Citation: Murugan Suvinthra, Krishnan Balachandran, Rajendran Mabel Lizzy. Large Deviations for Stochastic Fractional Integrodifferential Equations[J]. AIMS Mathematics, 2017, 2(2): 348-364. doi: 10.3934/Math.2017.2.348
[1] | Marimuthu Mohan Raja, Velusamy Vijayakumar, Anurag Shukla, Kottakkaran Sooppy Nisar, Wedad Albalawi, Abdel-Haleem Abdel-Aty . A new discussion concerning to exact controllability for fractional mixed Volterra-Fredholm integrodifferential equations of order r∈(1,2) with impulses. AIMS Mathematics, 2023, 8(5): 10802-10821. doi: 10.3934/math.2023548 |
[2] | Sumbal Ahsan, Rashid Nawaz, Muhammad Akbar, Saleem Abdullah, Kottakkaran Sooppy Nisar, Velusamy Vijayakumar . Numerical solution of system of fuzzy fractional order Volterra integro-differential equation using optimal homotopy asymptotic method. AIMS Mathematics, 2022, 7(7): 13169-13191. doi: 10.3934/math.2022726 |
[3] | Clement Manga, Alioune Coulibaly, Alassane Diedhiou . On some stochastic differential equations with jumps subject to small positives coefficients. AIMS Mathematics, 2019, 4(5): 1369-1385. doi: 10.3934/math.2019.5.1369 |
[4] | Subramanian Muthaiah, Manigandan Murugesan, Muath Awadalla, Bundit Unyong, Ria H. Egami . Ulam-Hyers stability and existence results for a coupled sequential Hilfer-Hadamard-type integrodifferential system. AIMS Mathematics, 2024, 9(6): 16203-16233. doi: 10.3934/math.2024784 |
[5] | Chao Wei . Parameter estimation for partially observed stochastic differential equations driven by fractional Brownian motion. AIMS Mathematics, 2022, 7(7): 12952-12961. doi: 10.3934/math.2022717 |
[6] | Manal Elzain Mohamed Abdalla, Hasanen A. Hammad . Solving functional integrodifferential equations with Liouville-Caputo fractional derivatives by fixed point techniques. AIMS Mathematics, 2025, 10(3): 6168-6194. doi: 10.3934/math.2025281 |
[7] | Thabet Abdeljawad, Sabri T. M. Thabet, Imed Kedim, Miguel Vivas-Cortez . On a new structure of multi-term Hilfer fractional impulsive neutral Levin-Nohel integrodifferential system with variable time delay. AIMS Mathematics, 2024, 9(3): 7372-7395. doi: 10.3934/math.2024357 |
[8] | Mouffak Benchohra, Zohra Bouteffal, Johnny Henderson, Sara Litimein . Measure of noncompactness and fractional integro-differential equations with state-dependent nonlocal conditions in Fréchet spaces. AIMS Mathematics, 2020, 5(1): 15-25. doi: 10.3934/math.2020002 |
[9] | Guowei Zhang . The lower bound on the measure of sets consisting of Julia limiting directions of solutions to some complex equations associated with Petrenko's deviation. AIMS Mathematics, 2023, 8(9): 20169-20186. doi: 10.3934/math.20231028 |
[10] | Saima Rashid, Fahd Jarad, Khadijah M. Abualnaja . On fuzzy Volterra-Fredholm integrodifferential equation associated with Hilfer-generalized proportional fractional derivative. AIMS Mathematics, 2021, 6(10): 10920-10946. doi: 10.3934/math.2021635 |
The subject of fractional calculus deals with the investigations of derivatives and integrals, of any arbitrary real or complex order, which unify and extend the notions of integer-order derivative and n-fold integral. It can be considered as a branch of mathematical analysis which deals with integrodifferential operators and equations where the integrals are of convolution type and exhibit (weakly singular) kernels of power-law type. It is strictly related to the theory of pseudo-differential operators. Fractional order models have the tendency to capture non-local relations in space and time, thus forming an improvised model for analyzing complex phenomena. It is a successful tool for describing complex quantum field dynamical systems, dissipation and long-range phenomena that cannot be well illustrated using ordinary differential and integral operators. For an introductory study on fractional calculus and fractional derivatives, see the literatures [19,21,25].
Inducing randomness into the model helps us to analyze better by taking into consideration the effect of uncertainty, thus leading to stochastic fractional differential equations (refer [24] and references therein). The theory of existence, controllability and stability of fractional differential equations has been studied by many authors (for instance, see [1,2,15,16]). However there seems to be possibly limited literature to the study of large deviations for stochastic fractional differential equations.
Large deviation theory is a branch of probability theory that deals with the study of rare events. Though the probability of occurrence of rare events is too small, their impact may be large and so it is significant to study such rare events. Large deviation theory finds its application in many areas such as mathematical finance, statistical mechanics and various fields ranging from physics to biology. The origin of large deviations dates back to the 1930s where there was a necessity to solve the problem of total claim exceeding the reserve fund set aside in an insurance company. The solution was discovered by the Swedish mathematician Cramer via refinement of the central limit theorem. Subsequent developments has been made since then and there was major breakthrough into the subject after Varadhan [31] established a general framework for large deviation principle and formulated the Varadhan's lemma in 1966. In 1970, Wentzell and Freidlin [13] developed a theory to enhance the large deviation principle for differential equations with small stochastic perturbations, which involves time discretization of the original problem and then analyzing the large deviation principle in the limit. Fleming [12] developed a stochastic control approach to establish large deviation principle and then Dupuis and Ellis [11] combined the weak convergence approach with the theory of Fleming. These developments indeed explore the close association of large deviation theory with optimal controllability problems.
Using the weak convergence approach, the large deviations for homeomorphism flows of non-Lipschitz Stochastic Differential Equations (SDEs) was studied by Ren and Zhang [27]; the large deviations for two-dimensional stochastic Navier-Stokes equations by Sritharan and Sundar [28], and for stochastic evolution equations with small multiplicative noise by Liu [18]. For more references on this approach, one may refer [5,6,11,14,26]. By using the approximating method, Mohammed and Zhang [23] established a Freidlin-Wentzell type large deviation principle for the stochastic delay differential equations. Mo and Luo [22] also studied the large deviations for the stochastic delay differential equations by employing the weak convergence approach. Bo and Jiang [4] analyzed the large deviation for Kuramoto-Sivashinsky stochastic partial differential equation. A large deviation principle for stochastic differential equations with deviating arguments is dealt with in [30].
A Freidlin-Wentzell type large deviation principle is discussed in Dembo and Zeitouni [8] for the following stochastic differential equation:
dX(t)=b(t,X(t))dt+√ϵσ(t,X(t))dW(t),t∈(0,T],X(0)=X0.} | (1) |
In the case that the system is affected by hereditary influences, the drift and diffusion coefficients (b and σ) also depend on an integral component, thus giving rise to stochastic integrodifferential equations. The large deviations for stochastic integrodifferential equations has been carried out in [29]. In this paper, we consider the stochastic fractional integrodifferential equations with Gaussian noise perturbation of multiplicative type and establish the large deviation principle by using the results developed by Budhiraja and Dupuis [7]. The compactness argument is done with the associated control equation and weak convergence result is obtained by observing the nature of the solution of the stochastic control equation as the perturbation of the noise term tends to zero.
Let X and H be separable Hilbert spaces. Denote by L(X) the space of all bounded linear operators from X to X. Denote by J the time interval [0,T]. Let {Ω,F,P} be a complete filtered probability space equipped with a complete family of right continuous increasing sub σ-algebras {Ft,t∈J} satisfying {Ft⊂F}. Let Q be a symmetric, positive, trace class operator on H and W(⋅) be a H-valued Wiener process with covariance operator Q. Denote the space H0:=Q1/2H. Then H0 is a Hilbert space with the inner product (X,Y)0:=(Q−1/2X,Q−1/2Y) for all X,Y∈H0 and the corresponding norm is denoted by ‖⋅‖0. Let LQ denote the space of all Hilbert-Schmidt operators from H0 to X. Consider the nonlinear stochastic fractional integrodifferential equation in X of the form
CDαX(t)=AX(t)+b(t,X(t),∫t0f(t,s,X(s))ds)+σ(t,X(t),∫t0g(t,s,X(s))ds)dW(t)dt, t∈J,X(0)=X0,} | (2) |
where 1/2<α≤1,X0∈X and A:X→X is a bounded linear operator. Also the drift coefficient b:J×X×X→X, the noise coefficient σ:J×X×X→LQ(H0;X) and f,g:J×J×X→X. Assume the following Lipschitz conditions on the drift and noise coefficients: For all x1,x2,y1,y2∈X and 0≤s≤t≤T, there exist constants Lb,Lσ,Lf,Lg>0 such that
‖b(t,x1,y1)−b(t,x2,y2)‖X≤Lb[‖x1−x2‖X+‖y1−y2‖X],‖σ(t,x1,y1)−σ(t,x2,y2)‖LQ≤Lσ[‖x1−x2‖X+‖y1−y2‖X],‖f(t,s,x1)−f(t,s,x2)‖X≤Lf‖x1−x2‖X,‖g(t,s,x1)−g(t,s,x2)‖X≤Lg‖x1−x2‖X.} | (3) |
Also assume the following linear growth assumptions on the coefficients: For all x,y∈X and 0≤s≤t≤T, there exist positive constants Kb,Kσ,Kf,Kg>0 such that
‖b(t,x,y)‖2X≤Kb[1+‖x‖2X+‖y‖2X],‖σ(t,x,y)‖2LQ≤Kσ[1+‖x‖2X+‖y‖2X],‖f(t,s,x)‖2X≤Kf[1+‖x‖2X],‖g(t,s,x)‖2X≤Kg[1+‖x‖2X].} | (4) |
Let us first quote some basic definitions from fractional calculus. For α,β>0, with n−1<α<n, n−1<β<n and n∈N, D is the usual differential operator and suppose f∈L1(R+), R+=[0,∞).
(ⅰ) Caputo Fractional Derivative:
The Riemann Liouville fractional integral of a function f is defined as
Iαf(t)=1Γ(α)∫t0(t−s)α−1f(s)ds, |
and the Caputo derivative of f is CDαf(t)=In−αf(n)(t), that is,
CDαf(t)=1Γ(n−α)∫t0(t−s)n−α−1f(n)(s)ds, |
where the function f(t) has absolutely continuous derivatives up to order n−1.
(ⅱ) Mittag-Leffler Operator Function: Two parameter family of Mittag-Leffler operator functions is defined as
Eα,β(A)=∞∑k=0AkΓ(kα+β),α,β>0. |
Here A is the bounded linear operator. In particular, for β=1, the one parameter Mittag-Leffler operator function is
Eα(A)=∞∑k=0AkΓ(kα+1). |
The Mittag-Leffler functions are in fact generalizations of the exponential function and are applicable in varied situations involving fractional derivatives, see for example [9]. Assume the following boundedness on the Mittag-Leffler operator functions with one and two parameters:
M1=supt∈J‖Eα(Atα)‖L(X), M2=supt∈J‖Eα,α(Atα)‖L(X). | (5) |
In order to find the solution representation, we need the following hypothesis and make use of the Lemma that follows.
(H1) The operator A∈L(X) commutes with the fractional integral operator Iα on X and ‖A‖2L(X)<(2α−1)(Γ(α))2T2α.
Lemma 2.1. [17] Suppose that A is a linear bounded operator defined on X (more generally, X may be a Banach space) and assume that ‖A‖L(X)<1. Then (I−A)−1 is linear and bounded. Also
(I−A)−1=∞∑k=0Ak. |
The convergence of the above series is in the operator norm and ‖(I−A)−1‖L(X)≤(1−‖A‖L(X))−1.
We next show that ‖IαA‖L(X)<1 and, by the Lemma, we obtain (I−IαA)−1 is bounded and linear. Let X∈X; then by (H1), we have
E[‖(IαA)X‖2C(J;X)]≤T(Γ(α))2E[supt∈J∫t0(t−s)2α−2‖AX(s)‖2Xds]≤T2α(2α−1)(Γ(α))2E[supt∈J‖AX(t)‖2X]<E‖X‖2C(J;X), |
hence yielding the desired inequality. On the other hand, defining the random differential operator
dF(t,X(t)):= b(t,X(t),∫t0f(t,s,X(s))ds)dt+σ(t,X(t),∫t0g(t,s,X(s))ds)dW(t) |
and operating by Iα on both sides of (2), we have
X(t)=X0+IαAX(t)+IαdF(t,X(t))dt,X(t)=(I−IαA)−1(X0+IαdF(t,X(t))dt). |
Therefore, using Lemma 2.1 and the fact that Iα commutes with A, we obtain (see [3,20])
X(t)=∞∑k=0(IαA)k(X0+IαdF(t,X(t))dt)=∞∑k=0IkαAkX0+IkαAkIαdF(t,X(t))dt=∞∑k=0IkαAkX0+Ikα+αAkdF(t,X(t))dt=∞∑k=0AktαkΓ(kα+1)X0+∫t0(t−s)α−1(∞∑k=0Ak(t−s)αkΓ(kα+α))dF(s,X(s)),=Eα(Atα)X0+∫t0(t−s)α−1Eα,α(A(t−s)α)dF(s,X(s)). |
Thus we obtain the solution representation of (2) as
X(t)=Eα(Atα)X0+∫t0(t−s)α−1Eα,α(A(t−s)α)b(s,X(s),∫s0f(s,τ,X(τ))dτ)ds+∫t0(t−s)α−1Eα,α(A(t−s)α)σ(s,X(s),∫s0g(s,τ,X(τ))dτ)dW(s). | (6) |
We now present some basic definitions and results from large deviation theory. For this, let {Xϵ} be a family of random variables defined on the space X and taking values in a Polish space Z (i.e., a complete separable metric space Z).
Definition 2.1. (Rate Function). A function I:Z→[0,∞] is called a rate function if I is lower semicontinuous. A rate function I is called a good rate function if for each N<∞, the level set KN={f∈Z:I(f)≤N} is compact in Z.
Definition 2.2. (Large Deviation Principle). Let I be a rate function on Z. We say the family {Xϵ} satisfies the large deviation principle with rate function I if the following two conditions hold:
(ⅰ) Large deviation upper bound. For each closed subset F of Z,
lim supϵ→0ϵlogP(Xϵ∈F)≤−I(F). |
(ⅱ) Large deviation lower bound. For each open subset G of Z,
lim infϵ→0ϵlogP(Xϵ∈G)≥−I(G). |
Definition 2.3. (Laplace Principle). Let I be a rate function on Z. We say the family {Xϵ} satisfies the Laplace principle with rate function I if for all real-valued bounded continuous functions h defined on Z,
limϵ→0ϵlogE{exp[−1ϵh(Xϵ)]}=−inff∈Z{h(f)+I(f)}. |
One of the main results of the theory of large deviations is the equivalence between the Laplace principle and the large deviation principle when the underlying space is Polish. For a proof we refer the reader to Theorem 1.2.1 and Theorem 1.2.3 in [11].
Theorem 2.1. The family {Xϵ} satisfies the Laplace principle with good rate function I on a Polish space Z if and only if {Xϵ} satisfies the large deviation principle with the same rate function I.
In this section, we consider the stochastic fractional integrodifferential equation (2) with the random noise term being perturbed by a small parameter ϵ>0 in the form
CDαXϵ(t)=AXϵ(t)+b(t,Xϵ(t),∫t0f(t,s,Xϵ(s))ds)+√ϵσ(t,Xϵ(t),∫t0g(t,s,Xϵ(s))ds)dW(t)dt, t∈(0,T],Xϵ(0)=X0.} | (7) |
Let Gϵ:C(J:H)→Z be a measurable map defined by Gϵ(W(⋅)):=Xϵ(⋅), where Xϵ is the solution of the above equation (7). We implement the variational representation developed by Budhiraja and Dupuis to study the large deviation principle for the solution processes {Xϵ}. Let
A={v:visH0- valuedFt- predictable process and∫T0‖v(s,ω)‖20ds<∞ a.s. }, |
SN={v∈L2(J;H0):∫T0‖v(s)‖20ds≤N}, |
where L2(J;H0) is the space of all H0 -valued square integrable functions on J. Then SN endowed with the weak topology in L2(J;H0) is a compact Polish space (see [10]). Let us also define
AN={v∈A:v(ω)∈SN P−a.s}. |
We now state the variational representation developed by Budhiraja and Dupuis [7,Theorem 4.4] that provides sufficient conditions under which Laplace principle (equivalently, large deviation principle) holds for the family {Xϵ}:
Proposition 3.1. Suppose that there exists a measurable map G0:C(J:H)→Z such that the following hold:
(ⅰ) Let {vϵ:ϵ>0}⊂AN for some N<∞. Let vϵ converge in distribution as SN-valued random elements to v. Then Gϵ(W(⋅)+1√ϵ∫⋅0vϵ(s)ds) converges in distribution to G0(∫⋅0v(s)ds).
(ⅱ) For every N<∞, the set
KN:={G0(.∫0v(s)ds):v∈SN} |
is a compact subset of Z.
For each h∈Z, define
I(h):=inf{v∈L2(J:H0):h=G0(.∫0v(s)ds)}{12∫T0‖v(s)‖20ds}, | (8) |
where the infimum over an empty set is taken as ∞. Then the family {Xϵ:ϵ>0}=Gϵ(W(⋅)) satisfies the Laplace principle in Z with the rate function I given by (8).
In Proposition 3.1, (ⅱ) is a compactness criterion and it is to be noticed that it has a coincidence with the fact that the level set for a good rate function is compact. Thanks to the variational representation prescribed by Budhiraja and Dupuis, the study of large deviation principle for any stochastic differential equation can now be simplified to the problem of identifying Borel measurable function G0 so that the hypothesis in the above proposition is satisfied.
Consider the controlled equation associated to (7) with control v∈SN.
CDαXv(t)=AXv(t)+b(t,Xv(t),∫t0f(t,s,Xv(s))ds)+ σ(t,Xv(t),∫t0g(t,s,Xv(s))ds)v(t), t∈(0,T],Xv(0)=X0,} | (9) |
and let Xv(t) denote the solution of the equation (9). The main result in this chapter is the following Freidlin-Wentzell type theorem:
Theorem 3.1. With the assumption (H1) on the bounded linear operator A, the family {Xϵ(t)} of solutions of (7) satisfies the large deviation principle (equivalently, Laplace principle) in C(J;X) with the good rate function
I(h):=inf{12∫T0‖v(t)‖20dt;Xv=h}, | (10) |
where v∈L2(J;H0) and Xv denotes the solution of the control equation (9) with the convention that the infimum of an empty set is infinity.
In order to prove the theorem, the main work is to verify the sufficient conditions in Proposition 3.1esponding to (7):
CDαXϵv(t)=AXϵv(t)+b(t,Xϵv(t),∫t0f(t,s,Xϵv(s))ds)+ σ(t,Xϵv(t),∫t0g(t,s,Xϵv(s))ds)v(t) + √ϵσ(t,Xϵv(t),∫t0g(t,s,Xϵv(s))ds)dW(t)dt, t∈(0,T],Xϵv(0)=X0.} | (11) |
The solution representation is given by
Xϵv(t)=Eα(Atα)X0+∫t0(t−s)α−1Eα,α(A(t−s)α)b(s,Xϵv(s),∫s0f(s,τ,Xϵv(τ))dτ)ds+∫t0(t−s)α−1Eα,α(A(t−s)α)σ(s,Xϵv(s),∫s0g(s,τ,Xϵv(τ))dτ)v(s)ds+√ϵ∫t0(t−s)α−1Eα,α(A(t−s)α)σ(s,Xϵv(s),∫s0g(s,τ,Xϵv(τ))dτ)dW(s). | (12) |
Before proceeding further analysis, we show that the solution Xϵv(t) obeys the following energy estimate:
Theorem 3.2. The solution Xϵv(t) of (11) is bounded in the space L2(Ω;C(J;X)), that is, there exists a positive constant K>0 such that
E[supt∈J‖Xϵv(t)‖2X]≤K. | (13) |
Proof. First we define the stopping time τN:=inf{t:‖Xϵv(t)‖2≥N}. And, for any t∈[0,T∧τN], consider the solution representation of (11) given by (12), take ‖⋅‖2X on both sides and use the algebraic identity (a+b+c+d)2≤4(a2+b2+c2+d2) to get
‖Xϵv(t)‖2X≤ 4‖Eα(Atα)‖2L(X)‖X0‖2X+ 4‖∫t0(t−s)α−1Eα,α(A(t−s)α)b(s,Xϵv(s),∫s0f(s,τ,Xϵv(τ))dτ)ds‖2X+ 4‖∫t0(t−s)α−1Eα,α(A(t−s)α)σ(s,Xϵv(s),∫s0g(s,τ,Xϵv(τ))dτ)v(s)ds‖2X+ 4ϵ‖∫t0(t−s)α−1Eα,α(A(t−s)α)σ(s,Xϵv(s),∫s0g(s,τ,Xϵv(τ))dτ)dW(s)‖2X. |
Using the Holder inequality and the bounds on ‖Eα(⋅)‖L(X) and ‖Eα,α(⋅)‖L(X) given by (5), we obtain the estimate
‖Xϵv(t)‖2X≤4M21‖X0‖2X+4M22∫t0(t−s)2α−2ds∫t0‖b(s,Xϵv(s),∫s0f(s,τ,Xϵv(τ))dτ)‖2Xds+ 4M22∫t0(t−s)2α−2‖σ(s,Xϵv(s),∫s0g(s,τ,Xϵv(τ))dτ)‖2LQds ∫t0‖v(s)‖20ds+ 4ϵ‖∫t0(t−s)2α−2Eα,α(A(t−s)α)σ(s,Xϵv(s),∫s0g(s,τ,Xϵv(τ))dτ)dW(s)‖2X. |
Now using the linear growth property of 'b' and 'σ' given by (3) results in
‖Xϵv(t)‖2X≤4M21‖X0‖2X+4KbM22T2α−12α−1∫t0[1+‖Xϵv(s)‖2X+‖∫s0f(s,τ,Xϵv(τ))dτ‖2X]ds+ 4KσM22N∫t0(t−s)2α−2[1+‖Xϵv(s)‖2X+‖∫s0g(s,τ,Xϵv(τ))dτ‖2X]ds+ 4ϵ‖∫t0(t−s)α−1Eα,α(A(t−s)α)σ(s,Xϵv(s),∫s0g(s,τ,Xϵv(τ))dτ)dW(s)‖2X. |
Using Holder's inequality for the integrands ‖∫s0f(s,τ,Xϵv(τ))dτ‖2X and ‖∫s0g(s,τ,Xϵv(τ))‖2X and also making use of the linear growth property of 'f' and 'g' given by (4), we get, on simplifying,
‖Xϵv(t)‖2X≤4M21‖X0‖2X+4KbM22T2α−12α−1∫t0[1+‖Xϵv(s)‖2X+KfT∫t0[1+‖Xϵv(τ)‖2X]dτ]ds+ 4KσM22N∫t0(t−s)2α−2[1+‖Xϵv(s)‖2X+KgT∫t0[1+‖Xϵv(τ)‖2X]dτ]ds+ 4ϵ‖∫t0(t−s)α−1Eα,α(A(t−s)α)σ(s,Xϵv(s),∫s0g(s,τ,Xϵv(τ))dτ)dW(s)‖2X. | (14) |
The stochastic integral term can be estimated by means of the Burkholder-Davis-Gundy inequality as
E{sup0≤t≤T∧τN[‖∫t0(t−s)α−1Eα,α(A(t−s)α)σ(s,Xϵv(s),∫s0g(s,τ,Xϵv(τ))dτ)dW(s)‖2X]}≤M22∫T0(T−s)2α−2‖σ(s,Xϵv(s),∫s0g(s,τ,Xϵv(τ))dτ)‖2LQds≤KσM22∫T0(T−s)2α−2[1+‖Xϵv(s)‖2X+KgT∫T0[1+‖Xϵv(τ)‖2X]dτ]ds≤KσM22∫T0(T−s)2α−2[1+‖Xϵv(s)‖2X]ds+KσM22KgTT2α−12α−1∫T0[1+‖Xϵv(s)‖2X]ds. |
Hence (14) becomes, after taking supremum and expectation on both sides and simplifying,
E[sup0≤t≤T∧τN‖Xϵv(t)‖2X]≤4M21E‖X0‖2X+4KbM22(1+KfT2)T2α−12α−1E∫T0[1+‖Xϵv(s)‖2X]ds+ 4KσM22(N+ϵ)E∫T0(T−s)2α−2[1+‖Xϵv(s)‖2X]ds+ 4KσM22KgTT2α−12α−1(N+ϵ)E∫T0[1+‖Xϵv(s)‖2X]ds. |
Further simplifying and applying the well known Gronwall inequality, we end up with
E[sup0≤t≤T∧τN‖Xϵv(t)‖2X]≤ (4M21E‖X0‖2X+CT)eCT=K, | (15) |
where CT=4M22T2α−12α−1[Kb(1+KfT2)T+Kσ(1+KgT)(N+ϵ)]. Observe that T∧τN→T as N→∞, hence resulting in (13).
Lemma 3.1 (Compactness). Define G0:C(J;H)→C(J;X) by
G0(h):={Xv, if h=.∫0v(s)ds for some v∈SN,0, otherwise. |
Then, for each N<∞, the set
KN={G0(.∫0v(s)ds):v∈SN} |
is a compact subset of C(J;X).
Proof. Let {vn} be a sequence of controls from SN that converge weakly to v in L2(J;H0) and let Xvn(t) denote the solution of (9) with control v replaced by vn. Take Yn(t)=Xvn(t)−Xv(t). Then the equation corresponding to Yn(t) would be
CDαYn(t)=AYn(t)+b(t,Xvn(t),∫t0f(t,s,Xvn(s))ds)−b(t,Xv(t),∫t0f(t,s,Xv(s))ds)+σ(t,Xvn(t),∫t0g(t,s,Xvn(s))ds)vn(t)−σ(t,Xv(t),∫t0g(t,s,Xv(s))ds)v(t),Yn(0)=0.} | (16) |
The solution representation is
Yn(t)= ∫t0(t−s)α−1Eα,α(A(t−s)α)[b(s,Xvn(s),∫s0f(s,τ,Xvn(τ))dτ)−b(s,Xv(s),∫s0f(s,τ,Xv(τ))dτ)]ds+∫t0(t−s)α−1Eα,α(A(t−s)α)[σ(s,Xvn(s),∫s0g(s,τ,Xvn(τ))dτ)vn(s)−σ(s,Xv(s),∫s0g(s,τ,Xv(τ))dτ)v(s)]ds=: I1(t)+I2(t)+I3(t), | (17) |
where
I1(t):= ∫t0(t−s)α−1Eα,α(A(t−s)α)[b(s,Xvn(s),∫s0f(s,τ,Xvn(τ))dτ)−b(s,Xv(s),∫s0f(s,τ,Xv(τ))dτ)]ds, | (18) |
I2(t):= ∫t0(t−s)α−1Eα,α(A(t−s)α)[σ(s,Xvn(s),∫s0g(s,τ,Xvn(τ))dτ)−σ(s,Xv(s),∫s0g(s,τ,Xv(τ))dτ)]vn(s)ds, | (19) |
I3(t):=∫t0(t−s)α−1Eα,α(A(t−s)α)σ(s,Xv(s),∫s0g(s,τ,Xv(τ))dτ)(vn(s)−v(s))ds. | (20) |
First consider the integral I1(t) and taking ‖⋅‖X on both sides, we get
‖I1(t)‖X≤∫t0(t−s)α−1‖Eα,α(A(t−s)α)‖L(X) ‖b(s,Xvn(s),∫s0f(s,τ,Xvn(τ))dτ)−b(s,Xv(s),∫s0f(s,τ,Xv(τ))dτ)‖Xds. |
Using the boundedness of ‖Eα,α(⋅)‖ given by (5) and Lipschitz continuity of 'b' from (3), we obtain
‖I1(t)‖X≤ LbM2∫t0(t−s)α−1[‖Yn(s)‖X+∫s0‖f(s,τ,Xvn(τ))−f(s,τ,Xv(τ))‖X dτ]ds. |
Using the Lipschitz continuity of 'f', we get subsequently
‖I1(t)‖X≤ LbM2∫t0(t−s)α−1[‖Yn(s)‖X+Lf∫s0‖Yn(τ)‖X dτ]ds≤ LbM2∫t0(t−s)α−1‖Yn(s)‖Xds+LbLfM2∫t0(t−s)α−1∫t0‖Yn(τ)‖Xdτds= LbM2∫t0(t−s)α−1‖Yn(s)‖Xds+LbLfM2Tαα∫t0‖Yn(s)‖Xds. | (21) |
In a similar way, consider the integral I2(t) and estimating using the boundedness of ‖Eα,α(⋅)‖ and Lipschitz continuity of 'σ', we get
‖I2(t)‖X≤LσM2∫t0(t−s)α−1[‖Yn(s)‖X+∫s0‖g(s,τ,Xvn(τ))−g(s,τ,Xv(τ))‖X dτ]‖vn(s)‖0ds. |
Now, using the Lipschitz continuity of 'g', we obtain
‖I2(t)‖X≤ LσM2∫t0(t−s)α−1‖Yn(s)‖X‖vn(s)‖0ds +LσLgM2∫t0‖Yn(τ)‖Xdτ∫t0(t−s)α−1‖vn(s)‖0ds. | (22) |
For the third integral, applying Holder's inequality, one gets
‖I3(t)‖X≤ ∫t0(t−s)α−1 ‖Eα,α(A(t−s)α)‖L(X)‖σ(s,Xv(s),∫s0g(s,τ,Xv(τ))dτ)(vn(s)−v(s))‖Xds≤ M2(∫t0(t−s)2α−2ds)1/2(∫t0‖σ(s,Xv(s),∫s0g(s,τ,Xv(τ))dτ)(vn(s)−v(s))‖2Xds)1/2≤ M2Tα(∫t0‖σ(s,Xv(s),∫s0g(s,τ,Xv(τ))dτ)(vn(s)−v(s))‖2Xds)1/2, | (23) |
where Tα=Tα−1/2√2α−1. Now (17) becomes, after substituting (21 -(23) and applying Gronwall's inequality,
‖Yn(t)‖X≤ M2Tα(∫t0‖σ(s,Xv(s),∫s0g(s,τ,Xv(τ))dτ)(vn(s)−v(s))‖2Xds)1/2×exp{LbM2∫t0(t−s)α−1ds+LbLfM2Tα+1α+LσM2∫t0(t−s)α−1‖vn(s)‖0ds+LσLgM2T∫t0(t−s)α−1‖vn(s)‖0ds}. |
Applying Holder's inequality to the last two integral terms on the exponential index, one gets
‖Yn(t)‖X≤ M2Tα[∫t0‖σ(s,Xv(s),∫s0g(s,τ,Xv(τ))dτ)(vn(s)−v(s))‖2Xds]1/2×exp{LbM2Tαα+LbLfM2Tα+1α+(LσM2+LσLgM2T)(∫t0(t−s)2α−2ds)1/2(∫t0‖vn(s)‖20ds)1/2}. |
On simplifying and taking supremum over t∈J, we get
supt∈J‖Yn(t)‖X≤ M2Tα[∫T0‖σ(s,Xv(s),∫s0g(s,τ,Xv(τ))dτ)(vn(s)−v(s))‖2Xds]1/2×exp{LbM2Tαα(1+LfT)+LσM2T2α−12α−1(1+LgT)√N}. | (24) |
Since vn⇀v weakly in L2(J;H0) and σ is a Hilbert-Schmidt operator and hence compact, we have that σvn→σv strongly in L2(J;X) and so Yn=Xvn−Xv→0 in C(J;X), thereby proving the compactness.
Lemma 3.2 (Weak Convergence). Let {vϵ:ϵ>0}⊂AN for some N<∞. Assume that vϵ converge to v in distribution as SN-valued random elements; then
Gϵ(W(⋅)+1√ϵ.∫0vϵ(s)ds)→G0(.∫0v(s)ds) |
in distribution as ϵ→0.
Proof. Consider the nonlinear stochastic fractional integrodifferential equation (11) with control vϵ∈L2(J;H0) and let the solution be denoted by Xϵvϵ(t). Take Yϵ(t)=Xϵvϵ(t)−Xv(t). Then
Yϵ(t)= ∫t0(t−s)α−1Eα,α(A(t−s)α)[b(s,Xϵvϵ(s),∫s0f(s,τ,Xϵvϵ(τ))dτ)−b(s,Xv(s),∫s0f(s,τ,Xv(τ))dτ)]ds+∫t0(t−s)α−1Eα,α(A(t−s)α)[σ(s,Xϵvϵ(s),∫s0g(s,τ,Xϵvϵ(τ))dτ)−σ(s,Xv(s),∫s0g(s,τ,Xv(τ))dτ)]vϵ(s)ds+∫t0(t−s)α−1Eα,α(A(t−s)α)σ(s,Xv(s),∫s0g(s,τ,Xv(τ))dτ)(vϵ(s)−v(s))ds+√ϵ∫t0(t−s)α−1Eα,α(A(t−s)α)σ(s,Xϵvϵ(s),∫s0g(s,τ,Xv(τ))dτ)dW(s). |
Taking ‖⋅‖2 on both sides and using the algebraic inequality (a+b+c+d)2≤4(a2+b2+c2+d2), we obtain
‖Yϵ(t)‖2X≤I1(t)+I2(t)+I3(t)+I4(t), | (25) |
where
I1(t):= 4‖∫t0(t−s)α−1Eα,α(A(t−s)α)[b(s,Xϵvϵ(s),∫s0f(s,τ,Xϵvϵ(τ))dτ)−b(s,Xv(s),∫s0f(s,τ,Xv(τ))dτ)]ds‖2X, | (26) |
I2(t):= 4‖∫t0(t−s)α−1Eα,α(A(t−s)α)[σ(s,Xϵvϵ(s),∫s0g(s,τ,Xϵvϵ(τ))dτ)−σ(s,Xv(s),∫s0g(s,τ,Xv(τ))dτ)]vϵ(s)ds‖2X, | (27) |
I3(t):= 4 ‖∫t0(t−s)α−1Eα,α(A(t−s)α)σ(s,Xv(s),∫s0g(s,τ,Xv(τ))dτ)(vϵ(s)−v(s))ds‖2X, | (28) |
I4(t):= 4 ϵ‖∫t0(t−s)α−1Eα,α(A(t−s)α)σ(s,Xϵvϵ(s),∫s0g(s,τ,Xv(τ))dτ)dW(s)‖2X. | (29) |
First consider the integral I1(t) and applying Holder's inequality along with the bound for ‖Eα,α(⋅)‖L(X) given by (5) and the Lipschitz continuity of 'b' given by (3), one gets
I1(t)≤ 4∫t0(t−s)2α−2‖Eα,α(A(t−s)α)‖2L(X)ds×∫t0‖b(s,Xϵvϵ(s),∫s0f(s,τ,Xϵvϵ(τ))dτ)−b(s,Xv(s),∫s0f(s,τ,Xv(τ))dτ)‖2Xds≤4L2bM22∫t0(t−s)2α−2ds∫t0[‖Yϵ(s)‖X+∫s0‖f(s,τ,Xϵvϵ(τ))−f(s,τ,Xv(τ))‖dτ]2ds. |
Using the algebraic identity (a+b)2≤2(a2+b2) and Holder's inequality to the last integral term on the right hand side and then using the Lipschitz continuity of 'f', we obtain simultaneously
I1(t)≤ 8L2bM22T2α−12α−1∫t0[‖Yϵ(s)‖2X+T∫s0‖f(s,τ,Xϵvϵ(τ))−f(s,τ,Xv(τ))‖2X]ds≤ 8L2bM22T2α−12α−1∫t0[‖Yϵ(s)‖2X+L2fT∫s0‖Yϵ(τ)‖2Xdτ]ds. |
On simplifying, the integral I1(t) can be estimated as
I1(t)≤ 8L2bM22(1+L2fT2)T2α−12α−1∫t0‖Yϵ(s)‖2Xds. | (30) |
Similarly consider the integral I2(t), apply Holder's inequality followed by the bound for ‖Eα,α(⋅)‖X and the Lipschitz continuity of 'σ' to get
I2(t)≤4∫t0(t−s)2α−2‖Eα,α(A(t−s)α)‖2L(X)ds×∫t0‖σ(s,Xϵvϵ(s),∫s0g(s,τ,Xϵvϵ(τ))dτ)−σ(s,Xv(s),∫s0g(s,τ,Xv(τ))dτ)‖2LQ‖vϵ(s)‖20ds≤8L2σM22T2α−12α−1∫t0[‖Yϵ(s)‖2X+L2gT∫s0‖Yϵ(τ)‖2Xdτ]‖vϵ(s)‖20ds. |
On further simplifying and making use of the fact that the control variable v∈SN, we obtain
I2(t)≤8L2σM22T2α−12α−1[∫t0‖Yϵ(s)‖2X‖vϵ(s)‖20ds+L2gNT∫t0‖Yϵ(τ)‖2Xdτ]. | (31) |
Now consider the integral I3(t), apply Holder's inequality and the bound for ‖Eα,α(⋅)‖L(X) to obtain
I3(t)≤4M22T2α−12α−1∫t0‖σ(s,Xv(s),∫s0g(s,τ,Xv(τ))dτ)(vϵ(s)−v(s))‖2Xds. | (32) |
Finally consider the stochastic integral I4(t) and taking supremum and then taking expectation on both sides and making use of the Burkholder-Davis-Gundy inequality, we get
E[supt∈JI4(t)]=4ϵE{supt∈J‖∫t0(t−s)α−1Eα,α(A(t−s)α)σ(s,Xϵvϵ(s),∫s0g(s,τ,Xv(τ))dτ)dW(s)‖2X}≤4ϵM22E∫T0(T−s)2α−2‖σ(s,Xϵvϵ(s),∫s0g(s,τ,Xϵvϵ(τ))dτ)‖2LQds. |
Using the linear growth property of 'σ' and 'g' and simplifying, we get
E[supt∈JI4(t)]≤4ϵKσM22E∫T0(T−s)2α−2[1+‖Xϵvϵ(s)‖2X+KgT∫s0(1+‖Xϵvϵ(τ)‖2X)dτ]ds≤4ϵKσM22[E∫T0(T−s)2α−2[1+‖Xϵvϵ(s)‖2X]ds+KgTT2α−12α−1E∫T0[1+‖Xϵvϵ(s)‖2X]ds]≤4ϵKσM22T2α−12α−1(1+KgT2){1+E[supt∈J‖Xϵvϵ(t)‖2X]}. | (33) |
With all these estimates on the integrals Ii(t),i=1,2,3,4, given by (30) -(33), equation (25) becomes, after taking supremum over t∈J and then taking expectation,
E[supt∈J‖Yϵ(t)‖2X]≤8L2bM22(1+L2fT2)T2α−12α−1E∫T0‖Yϵ(s)‖2Xds+ 8L2σM22T2α−12α−1[E∫T0‖Yϵ(s)‖2X‖vϵ(s)‖20ds+L2gNTE∫T0‖Yϵ(τ)‖2Xdτ]+ 4M22T2α−12α−1E∫T0‖σ(s,Xv(s),∫s0g(s,τ,Xv(τ))dτ)(vϵ(s)−v(s))‖2Xds+ 4ϵKσM22T2α−12α−1(1+KgT2){1+E[supt∈J‖Xϵvϵ(t)‖2X]}. |
Applying Gronwall's inequality and further simplifying, we end up with
E[supt∈J‖Yϵ(t)‖2X]≤{4M22T2α−12α−1E∫T0‖σ(s,Xv(s),∫s0g(s,τ,Xv(τ))dτ)(vϵ(s)−v(s))‖2Xds+ 4ϵKσM22T2α−12α−1(1+KgT2)(1+E[supt∈J‖Xϵvϵ(t)‖2X])}×exp(8M22T2α−12α−1[L2b(1+L2fT2)T+L2σ(1+L2gT2)N]). | (34) |
Since σ is a Hilbert-Schmidt operator and hence compact and since vϵ⇀v weakly in L2(J;H0) as ϵ→0, we have that σvϵ→σv strongly in L2(J;X) and so Yϵ=Xϵvϵ−Xv→0 in probability in the space L2(Ω;C(J;X)). Since convergence in probability always implies convergence in expectation, we have finally proved the required weak convergence criterion.
Example 4.1. Consider the stochastic fractional integrodifferential equation with additive noise given by
CDαX(t)=∫t0X(s)ds+sin(X(t))+∫t0√1+X2(s)ds+√ϵdW(t)dt, t∈(0,T],X(0)= X0,} | (35) |
with X0∈R and 1/2<α≤1. The corresponding controlled equation with control v∈L2(0,T;R) takes the form
CDαXv(t)=∫t0Xv(s)ds+sin(Xv(t))+∫t0√1+X2v(s)ds+v(t),t∈(0,T],Xv(0)=X0. |
It is observed that if there exists a unique solution Xv(⋅) for the above mentioned equation, then the control v∈L2([0,T],R) with which the unique solution Xv is attained is also unique and hence the rate function I:C([0,T];R)→[0,∞] is given explicitly by
I(ϕ)=12∫T0|CDαϕ−sinϕ−∫t0(ϕ(s)+√1+ϕ2(s))ds|2dt, | (36) |
if ϕ satisfies (35) for appropriate control v, and ∞ otherwise.
Example 4.2. As an example for (7) with multiplicative type noise, consider the following stochastic equation:
CD3/4X(t)=β∫t0[X(s)+exp(11+X2(s))]ds+√ϵη∫t0X(s)dsdW(t)dt, t∈(0,1],X(0)=1,} | (37) |
where η,β>0 are positive constants. Then the rate function I:C([0,1];R)→[0,∞] is given by
I(ϕ)=inf{12∫10|v(t)|2dt:v∈L2([0,1],R) such that Xv=ϕ}, | (38) |
where inf∅=∞ and Xv is the unique solution of
Xv(t)=1+βΓ(34)∫t01(t−s)1/4∫s0[Xv(r)+exp(11+X2v(r))]drds+ηΓ(34)∫t0v(s)(t−s)1/4∫s0Xv(r)drds,t∈[0,1]. | (39) |
It is evident from (38) that estimating the rate function I(ϕ) is a problem of finding the minimal cost 12∫10|v(t)|2dt, out of all the controls v that steers the desired solution ϕ=Xv from (39).
The first author would like to thank the Department of Science and Technology, New Delhi for their financial support under the INSPIRE Fellowship Scheme. The work of the third author is supported by the University Grants Commission [grant number: MANF-2015-17-TAM-50645] from the Government of India.
All authors declare that there is no conflict of interest.
[1] | K. Balachandran, S. Divya, M. Rivero and J.J. Trujillo, Controllability of nonlinear implicit neutral fractional Volterra integrodifferential systems, Journal of Vibration and Control, 22 (2016), 2165-2172. |
[2] | K. Balachandran, V. Govindaraj, L. Rodrguez-Germa and J.J. Trujillo, Controllability results for nonlinear fractional-order dynamical systems, Journal of Optimization Theory and Applications, 156 (2013), 33-44. |
[3] | K. Balachandran, M. Matar and J. J. Trujillo, Note on controllability of linear fractional dynamical systems, Journal of Control and Decision, 3 (2016), 267-279. |
[4] | L. Bo and Y. Jiang, Large deviation for the nonlocal Kuramoto-Sivashinsky SPDE, Nonlinear Analysis: Theory, Methods and Applications, 82 (2013), 100-114. |
[5] | M. Boue and P. Dupuis, A variational representation for certain functionals of Brownian motion, Annals of Probability, 26(1998), 1641-1659. |
[6] | A. Budhiraja P. Dupuis and V. Maroulas, Large deviations for infinite dimensional stochastic dynamical systems, Annals of Probability, 36 (2008), 1390-1420. |
[7] | A. Budhiraja and P. Dupuis, A variational representation for positive functionals of infinite dimensional Brownian motion, Probability of Mathematics and Statistics, 20 (2000), 39-61. |
[8] | A. Dembo and O. Zeitouni, Large Deviations Techniques and Applications, New York: Springer, 2007. |
[9] | A. Di Crescenzo and A. Meoli, On a fractional alternating Poisson process, AIMS Mathematics, 1(2016), 212-224. |
[10] | N. Dunford and J. Schwartz, Linear Operators, Part Ⅰ, New York: Wiley-Interscience, 1958. |
[11] | P. Dupuis and R. S. Ellis, A Weak Convergence Approach to the Theory of Large Deviations, New York: Wiley-Interscience, 1997. |
[12] | W.H. Fleming, A stochastic control approach to some large deviations problems, Recent Mathematical Methods in Dynamic Programming, Springer Lecture Notes in Math., 1119 (1985), 52-66. |
[13] | M.I. Freidlin and A.D. Wentzell, On small random perturbations of dynamical systems, Russian Mathematical Surveys, 25 (1970), 1-55. |
[14] | M. I. Freidlin and A. D. Wentzell, Random Perturbations of Dynamical Systems, New York: Springer, 1984. |
[15] | R. Joice Nirmala and K. Balachandran, Controllability of nonlinear fractional delay integrodifferential system, Discontinuity, Nonlinearity, and Complexity, 5 (2016), 59-73. |
[16] | M. Kamrani, Numerical solution of stochastic fractional differential equations, Numerical Algorithms, 68 (2015), 81-93. |
[17] | E. Kreyszig, Introductory Functional Analysis with Applications, New York: JohnWiley and Sons Inc, 1978. |
[18] | W. Liu, Large deviations for stochastic evolution equations with small multiplicative noise, Applied Mathematics and Optimization, 61 (2010), 27-56. |
[19] | A. Kilbas, H. M. Srivastava, J. J. Trujillo, Theory and Applications of Fractional Differential Equations, Amsterdam: Elsevier, 2006. |
[20] | R. Mabel Lizzy, K. Balachandran and M. Suvinthra, Controllability of nonlinear stochastic fractional systems with distributed delays in control, Journal of Control and Decision, 4 (2017), 153-167. |
[21] | K. S. Miller and B. Ross, An Introduction to the Fractional Calculus and Fractional Differential Equations, New York: John-Wiley, 1993. |
[22] | C. Mo and J. Luo, Large deviations for stochastic differential delay equations, Nonlinear Analysis: Theory, Methods and Applications, 80 (2013), 202-210. |
[23] | S.A. Mohammed and T.S. Zhang, Large deviations for stochastic systems with memory, Discrete and Continuous Dynamical Systems Series B, 6 (2006), 881-893. |
[24] | J. C. Pedjeu and G. S. Ladde, Stochastic fractional differential equations: Modelling, method and analysis, Chaos, Solitons and Fractals, 45 (2012), 279-293. |
[25] | I. Podlubny, Fractional Differential Equations, London: Academic Press, 1999. |
[26] | J. Ren and X. Zhang, Schilder theorem for the Brownian motion on the diffeomorphism group of the circle, Journal of Functional Analysis, 224 (2005), 107-133. |
[27] | J. Ren and X. Zhang, Freidlin-Wentzell large deviations for homeomorphism flows of non-Lipschitz SDE, Bulletin of Science, 129 (2005), 643-655. |
[28] | S.S. Sritharan and P. Sundar, Large deviations for two dimensional Navier-Stokes equations with multiplicative noise, Stochastic Processes and their Applications, 116 (2006), 1636-1659. |
[29] | M. Suvinthra and K. Balachandran, Large deviations for nonlinear Itôtype stochastic integrodifferential equations, Journal of Applied Nonlinear Dynamics, 6 (2017), 1-15. |
[30] | M. Suvinthra, K. Balachandran and J.K. Kim, Large deviations for stochastic differential equations with deviating arguments, Nonlinear Functional Analysis and Applications, 20 (2015), 659-674. |
[31] | S.R.S. Varadhan, Asymptotic probabilities and differential equations, Communications on Pure and Applied Mathematics, 19 (1966), 261-286. |
1. | Alagesan Siva Ranjani, Murugan Suvinthra, Krishnan Balachandran, Yong-Ki Ma, Analysis of stochastic neutral fractional functional differential equations, 2022, 2022, 1687-2770, 10.1186/s13661-022-01628-8 | |
2. | Khalil S. Al-Ghafri, Awad T. Alabdala, Saleh S. Redhwan, Omar Bazighifan, Ali Hasan Ali, Loredana Florentina Iambor, Symmetrical Solutions for Non-Local Fractional Integro-Differential Equations via Caputo–Katugampola Derivatives, 2023, 15, 2073-8994, 662, 10.3390/sym15030662 |