Abstract
We investigate a semilinear stochastic time-space fractional subdiffusion equation driven by fractionally integrated multiplicative noise. The equation involves the \(\psi \)-Caputo derivative of order \(\alpha \in (0,1)\) and the spectral fractional Laplacian of order \(\beta \in (\frac{1}{2},1]\). The existence and uniqueness of the mild solution are proved in a suitable Banach space by using the Banach contraction mapping principle. The spatial and temporal regularities of the mild solution are established in terms of the smoothing properties of the solution operators.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
1 Introduction
In this paper, we consider the following semilinear stochastic time-space subdiffusion equation driven by fractionally integrated multiplicative noise, with \( a \in \mathbb {R}\),
where \(_{C\psi }{\textrm{D}}_{a,t}^{\alpha } u (t)\) and \(_{\psi }{\textrm{D}}_{a,t}^{-\gamma } u (t)\) denote the \(\psi \)-Caputo fractional derivative of order \(\alpha \in (0,1)\) and the \(\psi \)-fractional integral of order \(\gamma \in [0,1]\) defined in Definitions 2.3 and 2.1, respectively.
Let \(H=L^{2}(\mathcal {D}), H_{0}^{1}(\mathcal {D})\) and \(H^{2}( \mathcal {D})\) be the standard Sobolev spaces. Let \(A=-\Delta \) with \(D(A)=H_{0}^{1}(\mathcal {D})\cap H^{2}(\mathcal {D})\) where \(\mathcal {D} \subset \mathbb {R}^{d}, d=1, 2, 3\) is a bounded domain with the smooth boundary and \(\Delta \) denotes the Laplacian. Let \(A^{\beta }, \, \beta \in (1/2,1]\) be the fractional power of A. Further, W denotes an H-valued Wiener process with covariance operator Q. The nonlinear functions f and g satisfy some properties in Assumptions 2.2, 2.4, 5.1, 5.2. The initial value \(u_{a} \in H\) is a random variable.
Fractional calculus, which includes fractional integrals and derivatives, has various practical and theoretical applications in fields such as physics, chemistry, biology, automatic control systems, anomalous diffusion, stochastic processes, and contaminant transport in underground water flows, among others [4, 8, 13, 18, 34, 35, 37, 41]. The Riemann-Liouville-type fractional calculus, which includes the Riemann-Liouville integral, Riemann-Liouville derivative, and Caputo derivative, has been extensively studied [14, 18, 37, 38]. Some researchers have also focused on the Hadamard-type fractional calculus [15, 17, 27,28,29], which includes the Hadamard integral, Hadamard derivative, and Caputo-Hadamard derivative. These two types of fractional integrals and derivatives have the different convolutional kernels. The Riemann-Liouville fractional integral has a power kernel and the Hadamard fractional integral has a logarithmic kernel. Using a general function \(\psi \) as a convolutional kernel instead of a specific function has significant implications, leading to the development of fractional integrals and derivatives of a function with respect to another function, commonly known as \(\psi \)-fractional calculus (\(\psi \)-fractional integral, \(\psi \)-Riemann-Liouville derivative, and \(\psi \)-Caputo derivative). Recently, some authors have investigated these types of fractional integrals and derivatives [2, 3, 11, 16, 30, 31, 33].
Deterministic fractional differential equations (FDEs) and fractional partial differential equations (FPDEs) have been extensively studied in the literature, with numerous theoretical and numerical analyses [9, 12, 14, 18, 24,25,26]. However, in many practical problems, stochastic perturbations from natural sources cannot be ignored, requiring the consideration of stochastic fractional differential equations (SFDEs) and stochastic fractional partial differential equations (SFPDEs). In this regard, several theoretical results have been achieved for SFPDEs. Cui and Yan [5] discussed the existence of solutions for fractional neutral stochastic integro-differential equations with infinite delay. Sakthivel et al. [39] derived the existence of mild solutions for a class of fractional stochastic differential equations with impulses in Hilbert spaces, utilizing fixed point techniques. In [36], the authors proved the existence and uniqueness of mild solutions for space-time fractional stochastic partial differential equations. Liu et al. [23] demonstrated the existence and uniqueness of solutions to stochastic quasi-linear partial differential equations with time-fractional derivatives. For additional references on this subject, we refer to [1, 6, 7, 20] and the references therein.
In [19], the author studied the following semilinear stochastic evolution equation
and proved the existence and uniqueness of the mild solution to Eq. (1.2) and derived spatial and temporal regularity results. The authors in [20] extended the model (1.2) and considered the following stochastic semilinear space and time fractional subdiffusion problem driven by fractionally integrated additive noise with \(0<\alpha <1\), \(\frac{1}{2}<\beta \le 1\), \(0\le \gamma \le 1\),
where \(^{C}_{0}{\textrm{D}}_{t}^{\alpha }\), \(^{R}_{0}{\textrm{D}}_{t}^{-\gamma }\) and \( A^{\beta }\) denote the Caputo fractional derivative operator, the Riemann-Liouville fractional integral operator and the spectral fractional Laplace operator, respectively. They shown the existence and uniqueness of mild solution to Eq. (1.3) by the Banach contraction mapping principle and further obtained some regularity results of the mild solution. Yang [42] investigated the following stochastic evolution equations with \(\psi \)-Caputo derivative and varying-time delay driven by fractional Brownian motion, with \( \tau >0\),
where \(_{C\psi }{\textrm{D}}_{a,t}^{\alpha } u (t)\) is the \(\psi \)-Caputo fractional derivative operator of order \(\frac{1}{2}<\alpha \le 1\) and \(W^{H_{1}} (t) \) is a H-valued Wiener process with Hurst parameter \(H_{1} \in (\frac{1}{2},1)\). The existence and uniqueness of the mild solution of (1.4) is proved by using the Banach fixed point theorem.
In this paper, we expand upon the equations presented in (1.2)- (1.4) by further exploring the equation in (1.1), which includes the \(\psi \)-Caputo fractional derivative and fractionally integrated multiplicative noise. This equation has significant mathematical and practical applications. Our methodology follows the same idea as Chapter 2 in [19]. We establish a \(L_{p}(\Omega ), p \ge 2\) integrable mild solution to Eq. (1.1) using the well-known Mittag-Leffler functions as solution operators. We investigate the properties of the solution operators and demonstrate the existence and uniqueness of the mild solution to Eq. (1.1) using the Banach contraction mapping principle. Furthermore, we present the spatial and temporal regularity of the mild solution to Eq. (1.1). The challenge of this paper is to develop various smoothing properties associated with Mittag-Leffler operators that involve the general function \(\psi \) and understand the relationships between the parameters \(\alpha , \beta , \gamma \) and the spatial as well as temporal regularities inherent in the mild solution of (1.1).
The paper is organized as follows. In Section 2, we provide some background information, including notations, definitions, and lemmas. In Section 3, we establish the existence, uniqueness of the mild solution to Eq. (1.1). In Section 4, we consider the spatial and temporal regularities of the mild solution. In Section 5, we discuss the further spatial and temporal regularities of the mild solution under strong assumptions for f and g. The conclusion is presented in Section 6 and the Appendix is in final Section 1.
Throughout the paper, we use C to denote a generic positive constant, which may vary at different occurrences and may depend on a and T.
2 Preliminaries
In this section, we shall introduce some notations, definitions, assumptions, and lemmas.
Denote by \((\Omega , \mathcal {F}, \mathbb {P})\) the complete probability space with filtration \((\mathcal {F}_{t})_{t\in [a,T]}\) and by \(W: [a, T]\times \rightarrow H\) a Q-Wiener process on H. Let \( \mathcal {L} (H)=\mathcal {L}(H,H)\) denote the space of bounded linear operators from H to H. Let \(Q\in \mathcal {L} (H)\). The Hilbert space of all Hilbert-Schmidt operators from \(Q^{\frac{1}{2}} (H)\) to H is defined by
where \(\{\varphi _{j}\}_{j=1}^{\infty }\) is the orthonormal basis in H.
Assume that \((\lambda _{k}, \varphi _{k})_{k=1}^{\infty }\) is a sequence of the eigenpairs of \(A: D (A) = H_{0}^{1} ( \mathcal {D}) \cap H^{2} (\mathcal {D}) \rightarrow H\). Let \( r \in \mathbb {R}\) and denote \(\dot{H}^{r}:= D (A^{\frac{r}{2}})\). The norm for \( \upsilon \in \dot{H}^{r}\) is defined by
Let \(L^{p}(\Omega ;\dot{H}^{r}), \, p\ge 2\) be the Hilbert space of \(L^{p} (\Omega )\)-integrable random variables with values in \(\dot{H}^{r}\) such that \(||v ||^{p}_{L^{p}(\Omega ;\dot{H}^{r})}= \mathbb {E}|v|_{r}^{p} <\infty \), where \(\mathbb {E}\) denotes the expectation.
Let
endowed with the norm
It is obvious that \((\mathcal {H}_{p}, ||\cdot ||_{\mathcal {H}_{p}})\) is a Banach space.
Next we introduce the \(\psi \)-fractional integral and derivatives.
Definition 2.1
[31] Let \((a,b)\,(-\infty \le a<b\le \infty )\) be a finite or infinite interval. Suppose that \(\psi (t)\) is a strictly monotone increasing function with a continuous derivative \(\psi '(t)\) and \(\psi (t)\rightarrow +\infty \) as \(t\rightarrow +\infty \). Assume that \( f \in L^{1}(a, b)\). The \(\psi \)-fractional integral of order \(\alpha >0\) is defined by
In particular, if \(\psi (t)=t, \, \log t\), then the \(\psi \)-fractional integral is reduced to the Riemann-Liouville and Hadamard fractional integrals, respectively [18]; If \(\psi (t)=e^{t}\), then it is reduced to the exponential fractional integral [32].
Let AC[a, b] be the space of absolutely continuous function defined on [a, b]. Denote \(AC^{n}_{\delta _{\psi }}[a,b]=\{f:[a,b]\rightarrow \mathbb {R}, \delta _{\psi }^{n-1}f(t)\in AC[a,b]\}\) with \(n=1,2,\ldots \), where \(\delta _{\psi }f(t)= \big ( \frac{1}{\psi '(t)}\frac{{\textrm{d}}}{{\textrm{d}}t} \big ) f(t)\), \(\delta _{\psi }^{n}f(t)=\delta _{\psi }(\delta _{\psi }^{n-1}f(t))\), and \(\delta _{\psi }^{0}f(t)=f(t)\).
Definition 2.2
[31] Let \((a,b)\,(-\infty \le a<b\le \infty )\) be a finite or infinite interval and \(n-1<\alpha <n\in \mathbb {N}\). Suppose that \(\psi (t) \) is a strictly monotone increasing function with \(\psi '(t)\ne 0\) and \(\psi (t)\rightarrow +\infty \) as \(t\rightarrow +\infty \). Assume that \(f \in AC^{n}_{\delta _{\psi }}[a,b]\). The \(\psi \)-Riemann–Liouville fractional derivative of order \(\alpha \) is defined by
Definition 2.3
[31] Let \((a,b)\,(-\infty \le a<b\le \infty )\) be a finite or infinite interval and \(n-1<\alpha <n\in \mathbb {N}\). Suppose that \(\psi (t)\) is a strictly monotone increasing function with \(\psi '(t)\ne 0\) and \(\psi (t)\rightarrow +\infty \) as \(t\rightarrow +\infty \). Assume that \(f \in AC^{n}_{\delta _{\psi }}[a,b]\). The \(\psi \)-Caputo fractional derivative of order \(\alpha \) is defined as
Now we introduce some assumptions. All the parameters \(\alpha , \beta , \gamma \), the functions \(\psi , f\) and the operators A, g in the Assumptions below are related to Eq. (1.1).
Assumption 2.1
The linear operator \(A: D( A)\subset H\rightarrow H\) is densely defined, self-adjoint and positive definite with compact inverse.
We introduce two parameters \(\kappa \) and \(\kappa _{1}\) which are related to the spatial and temporal regularities of the mind solution to Eq. (1.1).
Let \(\alpha \in (0,1)\), \(\beta \in (\frac{1}{2}, 1]\), \(\gamma \in [0,1]\), \(\alpha +\gamma >\frac{1}{2}\) and denote
and
Remark 2.1
We shall prove the solution \( u \in L^{p} (\Omega ; \dot{H}^{s})\) for \(s \in [0, \kappa )\) in Theorem 4.1 where \(\kappa \) is defined by (2.5). When \(\alpha =1, \, \beta =1, \, \gamma =0\), the value \(\kappa \) is reduced to \(\kappa =1\) which is the value used in [19, Theorem 2.25] for the stochastic heat equation.
Remark 2.2
We shall prove the solution u in Theorem 4.2 has the following temporal regularity
for all \( t_{1}, t_{2} \in [a, T]\) where \(\kappa _{1}\) is defined by (2.6). When \(\alpha =1, \, \beta =1, \, \gamma =0\) and \( \kappa =1\), the value \(\kappa _{1}\) is reduced to \(\kappa _{1}=\frac{1}{2}\) which is the value used in [19, Theorem 2.25] for the stochastic heat equation.
Assumption 2.2
Let \(\alpha \in (0,1)\), \(\beta \in (\frac{1}{2}, 1]\), \(\gamma \in [0,1]\) and \(\alpha +\gamma >\frac{1}{2}\). The nonlinear function \(f: [a,T]\times H\rightarrow \dot{H}^{-1}\) satisfies \(|f(a, 0)|_{-1}\le C\) and
for any \(t\in [a,T], h_{1}, h_{2}\in H\). Moreover,
for any \(t_{1}, t_{2}\in [a, T], \, h\in H\), where \(\kappa _{1}\) is defined by (2.6).
For example, we may choose \( f(t, h) = h\) or \( f(t, h) = \sin (h)\).
Remark 2.3
When \( \alpha =1, \, \beta =1, \gamma =0, \kappa =1\) and \( \psi (t)=t\), the power \( \kappa _{1} \) in (2.8) is reduced to \(\frac{1}{2}\) which is the power in [19, Assumption 2.14] for the stochastic heat equation. We may choose \(f(t, h) = f_{1} (\psi (t), h)\) where \(f_{1}(s, h)\) satisfies the Hölder condition with order \(\kappa _{1}\) for the variable s. We need such kind of assumption for f in the proofs of existence, uniqueness and regularities of the mild solution to Eq. (1.1).
Assumption 2.3
The covariance operator \(Q\in \mathcal {L} (H)\) is self-adjoint and positive semidefinite but not necessarily of finite trace.
Assumption 2.4
Let \(\alpha \in (0,1)\) and \(\beta \in (\frac{1}{2}, 1]\). The nonlinear function \(g: [a,T] \times H \rightarrow \mathcal {L}_{2}^{0}\) satisfies \(||g(a, 0)||_{\mathcal {L}_{2}^{0}}\le C\) and
for any \(t\in [a,T], h_{1}, h_{2}\in H\),.
Moreover,
for any \(t_{1}, t_{2}\in [a, T], h\in H\), where \(\kappa _{1}\) is defined in (2.6).
For example, we may choose \( g (t, h) = h\) or \( g(t, h) = \sin (h)\).
Assumption 2.5
Let \(\alpha \in (0,1)\), \(\beta \in (\frac{1}{2}, 1]\), \(\gamma \in [0,1]\), \(p\in [2,\infty )\), and \(\alpha +\gamma >\frac{1}{2}\). Let \(\kappa \) be defined by (2.5). The initial value \(u_{a}: \Omega \rightarrow \dot{H}^{\kappa }\) is a random variable and satisfies
We now introduce the definition and some properties of the Mittag-Leffler function.
Definition 2.4
[37] For \(\sigma >0\) and \(\rho \in \mathbb {R}\), the Mittag-Leffler function is defined as
Lemma 2.1
[37, Theorem 1.4] [40, Theorem 4] For \(0<\sigma <2\), \(\rho \in \mathbb {R}\), and \(\frac{\sigma \pi }{2}<\theta <\min (\pi , \sigma \pi )\), there exists a constant \(C>0\) such that, for \(\theta \le |\arg (z)|\le \pi \),
Below is the definition of the mild solution for Eq. (1.1).
Definition 2.5
A predictable stochastic process \(u: [a,T] \rightarrow H\) is called a \(L_{p}(\Omega )\)-integrable mild solution to Eq. (1.1) with \(p\ge 2\) if
and, for any \(t\in [a,T]\), it holds that \(\mathbb {P}\)-a.s.
where, for any \(w \ge 0\),
Here \(E_{\alpha ,1}( \cdot )\), \(E_{\alpha ,\alpha }( \cdot )\) and \(E_{\alpha ,\alpha +\gamma }( \cdot )\) are the Mittag-Leffler functions.
The following lemma examines the smoothing properties of the solution operators, which will be frequently utilized in the subsequent proofs of lemmas and theorems.
Lemma 2.2
Let \(\alpha \in (0,1)\), \(\beta \in (\frac{1}{2},1]\) and \(\gamma \in [0,1]\). Let \(|\cdot |_{r}, \, r>0\) be the norm defined in (2.1). Then there hold, for any \( \mu \in \mathbb {R}, \nu \in \mathbb {R}\) and \( w\ge 0\),
and
Moreover,
and
where
and
Proof
We only give the proofs for (2.14) and (2.17). The proofs of other inequalities are similar. We first prove (2.14). According to the definition of Mittag-Leffler function, we have
Applying Lemma 2.1 with \(\sigma =\alpha , \rho =1\), we arrive at,
which shows (2.14) where we use the condition \(0\le \mu -\nu \le 2\beta \) in the third inequality.
We now turn to the proof of (2.17). Following the same argument as in the proof of (2.14), we have
Applying Lemma 2.1 with \(\sigma =\alpha , \rho =\alpha -1\) which implies \( \rho - \sigma =-1 \in \mathbb {Z}^{-}\), we arrive at
which completes the proof of (2.17) where we use the condition \(0\le \mu -\nu \le 4 \beta \) in the third inequality. The proof of Lemma 2.2 is now complete. \(\square \)
We also need the following properties for the derivative of solution operators.
Lemma 2.3
Let \( \alpha \in (0, 1)\), \(\beta \in (\frac{1}{2},1]\) and \( \gamma \in [0, 1]\). We then have
and
Proof
We only prove (2.19) here. Similarly one may show (2.20).
which shows (2.19). \(\square \)
We close this section with a Burkholder-Davis-Gundy-type inequality.
Lemma 2.4
[10, Theorem 7.2] For any \(r\ge 1\) and for arbitrary \(\mathcal {L}_{2}^{0}\)-valued predictable process \(\Phi (t), t\in [0, T]\), we have
where
Proof
For the convenience of the readers, we give the sketch of the proof here. By the martingale [10, ineqality (3.8), p. 79] and [10, Corollary 4.14, p. 103] the result is true for \(r=1\). Assume now that \(r>1\), set \(Z(t)=\int _{0}^{t}\Phi (s){\textrm{d}}W(s), t\ge 0\), and apply Ito’s formula to \(f(Z(\cdot ))\) where \(f(x)=|x|^{2r}, x\in H\). Since
and
therefore
By taking the expectation, we obtain
By the Hölder inequality with \(p=r/(r-1)\),
Using the martingales inequality (2.21) [10, p.50] one arrives at
Dividing both sides by \((\mathbb {E}|Z(t)|^{2r})^{1-\frac{1}{r}}\) one gets
which completes the proof of Lemma 2.4. \(\square \)
3 Existence, uniqueness of the solution of Eq. (1.1)
In this section, we shall prove the existence and uniqueness of the mild solution to the semilinear stochastic time-space fractional subdiffusion Eq. (1.1) driven by fractionally integrated multiplicative noise by using the Banach contraction mapping principle.
We first introduce the following lemma which are related to the properties of f and g.
Lemma 3.1
Let \(\alpha \in (0,1)\), \(\beta \in (\frac{1}{2}, 1]\), and \(p\in [2,\infty )\). Let Assumptions 2.2 and 2.4 be fulfilled. Let \(\kappa _{1}\) be defined by (2.6). Then, for any \(t_{1}, t_{2}\in [a,T]\) and \(X, Y \in \mathcal {H}_{p}\), there exists a constant \(C>0\) such that
In particular, there holds
for any \(t\in [a,T]\) and all \(X\in \mathcal {H}_{p}\).
Proof
Note that
Applying Assumption 2.2 it follows that
Using Assumption 2.2 yields
Combining the estimates (3.3) and (3.4) gives the proof of the function f in (3.1). Further Assumption 2.2 implies \(||f(a,0)||_{L^{p} (\Omega ;\dot{H}^{-1})}<\infty \), we then obtain
which gives the estimate for the function f in (3.2). Similarly, we can derive the estimates with respect to the operator g under Assumption 2.4. The proof is completed. \(\square \)
The following theorem establishes existence and uniqueness of the mild solution to Eq. (1.1).
Theorem 3.1
Let \(\alpha \in (0,1)\), \(\beta \in (\frac{1}{2}, 1]\), \(\gamma \in [0,1]\), and \(\alpha +\gamma >\frac{1}{2}\). Let Assumptions 2.1-2.5 be satisfied and \(p\in [2,\infty )\). Then there exists a unique \(L^{p} (\Omega )\)-integrable mild solution \(u: [a,T] \rightarrow H\) to Eq. (1.1) such that \(\mathbb {P}(u(t)\in H)=1\) and
for every \(t\in [a,T]\).
Proof
We shall apply the Banach contraction mapping principle to prove this theorem. We first construct a mapping on the space \(\mathcal {H}_{p}\) and prove that it is well-defined. Then, we introduce a equivalent norm on the space \(\mathcal {H}_{p}\) to show that the mapping is contractive.
Define a mapping \(\mathcal {F}: \mathcal {H}_{p}\rightarrow \mathcal {H}_{p}\) given by
for \(t\in [a,T]\) and \(u\in \mathcal {H}_{p}\).
Step 1: Prove that the mapping \(\mathcal {F}: \mathcal {H}_{p}\rightarrow \mathcal {H}_{p}\) is well-defined.
In view of (2.14) in Lemma 2.2 with \(\mu =\nu =0\) and Assumption 2.5, one has
This implies \(\mathcal {F}_{0}\in \mathcal {H}_{p}\) and \(\mathcal {F}_{0}(t)\) takes almost surely values in H for any \(t\in [a,T]\).
Next, we consider \(\mathcal {F}_{1}(u)(t)\). It follows from (2.15) (\(\mu =1\) and \(\nu =0\)) in Lemma 2.2 and Lemma 3.1 that, noting \(\beta >\frac{1}{2}\),
This states that \(\mathcal {F}_{1}(u)\) is an adapted stochastic process such that
for any \(u\in \mathcal {H}_{p}\) with \(t\in [a,T]\). Therefore,
We continue to investigate \(\mathcal {F}_{2}(u)(t)\). Note that (2.16) (\(\mu =0\) and \(\nu =0\)) in Lemma 2.2, Lemma 2.4, and Lemma 3.1 it follows that, noting \( \alpha + \gamma > \frac{1}{2}\),
which implies
Hence, the random variable \(\mathcal {F}_{2}(u)(t)\) takes almost surely values in H for any \(t\in [a,T]\).
Step 2: Prove that the mapping \(\mathcal {F}: \mathcal {H}_{p}\rightarrow \mathcal {H}_{p}\) is a contraction mapping.
We introduce the following norm, with \(\rho \in \mathbb {R}\),
on \(\mathcal {H}_{p}\). It is easy to verify that the norm \(||\cdot ||_{\mathcal {H}_{p},\rho }\) is equivalent to \(||\cdot ||_{\mathcal {H}_{p}}\) since \(\psi (t)\) is a bounded function on [a, T]. Such equivalent norm was used in [20] when \( \psi (t)= t\).
Note that, by Lemma 3.1,
for all \(\tau \in [a,T]\). Then, for any \(u, v\in \mathcal {H}_{p}\) we obtain by (4.1)
Thus the first summand in (3.11) is estimated by
where we have used (2.15) (\(\mu =0\) and \(\nu =-1\)) in Lemma 2.2 and (3.10). Analogously, the second summand can also be estimated by applying Lemma 2.4, (2.16) (\(\mu =0\) and \(\nu =0\)) in Lemma 2.2 and (3.10)
Therefore we derive
For the first integral above, using Hölder inequality with \(\frac{1}{\mu }+\frac{1}{\mu '}=1\) yields
where we can always choose \(1<\mu '<\frac{1}{1+\frac{\alpha }{2\beta }-\alpha }\) such that \((\alpha -1-\frac{\alpha }{2\beta })\mu '>-1\) due to \(\beta >\frac{1}{2}\). As the second integral, by Hölder inequality with \(\frac{1}{\nu }+\frac{1}{\nu '}=1\), one has
where we can also take \(\nu '>1\) due to \(\alpha +\gamma >\frac{1}{2}\).
Hence we obtain for \(\rho >0\)
Choose \(\rho >0\) large enough, the mapping \(\mathcal {F}\) is a contraction on the space \(\mathcal {H}_{p}\) with respect to the norm \(||\cdot ||_{\mathcal {H}_{p, \rho }}\) and thus there exists a unique fixed point \(u\in \mathcal {H}_{p}\), which is a unique mild solution to Eq. (1.1). The proof of this theorem is now completed. \(\square \)
4 Regularities of the mild solution to Eq. (1.1)
In this section, we shall prove the spatial and temporal regularities for the mild solution to the semilinear stochastic time-space fractional subdiffusion Eq. (1.1) under Assumptions 2.1-2.5.
4.1 Spatial regularity
In this subsection, we shall prove the spatial regularity for the mild solution to Eq. (1.1) under Assumptions 2.1-2.5.
Theorem 4.1
Let \(\alpha \in (0,1)\), \(\beta \in (\frac{1}{2}, 1]\), \(\gamma \in [0,1]\), and \(\alpha +\gamma >\frac{1}{2}\). Let Assumptions 2.1-2.5 be satisfied and \(p\in [2,\infty )\). Let \(\kappa \) be defined in (2.5). Then for any \( s \in [0, \kappa )\) the mild solution u of Eq. (1.1) determined in Theorem 3.1 satisfies
for any \(t \in [a, T]\).
Proof
By Theorem 3.1, Eq. (1.1) admits the following unique mild solution
for \(t\in [a,T]\) and \(u\in \mathcal {H}_{p}\).
In view of (2.14) in Lemma 2.2 with \(\mu =\nu =s\) and Assumption 2.5, one has
This implies \(\mathcal {F}_{0}(t)\) takes almost surely values in \(\dot{H}^{s}\) for any \(s\in [0,\kappa )\) and \(t\in [a,T]\).
Next, we consider \(\mathcal {F}_{1}(u)(t)\). It follows from (2.15)(\(\mu =s+1\) and \(\nu =0\)) in Lemma 2.2 and Lemma 3.1 that, for \( s < 2 \beta -1\),
This states that \(\mathcal {F}_{1}(u)\) is an adapted stochastic process such that
for any \(s\in [0,\kappa )\) and \(t\in [a,T]\). Therefore, for any \(s\in [0,\kappa )\),
We continue to investigate \(\mathcal {F}_{2}(u)(t)\). Note that (2.16) (\(\mu =s\) and \(\nu =0\)) in Lemma 2.2, Lemma 2.4, and Lemma 3.1 it follows that, for \( s < \frac{2 \beta }{\alpha } ( \alpha + \gamma - \frac{1}{2})\),
which implies
Hence, \(\mathcal {F}_{2}(u)(t)\) takes almost surely values in \(\dot{H}^{s}\) for \(s\in [0,\kappa )\) and for any \(t\in [a,T]\).
Together these estimates complete the proof of Theorem 4.1. \(\square \)
4.2 Temporal regularity
In this subsection, we shall prove the temporal regularity for the mild solution to Eq. (1.1) under Assumptions 2.1-2.5.
Theorem 4.2
Let \(\alpha \in (0,1)\), \(\beta \in (\frac{1}{2}, 1]\), \(\gamma \in [0,1]\), and \(\alpha +\gamma >\frac{1}{2}\). Let Assumptions 2.1-2.5 be satisfied and \(p\in [2,\infty )\). Let \(\kappa _{1}\) be defined in (2.6). Then the mild solution u determined in Theorem 3.1 satisfies
for all \(t_{1}, t_{2}\in [a,T]\).
Proof
We again use the mild solution formula defined in (4.1). Applying (2.19), (2.15) (\(\mu =2\beta \) and \(\nu =\kappa \)) and Assumption 2.5 gives
Further we have
By using (2.15) (\(\mu =0\) and \(\nu =-1\)) in Lemma 2.2 and Lemma 3.1 we obtain, for \(\beta > \frac{1}{2}\),
From (2.20) for \( \gamma =0\), (2.17) (\(\mu =0\) and \(\nu =-1\)) in Lemma 2.2, and Lemma 3.1 we arrive at, for \( \beta > \frac{1}{2}\),
Together with the above two estimates yields, for \(\beta > \frac{1}{2}\),
for all \(a\le t_{1}<t_{2}\le T\).
We continue to investigate \(\mathcal {F}_{2}(u)(t)\) in \(\Vert \cdot \Vert _{L^{p}(\Omega ;H)}\) norm. Applying (2.20), (2.16) (\(\mu =0\) and \(\nu =0\)) and (2.18) (\(\mu =0\) and \(\nu =0\)), Lemma 2.4, and Lemma 3.1, it holds that, for \( \alpha + \gamma >\frac{1}{2}\),
that is,
Combining (4.6), (4.10) with (4.11) completes the proof of Theorem 4.2. \(\square \)
5 Further regularities of the mild solution to Eq. (1.1)
Under the following further assumptions for f, g and \(u_{a}\), we may obtain better regularity results of the mild solution to Eq. (1.1) than those obtained in Section 4.
Assumption 5.1
Let \( r \in [0, \kappa )\) where \( \kappa \) is defined by (2.5). The nonlinear function \(f: [a,T] \times H\rightarrow \dot{H}^{-1}\) satisfies
for every \(t\in [a,T], \, h\in \dot{H}^{r}\).
Assumption 5.2
Let \( r \in [0, \kappa )\) where \( \kappa \) is defined by (2.5). The nonlinear operator \(g: [a,T] \times H\rightarrow \mathcal {L}_{2}^{0}\) satisfies
for every \(t\in [a,T], h\in \dot{H}^{r}\).
Assumption 5.3
Let \(\alpha \in (0,1)\), \(\beta \in (\frac{1}{2}, 1]\), \(\gamma \in [0,1]\), \(p\in [2,\infty )\), and \(\alpha +\gamma >\frac{1}{2}\). Let \(\kappa \) be defined by (2.5) and \(r\in [0,\kappa ]\). The initial value \(u_{a}: \Omega \rightarrow \dot{H}^{\kappa +r}\) is a random variable and satisfies
We first give the following two technical lemmas.
Lemma 5.1
Let \(\alpha \in (0,1)\), \(\beta \in (\frac{1}{2}, 1]\), and \(p\in [2,\infty )\). Let Assumptions 2.1-2.4, and Assumptions 5.1-5.3 hold. Let \(\kappa \) and \(\kappa _{1} \) be defined by (2.5) and (2.6), respectively. Given a predictable stochastic process \(X:[a,T]\times \Omega \rightarrow H\) with \(\mathbb {P}(X(t)\in \dot{H}^{r})=1\) for any \(t\in [a,T]\) and
Then for all \(s\in \Big [0, (2\beta -1) + \min (r, \frac{2 \beta }{\alpha }\kappa _{1}) \Big ) \) there exists a constant \(C>0\) such that
for \(a\le t_{1}<t_{2}\le T\). Moreover, if there exists a constant \(C>0\) such that
for all \(t_{1}, t_{2}\in [a,T]\), then one has
Further we have
Proof
We first prove (5.4). By using (2.15) with \( \mu = s+1-r, \, \nu =0\) (note that \(\mu =s+1-r>0\) for any \(s>0\)), and Assumption 5.1, it follows that
which is the desired form (5.4), where we require \( \alpha - \frac{s+1}{2 \beta } \alpha + \frac{\alpha }{2 \beta } r>0\), that is, \(s \in [0, (2\beta -1)+ r)\).
We now show (5.6). In terms of (2.15) (\(\mu =s+1\) and \(\nu =0\)) in Lemma 2.2, Lemma 3.1 and the condition (5.5) we derive
where we require \( \alpha - \frac{s+1}{2 \beta } \alpha + \kappa _{1} >0\), that is, \(s \in \Big [0, (2\beta -1) + \frac{2 \beta }{\alpha }\kappa _{1} \Big )\).
Finally, combining (5.4) with (5.6), we obtain
where we require \(s \in \Big [0, (2\beta -1) + \min ( \kappa _{1}, \frac{\alpha }{2\alpha }r \Big )\). The proof of Lemma 5.1 is now comlete. \(\square \)
Lemma 5.2
Let \(\alpha \in (0,1)\), \(\beta \in (\frac{1}{2}, 1]\), and \(p\in [2,\infty )\). Let Assumptions 2.1-2.4, and Assumptions 5.1-5.3 hold. Let \(\kappa \) and \(\kappa _{1} \) be defined by (2.5) and (2.6), respectively. Given a predictable stochastic process \(X:[a,T]\times \Omega \rightarrow H\) with \(\mathbb {P}(X(t)\in \dot{H}^{r})=1\) for any \(t\in [a,T]\) and
Then for all \(s\in \Big [r, \frac{2 \beta }{\alpha }(\alpha + \gamma - \frac{1}{2}) + \min (r, \frac{2 \beta }{\alpha }\kappa _{1}) \Big ) \) there exists a constant \(C>0\) such that
for any \(a \le t_{1}<t_{2}\le T\). Moreover, if the condition (5.5) is fulfilled for some \( \kappa _{1} \) and for all \(t_{1}, t_{2}\in [a,T]\), then it holds that
Further one has
Proof
First we show (5.8). For any \(a\le t_{1}<t_{2}\le T\), by applying (2.16) with \( \mu =s-r, \, \nu =0\) (note that \( \mu = s-r \ge 0\) since \( s \ge r\)) and Assumption 5.2, it yields that
where we require \( 2 [ (\alpha + \gamma -1) - \frac{s-r}{2 \beta } \alpha ] +1 >0\), that is, \( s \in [r, \frac{2 \beta }{\alpha } ( \alpha + \gamma - \frac{1}{2}) +r)\).
Hence, the proof of (5.8) can be obtained from Assumption 5.2
where we require \( \alpha +\gamma -\frac{1}{2}- \frac{\alpha s}{2 \beta } + \frac{\alpha r}{2 \beta } >0\), that is, \( s \in [r, \frac{2 \beta }{\alpha } ( \alpha + \gamma - \frac{1}{2}) +r)\).
We next estimate (5.9). Based on (2.16) (\(\mu =s\) and \(\nu =0\)) in Lemma 2.2, Lemma 3.1 , the condition (5.5), and Assumption 5.2, we conclude that, with \(g_{2}( \tau )=g(t_{2}, X(t_{2}))\!-\!g(\tau , X(\tau )\),
where we require \( \alpha +\gamma -\frac{1}{2}- \frac{\alpha s}{2 \beta } + \kappa _{1} >0\), that is, \( s \in [r, \frac{2 \beta }{\alpha } ( \alpha + \gamma - \frac{1}{2}) + \frac{2 \beta }{\alpha }\kappa _{1} )\).
Finally, we can use (5.8) and (5.9) to obtain
where we require \( s \in \Big [r, \frac{2 \beta }{\alpha } ( \alpha + \gamma - \frac{1}{2}) + \min (r, \frac{2 \beta }{\alpha }\kappa _{1}) \Big )\). The proof is thus completed. \(\square \)
Now we present the spatial and temporal regularity results of the mild solution (2.13) to Eq. (1.1). The proofs are included in the Appendix.
Theorem 5.1
Let \(\alpha \in (0,1)\), \(\beta \in (\frac{1}{2}, 1]\), and \(p\in [2,\infty )\). Let Assumptions 2.1-2.4, and Assumptions 5.1-5.3 hold. Let \(\kappa \) and \(\kappa _{1} \) be defined by (2.5) and (2.6), respectively. Let \( r_{1} = \min ( r, \frac{2 \beta }{\alpha }\kappa _{1})\). Then the unique mild solution u to Eq. (1.1) satisfies
for any \(t\in [a,T]\). Moreover, there exists a constant \(C>0\) such that
Theorem 5.2
Let \(\alpha \in (0,1)\), \(\beta \in (\frac{1}{2}, 1]\), and \(p\in [2,\infty )\). Let Assumptions 2.1-2.4, and Assumptions 5.1-5.3 hold. Let \(\kappa \) and \(\kappa _{1} \) be defined by (2.5) and (2.6), respectively. Then, for every \(s\in [r,r+\kappa )\), the unique mild solution u(t) to Eq. (1.1) is Hölder continuous with respect to the norm \(||\cdot ||_{L^{p}(\Omega ;\dot{H}^{s})}\) and satisfies
6 Conclusions
This paper deals with a semilinear stochastic time-space fractional evolution equation driven by fractionally integrated multiplicative noise. The equation involves a more general \(\psi \)-Caputo temporal derivative and a spectral fractional Laplacian. We establish the existence and uniqueness of the mild solution using the Banach contraction mapping principle under appropriate assumptions. Based on this, we derive spatial and temporal regularity results for the mild solution. When we set \(\psi (t)=t\), \(a=0\), \(\beta =1\), \(\gamma =0\), and let \(\alpha \rightarrow 1\) in Eq. (1.1), the results presented in this paper reduce to the results in Chapter 2 in [19] for the stochastic heat equation. The model (1.1) with \(\psi \)-Caputo temporal derivative has more applications, for example, describing subdiffusion in a medium having a structure evolving over time [21] or modeling diffusion in a complex system consisting of a matrix and channels [22].
Change history
12 April 2024
A Correction to this paper has been published: https://doi.org/10.1007/s13540-024-00280-6
References
Anh, V.V., Leonenko, N.N., Ruiz-Medina, M.: Space-time fractional stochastic equations on regular bounded open domains. Fract. Calc. Appl. Anal. 19, 1161–1199 (2016). https://doi.org/10.1515/fca-2016-0061
Almeida, R.: A Caputo fractional derivative of a function with respect to another function. Commun. Nonlinear Sci. Numer. Simul. 44, 460–481 (2017)
Almeida, R., Malinowska, A., Monteiro, M.: Fractional differential equations with a Caputo derivative with respect to a kernel function and their applications. Math. Methods Appl. Sci. 41(1), 336–352 (2018)
Berkowitz, B., Klafter, J., Metzler, R., Scher, H.: Physical pictures of transport in heterogeneous media: advection-dispersion, random-walk, and fractional derivative formulations. Water Resour. Res. 38(10), 1191 (2002)
Cui, J., Yan, L.: Existence result for fractional neutral stochastic integro-differential equations with infinite delay. J. Phys. A 44(33), 335201 (2011)
Chen, Z.-Q., Kim, K.H., Kim, P.: Fractional time stochastic partial differential equations. Stoch. Process Appl. 125, 1470–1499 (2015)
Chen, L., Hu, Y., Nualart, D.: Nonlinear stochastic time-fractional slow and fast diffusion equations on \({\mathbb{R} }^{d}\). Stoch. Process Appl. 129, 5073–5112 (2019)
Diethelm, K., Kiryakova, V., Luchko, Y., Machado, J.A.T., Tarasov, V.E.: Trends, directions for further research, and some open problems of fractional calculus. Nonlinear Dyn. 107(4), 3245–3270 (2022)
Diethelm, D.: The Analysis of Fractional Differential Equations. Springer, Berlin (2010)
Da Prato, G., Zabczyk, J.: Stochastic Equations in Infinite Dimensions. Cambridge University Press, Cambridge (1992)
Fan, E.Y., Li, C.P., Stynes, M.: Discretised general fractional derivative. Math. Comput. Simul. 208, 501–534 (2023)
Gao, G.H., Sun, Z.Z., Zhang, H.W.: A new fractional numerical differentiation formula to approximate the Caputo fractional derivative and its applications. J. Comput. Phys. 259, 33–50 (2014)
Hilfer, R.: Applications of Fractional Calculus in Physics. World Scientific, Singapore (2000)
Jin, B.: Fractional Differential Equations. Springer, Cham (2021)
Jarad, F., Abdeljawad, T., Baleanu, D.: Caputo-type modification of the Hadamard fractional derivatives. Adv. Differ. Equ. 2012, 142 (2012)
Jarad, F., Abdeljawad, T.: Generalized fractional derivatives and Laplace transform. Discret. Contin. Dyn. Syst. S 13(3), 709–722 (2020)
Kilbas, A.A.: Hadamard-type fractional calculus. J. Korean Math. Soc. 38(6), 1191–1204 (2001)
Kilbas, A.A., Srivastava, H.M., Trujillo, J.J.: Theory and Applications of Fractional Differential Equations. Elsevier Science, Amsterdam (2006)
Kruse, R.: Strong and Weak Approximation of Semilinear Stochastic Evolution Equations. Springer, Cham (2014)
Kang, W., Egwu, B.A., Yan, Y., Pani, A.K.: Galerkin finite element approximation of a stochastic semilinear fractional subdiffusion with fractionally integrated additive noise. IMA J. Numer. Anal. 42(3), 2301–2335 (2022)
Kosztołowicz, T., Dutkiewicz, A.: Subdiffusion equation with Caputo fractional derivative with respect to another function. Phys. Rev. E 104(1), 014118 (2021)
Kosztołowicz, T., Dutkiewicz, A., Lewandowska, K.D., Wa̧sik, S., Arabski, M.: Subdiffusion equation with Caputo fractional derivative with respect to another function in modeling diffusion in a complex system consisting of a matrix and channels. Phys. Rev. E 106(4), 044138 (2022)
Liu, W., Röckner, M., Da Silva, J.L.: Quasi-linear (stochastic) partial differential equations with time-fractional derivatives. SIAM J. Math. Anal. 50, 2588–2607 (2017)
Lin, Y., Xu, C.: Finite difference/spectral approximations for the time-fractional diffusion equation. J. Comput. Phys. 225, 1533–1552 (2007)
Li, Z.Q., Liang, Z., Yan, Y.: High-order numerical methods for solving time fractional partial differential equations. J. Sci. Comput. 71, 785–803 (2017)
Li, C.P., Cai, M.: Theory and Numerical Approximations of Fractional Integrals and Derivatives. SIAM, Philadelphia (2019)
Li, C.P., Li, Z.Q.: Asymptotic behaviors of solution to Caputo-Hadamard fractional partial differential equation with fractional Laplacian. Int. J. Comput. Math. 98(2), 305–339 (2021)
Li, C.P., Li, Z.Q.: The blow-up and global existence of solution to Caputo-Hadamard fractional partial differential equation with fractional Laplacian. J. Nonlinear Sci. 31(5), 80 (2021)
Li, C.P., Li, Z.Q.: Stability and logarithmic decay of the solution to Hadamard-type fractional differential equation. J. Nonlinear Sci. 31(2), 31 (2021)
Li, C.P., Li, Z.Q.: The finite-time blow-up for semilinear fractional diffusion equations with time \(\psi \)-Caputo derivative. J. Nonlinear Sci. 31(6), 82 (2022)
Li, C.P., Li, Z.Q.: Stability and \(\psi \)-algebraic decay of the solution to \(\psi \)-fractional differential system. Int. J. Nonlinear Sci. Numer. Simul. 24(2), 695–733 (2023)
Li, C.P., Li, Z.Q.: On blow-up for a time-space fractional partial differential equation with exponential kernel in temporal derivative. J. Math. Sci. 266(3), 381–394 (2022)
Li, Z.Q.: Asymptotics and large time behaviors of fractional evolution equations with temporal \(\psi \)-Caputo derivative. Math. Comput. Simul. 196, 210–231 (2022)
Metzler, R., Klafter, J.: The random walk’s guide to anomalous diffusion: a fractional dynamics approach. Phys. Rep. 339(1), 1–77 (2000)
Monje, C.A., Chen, Y.Q., Vinagre, B.M., Xue, D., Feliu, V.: Fractional-Order Systems and Controls: Fundamentals and Applications. Springer, London (2010)
Mijena, J., Nane, E.: Space-time fractional stochastic partial differential equations. Stoch. Process Appl. 125, 3301–3326 (2015)
Podlubny, I.: Fractional Differential Equations. Academic Press, New York (1999)
Samko, S.G., Kilbas, A.A., Marichev, O.I.: Fractional Integrals and Derivatives: Theory and Applications. Gordon and Breach Science Publishers, Amsterdam (1993)
Sakthivel, R., Revathi, P., Ren, Y.: Existence of solutions for nonlinear fractional stochastic differential equations. Nonlinear Anal. 81, 70–86 (2013)
Simon, T.: Comparing Fréchet and positive stable laws. Electron. J. Probab. 19, 1–25 (2014)
Tien, D.N.: Fractional stochastic differential equations with applications to finance. J. Math. Anal. Appl. 397(1), 334–348 (2013)
Yang, M.: Existence uniqueness of mild solutions for \(\psi \)-Caputo fractional stochastic evolution equations driven by fBm. J. Inequal. Appl. 2021, 170 (2021)
Acknowledgements
The authors sincerely thank the anonymous referees for carefully reading the manuscript and their constructive comments and suggestions that improved the quality of the manuscript. The research of Z. Li was supported by the key research and development project of Lvliang city (No. 2022RC11).
Author information
Authors and Affiliations
Corresponding author
Ethics declarations
Conflict of interest
The authors declare no conflicts of interest.
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Appendix
Appendix
1.1 Proof of Theorem 5.1
According to the mild solution formula (2.13) to Eq. (1.1), we have
We now estimate these three terms, respectively. For the first term, we use (2.14) (\(\mu =\nu =r_{1}+\kappa \)) in Lemma 2.2 and Assumption 5.3 to derive
Applying (5.7) in Lemma 5.1 with \(t_{1}=a, t_{2}=t, s=r_{1}+\kappa \le r_{1}+2\beta -1\) and \(X=u\) we can obtain
For the last term, since \(s=r_{1} +\kappa \le r_{1} +\frac{2\beta }{\alpha }(\alpha +\gamma -\frac{1}{2})\), we take \(t_{1}=a, t_{2}=t, X=u\) in (5.10) of Lemma 5.2 and hence get
Combining the above estimates for I, II and III yields (5.11) and the proof of the theorem is thus completed.
1.2 Proof of Theorem 5.2
Applying the mild solution formula (2.13) to Eq. (1.1), for \(a\le t_{1}<t_{2}\le T\) and \(s\in [r,r+\kappa )\), one has
We now estimate these five terms respectively. From (2.15) (\(\mu =s+2\beta \) and \(\nu =r+\kappa \)) in Lemma 2.2, (2.19) in Lemma 2.3 and Assumption 5.3, the first term \(I_{1}\) is estimated by
where we require \(r+\kappa -s >0\), that is, \( s < r + \kappa \).
The second term \(I_{2}\) follows from (5.7) of Lemma 5.1, that is,
where we require \( \alpha -\frac{\alpha }{2\beta }(s+1-r) >0\), that is, \( s < r+ 2 \beta -1\).
For the third term \(I_{3}\), it follows from (2.20) in Lemma 2.3, (2.17) (\(\mu =s\) and \(\nu =-1+r\)) in Lemma 2.2 and Assumption (5.1) that
where we require \( s < r + (2 \beta -1)\).
A straightforward application of (5.10) in Lemma 5.2 gives
where we require \( s < r + \frac{2 \beta }{\alpha } ( \alpha + \gamma - \frac{1}{2})\).
To estimate the last term \(I_{5}\), applying (2.20) in Lemma 2.3, (2.18) in Lemma 2.2 with \(\mu =s-r, \, \nu =0\) ( note that \(\mu =s-r \ge 0\)), and Assumption (5.2), we arrive at
where we require \(\alpha +\gamma -\frac{1}{2}-\frac{\alpha }{2\beta } (s-r) >0\), that is, \( s < r+ \frac{2 \beta }{\alpha } (\alpha + \gamma - \frac{1}{2})\).
Combining the estimates \(I_{1}\)-\(I_{5}\), we obtain the desired result (5.12), and thus conclude the proof.
Rights and permissions
This article is published under an open access license. Please check the 'Copyright Information' section either on this page or in the PDF for details of this license and what re-use is permitted. If your intended use exceeds what is permitted by the license or if you are unable to locate the licence and re-use information, please contact the Rights and Permissions team.
About this article
Cite this article
Li, Z., Yan, Y. Existence, uniqueness and regularity for a semilinear stochastic subdiffusion with integrated multiplicative noise. Fract Calc Appl Anal 27, 487–518 (2024). https://doi.org/10.1007/s13540-024-00244-w
Received:
Revised:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s13540-024-00244-w
Keywords
- \(\psi \)-Caputo derivative
- Spectral fractional Laplacian
- Fractionally integrated multiplicative noise
- existence and uniqueness
- Regularity