Existence and uniqueness of solution of mixed stochastic differential equation driven by fractional Brownian motion and wiener process

The existence and uniqueness of solution of stochastic differential equation driven by standard Brownian motion and fractional Brownian motion with Hurst parameter H belongs (3/4, 1) is established.

Saved in:
Bibliographic Details
Date:2007
Main Authors: Mishura, Y., Posashkov, S.
Format: Article
Language:English
Published: Інститут математики НАН України 2007
Online Access:https://nasplib.isofts.kiev.ua/handle/123456789/4486
Tags: Add Tag
No Tags, Be the first to tag this record!
Journal Title:Digital Library of Periodicals of National Academy of Sciences of Ukraine
Cite this:Existence and uniqueness of solution of mixed stochastic differential equation driven by fractional Brownian motion and wiener process / Y. Mishura, S. Posashkov // Theory of Stochastic Processes. — 2007. — Т. 13 (29), № 1-2. — С. 152-165. — Бібліогр.: 8 назв.— англ.

Institution

Digital Library of Periodicals of National Academy of Sciences of Ukraine
_version_ 1859593979855634432
author Mishura, Y.
Posashkov, S.
author_facet Mishura, Y.
Posashkov, S.
citation_txt Existence and uniqueness of solution of mixed stochastic differential equation driven by fractional Brownian motion and wiener process / Y. Mishura, S. Posashkov // Theory of Stochastic Processes. — 2007. — Т. 13 (29), № 1-2. — С. 152-165. — Бібліогр.: 8 назв.— англ.
collection DSpace DC
description The existence and uniqueness of solution of stochastic differential equation driven by standard Brownian motion and fractional Brownian motion with Hurst parameter H belongs (3/4, 1) is established.
first_indexed 2025-11-27T19:28:04Z
format Article
fulltext Theory of Stochastic Processes Vol.13 (29), no.1-2, 2007, pp.152-165 YULIA MISHURA AND SERGIY POSASHKOV EXISTENCE AND UNIQUENESS OF SOLUTION OF MIXED STOCHASTIC DIFFERENTIAL EQUATION DRIVEN BY FRACTIONAL BROWNIAN MOTION AND WIENER PROCESS The existence and uniqueness of solution of stochastic differential equation driven by standard Brownian motion and fractional Brow- nian motion with Hurst parameter H ∈ (3/4, 1) is established. 1. Introduction We will consider the following stochastic differential equation with non- homogeneous coefficients, defined on the stochastic basis (Ω,F , (Ft, t ∈ [0, T ]),P) : Xt = X0 + ∫ t 0 a(s, Xs)ds + ∫ t 0 b(s, Xs)dWs + ∫ t 0 c(s, Xs)dBH s , t ∈ [0, T ], (1) where X0 is F0-measurable random variable, EX2 0 < ∞, W = (Wt, t ∈ [0, T ]) is standard Brownian motion (sBm), BH = (BH t , t ∈ [0, T ]) is frac- tional Brownian motion (fBm) with Hurst parameter H ∈ (3/4, 1). Coef- ficients a, b, c : [0, T ] × R −→ R are deterministic functions, which satisfy well-known conditions of existence of integrals in the right-hand of (1), where the integral ∫ t 0 c(s, Xs)dBH s is considered in pathwise sense. If b = 0 we obtain the equation, which contains only fBm. The condi- tions of existence and uniqueness of solution of such equations were shown in [1]. The case b(t, x) = bx and c(t, x) = cx was considered in [7]. Using the similarity to sBm of process MH,ε t := Vt + 1/εBH t , ε > 0, for H ∈ (3/4, 1), which was proven in [2], and the representation of processes similar to sBm from [8], the existence and uniqueness of solution of auxiliary stochastic differential equation XN t = X0 + ∫ t 0 a(s, XN s )dt + ∫ t 0 b(s, XN s )dWt + ∫ t 0 c(s, XN s )dBH s + 1 N ∫ t 0 c(s, XN s )dVs, t ∈ [0, T ], (2) 2000 Mathematics Subject Classifications. 60G15, 60H05, 60H10 Key words and phrases. Stochastic differential equation, fractional Brownian motion 152 SOLUTION OF STOCHASTIC DIFFERENTIAL EQUATION 153 where V = (Vt, t ∈ [0, T ]) is another standard Brownian motion, indepen- dent on W and BH , for any N ∈ N, was proved in the paper [3]. The main aim of present work is to find the solution of equation (1) as the limit in some complete space of solutions of (2) when N tends to ∞. 2. Main part Let the coefficients of equation (1) satisfy, in addition, the following assumptions (A) There exists A > 0 such that |a(t, x)| ≤ A, |b(t, x)| ≤ A, |c(t, x)| ≤ A ∀t ∈ [0, T ], ∀x ∈ R. (B) There exists L > 0 such that (a(t, x)−a(t, y))2 +(b(t, x)−b(t, y))2 +(c(t, x)−c(t, y))2 ≤ L2(x−y)2, for ∀t ∈ [0, T ], ∀x, y ∈ R. (C) The function c(t, x) is differentiable by c, and there exist constants B > 0 and β ∈ (1 − H, 1) such that ∀s, t ∈ [0, T ], ∀x ∈ R |c(s, x) − c(t, x)| + |∂xc(s, x) − ∂xc(t, x)| ≤ Bs − t|β. (D) Holder continuity of ∂xc(t, x) in x : |∂xc(t, x) − ∂xc(t, y)| ≤ D|x − y|ρ, for ∀t ∈ [0, T ], ∀x, y ∈ R with some parameter ρ ∈ (3/2 − H, 1). Note that this conditions imply the conditions from [3] of existence and uniqueness of solution of equation (2). Let consider for any 1−H < α < min(1/2, β, ρ−1/2) the space of Besov type Wα([0, T ]) := {Y = Yt(ω) : (t, ω) ∈ [0, T ] × Ω, ||Y ||α < ∞} (3) with the norm ||Y ||α := sup t∈[0,T ] ( E(Yt) 2 + E (∫ t 0 |Yt − Ys| (t − s)1+α ds )2 ) , (4) and prove that the solution of SDE (2) belongs to this space for any N ∈ N. We shall denote different constants as C if it is unimportant for stated results. From [5] we have for pathwise integral ∫ t 0 f(s)dBH s the representation∫ t 0 f(s)dBH s = ∫ t 0 Dα 0+f(s)D1−α t− BH t−(s)ds, 154 YULIA MISHURA AND SERGIY POSASHKOV therefore this integral can be estimated as∣∣∣∣∫ t 0 f(s)dBH s ∣∣∣∣ ≤ Ct(ω) (∫ t 0 |f(s)| sα ds + ∫ t 0 ∫ r 0 |f(r) − f(u)| (r − u)1+α dudr ) , (5) where Ct(ω) = sup 0≤u≤s≤t |D1−α s− BH s−(u)|. (6) The existence of all moments E|Ct(ω)|p for any p ≥ 1 has also been proved in [1]. Lemma 1. The process |Ct(ω)| is dominated by some continuous process. Proof. To prove this statement we estimate |D1−α t− BH t−(s)|: |D1−α t− BH t−(s)| ≤ 1 Γ(α) ( |BH t − BH s | (t − s)1−α + (1 − α) ∫ t s |BH s − BH u | (s − u)2−α du ) . Using Garsia - Rodemich - Rumsey inequality from [1] that has a form |f(t) − f(s)|p ≤ Cλ,p|t − s|λp−1 ∫ t 0 ∫ t 0 |f(x) − f(y)|p |x − y|λp+1 dxdy (7) with f ∈ C[0, T ], t, s ∈ [0, T ], p ≥ 1 and λ > p−1, we obtain |BH t − BH s | ≤ Cλ,p|t − s|λ−1/p (∫ t 0 ∫ t 0 |BH u − BH v |p |u − v|λp+1 dudv )1/p . Let λ = H − ε/2, p = 2/ε for 0 < ε < H + α − 1, then |BH t − BH s | ≤ Cε|t − s|H−ε (∫ t 0 ∫ t 0 |BH u − BH v |2/ε |u − v|2H/ε dudv )ε/2 , note that 1 − α < H − ε. Denote ψt,ε = (∫ t 0 ∫ t 0 |BH u − BH v |2/ε |u − v|2H/ε dudv )ε/2 , (8) The process ψt,ε is continuous and nondecreasing in t, and Eψq t,ε < ∞ for any q > 0 and any t ∈ [0, T ] [1]. So, |BH t − BH s | (t − s)1−α ≤ CH,ε(t − s)H−ε−1+αψt,ε ≤ Cψt,ε, and∫ t s |BH u − BH s | (u − s)2−α du ≤ CH,ε ∫ t s (u − s)H−ε−2+αψt,εdu ≤ CH,ε(t − s)H+α−1−εψt,ε ≤ Cψt,ε, SOLUTION OF STOCHASTIC DIFFERENTIAL EQUATION 155 where H + α − 1 − ε > 0. So sup 0≤u≤s≤t |D1−α s− BH s−(u)| ≤ Cψt,ε, and the statement of lemma follows from continuity and strictly increasing property of ψt,ε. Introduce the random variable C(ω) := sup 0≤t≤T Ct(ω). (9) Note that C(ω) ≤ ψT,ε and E|C(ω)|q < ∞ for any q ≥ 1. First of all we prove the Hölder continuity of the solution of equation (2). Theorem 2. For any δ ∈ (0, 1/2) the solution of equation (2) is Hölder continuous with parameter 1/2 − δ. Proof. At first, establish Hölder properties of the integral {∫ t 0 bsdWs, t ∈ [0, T ]}, where bs is a predictable bounded process. For any 0 < δ < 1/4 put p = 2 δ , θ = 1/2 − δ/2 in Garsia - Rodemich - Rumsey inequality (7). Then∣∣∣∣∫ t s budWu ∣∣∣∣ ≤ Cδ|t − s|1/2−δξb t,δ, where ξb t,δ := (∫ t 0 ∫ t 0 | ∫ y x budWu|p |x − y|θp+1 dxdy )1/p = (∫ t 0 ∫ t 0 | ∫ y x budWu|2/δ |x − y|1/δ dxdy )δ/2 (10) and for any q > p from Hölder and Burkholder inequalities E(ξb t,δ) q = E (∫ t 0 ∫ t 0 | ∫ y x budWu|2/δ |x − y|1/δ dxdy )qδ/2 ≤ (∫ t 0 ∫ t 0 E| ∫ y x budWu|qdxdy |x − y|q/2 ) tqδ−2 ≤ Cq (∫ t 0 ∫ t 0 | ∫ y x b2 udu|q/2dxdy |x − y|q/2 ) tqδ−2 ≤ Θt,q. So, the process ξb t,δ from (10) is continuous, strictly increasing and has the moments of any order. 156 YULIA MISHURA AND SERGIY POSASHKOV Now consider |XN r − XN z | for 0 < z < r < T : |XN r −XN z | ≤ | ∫ r z a(u, Xu)du|+ | ∫ r z b(u, Xu)dWu|+ 1 N | ∫ r z c(u, Xu)dVu| + | ∫ r z c(u, Xu)dBH u | ≤ A(r − z) + Cξb r,δ|r − z|1/2−δ + C N ξc r,δ|r − z|1/2−δ + Cr(ω) ∫ r z |c(u, XN u )|du uα + Cr(ω) ∫ r z ∫ u z |c(u, XN u ) − c(v, XN v )| (u − v)1+α dvdu ≤ C ′ r(ω)(r − z)1/2−δ + C ′ r(ω) ∫ r z ∫ u z |XN u − XN v | (u − v)1+α dvdu, where C ′ t(ω) := C(ψt,ε ∨ ξb t,δ ∨ ξc t,δ ∨ 1), (11) ψt,ε is defined by (8), C ′ t(ω) ≤ C ′ T (ω) and C ′ T (ω) has the moments of any order. Therefore, for δ < 1/2 − α φr,s := ∫ r s |XN r − XN z | (r − z)1+α dz ≤ C ′ r(ω) ∫ r s (r − z)−1/2−δ−αdz + C ′ r(ω) ∫ r s 1 (r − z)1+α ∫ r z ∫ u z |XN u − XN v | (u − v)1+α dvdudz = C ′ r(ω)(r − s)1/2−α−δ + C ′ r(ω) ∫ r s (r − u)−α ∫ u s |XN u − XN v | (u − v)1+α dvdu ≤ C ′ r(ω)(r − u)1/2−α−δ + C ′ r(ω) ∫ r s (r − u)−αφu,sdu. From modified Gronwall inequality (Lemma 7.6 [1]) φr,s ≤ C ′ r(ω)(r − s)1/2−α−δ exp[C ′ r(ω) 1 1−α ]. Return to |XN r − XN z |: |XN r − XN z | ≤ C ′ r(ω)(r − z)1/2−δ + C ′ r(ω) exp[C ′ r(ω) 1 1−α ] ∫ r z (v − z)1/2−α−δdv ≤ C̃r(ω)(r − z)1/2−δ, (12) where C̃r(ω) = C ′ r(ω) exp[C ′ r(ω) 1 1−α ], and the theorem is proved for 0 < δ < 1/2 − α consequently for 0 < δ < 1/2. Introduce the random variable C̃(ω) := sup 0≤t≤T C̃t(ω), (13) SOLUTION OF STOCHASTIC DIFFERENTIAL EQUATION 157 it also has moments of any order. Now prove that the solution of (2) belongs to space (3) with norm (4) for all N ∈ N. Theorem 3. Under assumptions (A) - (D) the solution of equation (2) belongs to space of Besov type (3) with norm (4) for all N ∈ N. Proof. To prove the statement of this theorem we estimate E(XN t )2 + E (∫ t 0 |XN t − XN s | (t − s)1+α ds )2 =: A1(t) + A2(t). (14) At first, for E(XN t )2 we have E(XN t )2 ≤ 5E(X0) 2 + 5E( ∫ t 0 a(s, XN s )ds)2 + 5E( ∫ t 0 b(s, XN s )dWs) 2 + 5E( ∫ t 0 c(s, XN s )dBH s )2 + 5E( 1 N ∫ t 0 c(s, XN s )dVs) 2. Evidently E (∫ t 0 a(s, XN s )ds )2 ≤ A2T 2, E (∫ t 0 b(s, XN s )dWs )2 ≤ A2T, E ( 1 N ∫ t 0 c(s, XN s )dVs )2 ≤ A2T N2 ≤ A2T . Further, using the estimate (5), (12) and with the help of random variables defined by (9), (13), E (∫ t 0 c(s, XN s )dBH s )2 can be estimated as E (∫ t 0 c(s, XN s )dBH s )2 ≤ E ( C 2 (ω) (∫ t 0 c(s, XN s ) sα ds + ∫ t 0 ∫ s 0 |c(s, XN s ) − c(u, XN u )| (s − u)1+α duds )2 ) ≤ CE ( C 2 (ω) ( t ∫ t 0 A2 s2α ds + (∫ t 0 ∫ s 0 B(s − u)β + LC̃(ω)(s − u)1/2−δ (s − u)1+α duds )2 ⎞⎠⎞⎠ ≤ C(A2t2−2αEC 2 (ω)+B2EC 2 (ω)t2(1−α+β) +L2E(C̃2(ω)C 2 (ω))T 3−2α−2δ)). 158 YULIA MISHURA AND SERGIY POSASHKOV So, we have A1(t) ≤ C(A2T 2 + 2A2T + A2T 2−2αEC 2 (ω) + B2EC 2 (ω)T 2(1−α+β) + L2E(C̃2(ω)C 2 (ω))T 3−2α−2δ) < ∞. (15) Consider now A2(t). We have that A2(t) ≤ 4E (∫ t 0 | ∫ t s a(u, XN u )du| (t − s)1+α ds )2 + 4E (∫ t 0 | ∫ t s b(u, XN u )dWu| (t − s)1+α ds )2 + 4N−2E (∫ t 0 | ∫ t s c(u, XN u )dVu| (t − s)1+α ds )2 + 4E (∫ t 0 | ∫ t s c(u, XN u )dBH u | (t − s)1+α ds )2 . Evidently, E (∫ t 0 | ∫ t s a(u, Xu)du| (t − s)1+α ds )2 ≤ CA2t2−2α. Now, let γ ∈ (α, 1/2), then E (∫ t 0 | ∫ t s b(u, Xu)dWu| (t − s)1+α ds )2 ≤ Ct1−2γ ∫ t 0 E| ∫ t s b(u, Xu)dWu|2 (t − s)2+2α−2γ ds ≤ Ct1−2γ ∫ t 0 ∫ t s b2(u, Xu)du (t − s)2+2α−2γ ds ≤ Ct1−2γ ∫ t 0 A2 (t − s)1+2α−2γ ds ≤ CA2t1−2α, and similarly E (∫ t 0 | ∫ t s c(u, Xu)dVu| (t − s)1+α ds )2 ≤ CA2t1−2α. SOLUTION OF STOCHASTIC DIFFERENTIAL EQUATION 159 Now we estimate E (∫ t 0 | t s c(u,Xu)dBH u | (t−s)1+α ds )2 . E (∫ t 0 | ∫ t s c(u, Xu)dBH u | (t − s)1+α ds )2 ≤ E ⎛⎝C(ω) ∫ t 0 ∫ t s |c(u,Xu)| (u−s)α du + ∫ t s ∫ u s |c(u,xN u )−c(r,XN r )| (u−r)1+α drdu (t − s)1+α ds ⎞⎠2 ≤ E ⎛⎝C(ω) ∫ t 0 ∫ t s |c(u,Xu)| (u−s)α du + ∫ t s ∫ u s B(u−r)β+LC(ω)(u−r)1/2−δ (u−r)1+α drdu (t − s)1+α ds ⎞⎠2 ≤ E ( C(ω) ∫ t 0 A(t − s)1−α + B(t − s)1+β−α + LC̃(ω)(t − s)3/2−δ−α (t − s)1+α ds )2 ≤ C(A2t2−4αEC 2 (ω) + B2t2+2β−4αEC 2 (ω) + L2t3−2δ−4αEC 2 (ω)C̃2(ω)). Therefore A2(t) satisfies the inequality A2(t) ≤ C(A2T 2−2α + A2T 1−2α + A2T 1−2α +(A2T 2−4αEC 2 (ω)+B2T 2+2β−4αEC 2 (ω)+L2T 3−2δ−4αEC 2 (ω)C̃2(ω)) < ∞. (16) At last, the statement of our theorem follows from inequalities (15) and (16). Introduce for any R > 1 the stopping time τR by τR := inf{t : C ′ t(ω) ≥ R} ∧ T, (17) where C ′ t(ω) is defined by (11). For any ω ∈ Ω τR = T , for any R > R(ω). Define the processes {XN t∧τR , N ∈ N, t ∈ [0, T ]} as the solutions of equa- tion (2) stopped an the moment τR, and prove that they are fundamental in the norm (4) of the space (3). Theorem 4. Under assumptions (A) - (D) the sequence {XN t∧τR , N ≥ 1, t ∈ [0, T ]} of solutions of equations (2) is fundamental in the norm (4). Proof. Consider E(XN t∧τR − XM t∧τR )2 + E (∫ t 0 |XN t∧τR − XM t∧τR − XN s∧τR + XM s∧τR | (t − s)1+α ds )2 = E(XN t∧τR − XM t∧τR )2 + E (∫ t∧τR 0 |XN t∧τR − XM t∧τR − XN s + XM s | (t − s)1+α ds )2 =: AN,M 1 (t) + AN,M 2 (t). (18) 160 YULIA MISHURA AND SERGIY POSASHKOV First, for AN,M 1 (t) we have AN,M 1 (t) ≤ 4E (∫ t∧τR 0 (a(s, XN s ) − a(s, XM s ))ds )2 + 4E (∫ t∧τR 0 (b(s, XN s ) − b(s, XM s ))dWs )2 + 4E (∫ t∧τR 0 (c(s, XN s ) − c(s, XM s ))dBH s )2 + 4E (∫ t∧τR 0 ( c(s, XN s ) N − c(s, XM s ) M ) dVs )2 =: 4(I1 + I2 + I3 + I4). Then I1 ≤ CTL2 ∫ t 0 E(XN s∧τR − XM s∧τR )2ds, I2 ≤ CL2 ∫ t 0 E(XN s∧τR − XM s∧τR )2ds, I4 ≤ CA2T (N−2 + M−2). Now we are in position to estimate I3 : I3 = E (∫ t∧τR 0 (c(s, XN s ) − c(s, XM s ))dBH s )2 ≤ R2E (∫ t∧τR 0 |c(s, XN s ) − c(s, XM s )| sα ds + ∫ t∧τR 0 ∫ s 0 |c(s, XN s ) − c(s, XM s ) − c(u, XN u ) + c(u, XM u )| (s − u)1+α duds )2 ≤ 2R2 ( E (∫ t∧τR 0 |c(s, XN s ) − c(s, XM s )| sα ds )2 + E (∫ t∧τR 0 ∫ s 0 |c(s, XN s ) − c(s, XM s ) − c(u, XN u ) + c(u, XM u )| (s − u)1+α duds )2 ) = 2R2(I4 + I5). Further, I4 ≤ CL2T 1−2αE ∫ t∧τR 0 (XN s − XM s )2ds = CL2T 1−2α ∫ t 0 (AN,M 1 (s))2ds. SOLUTION OF STOCHASTIC DIFFERENTIAL EQUATION 161 Using Lemma 7.1 [1] we estimate I5 as I5 ≤ E (∫ t∧τR 0 ∫ s 0 A|XN s − XM s − XN u + XM u | (s − u)1+α duds + ∫ t∧τR 0 ∫ s 0 AB|XN s − XM s |(s − u)β (s − u)1+α duds + ∫ t∧τR 0 ∫ s 0 D|XN s − XM s |(|XN s − XN u |ρ + |XM s − XM u |ρ) (s − u)1+α duds )2 ≤ 3E (∫ t∧τR 0 ∫ s 0 A|XN s − XM s − XN u + XM u | (s − u)1+α duds )2 + 3E (∫ t∧τR 0 ∫ s 0 AB|XN s − XM s |(s − u)β (s − u)1+α duds )2 + 3E (∫ t∧τR 0 ∫ s 0 D|XN s − XM s |(|XN s − XN u |ρ + |XM s − XM u |ρ) (s − u)1+α duds )2 = 3(I6 + I7 + I8), where I6 ≤ CTA2 ∫ t 0 E (∫ s∧τR 0 |XN s − XM s − XN u + XM u | (s − u)1+α du )2 ds, I7 ≤ CTA2 ∫ t 0 s2(β−α)E (|XN s∧τR − XM s∧τR |)2 ds, I8 ≤ E (∫ t∧τR 0 ∫ s 0 B|XN s − XM s |(2R(s − u)ρ(1/2−δ)) (s − u)1+α duds )2 ≤ CTD2R2 ∫ t 0 sρ−2ρδ−2αE (|XN s∧τR − XM s∧τR |)2 ds, where we choose δ in such a way that ρ− 2ρδ − 2α > 0. It is possible since α < ρ − 1/2 so ρ − 2α > 1/2 − α > 0. At last, I5 ≤ C ∫ t 0 E (∫ s∧τR 0 |XN s − XM s − XN u + XM u | (s − u)1+α du )2 ds + C ∫ t 0 s2(β−α)E (|XN s∧τR − XM s∧τR |)2 ds + CR2 ∫ t 0 sρ−2ρδ−2αE (|XN s∧τR − XM s∧τR |)2 ds, 162 YULIA MISHURA AND SERGIY POSASHKOV and AN,M 1 (t) ≤ CR2 ∫ t 0 AN,M 1 (s)ds + CR2 ∫ t 0 AN,M 2 (s)ds + C(N−2 + M−2). (19) Return to AN,M 2 (t). It admits the following estimate AN,M 2 (t) ≤ C ⎛⎝E (∫ t∧τR 0 ∫ t∧τR s (a(u, XN u ) − a(u, XM u ))du (t − s)1+α ds )2 + E (∫ t∧τR 0 ∫ t∧τR s (b(u, XN u ) − b(u, XM u ))dWu (t − s)1+α ds )2 + E (∫ t∧τR 0 ∫ t∧τR s (c(u, XN u ) − c(u, XM u ))dBH u (t − s)1+α ds )2 + E (∫ t∧τR 0 ∫ t∧τR s ( c(u,XN u ) N − c(u,XM u ) M )dVu (t − s)1+α ds )2 ⎞⎠ ≤ C(I9 + I10 + I11 + I12). I9 ≤ CT 1−2γE ∫ t∧τR 0 (t − s) ∫ t∧τR 0 L2|XN u − XM u |2du (t − s)2+2α−2γ ds ≤ CT 1−2α ∫ t 0 E ( XN s∧τR − XM s∧τR )2 ds ≤ CT 1−2α ∫ t 0 AN,M 1 (s)ds, I10 ≤ CT 1−2γ ∫ t 0 ∫ t s E|XN u∧τR − XM u∧τR |2du (t − s)2+2α−2γ ds ≤ CT 1−2γ ∫ t 0 AN,M 1 (s) (t − s)1+2α−2γ ds, where we choose γ in such a way that α < γ < 1/2. For I12 we have I12 ≤ CT 1−2α(N−2 + M−2). SOLUTION OF STOCHASTIC DIFFERENTIAL EQUATION 163 Now consider I11 : I11 ≤ CR2T 1−2γ ⎛⎜⎝E ∫ t∧τR 0 (∫ t∧τR s c(u,XN u )−c(u,XM u ) (u−s)α du )2 (t − s)2+2α−2γ ds + E ∫ t∧τR 0 (∫ t∧τR s ∫ u s |c(u,XN u )−c(u,XM u )−c(v,XN v )+c(v,XM v )| (u−v)1+α dvdu )2 (t − s)2+2α−2γ ds ⎞⎟⎠ =: CR2T 1−2γ(I12 + I13). I12 ≤ C ∫ t 0 (t − s) ∫ t s E(XN u∧τr −XM u∧τr) 2 (u−s)2α du (t − s)2+2α−2γ ≤ C ∫ t 0 AN,M 1 (s) (t − s)1+2α−2γ ds. I13 ≤ C ⎛⎜⎝E ∫ t∧τR 0 (∫ t∧τR s ∫ u s L|XN u −XM u −XN v +XM v | (u−v)1+α dvdu )2 (t − s)2+2α−2γ ds + E ∫ t∧τR 0 (∫ t∧τR s ∫ u s D|XN u −XM u |(u−v)β (u−v)1+α dvdu )2 (t − s)2+2α−2γ ds + E ∫ t∧τR 0 (∫ t∧τR s ∫ u s B|XN u −XM u |(|XN u −XN v |ρ+|XM u −XM v |ρ) (u−v)1+α dvdu )2 (t − s)2+2α−2γ ds ⎞⎟⎠ =: C(I14 + I15 + I16). I14 ≤ CT 2γ−2α ∫ t 0 E( ∫ s∧τR 0 |XN s − XM s − XN u + XM u | (s − u)1+α du)2ds = CA2T 2γ−2α ∫ t 0 AN,M 2 (s)ds, I15 ≤ C ∫ t 0 E (∫ t∧τR s |XN u − XM u |(u − s)β−α )2 (t − s)2+2α−2γ ds ≤ C ∫ t 0 (t − s)1+2β−2α ∫ t s E ( XN u∧τR − XM u∧τR )2 du (t − s)2+2α−2γ ds ≤ CT 2β+2γ−4α ∫ t 0 AN,M 1 (s)ds, 164 YULIA MISHURA AND SERGIY POSASHKOV note that α < β. I16 ≤ CR2E ∫ t∧τR 0 (∫ t∧τR s ∫ u s |XN u −XM u |(u−v)ρ(1/2−δ) (u−v)1+α dvdu )2 (t − s)2+2α−2γ ds, where we chose 0 < δ < 1/2−α/ρ, note that α < ρ− 1/2. Similarly to I15, I16 ≤ CT 2γ−2α ∫ t 0 AN,M 1 (s)ds. Therefore we have I13 ≤ CR2( ∫ t 0 AN,M 1 (s)ds + ∫ t 0 AN,M 2 (s)ds). Hence I11 ≤ CR4( ∫ t 0 AN,M 1 (s) (t − s)1+2α−2γ ds + ∫ t 0 AN,M 2 (s)ds). At last, AN,M 2 (t) ≤ CR4 (∫ t 0 AN,M 1 (s) (t − s)1+2α−2γds + ∫ t 0 AN,M 2 (s)ds ) + C(N−2 + M−2). (20) From (19) and (20) we obtain that the sum AN,M 1 (t) + AN,M 2 (t) admits the same estimate as AN,M 2 (t), i.e. AN,M 1 (t) + AN,M 2 (t) ≤ CR4 (∫ t 0 AN,M 1 (s) (t − s)1+2α−2γds + ∫ t 0 AN,M 2 (s)ds ) + C(N−2 + M−2), (21) and from modified Gronwall lemma [1] AN,M 1 (t) + AN,M 2 (t) ≤ CR4(N−2 + M−2) exp{t(CR4)1/(2γ−2α)}, (22) taking, for example, γ := (1/2 + α)/2. As choosing N, M → 0 see that right-hand side of (22) tends to zero whence the proof follows. Theorem 5. The SDE (1) has the solution on interval [0, T ], and this solution is unique. SOLUTION OF STOCHASTIC DIFFERENTIAL EQUATION 165 Proof. Since the space (3) is complete, and from Theorem 4 we can define Xt∧τR := lim N→∞ XN t∧τR , (23) where the limit is taken in space Wα[0, T ] (in particular, we have that the limit exists in L2(Ω× [0, T ])).Using the similar estimates an Theorem 4, we can prove that Xt∧τR is the unique solution of the original equation (1) on interval [0, τR]. From definition (17) of τR we have τR1 ≤ τR2 for R1 ≤ R2. So XτR1 and XτR2 coincide a.s. on interval [0, τR1 ]. Where R → ∞ we obtain the existence and uniqueness of solution of SDE (1) on interval [0, T ]. 3. Conclusion So, we proved the existence and uniqueness of solution of stochastic differential equation driven by standard and fractional Brownian motions (1). Bibliography 1. Nualart, D., Răşcanu Differential equations driven by fractional Brownian motion, Collect. Math., 53, (2000), 55–81. 2. Cheridito, P. Regularizing fractional Brownian motion with a view towards stock price modeling. PhD thesis, Zurich (2001). 3. Mishura, Yu., Posashkov, S., Existence and uniqueness of solution of sto- chastic differential equation driven by fractional Brownian motion with sta- bilizing term Theory Prob. Math. Stat, 76, (2007). 4. Zähle, M. On the link between fractional and stochastic calculcus. In: Grauel, H., Gundlach, M. (eds.) Stochastic Dynamics, Springer, (1999), 305–325. 5. Zähle, M. Integration with respect to fractal functions and stochastic calcu- lus. I. Probab. Theory Rel. Fields, 111,(1998), 333–374. 6. Zähle, M. Integraton with respect to fractal functions and stochastic calcu- lus. II. Math. Nachr., 225,(2001), 145–183. 7. Krvavych, Yu., Mishura, Yu., Exponential formula and Girsanov theorem for mixed semilinear stochastic differential equatios, in: Mathematical Fi- nance (Trends in Mathematics), Basel, Birkhäuser, (2001), 230–39. 8. Hitsuda, M. Representation of Gaussian processes equivalent to Wiener process, Osaka Journal of Mathematics, 5, (1968), 299–312. Department of Probability Theory and Mathematical Statistics, Kyiv National Taras Shevchenko University, Kyiv, Ukraine E-mail address: myus@univ.kiev.ua Department of Probability Theory and Mathematical Statistics, Kyiv National Taras Shevchenko University, Kyiv, Ukraine E-mail address: corlagon@univ.kiev.ua
id nasplib_isofts_kiev_ua-123456789-4486
institution Digital Library of Periodicals of National Academy of Sciences of Ukraine
issn 0321-3900
language English
last_indexed 2025-11-27T19:28:04Z
publishDate 2007
publisher Інститут математики НАН України
record_format dspace
spelling Mishura, Y.
Posashkov, S.
2009-11-19T10:19:03Z
2009-11-19T10:19:03Z
2007
Existence and uniqueness of solution of mixed stochastic differential equation driven by fractional Brownian motion and wiener process / Y. Mishura, S. Posashkov // Theory of Stochastic Processes. — 2007. — Т. 13 (29), № 1-2. — С. 152-165. — Бібліогр.: 8 назв.— англ.
0321-3900
https://nasplib.isofts.kiev.ua/handle/123456789/4486
The existence and uniqueness of solution of stochastic differential equation driven by standard Brownian motion and fractional Brownian motion with Hurst parameter H belongs (3/4, 1) is established.
en
Інститут математики НАН України
Existence and uniqueness of solution of mixed stochastic differential equation driven by fractional Brownian motion and wiener process
Article
published earlier
spellingShingle Existence and uniqueness of solution of mixed stochastic differential equation driven by fractional Brownian motion and wiener process
Mishura, Y.
Posashkov, S.
title Existence and uniqueness of solution of mixed stochastic differential equation driven by fractional Brownian motion and wiener process
title_full Existence and uniqueness of solution of mixed stochastic differential equation driven by fractional Brownian motion and wiener process
title_fullStr Existence and uniqueness of solution of mixed stochastic differential equation driven by fractional Brownian motion and wiener process
title_full_unstemmed Existence and uniqueness of solution of mixed stochastic differential equation driven by fractional Brownian motion and wiener process
title_short Existence and uniqueness of solution of mixed stochastic differential equation driven by fractional Brownian motion and wiener process
title_sort existence and uniqueness of solution of mixed stochastic differential equation driven by fractional brownian motion and wiener process
url https://nasplib.isofts.kiev.ua/handle/123456789/4486
work_keys_str_mv AT mishuray existenceanduniquenessofsolutionofmixedstochasticdifferentialequationdrivenbyfractionalbrownianmotionandwienerprocess
AT posashkovs existenceanduniquenessofsolutionofmixedstochasticdifferentialequationdrivenbyfractionalbrownianmotionandwienerprocess