Discrete time approximation of decoupled Forward-Backward SDE with jumps

Size: px
Start display at page:

Download "Discrete time approximation of decoupled Forward-Backward SDE with jumps"

Transcription

1 Discree ime approximaion of decoupled Forward-Backward SD wih jumps Bruno BOUCHARD LPMA CNRS, UMR 7599 Universié Paris 6 bouchard@ccr.jussieu.fr and CRST Romuald LI CRMAD CNRS, UMR 7534 Universié Paris 9 elie@ensae.fr and CRST-NSA This version : March 27 Absrac We sudy a discree-ime approximaion for soluions of sysems of decoupled forward-backward sochasic differenial equaions wih jumps. Assuming ha he coefficiens are Lipschiz-coninuous, we prove he convergence of he scheme when he number of ime seps n goes o infiniy. The rae of convergence is a leas n 1/2+ε, for any ε >. When he jump coefficien of he firs variaion process of he forward componen saisfies a non-degeneracy condiion which ensures is inversibiliy, we achieve he opimal convergence rae n 1/2. The proof is based on a generalizaion of a remarkable resul on he pah-regulariy of he soluion of he backward equaion derived by Zhang 25 in he no-jump case. Key words : Discree-ime approximaion, forward-backward SD s wih jumps, Malliavin calculus. MSC Classificaion (2: 65C99, 6H7, 6J75. 1 Inroducion In his paper, we sudy a discree ime approximaion scheme for he soluion of a sysem of decoupled Forward-Backward Sochasic Differenial quaions (FBSD in shor wih jumps of he form { X = X + b(x rdr + σ(x rdw r + β(x r,e µ(de,dr, Y = g(x T + T h(θ r dr T Z r dw r T U (1.1 r(e µ(de,dr where Θ := (X,Y,Z,Γ wih Γ := ρ(eu(eλ(de. Here, W is a d-dimensional Brownian moion and µ an independen compensaed Poisson measure µ(de, dr = µ(de, dr λ(dedr. Such equaions naurally appear in hedging problems, see e.g. yraud-loisel 13, or in sochasic conrol, see e.g. Tang and Li 23 and he 1

2 recen paper Becherer 3 for an applicaion o exponenial uiliy maximizaion in finance. Under sandard Lipschiz assumpions on he coefficiens b, σ, β, g and h, exisence and uniqueness of he soluion have been proved by Tang and Li 23, hus generalizing he seminal paper of Pardoux and Peng 2. The main moivaion for sudying discree ime approximaions of sysems of he above form is ha hey provide an alernaive o classical numerical schemes for a large class of (deerminisic PD s of he form Lu(,x + h(,x,u(,x,σ(,x x u(,x, Iu(,x =, u(t,x = g(x, (1.2 where Lu(,x := Iu(,x := u (,x + xu(,xb(x + 1 d (σσ (x ij 2 u 2 x i x j (,x i,j=1 + {u(,x + β(x,e u(,x x u(,xβ(x,e}λ(de, {u(,x + β(x,e u(,x}ρ(eλ(de. Indeed, is well known ha, under mild assumpions on he coefficiens, he componen Y of he soluion can be relaed o he (viscosiy soluion u of (1.2 in he sense ha Y = u(,x, see e.g. 1 or 9. Thus solving (1.1 or (1.2 is essenially he same. In he so-called four-seps scheme, his relaion allows o approximae he soluion of (1.1 by firs esimaing numerically u, see e.g. 1. Here, we follow he converse approach. Since classical numerical schemes for PD s generally do no perform well in high dimension, we wan o esimae direcly he soluion of (1.1 so as o provide an approximaion of u. In he no-jump case, i.e. β =, he numerical approximaion of (1.1 has already been sudied in he lieraure, see e.g. Zhang 25, Bally and Pages 2, Bouchard and Touzi 7 or Gobe e al. 16. In 7, he auhors sugges he following implici scheme. Given a regular grid π = { = it/n, i =,...,n}, hey approximae X by is uler scheme X π and (Y,Z by he discree-ime process (Ȳ π i, Z π i i n defined backward by Z π i = n T Ȳ π i+1 W i+1 F i Ȳ π = Ȳ π i+1 F i + T n h( X π,ȳ π, Z π where Ȳ π n := g(x π n and W i+1 := W i+1 W i. In he no-jump case, i urns ou ha he discreizaion error rr n (Y,Z := { max i<n sup,+1 is inimaely relaed o he quaniy n 1 i+1 i= Y Ȳ π 2 + n 1 i+1 i= Z Z i 2 d where Zi := n 2 Z Z π 2 d i+1 Z d F i }1 2.

3 Under Lipschiz coninuiy condiions on he coefficiens, Zhang 24 was able o prove ha he laer is of order of n 1. This remarkable resul allows o derive he bound rr n (Y,Z Cn 1/2. Observe ha his rae of convergence can no be improved in general. Consider for example he case where X is equal o he Brownian moion W, g is he ideniy and h =. Then, Y = W and Ȳ π i = W i. In his paper, we exend he approach of Bouchard and Touzi 7 and approximae he soluion of (1.1 by he backward scheme Z π i = n T Ȳ π i+1 W i+1 F i Ȳ π = Ȳ π i+1 F i + T n h( X π,ȳ π, Z π, Γ π, Γ π = n T Ȳ π i+1 ρ(e µ(de,(,+1 F i where Ȳ π n := g(x π n. By adaping he argumens of Gobe e al. 16, we firs prove ha our discreizaion error rr n (Y,Z,U defined as { max i<n sup,+1 Y Ȳ π 2 + n 1 i+1 i= Z Z π 2 + Γ Γ π 2 d converges o as he discreizaion sep T/n ends o. We hen provide upper bounds on max i<n sup,+1 Y Y i 2 n 1 + i= i+1 Z Z i 2 + Γ Γ i 2 d, where Γ i+1 i := (n/t Γ d F i. When he coefficiens are Lipschiz coninuous, we obain and max i<n sup,+1 n 1 i+1 i= Y Y i 2 n 1 + i= i+1 } 1 2 Γ Γ i 2 d C n 1 Z Z i 2 d C ε n 1+ε, for any ε >. Under some addiional condiions on he inversibiliy of β + I d, see H, we hen prove ha he previous inequaliy holds rue for ε =. This exends o our framework he remarkable resul derived by Zhang 25 in he no-jump case. I allows us o show ha our discree-ime scheme achieves, under he sandard Lipschiz condiions, a rae of convergence of a leas n 1/2+ε, for any ε >, and he opimal rae n 1/2 under he addiional assumpion H. Observe ha, in opposiion o algorihms based on he approximaion of he Brownian moion by discree processes aking a finie number of possible values (see e.g. 17 and he references herein, our scheme does no provide a fully implemenable numerical procedure since involves he compuaion of a large number of condiional expecaions. However, he implemenaion of he above menioned schemes 3

4 in high dimension is quesionable and, in our seing, his issue could be solved by approximaing he condiional expecaion operaors numerically in an efficien way, see 2, 7, 16 and 11 for an adapaion o our seing of he echnics suggesed in 16. The res of he paper is organized as follows. In Secion 2, we describe he approximaion scheme and sae our main convergence resul. Secion 3 conains some resuls on he Malliavin derivaives of Forward and Backward SD s. Applying hese resuls in Secion 4, we derive some regulariy properies for he soluion of he backward equaion under addiional smoohness assumpions on he coefficiens. We finally use an approximaion argumen o conclude he proof of our main heorem. Noaions : Any elemen x R d will be idenified o a column vecor wih i-h componen x i and uclidian norm x. For x i R d i, i n and d i N, we define (x 1,...,x n as he column vecor associaed o (x 1 1,...,xd 1 1,...,x1 n,...,x dn n. The scalar produc on R d is denoed by x y. For a (m d-dimensional marix M, we noe M := sup{ Mx ; x R d, x = 1}, M is ranspose and we wrie M M d if m = d. Given p N and a measured space (A, A,µ A, we denoe by L p (A, A,µ A ; R d, or simply L p (A, A or L p (A if no confusion is possible, he se of p- inegrable R d -valued measurable maps on (A, A,µ A. For p =, L (A, A,µ A ; R d is he se of essenially bounded R d -valued measurable maps. The se of k-imes differeniable maps wih bounded derivaives up o order k is denoed by Cb k and Cb := k 1 Cb k. For a map b : Rd R k, we denoe by b is Jacobian marix whenever i exiss. In he following, we shall use hese noaions wihou specifying he dimension when is clearly given by he conex. 2 Discree ime approximaion of decoupled FBSD wih jumps 2.1 Decoupled forward backward SD s As in 5, we shall work on a suiable produc space Ω := Ω W Ω µ where Ω W is he se of coninuous funcions w from,t ino R d, and Ω µ is he se of ineger-valued measures on,t wih := R m for some m 1. For ω = (w,η Ω, we se W(w,η = w and µ(w,η = η and define F W = (F W T (resp. F µ = (F µ T as he smalles righ-coninuous filraion on Ω W (resp. Ω µ such ha W (resp. µ is opional. We le P W be he Wiener measure on (Ω W, FT W and P µ be he measure on (Ω µ, F µ T under which µ is a Poisson measure wih inensiy ν(d,de = λ(ded, for some finie measure λ on, endowed wih is Borel ribe. We hen define he probabiliy measure P := P W P µ on (Ω, FT W Fµ T. Wih his consrucion, W and µ are independen under P. Wihou loss of generaliy, we can assume ha he naural filraion F = (F T induced by (W,µ is complee. We denoe by 4

5 µ := µ ν he compensaed measure associaed o µ. Given K >, wo K-Lipschiz coninuous funcions b : R d R d and σ : R d M d, and a measurable map β : R d R d such ha sup e β(, e K and sup β(x,e β(x,e K x x x, x R d, (2.1 e we define X as he soluion on,t of X = X + b(x r dr + σ(x r dw r + β(x r,e µ(de,dr, (2.2 for some iniial condiion X R d. The exisence and uniqueness of such a soluion is well known under he above assumpions, see he Appendix for sandard esimaes for soluions of such SD. Before inroducing he backward SD, we need o define some addiional noaions. Given s and some real number p 2, we denoe by S p he se of real valued s, adaped càdlàg processes Y such ha Y S p s, 1 := sup Y r p p s r <, H p s, is he se of progressively measurable Rd -valued processes Z such ha Z H p s, := ( s p 1 p Z r 2 2 dr <, L p λ,s, is he se of P measurable maps U : Ω,T R such ha U L p λ,s, := s 1 U s (e p p λ(deds < wih P defined as he σ-algebra of F-predicable subses of Ω,T. The space B p s, := S p s, Hp s, Lp λ,s, is endowed wih he norm (Y,Z,U B p s, := ( 1 Y p S p + Z p H p + U p p L p. s, s, λ,s, In he sequel, we shall omi he subscrip s, in hese noaions when (s, = (,T. For ease of noaions, we shall someimes wrie ha an R n -valued process is in S p s, or L p λ,s, meaning ha each componens in he corresponding space. Similarly an elemen of M m is said o belong o H p s, if each column belongs o Hp s,. The norms are hen naurally exended o such processes. 5

6 The aim of his paper is o sudy a discree ime approximaion of he riple (Y,Z,U soluion on,t of he backward sochasic differenial equaion T T T Y = g(x T + h(θ r dr Z r dw r U r (e µ(de,dr, (2.3 where Θ := (X, Y, Z, Γ and Γ is defined by Γ := ρ(eu(eλ(de, for some measurable map ρ : R m saisfying sup ρ(e K. (2.4 e By a soluion, we mean a riple (Y,Z,U B 2 saisfying (2.3. In order o ensure he exisence and uniqueness of a soluion o (2.3, we assume ha he map g : R d R and h : R d R R d R m R are K-Lipschiz coninuous (see Lemma 5.2 in he Appendix. For ease of noaions, we shall denoe by C p a generic consan depending only on p and he consans K, λ(, b(, σ(, h(, g( and T. We wrie C p if i also depends on X. In his paper, p will always denoe a real number greaer han 2. Remark 2.1 For he convenience of he reader, we have colleced in he Appendix sandard esimaes for he soluions of Forward and Backward SD s. In paricular, hey imply (X,Y,Z,U p S p B p C p (1 + X p, p 2. (2.5 The esimae on X is sandard, see (5.2 of Lemma 5.1 in he Appendix. Plugging his in (5.6 of Lemma 5.2 leads o he bound on (Y,Z,U B p. Using (5.3 of Lemma 5.1, we also deduce ha sup X u X s p C p (1 + X p s, (2.6 s u while he previous esimaes on X combined wih (5.7 of Lemma 5.2 implies { } sup Y u Y s p C p (1 + X p s p + Z p H p + U p L p. (2.7 s u s, λ,s, 2.2 Discree ime approximaion We firs fix a regular grid π := { := it/n, i =,...,n} on,t and approximae X by is uler scheme X π defined by { X π := X X π i+1 := X π i + T n b(xπ + σ(x π i W i+1 + (2.8 β(xπ,e µ(de,(,+1 6

7 where W i+1 := W i+1 W i. Is well known ha max sup X X π i<n i 2 C2 n 1. (2.9,+1 We hen approximae (Y,Z,Γ by (Ȳ π, Z π, Γ π defined by he backward implici scheme Z π := n Ȳ T π i+1 W i+1 F i, Γ π := n T Ȳ π i+1 ρ(e µ(de,(,+1 F i Ȳ π π := Ȳi+1 F i + T n h( X π i,ȳ π i, Z π i, Γ π (2.1 on each inerval,+1, where Ȳ π n := g(x π n. Observe ha he resoluion of he las equaion in (2.1 may involve he use of a fixed poin procedure. However, h being Lipschiz and muliplied by 1/n, he approximaion error can be negleced for large values of n. Remark 2.2 The above backward scheme is a naural exension of he one considered in 7 in he case β =. By he represenaion heorem, see e.g. Lemma 2.3 in 23, here exis wo processes Z π H 2 and U π L 2 λ saisfying Ȳ π π i+1 Ȳi+1 F i = i+1 Z π s dw s + i+1 U π s (e µ(ds,de. Observe ha Z π and Γ π defined in (2.1 saisfy Z π i = n i+1 T Zs π ds F and Γπ i = n i+1 T Γ π s ds F (2.11 and herefore coincide wih he bes H 2,+1 -approximaions of (Zπ <+1 and (Γ π i <+1 := ( ρ(euπ (eλ(de i <+1 by F i -measurable random variables (viewed as consan processes on,+1. Finally, observe ha we can define Y π on,+1 by seing Y π := Ȳ π i ( h(x π i,ȳ π i, Z π i, Γ π + Zs π dw s + Us π (e µ(ds,de. 2.3 Convergence of he approximaion scheme In his subsecion, we show ha he approximaion error { rr n (Y,Z,U := sup }1 Y Ȳ π 2 + Z Z π 2 H + Γ Γ 2 π 2 2 H 2 T converges o. Le us firsnroduce he processes ( Z, Γ defined on each inerval,+1 by Z := n i+1 T Z s ds F i and Γ := n i+1 T Γ s ds F i. 7

8 Remark 2.3 Observe ha Z i and Γ i are he counerpars of Zπ i and Γ π for he original backward SD. They can also be inerpreed as he bes H 2,+1 - approximaions of (Z i <+1 and (Γ i <+1 by F i -measurable random variables (viewed as consan processes on,+1. Proposiion 2.1 We have n 1 i= i+1 Y Y i 2 d C 2 n 1 and Z Z H 2 + Γ Γ H 2 ǫ(n(2.12 where ǫ(n as n. Moreover, rr n (Y,Z,U C 2 so ha rr n (Y,Z,U n. Proof. Since Y solves (2.3, Y Y i 2 C 2 ( n 1/2 + Z Z H 2 + Γ Γ H 2, (2.13 h(x r,y r,z r,γ r 2 + Z r 2 + U r (e 2 λ(de dr. Combining he Lipschiz propery of h wih (2.5, i follows ha n 1 i+1 i= Y Y i 2 d C 2 n. This is exacly he firs par of (2.12. The proof of (2.13 hen follows exacly he same argumens as in 7, see p 99 in 11 for deails. I remains o prove he second par of (2.12. Since Z is F-adaped, here is a sequence of adaped processes (Z n n such ha Z n = Z n i on each,+1 and Z n converges o Z in H 2. By Remark 2.3, we observe ha Z Z 2 H Z Z n 2 2 H, and applying he same reasoning o Γ 2 concludes he proof. 2.4 Pah-regulariy and convergence rae In view of Proposiion 2.1, he discreizaion error converges o zero. In order o conrol is speed of convergence, i remains o sudy Z Z 2 H 2 + Γ Γ 2 H 2. Before saing our main resul, le us inroduce he following assumpion: H : For each e, he map x R d β(x,e admis a Jacobian marix β(x,e such ha he funcion (x,ξ R d R d a(x,ξ;e := ξ ( β(x,e + I d ξ saisfies one of he following condiion uniformly in (x,ξ R d R d a(x,ξ;e ξ 2 K 1 or a(x,ξ;e ξ 2 K 1. 8

9 Remark 2.4 Observe for laer use ha he condiion H implies ha, for each (x,e R d, he marix β(x,e + I d is inverible wih inverse bounded by K. This ensures he inversibiliy of he firs variaion process X of X, see Remark 3.6. Moreover, if q is a smooh densiy on R d wih compac suppor, hen he approximaing funcions β k, k N, defined by β k (x,e := R d k d β( x,eq(kx xd x are smooh and also saisfy H. In Secion 5 of 9, he auhors imposes a similar condiion: de( β(x,e + I d 1 δ, x R d λ(de a.e. for some δ >. Under mild addiional assumpions, his allows o prove he exisence of a bounded soluion, in a suiable weighed Sobolev space, o a PD of he form (1.2 which can hen be relaed o Y. Our main heorem is saed for a suiable version of (Z,U,Γ. Observe ha does no change he quaniy rr n (Y,Z,U. Theorem 2.1 The following holds. (i For all i < n, sup Y Y i 2 C2 n 1 and sup Γ Γ i 2 C2 n 1 (2.14,+1,+1 so ha Γ Γ 2 S 2 C 2 n 1 and Γ Γ 2 H 2 C 2 n 1. Moreover, for any ε >, (ii Assume ha H holds. Then Z Z 2 H 2 C ε n 1+ε. (2.15 Z Z 2 H 2 C 2 n 1. (2.16 This regulariy propery will be proved in he subsequen secions. Combined wih Proposiion 2.1, i provides an upper bound for he convergence rae of our backward implici scheme. Corollary 2.1 For any ε >, rr n (Y,Z,U C ε n 1/2+ε. If H holds, hen rr n (Y,Z,U C 2 n 1/2. Remark 2.5 One could also use an explici scheme as in e.g. 2 or 16. In his case, (2.1 has o be replaced by ρ(e µ(de,(,+1 F i Z π := n T Ỹ π i+1 W i+1 F i Ỹ π, Γ π := n T Ỹ π +1 π := Ỹi+1 F i + T ( h n X π i,ỹ π i+1, Z π i, Γ π F i (2.17 wih he erminal condiion Ỹ π n = g(x π n. The advanage of his scheme is ha does no require a fixed poin procedure. However, from a numerical poin of view, adding a erm in he condiional expecaion defining Ỹ π i makes i more difficul o esimae. We herefore hink ha he implici scheme may be more racable in pracice. The above convergence resuls can be easily exended o his scheme, see 11 for deails. 9

10 Remark 2.6 In he unpublished paper 9, he auhors discuss he regulariy of (X,Y,Z,U wih respec o he iniial condiion X in a case where he coefficiens b,σ,h and g are C 3, wih linear growh and bounded derivaives for he wo firs ones and derivaives having polynomial growh for he wo las ones. Under hese regulariy assumpions, hey show ha he map (,x u(,x = Y,x belongs o C,2 (,T R d wih 1 2-Hölder coninuiy in ime and derivaives having polynomial growh in space, see heir Proposiion 3.5 and heir Corollary 3.6. Similar resuls are obained for (,x (Z,x,U,x which can be idenified o ( u(,xσ(x,u(,x + β(x, u(,x. This readily implies he properies saed in Theorem 2.1 which can be seen as weak versions of he regulariy resuls of 9. The imporan poin here is ha: 1. Our resuls do no require all he regulariy assumpions of 9; 2. This is all we need o provide he convergence raes of Corollary 2.1. Remark 2.7 I will be clear from he proofs ha all he resuls of his paper hold if we le he maps b, σ, β, and h depend on whenever hese funcions are 1/2-Hölder in and he oher assumpions are saisfied uniformly in. The uler approximaion X π of X could also be replaced by any oher adaped approximaion saisfying (2.9. Remark 2.8 We refer o 11 for exensions o he approximaion of sysems of semilinear PDs hrough heir relaion wih BSDs wih jumps, see 21, and for similar convergence resuls wihou H bu under addiionnal regulariy assumpions. 3 Malliavin calculus for FBSD In his secion, we prove ha he soluion (Y,Z,U of (2.3 is smooh in he Malliavin sense under he addiional assumpions C X : b, σ and β(,e are C 1 b uniformly in e, CY : g and h are C 1 b. This will allow us o provide represenaion and regulariy resuls for Y, Z and U in Secion 4. Under C X -C Y, hese resuls will immediaely imply he firs asserion of (i of Theorem 2.1, while he second one (resp. (ii will be obained by adaping he argumens of 6 (resp. 25 under he addiional assumpion H. 3.1 Generaliies The consrucion of he Malliavin derivaives on he Wiener space is sandard, see e.g. 18, and can be easily exended o our seing by observing ha here is an isomery beween L 2 (Ω W Ω µ and L 2 (Ω W,L 2 (Ω µ, wih obvious noaions. Le S denoe he se of random variables of he form ( T T F = φ f 1 ( dw,..., f κ ( dw,µ, where κ 1, f i :,T R d is a bounded measurable map for each i κ, φ is a real-valued measurable map on R κ Ω µ and φ(,η C b, P µ(dη-a.e. 1

11 We denoe by D he Malliavin derivaive operaor wih respec o he Brownian moion. For F S as above and s T, is defined as D s F := ( T T i φ f 1 ( dw,..., f κ ( dw,µ f i (s, i κ where i φ is he derivaive of φ wih respec o is i-h argumen. We hen denoe by ID 1,2 he closure of S wih respec o he norm F ID 1,2 := { F 2 T } 1 + D s F 2 2 ds and define H 2 (ID 1,2 as he se of elemens ξ H 2 such ha ξ ID 1,2 for almos all T and such ha, afer possibly passing o a measurable version, T ξ 2 := ξ 2 H 2 (ID 1,2 H + D 2 s ξ 2 H ds <. 2, Observe ha for ψ in L 2 λ (Fµ, he se of elemens of L 2 λ which are independen of W, we have Dψ =. We finally define L 2 λ (ID1,2 as he closure of he se } L 2 λ (ID 1,2 := Vec {ψ = ξϑ : ξ H 2 P (ID1,2, F W, ϑ L 2 λ (Fµ, ψ L 2λ (ID 1,2 < for he norm T ψ 2 := ψ 2 L 2 λ (ID1,2 L + D 2 s ψ 2 λ L. 2 λds Here, H 2 P (ID1,2, F W denoes he se of F W -predicable elemens of H 2 (ID 1,2 and D s (ξϑ = (D s ξϑ for ξ H 2 P (ID1,2, F W, ϑ L 2 λ (Fµ. Here again, we exend he definiion of H 2 (ID 1,2 and L 2 λ (ID1,2 o processes wih values in Md and R d in a naural way. From now on, given a marix A, we shall denoe by A i is i-h column. For k d, we denoe by D k he Malliavin derivaive wih respec o W k, meaning ha D k F = (DF k for F ID 1,2. Remark 3.1 Wih his consrucion, he operaor D enjoys he usual properies of he Malliavin derivaive operaor on Wiener spaces. In paricular, if ξ H 2 (ID 1,2 and f C 1 b (Rd, hen D s ( T T T f(ξ d + ξ dw = s f(ξ D s ξ d + ξ s + d j=1 T s D s ξ j dw j for all s T. Here denoes ransposiion. I follows from he same argumen as in 18, which we refer o for more deails. 11

12 Remark 3.2 Fix ξ H 2 P (ID1,2, F W. By Lemma in 18, here exiss a family of deerminisic measurable kernels f m ( 1,..., m, in L 2 (,T m+1, m, such ha ξ = m I m(f m (, and D s ξ = m 1 mi m 1(f m (,s, where I m denoes he m-ieraed Wiener inegral, see Proposiion in 18. Therefore, if τ is a random ime bounded by T and independen of W, we have ξ τ = m I m(f m (,τ and, by he same argumen as in he proof of Proposiion in 18, ξ τ ID 1,2 whenever τ has a bounded densiy and D s (ξ τ = m 1 mi m 1(f m (,s,τ = (D s ξ τ. Remark 3.3 For ξ H 2 (ID 1,2 he inegrals T D sξ d and T D sξ dw have T o be undersood as inegrals wrien on he process D s ξ, i.e. (D sξ d and T (D sξ dw. Similarly, for ψ L 2 λ (ID1,2, T D sψ (e µ(de,d has o be undersood as T (D sψ (e µ(de,d. However, i follows from Remark 3.2 ha coincides wih T D s(ψ (e µ(de,d, so ha here is no ambiguiy. The wo following Lemmas are generalizaions of Lemma 3.3 and Lemma 3.4 in 21 which correspond o he case where is finie, see also Lemma 2.3 in 2 for he case of Iô inegrals. Lemma 3.1 Assume ha ψ L 2 λ (ID1,2. Then, H := T ψ (e µ(de,d ID 1,2 and D s H = T D sψ (e µ(de,d for all s T. Proof. Firs noice ha suffices o prove he required resul when ψ L 2 λ (ID 1,2. Indeed, we can hen rerieve he general case by considering a sequence (ψ n n in L 2 λ (ID 1,2 which converges o ψ in L 2 λ (ID1,2, so ha H n := T ψn (e µ(de,d is a Cauchy sequence in ID 1,2 which converges o H and (D s H n s T converges o ( T D sψ (e µ(de,d s T in H 2. We herefore assume ha ψ = ξϑ where ξ H 2 P (ID1,2, F W, ϑ L 2 λ (Fµ and ψ L 2 λ (ID 1,2 <. Then, T ψ (e µ(de,d = T T ξ ϑ (eµ(de,d ξ ϑ (eλ(ded, where, by Remark 3.1 and he fac ha ϑ (eλ(de is independen of W, T ( D s ξ I remains o prove ha T D s ϑ (eλ(de d = ξ ϑ (eµ(de,d = = T T T D s ξ ϑ (eλ(ded (D s ξ ϑ (eλ(ded. (D s ξ ϑ (eµ(de,d. 12

13 To see his, we define N by N := µ(,ds for T, (τ i i 1 as he sequence of jump imes of N and ( i i 1 by i := N τi N τi. Wih hese noaions, we have o show ha D s ξ τi ϑ τi ( i 1 τi T = s ξ τi ϑ τi ( i 1 τi T, (3.1 i 1(D i 1 see Remark 3.3. Using Remark 3.2, we now oberve ha D n s i=1 ξ τ i ϑ τi ( i 1 τi T = n i=1 (D sξ τi ϑ τi ( i 1 τi T, for each n 1. Passing o he limi leads o (3.1 and concludes he proof. Indeed, he previous ideniy implies ha he sequence (F n n 1 defined by F n := n i=1 ξ τ i ϑ τi ( i 1 τi T belongs o H 2 (ID 1,2 and saisfies F n i 1 ξ τ i ϑ τi ( i 1 τi T as well as D s F n i 1 (D sξ τi ϑ τi ( i 1 τi T which implies ha sup n 1 F n 2 H 2 (ID 1,2 is bounded by ( T 2 T 2 2 ξ ϑ (e µ(de,d + ξ ϑ (e λ(ded T +2 T +2 T T C 2 ψ 2 L 2 λ (ID1,2 <, 2 D s ξ ϑ (e µ(de,d ds 2 D s ξ ϑ (e λ(ded ds where he second and he fourh erms on he righ hand-side are bounded by using Jensen s inequaliy and he assumpion λ( <. Moreover, by dominaed convergence, F n F 2 + T D sf n D s F 2 ds where F := i 1 ξ τ i ϑ τi ( i 1 τi T and D s F := i 1 (D sξ τi ϑ τi ( i 1 τi T saisfy, by he same argumens as above, F 2 H 2 (ID 1,2 C 2 ψ 2 L 2 λ (ID1,2 <. Remark 3.4 Similar argumens as in he above proof shows ha for ψ L 2 λ (ID1,2 and f L (, we have, for almos every s T, ψ s(ef(eλ(de ID 1,2 and D ( ψ s (ef(eλ(de := D ψ s (ef(eλ(de. Lemma 3.2 Le S(W denoe he se of random variables of he form H W ( T T = φ f 1 ( dw,..., f κ ( dw where κ 1, φ Cb and f i :,T R d is a bounded measurable map for each i κ. Then, Vec{S(W L (Ω µ, F µ T } is dense in ID1,2 for he norm ID 1,2. 13

14 Proof. I suffices o prove ha Vec{S(W L (Ω µ, F µ T } is dense in S. Fix H S of he form ( T T H = φ f 1 ( dw,..., f κ ( dw,µ. Observe ha Ω µ can be idenified o he space of finie (possibly empy sequences (,e i 1 i n of,t, n, such ha ( i 1 is increasing. Le G n denoe he se of such sequences of lengh n and G := n G n. Given η Ω u, we denoe by ( η i,eη i i 1 he associaed sequence, and we idenify φ wih a measurable map on R κ G. We denoe by φ n is resricion o R κ G n, n. Le ψ n denoe he gradien ( of φ n wih respec o is firs κ componens and se f := (f 1,...,f κ, T G := f1 ( dw,..., T fκ ( dw. Since (H,D s H = n (φ n (G,( µ i,eµ i 1 i n,ψ n (G,( µ i,eµ i 1 i n f(s1 µ(,,t=n, i suffices o prove ha each H n := φ n (G,( µ i,eµ i 1 i n can be approximaed by linear combinaions of elemens of S(W L (Ω µ, F µ T. Moreover, we can always assume ha φ n is Cb on R κ G n. Indeed, φ is already Cb in is firs κ componens, a.e., and we can replace φ n by is convoluion wih a sequence of smooh kernels acing only is las n componens. Since boh funcions are coninuous, we can hen approximae (φ n,ψ n poinwise by linear combinaions of funcions of he form (φ n,ψ n (,(,e i 1 i n 1 A where A is a Borel se of G n and (,e i 1 i n G n. The required resul hen follows from he fac ha D s φ n (G,(,e i 1 i n 1 A (( µ i,eµ i 1 i n = (ψ n (G,(,e i 1 i n f(s1 A (( µ i,eµ i 1 i n. Lemma 3.3 Fix (ξ,ψ H 2 L 2 λ and assume ha T T H := ξ dw + ψ (e µ(de,d ID 1,2. Then, (ξ,ψ H 2 (ID 1,2 L 2 λ (ID1,2 and T D s H := ξs + d i=1 where ξ denoes he ranspose of ξ. T D s ξ dw + D s ψ (e µ(de,d, Proof. One easily deduce from Lemma 3.2 ha H := Vec{H W H µ : H W S(W, H µ L (Ω µ, F µ T, H W H µ = } is dense in ID 1,2 {H L 2 (Ω, F, P : H = } for ID 1,2. Thus, i suffices o prove he resul for H of he form H W H µ where H W S(W, H µ L (Ω µ, F µ T and H W H µ =. By he represenaion heorem, see e.g. 8, here exiss ψ L 2 λ such ha H µ = H µ + T ψ (e µ(de,d and by Ocone s formula, see e.g. Proposiion in 18, H W = H W + T D H W F W dw. Thus i follows from Iô s Lemma ha H = T H µ D H W F W dw + T HW ψ (e µ(de,d where H µ = H µ F and H W = H W F. Furhermore, easy compuaions show ha he wo inegrands belong respecively o H 2 (ID 1,2 and L 2 λ (ID1,2. Thus, Remark 3.1 and Lemma 3.1 conclude he proof. 14

15 3.2 Malliavin calculus on he Forward SD In his secion, we recall well-known properies concerning he differeniabiliy in he Malliavin sense of he soluion of a Forward SD. In he case where β = he following resuls saed in e.g. 18. The exension o he case β is easily obained by condiioning by µ, see e.g. 14 for explanaions in he case where is finie, or by combining Remark 3.1, Lemma 3.1 wih a fixed poin procedure as in he proof of Theorem in 18, see also Proposiion 3.2 below. Proposiion 3.1 Assume ha C X holds, hen X ID 1,2 for all T. For all s T and k d, D k sx admis a version χ s,k which solves on s,t χ s,k = σ k (X s + + s s b(x r χ s,k r dr + s β(x r,eχ s,k r µ(dr,de. d j=1 σ j (X r χ s,k r dw j r Remark 3.5 Fix p 2 and r s u T. Under C X, i follows from Lemma 5.1 applied o X and χ s ha χ s p S p C p (1 + X p, χ s u χ s p C p u (1 + X p and χ s χ r p S p C p s r (1 + X p. Remark 3.6 Under C X, we can define he firs variaion process X of X which solves on,t X = I d + + b(x r X r dr + d σ j (X r X r dwr j j=1 β(x r,e X r µ(dr,de. (3.2 Moreover, under H, see Remark 2.4, ( X 1 is well defined and solves on,t d ( X 1 = I d ( X 1 r b(x r σ j (X r σ j (X r dr + j=1 ( X 1 r β(x r,eλ(dedr d ( X 1 r σj (X r dwr j j=1 ( X 1 r ( β(x r,e + I d 1 β(x r,eµ(de,dr. (3.3 Remark 3.7 Fix p 2. Under H-C X, i follows from Remark 2.4 and Lemma 5.1 applied o X and ( X 1 ha X S p + ( X 1 S p C p. Remark 3.8 Assume ha H-C X holds and observe ha χ s = (χ s,k k d and X solve he same equaion up o he condiion a ime s. By uniqueness of he soluion on,t, i follows ha χ s r = X r ( X s 1 σ(x s 1 s r for all s,r T. 15

16 3.3 Malliavin calculus on he Backward SD In his secion, we generalize he resul of Proposiion 3.1 in 21. Le us denoe by B 2 (ID 1,2 he se of riples (Y,Z,U B 2 such ha Y ID 1,2 for all T and (Z,U H 2 (ID 1,2 L 2 λ (ID1,2. Proposiion 3.2 Assume ha C X -C Y holds. Then, he riples (Y,Z,U belongs o B 2 (ID 1,2. For each s T and k d, he equaion Υ s,k T T = g(x T χ s,k T + h(θ r Φ s,k r dr T ζr s,k dw r Vr s,k (e µ(de, dr (3.4 wih Φ s,k := (χ s,k,υ s,k,ζ s,k,γ s,k and Γ s,k := ρ(ev s,k (eλ(de, admis a unique soluion. Moreover, (Υ s,k,ζ s,k,v s,k s, T is a version of (DsY k,dsz k,dsu k s, T. Proof. Wih he help of Lemma 3.3 and he esimaes of Lemma 5.2, we can reproduce exacly he proof of Proposiion 5.3 in 12 up o minor modificaions. See 11 p 113 for deails. Proposiion 3.3 Assume ha C X -C Y holds. For each k d, he equaion T T T Y k = g(x T XT+ k h(θ r Φ k rdr Zr k dw r Ur k (e µ(de,dr (3.5 wih Φ k = ( X k, Y k, Z k, Γ k and Γ k := ρ(e Uk (eλ(de, admis a unique soluion ( Y k, Z k, U k. Moreover, here is a version of (ζ s,k,υ s,k,v s,k given by {( Y, Z, U ( X s 1 σ k (X s 1 s } s, T where Y s he marix whose k-column is given by Y k and Z, U are defined similarly. Proof. In view of Proposiion 3.2 and Remark 3.8, his follows immediaely from he uniqueness of he soluion of (3.4. Remark 3.9 Combined wih C X -C Y and Remark 3.5, Lemma 5.2 below implies ha sup (Υ s,ζ s,v s p B C p p (1 + X p for all p 2. (3.6 s T I follows from Lemma 5.2 and Remark 3.7 ha ( Y, Z, U B p C p for all p 2. (3.7 s, T 4 Represenaion resuls and pah regulariy for he BSD In his secion, we use he above resuls o obain some regulariy for he soluion of he BSD (2.3 under C X -C Y, C X -C Y -H. Similar resuls wihou C X -C Y will hen be obained by using an approximaion argumen. 16

17 Fix (u,s,,x,t 3 R d and k,l d. In he sequel, we shall denoe by X(,x he soluion of (2.2 on,t wih iniial condiion X(,x = x, and by (Y (, x, Z(, x, U(, x he soluion of (2.3 wih X(, x in place of X. We define similarly (Υ s,k (,x, ζ s,k (,x, V s,k (,x and ( Y (,x, Z(,x, U(,x. Observe ha, wih hese noaions, we have (X(,X,Y (,X, Z(,X, U(,X = (X,Y,Z, U. 4.1 Represenaion We sar his secion by proving useful bounds for he (deerminisic maps defined on,t R d by u(,x := Y (,x, u(,x := Y (,x, v s,k (,x := Υ s,k (,x, where s,t and k d. Proposiion 4.1 Assume ha C X and C Y hold, hen, u(,x + v s,k (,x C 2 (1 + x and u(,x C 2 (4.1 for all s, T, k d and x R d. Proof. When (,x = (,X, he resul follows from (2.5 in Remark 2.1, (3.6 and (3.7. The general case is obained similarly by changing he iniial condiion on X. Proposiion 4.2 Assume ha C X and C Y hold. Then, here is a version of Z given by (Υ T which saisfies Z p S p C p (1 + X p. Proof. Here again we only consider he case d = 1 and omi he indexes k,l. By Proposiion 3.2, (Y,Z,U belongs o B 2 (ID 1,2 and i follows from Lemma 3.3 ha D s Y = Z s s h(θ r D s Θ r dr + s D s Z r dw r + s D s U r (e µ(de,dr, (4.2 for < s T. Taking s = leads o he represenaion of Z. Thus, afer possibly passing o a suiable version, we have Z = D Y = Υ. By uniqueness of he soluion of (2.2-(2.3-(3.4 for any iniial condiion in L 2 (Ω, F a, we have Υ = v (,X. The bound on Z hen follows from Proposiion 4.1 combined wih (2.5 of Remark 2.1. Proposiion 4.3 (i Define Ũ by Ũ(e := u(,x + β(x,e lim r u(r,x r = Y Y. Then Ũ is a version of U and i saisfies sup Ũ(e p S C p p (1 + X p. (4.3 e (ii Assume ha C X and C Y hold. Define Ũ by Ũ(e := u(,x + β(x,e lim r u(r,x r. Then Ũ is a version of U and i saisfies sup Ũ(e p S C p p. (4.4 e 17

18 Remark 4.1 We will see in Proposiion 4.4 below ha u is coninuous under C X and C Y so ha Ũ(e = u(,x + β(x,e u(,x. A similar represenaion is derived in 21, in a case where is finie, and in 9, in he case where is no finie, bu under more regulariy assumpions on he coefficiens. Proof of Proposiion 4.3. We only provide he proof of (i, he oher asserion is proved similarly. 1. By uniqueness of he soluion of (2.2-(2.3 for any iniial condiion in L 2 (Ω, F a ime, one has Y = u(,x a.s. for each T. We shall prove in sep 2. below ha u is joinly coninuous in x and righ-coninuous in. This implies ha (u(,x T is righ-coninuous so ha Y = u(,x and Y = lim r u(r,x r for each T a.s., see Theorem I.2 in 22 and recall ha X and Y are càdlàg. Thus U (eµ(de, {} = Y Y = u(,x lim r u(r,x r = Ũ(eµ(de, {}, for each T a.s. and T Ũ(e U (e 2 µ(de,d =, which implies, by aking expecaion, Ũ(e U (e =. L 2 λ 2. We now prove ha u is coninuous in x and righ-coninuous on. Fix 1 2 T and (x 1,x 2 R 2d. For A denoing X,Y,Z or U, we se A i := A(,x i for i = 1,2 and δa := A 1 A 2. By (5.4 of Lemma 5.1, we derive δx 2 S 2 2,T C 2 { x1 x (1 + x }. (4.5 Plugging his esimae in (5.8 of Lemma 5.2 leads o (δy,δz,δu 2 B 2 2,T C 2 { x1 x (1 + x }. (4.6 Now, observe ha Y u( 1,x 1 u( 2,x 2 2 = Y 1 1 Y C 2 2 Y Y 1 2 Y (4.7. Since Y 1 is righ-coninuous and bounded in S 2, he firs erm on he righ-hand side goes o as 2 1, while he second is conrolled by ( Pah regulariy Proposiion 4.4 Assume ha C X and C Y hold. Then, for 1 2 T and (x 1,x 2 R 2d, u( 1,x 1 u( 2,x 2 2 C 2 { (1 + x x 1 x 2 2}. Proof. I suffices o plug he esimae of Proposiion 4.2 and (4.3 in (2.7, which is possible since he norms in (2.7 do no change afer passing o suiable versions, and appeal o (4.6 and (4.7. Remark 4.2 A similar resuls obained in 21 when λ has a finie suppor. The coninuiy of u is proved in 1 in a case where h is bounded, see also 9. 18

19 Corollary 4.1 Assume ha C X and C Y hold. Then, here is a version of U such ha for all s T sup Y r Y s 2 + sup sup U r (e U s (e 2 C 2 (1 + X 2 s. r s, e r s, Proof. Recall from he proof of Proposiion 4.3 ha Y = u(,x on,t. Thus, plugging (2.5 and (2.6 in Proposiion 4.4 gives he upper-bound on he quaniy sup r s, Y r Y s 2. The upper-bound on sup e sup r s, U r (e U s (e 2 is obained similarly by passing o he version of U given in Remark 4.1. Proposiion 4.5 Assume ha H-C X -C Y holds. Then here is a version of Z such ha, for all n 1, n 1 i+1 i= Z Z i 2 C2 n 1. Proof. 1. We denoe by x h (resp. y h, z h, γ h he gradien of h wih respec o is x variable ( (resp. y, z, γ. We firsnroduce he processes Λ and M defined by Λ := exp yh(θ r dr, M := 1+ M r z h(θ r dw r. Since h has bounded derivaives, i follows from Iô s Lemma and Proposiion 4.2 ha T Λ M Z = M T (Λ T g(x T χ T + ( x h(θ r χ r + γh(θ r Γ r Λr dr F By Remark 3.8 and Proposiion 3.3, we deduce ha ( T Λ M Z = M T Λ T g(x T X T + F r Λ r dr F ( X 1 σ(x where he process F is defined by F r := x h(θ r X r + γ h(θ r Γ r for r T. I follows ha { } Λ M Z = G F F r Λ r dr ( X 1 σ(x (4.8 ( where G := M T Λ T g(x T X T + T F r Λ r dr. By Remark 3.7 and (4.4, we deduce ha. G p Cp for all p 2. (4.9 Se m s := G F s and le ( ζ,ṽ H2 L 2 λ (wih values in Md R d be defined such ha m s = G T ζ s r dw r T s Ṽr(e µ(de,dr. Applying (4.9 and Lemma 5.2 o (m, ζ,ṽ implies ha (m, ζ,ṽ B p C p for all p 2. (4.1 Using C X, Remark 3.7, (4.4, (4.1, applying Lemma 5.1 o M 1 and using Iô s Lemma, we deduce from he las asserion ha Z := (ΛM (m 1 F r Λ r dr ( X 1 19

20 can be wrien as Z = Z + µ rdr + σ rdw r + β r (e µ(de,dr, where and µ, σ and β are adaped processes saisfying Z p S p C p for all p 2, (4.11 A p,t C p for all p 2 (4.12 where A p s, := µ p H p + σ p H p + β p L p, s T. s, s, λ,s, 2. Observe ha Z = Z σ(x P a.s. since he probabiliy of having a jump a ime s equal o zero. I follows ha, for all i n and,+1, and I 2 i, := σ(x σ(x i 2 Z 2. Ob- where I 1 i, := Z Z i 2 σ(x i 2 serving ha I 1, Z Z i 2 ( C 2 I 1 i, + I2, = Z Z i 2 F i σ(x i 2 ( i+1 C 2 µ r 2 + σ r 2 + (4.13 β r (e 2 λ(de dr σ(x i 2 we deduce from Hölder inequaliy, (2.5 and he linear growh assumpion on σ ha n 1 i+1 i= C 2 n 1 I 1,d ( T µ r 2 + σ r 2 + β r (e 2 λ(de dr sup σ(x 2 T C 2(A 4,T 1 2 n 1. (4.14 Using he Lipschiz coninuiy of σ, we obain I 2 i, C 2 X X i 2 Z 2. (4.15 Now observe ha for each k,l d (X k Xk i 2 ( Z ( l 2 C 2 Arguing as above, we obain n 1 i+1 i= ( Z l Z l 2 (X k 2 + (X k Z l Xk Zl i 2.(4.16 ( Z l Z l i 2 (X k i 2 ( C2 1 + (A 4,T 1 2 n 1. (4.17 Moreover, we deduce from he linear growh condiion on b, σ, β and (2.5, (4.11 and (4.12 ha X k Zl can be wrien as X k Z l = X k Z l + ˆµ kl r dr + ˆσ r kl dw r + 2 ˆβ kl r (e µ(de, dr,

21 where ˆµ kl, ˆσ kl and ˆβ kl are adaped processes saisfying ˆµ kl H 2+ ˆσ kl H 2+ ˆβ kl L 2 λ C2. I follows ha n 1 i+1 i= (X k Z l Xk Zl i 2 ( C 2 n 1 ˆµ kl 2 H + ˆσ kl 2 2 H + ˆβ kl 2 2 L 2 λ which combined wih (4.15, (4.16 and (4.17 leads o n 1 i+1 i= I 2,d C 2(1 + (A 4,T 1 2 n 1. (4.18 The proof is concluded by plugging (4.14-(4.18 in (4.13 and recalling (4.12. Proposiion 4.6 Assume ha C X -C Y holds. Then here is a version of Z such ha, for all ε > and n 1, n 1 i+1 i= Z Z i 2 Cε n 1+ε. Proof. We adap he argumens of 6. Le Λ and M be defined as in he proof of Proposiion 4.5 and recall ha, afer possibly passing o a suiable version, Z = I where, for s, T, T Is := MT (Λ T g(x T χ T Λ M + ( x h(θ r χ r + γh(θ r Γ r Λr dr F s For,+1, i n 1, we herefore have Z Z i 2 ( C 2 I I 2 + I I 2 where, by Remark 3.5, (5.8 below applied o (3.4, recall ha ρ is bounded, and sandard esimaions on ΛM, sup i n 1, i,+1 I I 2 C2 n 1. Thus i suffices o prove ha n 1 i+1 i= I I 2 C ε n 1+ε, where ε > is now fixed. To his purpose, we firs observe ha I is a maringale on,+1, which implies ha I I 2 I +1 2 I 2. (4.19 Remark ha n 1 i= I +1 2 I 2 = Z T 2 Z 2 + n i=1 I 1 2 I 2 which, combined wih Proposiion 4.2 and (4.19, leads o n 1 i+1 i= I I 2 = C 2n 1 (1 + n i=1 I 1 2 I 2.., To conclude he proof, i remains o show ha I 1 2 I 2 I 1 I I 1 + I C ε n 1+ε. 21

22 which follows from Hölder inequaliy, Remark 3.5 and Lemma 5.2 as above. We now complee he proof of Theorem 2.1. Proof of Theorem We sar wih (ii. We firs show ha (2.16 holds under H and C Y. We consider a Cb densiy q on R d wih compac suppor and se (b k,σ k,β k (,e(x = k d R d (b,σ,β(,e( xq (kx x d x. For large k N, hese funcions are bounded by 2K a. Moreover, hey are K-Lipschiz and C 1 b. Using he coninuiy of σ, one also easily checks ha σk is sill inverible. By H and Remark 2.4, for each e and x R d, I d + β k (x,e is inverible wih uniformly bounded inverse. We denoe by (X k,y k,z k,u k he soluion of (2.2-(2.3 wih (b,σ,β replaced by (b k,σ k,β k. Since (b k,σ k,β k converges poinwise o (b,σ,β, one easily deduces from Lemma 5.1 and Lemma 5.2 ha (X k,y k,z k,u k converges o (X,Y,Z,U in S 2 B 2. Since he resul of Proposiion 4.5 holds for (X k,y k,z k,u k uniformly in k, his shows ha (ii holds under H and C Y, recall Remark 2.3. We now prove ha (2.16 holds under H. Le (X,Y k,z k,u k be he soluion of (2.2-(2.3 wih h k insead of h, where h k is consruced by considering a sequence of molifiers as above. For large k, h k ( is bounded by 2K. By Lemma 5.2, (Y k,z k,u k converges o (Y,Z,U in S 2 B 2 which implies (ii by arguing as above. 2. Since ρ is bounded and λ( <, Corollary 4.1 and Proposiion 4.6 imply (2.14 and (2.15 under C X -C Y, recall Remark 2.3. Now observe ha sup Γ Γ i 2 2 sup Γ Γ i Γ i Γ i 2,+1,+1 where, by Jensen s inequaliy and he fac ha Γ i is F i -measurable, Γ i Γ i 2 n T i+1 2 (Γ i Γ s ds n T i+1 Γ i Γ s 2 ds. Thus, (2.14 implies Γ Γ 2 S C 2 2 n 1 and Γ Γ 2 H C 2 2 n 1. We conclude by he same approximaion argumen as above. 5 Appendix: A priori esimaes For sake of compleeness, we provide in his secion some a priori esimaes on soluions of forward and backward SD s wih jumps. In he following, we consider some measurable maps b i : Ω,T R d R d, σ i : Ω,T R d M d, β i : Ω,T R d R d and f i : Ω,T R 22

23 R d L 2 (,,λ; R, i = 1,2. Here L 2 (,,λ; R is endowed wih he naural norm ( a(e 2 λ(de 1 2. Omiing he dependence of hese maps wih respec o ω Ω, we assume ha for each T b i (,, σ i (,, β i (,,e and f i (, are a.s. K-Lipschiz coninuous uniformly in e for β i. We also assume ha ( f i (,, b i (, is F-progressively measurable, and ( σ i (,, β i (, is F-predicable, i = 1,2. Given some real number p 2, we assume ha b i (,, σ i (, and f i (, are in H p, and ha β i (,, is in L p λ. For 1 2 T, Xi L 2 (Ω, F i, P; R d for i = 1,2, we now denoe by X i he soluion on,t of X = X i + bi (s,xs i ds + σ i (s,xs i dw s + Lemma 5.1 β i (s,e,xs i µ(de,ds. (5.1 X 1 p S p 1,T C p X 1 p + b 1 (, p H p + σ 1 (, p H p + β 1 (,, p L p. 1,T 1,T λ, 1,T (5.2 Moreover, for all 1 s T, sup Xu 1 Xs 1 p s u C p A 1 p s, (5.3 where A 1 p is defined as X 1 p + sup b 1 (s, p + sup σ 1 (s, p + sup 1 s T 1 s T 1 s T { β 1 (s,,e p λ(de}, and, for 2 T, δx p S p C p ( X 1 X 2 p + A 1 p 2 1 2,T ( T p + C p ( δ b d + δ σ p H p + δ β p L p 2 2,T λ, 2,T (5.4 where δx := X 1 X 2, δ b = ( b 1 b 2 (,X 1 and δ σ, δ β are defined similarly. Lemma 5.2 (i Le f be equal o f 1 or f 2. Given Ỹ Lp (Ω, F T, P; R, he backward SD T T T Y = Ỹ + f(s,y s,z s,u s ds + Z s dw s + U s (e µ(de,ds (5.5 has a unique soluion (Y,Z,U in B 2. I saisfies ( T (Y,Z,U p B C p p Ỹ p + p f(, d. (5.6 23

24 Moreover, if A p := Ỹ p + sup T f(, p <, hen { } sup Y u Y s p C p A p s p + Z p H p + U p L p s u s, λ,s,. (5.7 (ii Fix Ỹ 1 and Ỹ 2 in L p (Ω, F T, P; R and le (Y i,z i,u i be he soluion of (5.6 wih (Ỹ i, f i in place of (Ỹ, f, i = 1,2. Then, for all T, ( T p (δy,δz,δu p B p C p δỹ p + δ f r dr (5.8,T where δỹ := Ỹ 1 Ỹ 2, δy := Y 1 Y 2, δz := Z 1 Z 2, δu := U 1 U 2 and δ f := ( f 1 f 2 (,Y 1,Z 1,U 1. The proof of he previous resuls is sandard, see e.g. 15 and 12. We refer o pages in 11 for more deails. The jump pars handled by using he following proposiion which plays he same role as he usual Burkholder-Davis- Gundy s Lemma. I follows from an inducion argumen already used in 4. The proof can be found in 11, see page 125. Proposiion 5.1 Given ψ L 2 λ, le M be defined by M = ψ s(e µ(ds,de on,t. Then, for all p 2, k p ψ p L p M p S p K p ψ p L p where k p, K p λ,,t,t λ,,t, are posiive numbers ha depend only on p, λ( and T. References 1 Barles G., R. Buckdahn and. Pardoux (1997. Backward sochasic differenial equaions and inegral-parial differenial equaions. Sochasics Sochasics Repors, 6, Bally V., and G. Pages (22. A quanizaion algorihm for solving discree ime mulidimensional opimal sopping problems. Bernoulli, 9 (6, Becherer D. (25. Bounded soluions o Backward SD s wih Jumps for Uiliy Opimizaion and Indifference Hedging. Preprin, Imperial College London. 4 Bicheler K. and J. Jacod (1983. Calcul de Malliavin pour des diffusions avec sau: exisence d une densié dans le cas unidimensionel. Séminaire de Probabilié, 17, Bicheler K., Gravereaux J.-B. and J. Jacod (1987. Malliavin calculus for processes wih jumps. Gordon and Breach Science Publishers, New York. 6 Bouchard B. and J.-F. Chassagneux (26. Discree ime approximaion for coninuously and discreely refleced BSD s. Preprin LPMA, Universiy Paris 6. 7 Bouchard B. and N. Touzi (24. Discree-Time Approximaion and Mone-Carlo Simulaion of Backward Sochasic Differenial quaions. Sochasic Processes and heir Applicaions, 111 (2, Brémaud P. (1981. Poin Processes and Queues - Maringale Dynamics, Springer- Verlag, New-York. 24

25 9 Buckdahn R. and. Pardoux (1994. BSD s wih jumps and associaed inegroparial differenial equaions. Preprin. 1 Douglas J. Jr., J. Ma, and P. Proer (1996. Numerical Mehods for Forward- Backward Sochasic Differenial quaions. Annals of Applied Probabiliy, 6, lie R. (26. Conrôle sochasique e méhodes numériques en finance mahémaique. PHD hesis, Universiy Paris-Dauphine. 12 l Karoui N., S. Peng and M.-C. Quenez (1997. Backward sochasic differenial equaions in finance. Mahemaical finance, 7 (1, yraud-loisel A. (25. Backward Sochasic Differenial quaions wih nlarged Filraion. Opion Hedging of an insider rader in a financial marke wih Jumps. Sochasic processes and heir Applicaions, 115 (11, Forser B.,. Lükebohmer and J. Teichmann (25. Calculaion of he greeks for jump-diffusions. Preprin. 15 Gikhman I. and A. Skorohod (1972. Sochasic differenial equaions. Springer. 16 Gobe., J.P. Lemor and X. Warin (25. Rae of convergence of empirical regression mehod for solving generalized BSD. To appear in Bernoulli. 17 Ma J., P. Proer, J. San Marin, and S. Torres (22. Numerical Mehod for Backward Sochasic Differenial quaions. Annals of Applied Probabiliy, 12 (1, Nualar D. (1995. The Malliavin Calculus and Relaed Topics. Springer Verlag, Berlin. 19 Nualar D. and. Pardoux (1988. Sochasic calculus wih anicipaing inegrands. Prob. Theory and Rel. Fields, 78, Pardoux. and S. Peng (1992. Backward sochasic differenial equaions and quasilinear parabolic parial differenial equaions. Lecure Noes in Conrol and Inform. Sci, 176, Pardoux., F. Pradeilles and Z. Rao (1997. Probabilisic inerpreaion for a sysem of semilinear parabolic parial differenial equaions. Ann. Ins. H. Poincare, 33 (4, Proer P. (199. Sochasic inegraion and differenial equaions. Springer Verlag, Berlin. 23 Tang S. and X. Li (1994. Necessary condiions for opimal conrol of sochasic sysems wih random jumps. SIAM J. Conrol Opim., 32 (5, Ma J. and Zhang J. (22. Pah Regulariy of Soluions o Backward Sochasic Differenial quaions. Probabiliy Theory and Relaed Fields, 122, Zhang J. (24. A numerical scheme for BSDs. Annals of Applied Probabiliy, 14 (1,

An Introduction to Malliavin calculus and its applications

An Introduction to Malliavin calculus and its applications An Inroducion o Malliavin calculus and is applicaions Lecure 5: Smoohness of he densiy and Hörmander s heorem David Nualar Deparmen of Mahemaics Kansas Universiy Universiy of Wyoming Summer School 214

More information

6. Stochastic calculus with jump processes

6. Stochastic calculus with jump processes A) Trading sraegies (1/3) Marke wih d asses S = (S 1,, S d ) A rading sraegy can be modelled wih a vecor φ describing he quaniies invesed in each asse a each insan : φ = (φ 1,, φ d ) The value a of a porfolio

More information

Utility maximization in incomplete markets

Utility maximization in incomplete markets Uiliy maximizaion in incomplee markes Marcel Ladkau 27.1.29 Conens 1 Inroducion and general seings 2 1.1 Marke model....................................... 2 1.2 Trading sraegy.....................................

More information

An Introduction to Backward Stochastic Differential Equations (BSDEs) PIMS Summer School 2016 in Mathematical Finance.

An Introduction to Backward Stochastic Differential Equations (BSDEs) PIMS Summer School 2016 in Mathematical Finance. 1 An Inroducion o Backward Sochasic Differenial Equaions (BSDEs) PIMS Summer School 2016 in Mahemaical Finance June 25, 2016 Chrisoph Frei cfrei@ualbera.ca This inroducion is based on Touzi [14], Bouchard

More information

Simulation of BSDEs and. Wiener Chaos Expansions

Simulation of BSDEs and. Wiener Chaos Expansions Simulaion of BSDEs and Wiener Chaos Expansions Philippe Briand Céline Labar LAMA UMR 5127, Universié de Savoie, France hp://www.lama.univ-savoie.fr/ Workshop on BSDEs Rennes, May 22-24, 213 Inroducion

More information

On a Fractional Stochastic Landau-Ginzburg Equation

On a Fractional Stochastic Landau-Ginzburg Equation Applied Mahemaical Sciences, Vol. 4, 1, no. 7, 317-35 On a Fracional Sochasic Landau-Ginzburg Equaion Nguyen Tien Dung Deparmen of Mahemaics, FPT Universiy 15B Pham Hung Sree, Hanoi, Vienam dungn@fp.edu.vn

More information

arxiv: v1 [math.pr] 19 Feb 2011

arxiv: v1 [math.pr] 19 Feb 2011 A NOTE ON FELLER SEMIGROUPS AND RESOLVENTS VADIM KOSTRYKIN, JÜRGEN POTTHOFF, AND ROBERT SCHRADER ABSTRACT. Various equivalen condiions for a semigroup or a resolven generaed by a Markov process o be of

More information

Lecture 20: Riccati Equations and Least Squares Feedback Control

Lecture 20: Riccati Equations and Least Squares Feedback Control 34-5 LINEAR SYSTEMS Lecure : Riccai Equaions and Leas Squares Feedback Conrol 5.6.4 Sae Feedback via Riccai Equaions A recursive approach in generaing he marix-valued funcion W ( ) equaion for i for he

More information

Simulation of BSDEs and. Wiener Chaos Expansions

Simulation of BSDEs and. Wiener Chaos Expansions Simulaion of BSDEs and Wiener Chaos Expansions Philippe Briand Céline Labar LAMA UMR 5127, Universié de Savoie, France hp://www.lama.univ-savoie.fr/ Sochasic Analysis Seminar Oxford, June 1, 213 Inroducion

More information

Dual Representation as Stochastic Differential Games of Backward Stochastic Differential Equations and Dynamic Evaluations

Dual Representation as Stochastic Differential Games of Backward Stochastic Differential Equations and Dynamic Evaluations arxiv:mah/0602323v1 [mah.pr] 15 Feb 2006 Dual Represenaion as Sochasic Differenial Games of Backward Sochasic Differenial Equaions and Dynamic Evaluaions Shanjian Tang Absrac In his Noe, assuming ha he

More information

Generalized Snell envelope and BSDE With Two general Reflecting Barriers

Generalized Snell envelope and BSDE With Two general Reflecting Barriers 1/22 Generalized Snell envelope and BSDE Wih Two general Reflecing Barriers EL HASSAN ESSAKY Cadi ayyad Universiy Poly-disciplinary Faculy Safi Work in progress wih : M. Hassani and Y. Ouknine Iasi, July

More information

Convergence of the Neumann series in higher norms

Convergence of the Neumann series in higher norms Convergence of he Neumann series in higher norms Charles L. Epsein Deparmen of Mahemaics, Universiy of Pennsylvania Version 1.0 Augus 1, 003 Absrac Naural condiions on an operaor A are given so ha he Neumann

More information

Finish reading Chapter 2 of Spivak, rereading earlier sections as necessary. handout and fill in some missing details!

Finish reading Chapter 2 of Spivak, rereading earlier sections as necessary. handout and fill in some missing details! MAT 257, Handou 6: Ocober 7-2, 20. I. Assignmen. Finish reading Chaper 2 of Spiva, rereading earlier secions as necessary. handou and fill in some missing deails! II. Higher derivaives. Also, read his

More information

Cash Flow Valuation Mode Lin Discrete Time

Cash Flow Valuation Mode Lin Discrete Time IOSR Journal of Mahemaics (IOSR-JM) e-issn: 2278-5728,p-ISSN: 2319-765X, 6, Issue 6 (May. - Jun. 2013), PP 35-41 Cash Flow Valuaion Mode Lin Discree Time Olayiwola. M. A. and Oni, N. O. Deparmen of Mahemaics

More information

Oscillation of an Euler Cauchy Dynamic Equation S. Huff, G. Olumolode, N. Pennington, and A. Peterson

Oscillation of an Euler Cauchy Dynamic Equation S. Huff, G. Olumolode, N. Pennington, and A. Peterson PROCEEDINGS OF THE FOURTH INTERNATIONAL CONFERENCE ON DYNAMICAL SYSTEMS AND DIFFERENTIAL EQUATIONS May 4 7, 00, Wilmingon, NC, USA pp 0 Oscillaion of an Euler Cauchy Dynamic Equaion S Huff, G Olumolode,

More information

Uniqueness of solutions to quadratic BSDEs. BSDEs with convex generators and unbounded terminal conditions

Uniqueness of solutions to quadratic BSDEs. BSDEs with convex generators and unbounded terminal conditions Recalls and basic resuls on BSDEs Uniqueness resul Links wih PDEs On he uniqueness of soluions o quadraic BSDEs wih convex generaors and unbounded erminal condiions IRMAR, Universié Rennes 1 Châeau de

More information

A proof of Ito's formula using a di Title formula. Author(s) Fujita, Takahiko; Kawanishi, Yasuhi. Studia scientiarum mathematicarum H Citation

A proof of Ito's formula using a di Title formula. Author(s) Fujita, Takahiko; Kawanishi, Yasuhi. Studia scientiarum mathematicarum H Citation A proof of Io's formula using a di Tile formula Auhor(s) Fujia, Takahiko; Kawanishi, Yasuhi Sudia scieniarum mahemaicarum H Ciaion 15-134 Issue 8-3 Dae Type Journal Aricle Tex Version auhor URL hp://hdl.handle.ne/186/15878

More information

Time discretization of quadratic and superquadratic Markovian BSDEs with unbounded terminal conditions

Time discretization of quadratic and superquadratic Markovian BSDEs with unbounded terminal conditions Time discreizaion of quadraic and superquadraic Markovian BSDEs wih unbounded erminal condiions Adrien Richou Universié Bordeaux 1, INRIA équipe ALEA Oxford framework Le (Ω, F, P) be a probabiliy space,

More information

Singular control of SPDEs and backward stochastic partial diffe. reflection

Singular control of SPDEs and backward stochastic partial diffe. reflection Singular conrol of SPDEs and backward sochasic parial differenial equaions wih reflecion Universiy of Mancheser Join work wih Bern Øksendal and Agnès Sulem Singular conrol of SPDEs and backward sochasic

More information

t is a basis for the solution space to this system, then the matrix having these solutions as columns, t x 1 t, x 2 t,... x n t x 2 t...

t is a basis for the solution space to this system, then the matrix having these solutions as columns, t x 1 t, x 2 t,... x n t x 2 t... Mah 228- Fri Mar 24 5.6 Marix exponenials and linear sysems: The analogy beween firs order sysems of linear differenial equaions (Chaper 5) and scalar linear differenial equaions (Chaper ) is much sronger

More information

Backward stochastic dynamics on a filtered probability space

Backward stochastic dynamics on a filtered probability space Backward sochasic dynamics on a filered probabiliy space Gechun Liang Oxford-Man Insiue, Universiy of Oxford based on join work wih Terry Lyons and Zhongmin Qian Page 1 of 15 gliang@oxford-man.ox.ac.uk

More information

Approximation of backward stochastic variational inequalities

Approximation of backward stochastic variational inequalities Al. I. Cuza Universiy of Iaşi, România 10ème Colloque Franco-Roumain de Mahémaiques Appliquées Augus 27, 2010, Poiiers, France Shor hisory & moivaion Re eced Sochasic Di erenial Equaions were rs sudied

More information

A FAMILY OF MARTINGALES GENERATED BY A PROCESS WITH INDEPENDENT INCREMENTS

A FAMILY OF MARTINGALES GENERATED BY A PROCESS WITH INDEPENDENT INCREMENTS Theory of Sochasic Processes Vol. 14 3), no. 2, 28, pp. 139 144 UDC 519.21 JOSEP LLUÍS SOLÉ AND FREDERIC UTZET A FAMILY OF MARTINGALES GENERATED BY A PROCESS WITH INDEPENDENT INCREMENTS An explici procedure

More information

LECTURE 1: GENERALIZED RAY KNIGHT THEOREM FOR FINITE MARKOV CHAINS

LECTURE 1: GENERALIZED RAY KNIGHT THEOREM FOR FINITE MARKOV CHAINS LECTURE : GENERALIZED RAY KNIGHT THEOREM FOR FINITE MARKOV CHAINS We will work wih a coninuous ime reversible Markov chain X on a finie conneced sae space, wih generaor Lf(x = y q x,yf(y. (Recall ha q

More information

t 2 B F x,t n dsdt t u x,t dxdt

t 2 B F x,t n dsdt t u x,t dxdt Evoluion Equaions For 0, fixed, le U U0, where U denoes a bounded open se in R n.suppose ha U is filled wih a maerial in which a conaminan is being ranspored by various means including diffusion and convecion.

More information

arxiv: v1 [math.pr] 28 Nov 2016

arxiv: v1 [math.pr] 28 Nov 2016 Backward Sochasic Differenial Equaions wih Nonmarkovian Singular Terminal Values Ali Devin Sezer, Thomas Kruse, Alexandre Popier Ocober 15, 2018 arxiv:1611.09022v1 mah.pr 28 Nov 2016 Absrac We solve a

More information

Heat kernel and Harnack inequality on Riemannian manifolds

Heat kernel and Harnack inequality on Riemannian manifolds Hea kernel and Harnack inequaliy on Riemannian manifolds Alexander Grigor yan UHK 11/02/2014 onens 1 Laplace operaor and hea kernel 1 2 Uniform Faber-Krahn inequaliy 3 3 Gaussian upper bounds 4 4 ean-value

More information

Optimal Control versus Stochastic Target problems: An Equivalence Result

Optimal Control versus Stochastic Target problems: An Equivalence Result Opimal Conrol versus Sochasic Targe problems: An quivalence Resul Bruno Bouchard CRMAD, Universié Paris Dauphine and CRST-NSA bouchard@ceremade.dauphine.fr Ngoc Minh Dang CRMAD, Universié Paris Dauphine

More information

T L. t=1. Proof of Lemma 1. Using the marginal cost accounting in Equation(4) and standard arguments. t )+Π RB. t )+K 1(Q RB

T L. t=1. Proof of Lemma 1. Using the marginal cost accounting in Equation(4) and standard arguments. t )+Π RB. t )+K 1(Q RB Elecronic Companion EC.1. Proofs of Technical Lemmas and Theorems LEMMA 1. Le C(RB) be he oal cos incurred by he RB policy. Then we have, T L E[C(RB)] 3 E[Z RB ]. (EC.1) Proof of Lemma 1. Using he marginal

More information

Existence and uniqueness of solution for multidimensional BSDE with local conditions on the coefficient

Existence and uniqueness of solution for multidimensional BSDE with local conditions on the coefficient 1/34 Exisence and uniqueness of soluion for mulidimensional BSDE wih local condiions on he coefficien EL HASSAN ESSAKY Cadi Ayyad Universiy Mulidisciplinary Faculy Safi, Morocco ITN Roscof, March 18-23,

More information

Chapter 6. Systems of First Order Linear Differential Equations

Chapter 6. Systems of First Order Linear Differential Equations Chaper 6 Sysems of Firs Order Linear Differenial Equaions We will only discuss firs order sysems However higher order sysems may be made ino firs order sysems by a rick shown below We will have a sligh

More information

arxiv: v1 [math.pr] 18 Feb 2015

arxiv: v1 [math.pr] 18 Feb 2015 Non-Markovian opimal sopping problems and consrained BSDEs wih jump arxiv:152.5422v1 [mah.pr 18 Feb 215 Marco Fuhrman Poliecnico di Milano, Diparimeno di Maemaica via Bonardi 9, 2133 Milano, Ialy marco.fuhrman@polimi.i

More information

Hamilton- J acobi Equation: Explicit Formulas In this lecture we try to apply the method of characteristics to the Hamilton-Jacobi equation: u t

Hamilton- J acobi Equation: Explicit Formulas In this lecture we try to apply the method of characteristics to the Hamilton-Jacobi equation: u t M ah 5 2 7 Fall 2 0 0 9 L ecure 1 0 O c. 7, 2 0 0 9 Hamilon- J acobi Equaion: Explici Formulas In his lecure we ry o apply he mehod of characerisics o he Hamilon-Jacobi equaion: u + H D u, x = 0 in R n

More information

ODEs II, Lecture 1: Homogeneous Linear Systems - I. Mike Raugh 1. March 8, 2004

ODEs II, Lecture 1: Homogeneous Linear Systems - I. Mike Raugh 1. March 8, 2004 ODEs II, Lecure : Homogeneous Linear Sysems - I Mike Raugh March 8, 4 Inroducion. In he firs lecure we discussed a sysem of linear ODEs for modeling he excreion of lead from he human body, saw how o ransform

More information

Lecture 10: The Poincaré Inequality in Euclidean space

Lecture 10: The Poincaré Inequality in Euclidean space Deparmens of Mahemaics Monana Sae Universiy Fall 215 Prof. Kevin Wildrick n inroducion o non-smooh analysis and geomery Lecure 1: The Poincaré Inequaliy in Euclidean space 1. Wha is he Poincaré inequaliy?

More information

The Asymptotic Behavior of Nonoscillatory Solutions of Some Nonlinear Dynamic Equations on Time Scales

The Asymptotic Behavior of Nonoscillatory Solutions of Some Nonlinear Dynamic Equations on Time Scales Advances in Dynamical Sysems and Applicaions. ISSN 0973-5321 Volume 1 Number 1 (2006, pp. 103 112 c Research India Publicaions hp://www.ripublicaion.com/adsa.hm The Asympoic Behavior of Nonoscillaory Soluions

More information

Ann. Funct. Anal. 2 (2011), no. 2, A nnals of F unctional A nalysis ISSN: (electronic) URL:

Ann. Funct. Anal. 2 (2011), no. 2, A nnals of F unctional A nalysis ISSN: (electronic) URL: Ann. Func. Anal. 2 2011, no. 2, 34 41 A nnals of F uncional A nalysis ISSN: 2008-8752 elecronic URL: www.emis.de/journals/afa/ CLASSIFICAION OF POSIIVE SOLUIONS OF NONLINEAR SYSEMS OF VOLERRA INEGRAL EQUAIONS

More information

Online Appendix to Solution Methods for Models with Rare Disasters

Online Appendix to Solution Methods for Models with Rare Disasters Online Appendix o Soluion Mehods for Models wih Rare Disasers Jesús Fernández-Villaverde and Oren Levinal In his Online Appendix, we presen he Euler condiions of he model, we develop he pricing Calvo block,

More information

Chapter 3 Boundary Value Problem

Chapter 3 Boundary Value Problem Chaper 3 Boundary Value Problem A boundary value problem (BVP) is a problem, ypically an ODE or a PDE, which has values assigned on he physical boundary of he domain in which he problem is specified. Le

More information

Representation of Stochastic Process by Means of Stochastic Integrals

Representation of Stochastic Process by Means of Stochastic Integrals Inernaional Journal of Mahemaics Research. ISSN 0976-5840 Volume 5, Number 4 (2013), pp. 385-397 Inernaional Research Publicaion House hp://www.irphouse.com Represenaion of Sochasic Process by Means of

More information

Asymptotic instability of nonlinear differential equations

Asymptotic instability of nonlinear differential equations Elecronic Journal of Differenial Equaions, Vol. 1997(1997), No. 16, pp. 1 7. ISSN: 172-6691. URL: hp://ejde.mah.sw.edu or hp://ejde.mah.un.edu fp (login: fp) 147.26.13.11 or 129.12.3.113 Asympoic insabiliy

More information

Course Notes for EE227C (Spring 2018): Convex Optimization and Approximation

Course Notes for EE227C (Spring 2018): Convex Optimization and Approximation Course Noes for EE7C Spring 018: Convex Opimizaion and Approximaion Insrucor: Moriz Hard Email: hard+ee7c@berkeley.edu Graduae Insrucor: Max Simchowiz Email: msimchow+ee7c@berkeley.edu Ocober 15, 018 3

More information

On R d -valued peacocks

On R d -valued peacocks On R d -valued peacocks Francis HIRSCH 1), Bernard ROYNETTE 2) July 26, 211 1) Laboraoire d Analyse e Probabiliés, Universié d Évry - Val d Essonne, Boulevard F. Mierrand, F-9125 Évry Cedex e-mail: francis.hirsch@univ-evry.fr

More information

Quadratic and Superquadratic BSDEs and Related PDEs

Quadratic and Superquadratic BSDEs and Related PDEs Quadraic and Superquadraic BSDEs and Relaed PDEs Ying Hu IRMAR, Universié Rennes 1, FRANCE hp://perso.univ-rennes1.fr/ying.hu/ ITN Marie Curie Workshop "Sochasic Conrol and Finance" Roscoff, March 21 Ying

More information

An introduction to the theory of SDDP algorithm

An introduction to the theory of SDDP algorithm An inroducion o he heory of SDDP algorihm V. Leclère (ENPC) Augus 1, 2014 V. Leclère Inroducion o SDDP Augus 1, 2014 1 / 21 Inroducion Large scale sochasic problem are hard o solve. Two ways of aacking

More information

EXISTENCE AND UNIQUENESS OF SOLUTIONS TO THE BACKWARD STOCHASTIC LORENZ SYSTEM

EXISTENCE AND UNIQUENESS OF SOLUTIONS TO THE BACKWARD STOCHASTIC LORENZ SYSTEM Communicaions on Sochasic Analysis Vol. 1, No. 3 (27) 473-483 EXISTENCE AND UNIQUENESS OF SOLUTIONS TO THE BACKWARD STOCHASTIC LORENZ SYSTEM P. SUNDAR AND HONG YIN Absrac. The backward sochasic Lorenz

More information

Backward doubly stochastic di erential equations with quadratic growth and applications to quasilinear SPDEs

Backward doubly stochastic di erential equations with quadratic growth and applications to quasilinear SPDEs Backward doubly sochasic di erenial equaions wih quadraic growh and applicaions o quasilinear SPDEs Badreddine MANSOURI (wih K. Bahlali & B. Mezerdi) Universiy of Biskra Algeria La Londe 14 sepember 2007

More information

MODULE 3 FUNCTION OF A RANDOM VARIABLE AND ITS DISTRIBUTION LECTURES PROBABILITY DISTRIBUTION OF A FUNCTION OF A RANDOM VARIABLE

MODULE 3 FUNCTION OF A RANDOM VARIABLE AND ITS DISTRIBUTION LECTURES PROBABILITY DISTRIBUTION OF A FUNCTION OF A RANDOM VARIABLE Topics MODULE 3 FUNCTION OF A RANDOM VARIABLE AND ITS DISTRIBUTION LECTURES 2-6 3. FUNCTION OF A RANDOM VARIABLE 3.2 PROBABILITY DISTRIBUTION OF A FUNCTION OF A RANDOM VARIABLE 3.3 EXPECTATION AND MOMENTS

More information

arxiv: v1 [math.fa] 9 Dec 2018

arxiv: v1 [math.fa] 9 Dec 2018 AN INVERSE FUNCTION THEOREM CONVERSE arxiv:1812.03561v1 [mah.fa] 9 Dec 2018 JIMMIE LAWSON Absrac. We esablish he following converse of he well-known inverse funcion heorem. Le g : U V and f : V U be inverse

More information

L p -L q -Time decay estimate for solution of the Cauchy problem for hyperbolic partial differential equations of linear thermoelasticity

L p -L q -Time decay estimate for solution of the Cauchy problem for hyperbolic partial differential equations of linear thermoelasticity ANNALES POLONICI MATHEMATICI LIV.2 99) L p -L q -Time decay esimae for soluion of he Cauchy problem for hyperbolic parial differenial equaions of linear hermoelasiciy by Jerzy Gawinecki Warszawa) Absrac.

More information

Couplage du principe des grandes déviations et de l homogénisation dans le cas des EDP paraboliques: (le cas constant)

Couplage du principe des grandes déviations et de l homogénisation dans le cas des EDP paraboliques: (le cas constant) Couplage du principe des grandes déviaions e de l homogénisaion dans le cas des EDP paraboliques: (le cas consan) Alioune COULIBALY U.F.R Sciences e Technologie Universié Assane SECK de Ziguinchor Probabilié

More information

Quasi-sure Stochastic Analysis through Aggregation

Quasi-sure Stochastic Analysis through Aggregation E l e c r o n i c J o u r n a l o f P r o b a b i l i y Vol. 16 (211), Paper no. 67, pages 1844 1879. Journal URL hp://www.mah.washingon.edu/~ejpecp/ Quasi-sure Sochasic Analysis hrough Aggregaion H. Mee

More information

f(s)dw Solution 1. Approximate f by piece-wise constant left-continuous non-random functions f n such that (f(s) f n (s)) 2 ds 0.

f(s)dw Solution 1. Approximate f by piece-wise constant left-continuous non-random functions f n such that (f(s) f n (s)) 2 ds 0. Advanced Financial Models Example shee 3 - Michaelmas 217 Michael Tehranchi Problem 1. Le f : [, R be a coninuous (non-random funcion and W a Brownian moion, and le σ 2 = f(s 2 ds and assume σ 2

More information

Hamilton- J acobi Equation: Weak S olution We continue the study of the Hamilton-Jacobi equation:

Hamilton- J acobi Equation: Weak S olution We continue the study of the Hamilton-Jacobi equation: M ah 5 7 Fall 9 L ecure O c. 4, 9 ) Hamilon- J acobi Equaion: Weak S oluion We coninue he sudy of he Hamilon-Jacobi equaion: We have shown ha u + H D u) = R n, ) ; u = g R n { = }. ). In general we canno

More information

Class Meeting # 10: Introduction to the Wave Equation

Class Meeting # 10: Introduction to the Wave Equation MATH 8.5 COURSE NOTES - CLASS MEETING # 0 8.5 Inroducion o PDEs, Fall 0 Professor: Jared Speck Class Meeing # 0: Inroducion o he Wave Equaion. Wha is he wave equaion? The sandard wave equaion for a funcion

More information

Matrix Versions of Some Refinements of the Arithmetic-Geometric Mean Inequality

Matrix Versions of Some Refinements of the Arithmetic-Geometric Mean Inequality Marix Versions of Some Refinemens of he Arihmeic-Geomeric Mean Inequaliy Bao Qi Feng and Andrew Tonge Absrac. We esablish marix versions of refinemens due o Alzer ], Carwrigh and Field 4], and Mercer 5]

More information

Differential Equations

Differential Equations Mah 21 (Fall 29) Differenial Equaions Soluion #3 1. Find he paricular soluion of he following differenial equaion by variaion of parameer (a) y + y = csc (b) 2 y + y y = ln, > Soluion: (a) The corresponding

More information

Example on p. 157

Example on p. 157 Example 2.5.3. Le where BV [, 1] = Example 2.5.3. on p. 157 { g : [, 1] C g() =, g() = g( + ) [, 1), var (g) = sup g( j+1 ) g( j ) he supremum is aken over all he pariions of [, 1] (1) : = < 1 < < n =

More information

Backward Stochastic Differential Equations with Nonmarkovian Singular Terminal Values

Backward Stochastic Differential Equations with Nonmarkovian Singular Terminal Values Backward Sochasic Differenial Equaions wih Nonmarkovian Singular Terminal Values Ali Sezer, Thomas Kruse, Alexandre Popier, Ali Sezer To cie his version: Ali Sezer, Thomas Kruse, Alexandre Popier, Ali

More information

DISCRETE GRONWALL LEMMA AND APPLICATIONS

DISCRETE GRONWALL LEMMA AND APPLICATIONS DISCRETE GRONWALL LEMMA AND APPLICATIONS JOHN M. HOLTE MAA NORTH CENTRAL SECTION MEETING AT UND 24 OCTOBER 29 Gronwall s lemma saes an inequaliy ha is useful in he heory of differenial equaions. Here is

More information

23.2. Representing Periodic Functions by Fourier Series. Introduction. Prerequisites. Learning Outcomes

23.2. Representing Periodic Functions by Fourier Series. Introduction. Prerequisites. Learning Outcomes Represening Periodic Funcions by Fourier Series 3. Inroducion In his Secion we show how a periodic funcion can be expressed as a series of sines and cosines. We begin by obaining some sandard inegrals

More information

arxiv: v1 [math.ca] 15 Nov 2016

arxiv: v1 [math.ca] 15 Nov 2016 arxiv:6.599v [mah.ca] 5 Nov 26 Counerexamples on Jumarie s hree basic fracional calculus formulae for non-differeniable coninuous funcions Cheng-shi Liu Deparmen of Mahemaics Norheas Peroleum Universiy

More information

Undetermined coefficients for local fractional differential equations

Undetermined coefficients for local fractional differential equations Available online a www.isr-publicaions.com/jmcs J. Mah. Compuer Sci. 16 (2016), 140 146 Research Aricle Undeermined coefficiens for local fracional differenial equaions Roshdi Khalil a,, Mohammed Al Horani

More information

Generalized Chebyshev polynomials

Generalized Chebyshev polynomials Generalized Chebyshev polynomials Clemene Cesarano Faculy of Engineering, Inernaional Telemaic Universiy UNINETTUNO Corso Viorio Emanuele II, 39 86 Roma, Ialy email: c.cesarano@unineunouniversiy.ne ABSTRACT

More information

Local Strict Comparison Theorem and Converse Comparison Theorems for Reflected Backward Stochastic Differential Equations

Local Strict Comparison Theorem and Converse Comparison Theorems for Reflected Backward Stochastic Differential Equations arxiv:mah/07002v [mah.pr] 3 Dec 2006 Local Sric Comparison Theorem and Converse Comparison Theorems for Refleced Backward Sochasic Differenial Equaions Juan Li and Shanjian Tang Absrac A local sric comparison

More information

THE WAVE EQUATION. part hand-in for week 9 b. Any dilation v(x, t) = u(λx, λt) of u(x, t) is also a solution (where λ is constant).

THE WAVE EQUATION. part hand-in for week 9 b. Any dilation v(x, t) = u(λx, λt) of u(x, t) is also a solution (where λ is constant). THE WAVE EQUATION 43. (S) Le u(x, ) be a soluion of he wave equaion u u xx = 0. Show ha Q43(a) (c) is a. Any ranslaion v(x, ) = u(x + x 0, + 0 ) of u(x, ) is also a soluion (where x 0, 0 are consans).

More information

4 Sequences of measurable functions

4 Sequences of measurable functions 4 Sequences of measurable funcions 1. Le (Ω, A, µ) be a measure space (complee, afer a possible applicaion of he compleion heorem). In his chaper we invesigae relaions beween various (nonequivalen) convergences

More information

Average Number of Lattice Points in a Disk

Average Number of Lattice Points in a Disk Average Number of Laice Poins in a Disk Sujay Jayakar Rober S. Sricharz Absrac The difference beween he number of laice poins in a disk of radius /π and he area of he disk /4π is equal o he error in he

More information

14 Autoregressive Moving Average Models

14 Autoregressive Moving Average Models 14 Auoregressive Moving Average Models In his chaper an imporan parameric family of saionary ime series is inroduced, he family of he auoregressive moving average, or ARMA, processes. For a large class

More information

GMM - Generalized Method of Moments

GMM - Generalized Method of Moments GMM - Generalized Mehod of Momens Conens GMM esimaion, shor inroducion 2 GMM inuiion: Maching momens 2 3 General overview of GMM esimaion. 3 3. Weighing marix...........................................

More information

Monotonic Solutions of a Class of Quadratic Singular Integral Equations of Volterra type

Monotonic Solutions of a Class of Quadratic Singular Integral Equations of Volterra type In. J. Conemp. Mah. Sci., Vol. 2, 27, no. 2, 89-2 Monoonic Soluions of a Class of Quadraic Singular Inegral Equaions of Volerra ype Mahmoud M. El Borai Deparmen of Mahemaics, Faculy of Science, Alexandria

More information

MATH 4330/5330, Fourier Analysis Section 6, Proof of Fourier s Theorem for Pointwise Convergence

MATH 4330/5330, Fourier Analysis Section 6, Proof of Fourier s Theorem for Pointwise Convergence MATH 433/533, Fourier Analysis Secion 6, Proof of Fourier s Theorem for Poinwise Convergence Firs, some commens abou inegraing periodic funcions. If g is a periodic funcion, g(x + ) g(x) for all real x,

More information

Weak error analysis via functional Itô calculus

Weak error analysis via functional Itô calculus Weak error analysis via funcional Iô calculus Mihály Kovács and Felix Lindner arxiv:163.8756v2 [mah.pr] 14 Jun 216 Absrac We consider auonomous sochasic ordinary differenial equaions SDEs and weak approximaions

More information

Chapter 2. First Order Scalar Equations

Chapter 2. First Order Scalar Equations Chaper. Firs Order Scalar Equaions We sar our sudy of differenial equaions in he same way he pioneers in his field did. We show paricular echniques o solve paricular ypes of firs order differenial equaions.

More information

Lecture 6: Wiener Process

Lecture 6: Wiener Process Lecure 6: Wiener Process Eric Vanden-Eijnden Chapers 6, 7 and 8 offer a (very) brief inroducion o sochasic analysis. These lecures are based in par on a book projec wih Weinan E. A sandard reference for

More information

2. Nonlinear Conservation Law Equations

2. Nonlinear Conservation Law Equations . Nonlinear Conservaion Law Equaions One of he clear lessons learned over recen years in sudying nonlinear parial differenial equaions is ha i is generally no wise o ry o aack a general class of nonlinear

More information

IMPLICIT AND INVERSE FUNCTION THEOREMS PAUL SCHRIMPF 1 OCTOBER 25, 2013

IMPLICIT AND INVERSE FUNCTION THEOREMS PAUL SCHRIMPF 1 OCTOBER 25, 2013 IMPLICI AND INVERSE FUNCION HEOREMS PAUL SCHRIMPF 1 OCOBER 25, 213 UNIVERSIY OF BRIISH COLUMBIA ECONOMICS 526 We have exensively sudied how o solve sysems of linear equaions. We know how o check wheher

More information

Feynman-Kac representation for Hamilton-Jacobi-Bellman IPDE

Feynman-Kac representation for Hamilton-Jacobi-Bellman IPDE Feynman-Kac represenaion for Hamilon-Jacobi-Bellman IPDE Idris KHRROUBI 1), Huyên PHM 2) December 11, 2012 revised version: November 27, 2013 1) CEREMDE, CNRS, UMR 7534 2) Laboraoire de Probabiliés e Universié

More information

Solutions from Chapter 9.1 and 9.2

Solutions from Chapter 9.1 and 9.2 Soluions from Chaper 9 and 92 Secion 9 Problem # This basically boils down o an exercise in he chain rule from calculus We are looking for soluions of he form: u( x) = f( k x c) where k x R 3 and k is

More information

and Applications Alexander Steinicke University of Graz Vienna Seminar in Mathematical Finance and Probability,

and Applications Alexander Steinicke University of Graz Vienna Seminar in Mathematical Finance and Probability, Backward Sochasic Differenial Equaions and Applicaions Alexander Seinicke Universiy of Graz Vienna Seminar in Mahemaical Finance and Probabiliy, 6-20-2017 1 / 31 1 Wha is a BSDE? SDEs - he differenial

More information

PENALIZED LEAST SQUARES AND PENALIZED LIKELIHOOD

PENALIZED LEAST SQUARES AND PENALIZED LIKELIHOOD PENALIZED LEAST SQUARES AND PENALIZED LIKELIHOOD HAN XIAO 1. Penalized Leas Squares Lasso solves he following opimizaion problem, ˆβ lasso = arg max β R p+1 1 N y i β 0 N x ij β j β j (1.1) for some 0.

More information

Hamilton Jacobi equations

Hamilton Jacobi equations Hamilon Jacobi equaions Inoducion o PDE The rigorous suff from Evans, mosly. We discuss firs u + H( u = 0, (1 where H(p is convex, and superlinear a infiniy, H(p lim p p = + This by comes by inegraion

More information

Mean-Variance Hedging for General Claims

Mean-Variance Hedging for General Claims Projekbereich B Discussion Paper No. B 167 Mean-Variance Hedging for General Claims by Marin Schweizer ) Ocober 199 ) Financial suppor by Deusche Forschungsgemeinschaf, Sonderforschungsbereich 33 a he

More information

Existence of positive solution for a third-order three-point BVP with sign-changing Green s function

Existence of positive solution for a third-order three-point BVP with sign-changing Green s function Elecronic Journal of Qualiaive Theory of Differenial Equaions 13, No. 3, 1-11; hp://www.mah.u-szeged.hu/ejqde/ Exisence of posiive soluion for a hird-order hree-poin BVP wih sign-changing Green s funcion

More information

Variational Iteration Method for Solving System of Fractional Order Ordinary Differential Equations

Variational Iteration Method for Solving System of Fractional Order Ordinary Differential Equations IOSR Journal of Mahemaics (IOSR-JM) e-issn: 2278-5728, p-issn: 2319-765X. Volume 1, Issue 6 Ver. II (Nov - Dec. 214), PP 48-54 Variaional Ieraion Mehod for Solving Sysem of Fracional Order Ordinary Differenial

More information

Robust estimation based on the first- and third-moment restrictions of the power transformation model

Robust estimation based on the first- and third-moment restrictions of the power transformation model h Inernaional Congress on Modelling and Simulaion, Adelaide, Ausralia, 6 December 3 www.mssanz.org.au/modsim3 Robus esimaion based on he firs- and hird-momen resricions of he power ransformaion Nawaa,

More information

The expectation value of the field operator.

The expectation value of the field operator. The expecaion value of he field operaor. Dan Solomon Universiy of Illinois Chicago, IL dsolom@uic.edu June, 04 Absrac. Much of he mahemaical developmen of quanum field heory has been in suppor of deermining

More information

A class of multidimensional quadratic BSDEs

A class of multidimensional quadratic BSDEs A class of mulidimensional quadraic SDEs Zhongmin Qian, Yimin Yang Shujin Wu March 4, 07 arxiv:703.0453v mah.p] Mar 07 Absrac In his paper we sudy a mulidimensional quadraic SDE wih a paricular class of

More information

REFLECTED SOLUTIONS OF BACKWARD SDE S, AND RELATED OBSTACLE PROBLEMS FOR PDE S

REFLECTED SOLUTIONS OF BACKWARD SDE S, AND RELATED OBSTACLE PROBLEMS FOR PDE S The Annals of Probabiliy 1997, Vol. 25, No. 2, 72 737 REFLECTED SOLUTIONS OF BACKWARD SDE S, AND RELATED OBSTACLE PROBLEMS FOR PDE S By N. El Karoui, C. Kapoudjian, E. Pardoux, S. Peng and M. C. Quenez

More information

Vanishing Viscosity Method. There are another instructive and perhaps more natural discontinuous solutions of the conservation law

Vanishing Viscosity Method. There are another instructive and perhaps more natural discontinuous solutions of the conservation law Vanishing Viscosiy Mehod. There are anoher insrucive and perhaps more naural disconinuous soluions of he conservaion law (1 u +(q(u x 0, he so called vanishing viscosiy mehod. This mehod consiss in viewing

More information

EXPLICIT TIME INTEGRATORS FOR NONLINEAR DYNAMICS DERIVED FROM THE MIDPOINT RULE

EXPLICIT TIME INTEGRATORS FOR NONLINEAR DYNAMICS DERIVED FROM THE MIDPOINT RULE Version April 30, 2004.Submied o CTU Repors. EXPLICIT TIME INTEGRATORS FOR NONLINEAR DYNAMICS DERIVED FROM THE MIDPOINT RULE Per Krysl Universiy of California, San Diego La Jolla, California 92093-0085,

More information

On Gronwall s Type Integral Inequalities with Singular Kernels

On Gronwall s Type Integral Inequalities with Singular Kernels Filoma 31:4 (217), 141 149 DOI 1.2298/FIL17441A Published by Faculy of Sciences and Mahemaics, Universiy of Niš, Serbia Available a: hp://www.pmf.ni.ac.rs/filoma On Gronwall s Type Inegral Inequaliies

More information

RANDOM LAGRANGE MULTIPLIERS AND TRANSVERSALITY

RANDOM LAGRANGE MULTIPLIERS AND TRANSVERSALITY ECO 504 Spring 2006 Chris Sims RANDOM LAGRANGE MULTIPLIERS AND TRANSVERSALITY 1. INTRODUCTION Lagrange muliplier mehods are sandard fare in elemenary calculus courses, and hey play a cenral role in economic

More information

Lecture 33: November 29

Lecture 33: November 29 36-705: Inermediae Saisics Fall 2017 Lecurer: Siva Balakrishnan Lecure 33: November 29 Today we will coninue discussing he boosrap, and hen ry o undersand why i works in a simple case. In he las lecure

More information

AMartingaleApproachforFractionalBrownian Motions and Related Path Dependent PDEs

AMartingaleApproachforFractionalBrownian Motions and Related Path Dependent PDEs AMaringaleApproachforFracionalBrownian Moions and Relaed Pah Dependen PDEs Jianfeng ZHANG Universiy of Souhern California Join work wih Frederi VIENS Mahemaical Finance, Probabiliy, and PDE Conference

More information

A remark on the H -calculus

A remark on the H -calculus A remark on he H -calculus Nigel J. Kalon Absrac If A, B are secorial operaors on a Hilber space wih he same domain range, if Ax Bx A 1 x B 1 x, hen i is a resul of Auscher, McInosh Nahmod ha if A has

More information

Approximating Random Variables by Stochastic Integrals

Approximating Random Variables by Stochastic Integrals Projekbereich B Discussion Paper No. B 6 Approximaing Random Variables by Sochasic Inegrals by Marin Schweizer November 993 Financial suppor by Deusche Forschungsgemeinschaf, Sonderforschungsbereich 33

More information

Essential Microeconomics : OPTIMAL CONTROL 1. Consider the following class of optimization problems

Essential Microeconomics : OPTIMAL CONTROL 1. Consider the following class of optimization problems Essenial Microeconomics -- 6.5: OPIMAL CONROL Consider he following class of opimizaion problems Max{ U( k, x) + U+ ( k+ ) k+ k F( k, x)}. { x, k+ } = In he language of conrol heory, he vecor k is he vecor

More information

System of Linear Differential Equations

System of Linear Differential Equations Sysem of Linear Differenial Equaions In "Ordinary Differenial Equaions" we've learned how o solve a differenial equaion for a variable, such as: y'k5$e K2$x =0 solve DE yx = K 5 2 ek2 x C_C1 2$y''C7$y

More information

Markov Processes and Stochastic Calculus

Markov Processes and Stochastic Calculus Markov Processes and Sochasic Calculus René Caldeney In his noes we revise he basic noions of Brownian moions, coninuous ime Markov processes and sochasic differenial equaions in he Iô sense. 1 Inroducion

More information