 11/04/2013. Ito integral. Properties

MASSACHUSETTS INSTITUTE OF TECHNOLOGY 6.265/15.070J Fall 2013 Lecture 16 11/04/2013. Ito integral. Properties MASSACHUSES INSIUE OF ECHNOLOGY 6.65/15.7J Fall 13 Lecture 16 11/4/13 Ito itegral. Propertie Cotet. 1. Defiitio of Ito itegral. Propertie of Ito itegral 1 Ito itegral. Exitece We cotiue with the cotructio

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

Suggested Solutions to Assignment 1 (REQUIRED)

Suggested Solutions to Assignment 1 (REQUIRED) EC 45 dvaced Macroecoomic Irucor: Sharif F ha Deparme of Ecoomic Wilfrid Laurier Uiveri Wier 28 Suggeed Soluio o igme (REQUIRED Toal Mar: 5 Par True/ Fale/ Ucerai Queio [2 mar] Explai wh he followig aeme

More information

Two Implicit Runge-Kutta Methods for Stochastic Differential Equation

Two Implicit Runge-Kutta Methods for Stochastic Differential Equation Alied Mahemaic, 0, 3, 03-08 h://dx.doi.org/0.436/am.0.306 Publihed Olie Ocober 0 (h://www.scirp.org/oural/am) wo mlici Ruge-Kua Mehod for Sochaic Differeial quaio Fuwe Lu, Zhiyog Wag * Dearme of Mahemaic,

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

Lecture 15: Three-tank Mixing and Lead Poisoning

Lecture 15: Three-tank Mixing and Lead Poisoning Lecure 15: Three-ak Miig ad Lead Poisoig Eigevalues ad eigevecors will be used o fid he soluio of a sysem for ukow fucios ha saisfy differeial equaios The ukow fucios will be wrie as a 1 colum vecor [

More information

1 Infinitesimal Generators

1 Infinitesimal Generators Ifiieimal Geeraor. Le {T : } i a emi-group of bouded operaor i Baach pace X, i.e., i aifie ha T T = T + for all, > ad T = I. Le f = l T. Suppoe ha f i bouded o [, a], how ha f i ub-addiive, i.e., f + f

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

Theoretical Physics Prof. Ruiz, UNC Asheville, doctorphys on YouTube Chapter Q Notes. Laplace Transforms. Q1. The Laplace Transform.

Theoretical Physics Prof. Ruiz, UNC Asheville, doctorphys on YouTube Chapter Q Notes. Laplace Transforms. Q1. The Laplace Transform. Theoreical Phyic Prof. Ruiz, UNC Aheville, docorphy o YouTue Chaper Q Noe. Laplace Traform Q1. The Laplace Traform. Pierre-Simo Laplace (1749-187) Courey School of Mhemic ad Siic Uiveriy of S. Adrew, Scolad

More information

Common Fixed Point Theorem in Intuitionistic Fuzzy Metric Space via Compatible Mappings of Type (K)

Common Fixed Point Theorem in Intuitionistic Fuzzy Metric Space via Compatible Mappings of Type (K) Ieraioal Joural of ahemaics Treds ad Techology (IJTT) Volume 35 umber 4- July 016 Commo Fixed Poi Theorem i Iuiioisic Fuzzy eric Sace via Comaible aigs of Tye (K) Dr. Ramaa Reddy Assisa Professor De. of

More information

Introduction to the Mathematics of Lévy Processes

Introduction to the Mathematics of Lévy Processes Iroducio o he Mahemaics of Lévy Processes Kazuhisa Masuda Deparme of Ecoomics The Graduae Ceer, The Ciy Uiversiy of New York, 365 Fifh Aveue, New York, NY 10016-4309 Email: maxmasuda@maxmasudacom hp://wwwmaxmasudacom/

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

NON-EXPONENTIAL SANOV AND SCHILDER THEOREMS ON WIENER SPACE: BSDES, SCHRÖDINGER PROBLEMS AND CONTROL. October 3, 2018

NON-EXPONENTIAL SANOV AND SCHILDER THEOREMS ON WIENER SPACE: BSDES, SCHRÖDINGER PROBLEMS AND CONTROL. October 3, 2018 NON-EXPONENTIAL SANOV AND SCHILDER THEOREMS ON WIENER SPACE: BSDES, SCHRÖDINGER PROBLEMS AND CONTROL. JULIO BACKHOFF-VERAGUAS, DANIEL LACKER, AND LUDOVIC TANGPI Ocober 3, 28 Absrac. We derive ew limi heorems

More information

A Comparative Study of Adomain Decompostion Method and He-Laplace Method

A Comparative Study of Adomain Decompostion Method and He-Laplace Method Applied Mahemaic,, 5, 5-6 Publihed Olie December i SciRe. hp://www.cirp.org/joural/am hp://d.doi.org/.6/am..5 A Comparaive Sudy of Adomai Decompoio Mehod ad He-Laplace Mehod Badradee A. A. Adam, Deparme

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

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

Four equations describe the dynamic solution to RBC model. Consumption-leisure efficiency condition. Consumption-investment efficiency condition

Four equations describe the dynamic solution to RBC model. Consumption-leisure efficiency condition. Consumption-investment efficiency condition LINEARIZING AND APPROXIMATING THE RBC MODEL SEPTEMBER 7, 200 For f( x, y, z ), mulivariable Taylor liear expasio aroud ( x, yz, ) f ( x, y, z) f( x, y, z) + f ( x, y, z)( x x) + f ( x, y, z)( y y) + f

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

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

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

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

Stability. Outline Stability Sab Stability of Digital Systems. Stability for Continuous-time Systems. system is its stability:

Stability. Outline Stability Sab Stability of Digital Systems. Stability for Continuous-time Systems. system is its stability: Oulie Sabiliy Sab Sabiliy of Digial Syem Ieral Sabiliy Exeral Sabiliy Example Roo Locu v ime Repoe Fir Orer Seco Orer Sabiliy e Jury e Rouh Crierio Example Sabiliy A very impora propery of a yamic yem

More information

Solutions to Problems 3, Level 4

Solutions to Problems 3, Level 4 Soluios o Problems 3, Level 4 23 Improve he resul of Quesio 3 whe l. i Use log log o prove ha for real >, log ( {}log + 2 d log+ P ( + P ( d 2. Here P ( is defied i Quesio, ad parial iegraio has bee used.

More information

arxiv: v1 [math.st] 6 Jun 2013

arxiv: v1 [math.st] 6 Jun 2013 Eimaio of he fiie righ edpoi i he Gumbel domai Iabel Fraga Alve CEAUL ad DEIO Cláudia Neve CEAUL ad Uiveriy of Aveiro arxiv:306.45v [mah.st] 6 Ju 03 FCUL, Uiveriy of Libo Abrac A imple eimaor for he fiie

More information

PIECEWISE N TH ORDER ADOMIAN POLYNOMIAL STIFF DIFFERENTIAL EQUATION SOLVER 13

PIECEWISE N TH ORDER ADOMIAN POLYNOMIAL STIFF DIFFERENTIAL EQUATION SOLVER 13 Abrac PIECEWISE N TH ORDER ADOMIAN POLYNOMIAL A piecewie h order Adomia polyomial olver for iiial value differeial equaio capable of olvig highly iff problem i preeed here. Thi powerful echique which employ

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

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

Laplace Transform. Inverse Laplace Transform. e st f(t)dt. (2)

Laplace Transform. Inverse Laplace Transform. e st f(t)dt. (2) Laplace Tranform Maoud Malek The Laplace ranform i an inegral ranform named in honor of mahemaician and aronomer Pierre-Simon Laplace, who ued he ranform in hi work on probabiliy heory. I i a powerful

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

LIMITS OF FUNCTIONS (I)

LIMITS OF FUNCTIONS (I) LIMITS OF FUNCTIO (I ELEMENTARY FUNCTIO: (Elemeary fucios are NOT piecewise fucios Cosa Fucios: f(x k, where k R Polyomials: f(x a + a x + a x + a x + + a x, where a, a,..., a R Raioal Fucios: f(x P (x,

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

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

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

Explicit form of global solution to stochastic logistic differential equation and related topics

Explicit form of global solution to stochastic logistic differential equation and related topics SAISICS, OPIMIZAION AND INFOMAION COMPUING Sa., Opim. Inf. Compu., Vol. 5, March 17, pp 58 64. Publihed online in Inernaional Academic Pre (www.iapre.org) Explici form of global oluion o ochaic logiic

More information

Introduction to Hypothesis Testing

Introduction to Hypothesis Testing Noe for Seember, Iroducio o Hyohei Teig Scieific Mehod. Sae a reearch hyohei or oe a queio.. Gaher daa or evidece (obervaioal or eerimeal) o awer he queio. 3. Summarize daa ad e he hyohei. 4. Draw a cocluio.

More information

Introduction to SLE Lecture Notes

Introduction to SLE Lecture Notes Inroducion o SLE Lecure Noe May 13, 16 - The goal of hi ecion i o find a ufficien condiion of λ for he hull K o be generaed by a imple cure. I urn ou if λ 1 < 4 hen K i generaed by a imple curve. We will

More information