Well-posedness of the stochastic Boussinesq equation driven by Levy processes
- Open Access
- 01-12-2019
- Research
Published in:
Activate our intelligent search to find suitable subject content or patents.
Select sections of text to find matching patents with Artificial Intelligence. powered by
Select sections of text to find additional relevant content using AI-assisted search. powered by (Link opens in a new window)
Abstract
1 Introduction
To model various phenomena in climate systems, geophysical, environmental, and so on, the Navier–Stokes equations may be coupled with other equations. Coupled with Navier–Stokes equations and the transport equations for temperature, the Boussinesq equation as a two-dimensional coupled system of Navier–Stokes equations and temperature-dependent transport equations can be used to describe the oceanic gravity currents [3, 9].
Recently there have been a few works related to the stochastic Boussinesq equation concentrating on various topics. For example, [13] and [18] are devoted to well-posedness problems while using different types of solution to deal with low regularity. [1, 19], and [15] study the random dynamics of the Boussinesq systems under diverse conditions on noise or random boundary and show the existence of random attractor. [10] considers the existence, uniqueness, and attraction properties of an ergodic invariant measure for the Boussinesq equations with degenerate Gaussian noise. [7, 16], and [20] progress gradually in the technical level and provide the large deviations principle for the Boussinesq equations with different random conditions. It is worth mentioning that all but a few works above are done under various Gaussian noises. Due to low regularity caused by noise or nonlinear terms, they may use weak, mild, or martingale solutions. However, we find that Levy noise with the characteristic of jump is really special in influencing the driven Boussinesq system, and the classical solution notion may not be appropriate, despite of [13] and [20], especially for further problems like ergodicity. So we adopt the original idea of [8] and put forward with “fragile solution”.
Advertisement
To be specific, this article is concerned with the Boussinesq equations perturbed by Levy noise. The aim is to prove the existence and uniqueness of a special type of global solutions on the condition that Poisson measure possesses a σ-finite characteristic measure. Then we show that the related system possesses invariant measures. The novelty is that we adopt a special type of solution definition (fragile solution) to conquer low regularity, then we apply it successfully by obtaining well-posedness and studying statistically asymptotic behavior. To accomplish this, we develop a novel progressive stopping time technique to obtain the necessary a priori estimates. The critical point is that this method enables us to decompose the difficulties into several parts and tackle with the crucial regular problems. We believe that the current results can be generalized to other nonlocal models or the ones driven by more irregular noise. To illustrate this, in the last section we generalize the argument for stochastic Boussinesq equations to a class of abstract stochastic 2D hydrodynamical type systems driven by Levy noise with suitable modification. Roughly speaking, we can also obtain well-posedness for 2D magneto-hydrodynamic equation, 2D Boussinesq model for the Benard convection, 2D magnetic Benard problem, and 3D Leray α model for Navier–Stokes equations driven by Levy noise.
The rest of the paper is organized as follows. Section 2 is devoted to some preliminaries on notations, lemmas, and especially the definition of fragile solutions. In Sect. 3, we prove the existence and uniqueness of global fragile solution for the Boussinesq equation perturbed by Levy noise. In Sect. 4, we obtain the existence of equilibrium for the considered system. Finally, some discussions on the generalizations to abstract stochastic hydrodynamical equations driven by Levy noise are presented in Sect. 5.
2 Preliminary
In the current paper, we study the stochastic Boussinesq equations driven by Levy white noises
with velocity \(u=u(x,t)=(u_{1},u_{2})\in R^{2}\), salinity \(\theta=\theta (x,t)\in R\), pressure p, \(x=(\xi,\eta)\in D\subset R^{2}\). \(e_{2}\in R^{2}\) is a unit vector in the upward vertical direction. \(Q_{1}\) and \(Q_{2}\) are of trace class. \(W_{1}(\cdot)\) (\(W_{2}(\cdot)\), respectively) is an \((L^{2}(D))^{2}\) (\(L^{2}(D)\), respectively)-valued cylindrical Wiener process. f (g, respectively) is a measurable mapping from some measurable space U to \((L^{2}(D))^{2}\) (\(L^{2}(D)\), respectively). \(\tilde{N}_{1}\) and \(\tilde{N}_{2}\) are compensated Poisson measures on \([0,\infty)\times U\) with intensity measures \(\lambda_{1}\) an \(\lambda _{2}\) being σ-finite measures on \(\mathscr {B}(U)\). Assume that \(W_{i}(t)\) and \(\tilde{N}_{i}(dt, du)\) are independent.
$$ \textstyle\begin{cases} \frac{\partial u}{\partial t}+(u\cdot\nabla)u-\nu\Delta u+\nabla p =\theta e_{2} +Q_{1}\, dW_{1}(t)+\int_{U} f(u(t-),w)\tilde{N}_{1}(dt,dw), \\ \frac{\partial\theta}{\partial t}+(u\cdot\nabla)\theta-k \Delta \theta=u_{2}+Q_{2}\, dW_{2}(t)+\int_{U} g(\theta(t-),w)\tilde{N}_{2}(dt,dw), \\ \nabla\cdot u=0, \\ u|_{\partial D}=0,\qquad u(0)=u_{0},\qquad \theta(0)=\theta_{0}, \end{cases} $$
(2.1)
Let \(L^{2}(D)\) be the Hilbert space. Denote \(\mathbf{H}=(L^{2}(D))^{2}\times L^{2}(D)\) with the scalar product and the induced norm
Set \(\mathbf{V}=V_{1}\times V_{2}=(H^{1}(D))^{2}\times H^{1}(D)\), then V is a product Hilbert space, of which the scalar product and the induced norm are given by
By the classical interpolation inequality, there exists \(C_{1}>0\) such that, for \(u\in V_{1}, \theta\in V_{2}\),
$$(\phi,\psi)= \int_{D}\phi(x)\psi(x)\,dx,\qquad |\phi|^{2}=(\phi, \phi),\quad \forall\phi \in\mathbf{H}. $$
$$(\phi,\psi)_{\mathbf{V}}= \int_{D} \nabla\phi\cdot\nabla\psi \,dx,\qquad \|\phi\| ^{2}_{\mathbf{V}}=(\phi,\phi)_{\mathbf{V}}=\|\phi_{1} \|_{V_{1}}^{2}+\|\phi_{2}\|_{V_{2}}^{2}. $$
$$ |u|^{2}_{(L^{4}(D))^{2}}\leq C_{1} |u|\|u \|_{V_{1}},\qquad |\theta|^{2}_{L^{4}(D)}\leq C_{1}|\theta|\|\theta\|_{V_{2}}. $$
(2.2)
Advertisement
Define an unbounded linear operator \(A=(\nu A_{1},kA_{2}):\mathbf{H}\to \mathbf{H}\) by
where \(D(A_{1})=V_{1}\cap(H^{2}(D))^{2}\), \(D(A_{2})=V_{2}\cap H^{2}(D)\), and \(D(A)=D(A_{1})\times D(A_{2})\).
$$(A_{1}u,v)=(u,u)_{V_{1}}, \qquad (A_{2}\theta,\eta)=( \theta,\eta)_{V_{2}},\quad \forall u,v\in D(A_{1}), \forall \theta,\eta\in D(A_{2}), $$
Introduce the following bilinear operators \(B_{1},B_{2}\) for every \(u,v,w\in V_{1}\), \(\theta,\eta\in V_{2}\):
Denote
$$\begin{aligned}& \bigl(B_{1}(u,v),w \bigr)= \int_{D}[u\cdot\nabla v]w\, dx=\sum _{i,j=1,2} \int_{D}u_{i}\partial _{i}v_{j}w_{j} \, dx, \\& \bigl(B_{2}(u,\theta),\eta \bigr)= \int_{D}[u\cdot\nabla\theta]\eta \,dx=\sum _{i,j=1,2} \int_{D}u_{i}\partial_{i} \theta_{j} \eta_{j}\, dx. \end{aligned}$$
We will need some lemmas with regard to the properties of A and B.
Lemma 2.1
([1, Lemma 2.2])
A
is a positive, self-adjoint operator satisfying
$$(A\phi,\phi)\geq\rho\|\phi\|_{\mathbf{H}}^{2}, \quad \rho=\min( \nu,k), \phi\in D(A). $$
Lemma 2.2
([7, Lemma 3.4])
For
\(u,v,w\in V_{1}\), \(\theta,\eta\in V_{2}\), there holds
(1)
\((B_{1}(u,v),v)=0\), \((B_{2}(u,\theta),\theta)=0\),
(2)
\((B_{1}(u,v),w)=-(B_{1}(u,w),v)\), \((B_{2}(u,\theta),\eta )=-(B_{2}(u,\eta),\theta)\).
Lemma 2.3
([7, Lemma 3.5])
For
\(u\in V_{1}\), \(\theta,\eta\in V_{2}\), \(\phi=(u,\theta)\), there holds
(1)
\(|B_{1}(u,v)|_{V_{1}'}\leq|u|^{2}_{L^{4}}\leq C_{1}\|u\||u|\),
(2)
\(|(B_{2}(u,\theta),\eta)|\leq\|\eta\|\cdot|u|_{L^{4}}\cdot |\theta|_{L^{4}} \leq C_{1}\|\eta\|\cdot|u|^{\frac{1}{2}}\|u\|^{\frac{1}{2}}\cdot|\theta |^{\frac{1}{2}}\|\theta\|^{\frac{1}{2}}\).
To ease notation, set
Note that \(\tilde{N}(ds,dw)\) is the compensated Poisson measure with intensity measure λ. We will consider the following equivalent abstract stochastic evolution equation in the sequel:
$$\begin{aligned} X_{t} =&x- \int^{t}_{0}A X_{s}\, ds- \int^{t}_{0}B(X_{s})\,ds- \int^{t}_{0}R(X_{s})\,ds \\ &{}+ \int^{t}_{0}Q\, dW_{s}+ \int^{t}_{0} \int _{U}F(X_{s-},w)\tilde{N}(ds,dw). \end{aligned}$$
(2.3)
Due to the low regularity of the noise, we cannot expect that the solution X to equation (2.3) is square integrable in time with values in V. So we introduce the following unconventional definition.
Definition 2.1
A fragile solution to equation (2.3) is a progressively measurable stochastic process \(X_{t}\) in \([0,T]\) with
for a.e. \(\omega\in\varOmega\), such that
a.e. for all \(t\in[0,T]\) and all \(\phi\in \mathscr {D}(A)\).
$$X(\omega)\in C\bigl([0,T];\mathbf{H}\bigr)\cap L^{2}\bigl(0,T; \mathscr {D}\bigl(A^{\frac{1}{4}}\bigr)\bigr) $$
$$\begin{aligned}& (X_{t},\phi)+ \int^{t}_{0}(X_{s},A\phi)\,ds+ \int^{t}_{0} \bigl(B(X_{s}, \phi),X_{s} \bigr)\,ds+ \int ^{t}_{0} \bigl(R(X_{s}, \phi),X_{s} \bigr)\,ds \\& \quad =(x,\phi)+ \int^{t}_{0} \int_{U} \bigl(F(X_{s-},u),\phi \bigr) \tilde{N}(ds,du)+(W_{t},Q\phi) \end{aligned}$$
Note that this definition coincides with the generalized solution introduced by [8]. Here we adopt the name “fragile” to distinguish from “weak”.
Now we impose some hypotheses on F, which are also included in [5]. Suppose that \(\{U_{k}\}_{k\geq1}\) are the measurable subsets of U on the condition that \(U_{k}\nearrow U\) and \(\lambda(U_{k})<\infty\). There exist positive constants C and K such that, for some \(\alpha\in[1/4,1/2)\),
(\(\mathrm{H}_{1}\))
\(Q:\mathbf{H}\rightarrow \mathbf{H}\) is a linear bounded operator, whose range \(\mathscr {R}(Q)\) is dense in \(\mathscr {D}(A^{\frac{1}{4}+\frac{\alpha}{2}})\) and \(\mathscr {D}(A^{2\alpha})\subset \mathscr {R}(Q)\subset \mathscr {D}(A^{\frac{1}{4}+\frac{\alpha}{2}+\varepsilon })\) for some \(\varepsilon >0\);
(\(\mathrm{H}_{2}\))
\(\int_{U}|A^{\alpha}F(0,u)|^{2}\lambda(du)=C\);
(\(\mathrm{H}_{3}\))
\(\int_{U}|A^{\alpha}(F(x,u)-F(y,u))|^{2}\lambda (du)\leq K|A^{\alpha}(x-y)|^{2}\);
(\(\mathrm{H}_{4}\))
\(\sup_{x\in H}\int_{U^{c}_{m}}|A^{\alpha}F(x,u)|^{2}\lambda(du)\rightarrow0\), as \(m\rightarrow\infty\).
3 Global solutions
3.1 \(\lambda(U)<\infty\)
Throughout this subsection we suppose that \(\lambda(U)<\infty\). Noting that the character measure \(\lambda(U)\) is finite, we will rearrange the jump times of \(N(dt,du)\). Let the jump times of \(N(dt,du)\) be \(\sigma _{1}(\omega)<\sigma_{2}(\omega)<\cdots\) , then on \([0,\sigma_{1})\), equation (2.3) can be regarded as the equivalence of the following deterministic integral equation:
$$\begin{aligned} X_{t} =&x- \int^{t}_{0}A X_{s}\,ds- \int^{t}_{0}B(X_{s})\,ds- \int^{t}_{0}R(X_{s})\,ds \\ &{}+ \int^{t}_{0}Q\,dW_{s}+ \int^{t}_{0} \int _{U}F(X_{s-},z)\lambda(ds,dz). \end{aligned}$$
(3.1)
It is well known that the Ornstein–Uhlenbeck process z is the solution of
$$\textstyle\begin{cases} dz_{t}+Az_{t}\,dt=Q\,dW_{t}, \\ z_{0}=0. \end{cases} $$
Lemma 3.1
If hypothesis (\(\mathrm{H}_{1}\)) holds, then
\(z\in C_{0}([0,T],\mathscr {D}(A^{1/4+\alpha/2}))\)
a.s.
Proof
Since \(A^{\frac{1}{4}+\frac{\alpha}{2}+\varepsilon }\) is continuous in H by the closed graph theorem and \(A^{-\frac{1}{2}-\frac {\varepsilon }{2}}\) is Hilbert–Schmidt in H, \(A^{-\frac {1}{4}+\frac{\alpha}{2}+\frac{\varepsilon }{2}}Q=A^{-\frac{1}{2}-\frac {\varepsilon }{2}}(A^{\frac{1}{4}+\frac{\alpha}{2}+\varepsilon }Q)\) is a Hilbert–Schmidt operator on H. Let \({e}_{k}\), \(k\in\mathbf{N}\) be the normalized eigenfunctions corresponding to the eigenvalues \(0<\lambda_{1}\leq\lambda_{2}\leq\cdots\) of A. Letting \(Qe_{n}=\sum^{\infty}_{k=1}q_{nk}e_{k}\) and \(\sigma^{2}_{k}=\sum^{\infty}_{n=1}q^{2}_{nk}\), one finds that
With this inequality, one can proceed similarly as [8, p. 411] did to obtain
for some \(\varepsilon _{0}<\frac{1}{2}\varepsilon \). □
$$\sum^{\infty}_{k=1}\sigma^{2}_{k} \lambda_{k}^{-\frac{1}{2}+\alpha+\varepsilon }=\sum^{\infty}_{k=1} \sum^{\infty}_{n=1}q^{2}_{nk} \lambda_{k}^{-\frac {1}{2}+\alpha+\varepsilon }= \bigl\vert A^{-\frac{1}{4}+\frac{\alpha}{2}+\varepsilon }Q \bigr\vert _{\mathrm{HS}}< \infty. $$
$$E|z_{t}|^{4}_{\mathscr {D}(A^{\frac{1}{4}+\frac{\alpha}{2}+\varepsilon _{0}})}\leq 2 \Biggl(\sum ^{\infty}_{n=1}\sigma^{2}_{n} \lambda_{n}^{-\frac{1}{2}+\alpha +2\varepsilon _{0}} \Biggr)^{2} $$
By the change of variable \(Y_{t}=X_{t}-z_{t}\), we transform equation (2.3) into
$$\begin{aligned} Y_{t} =&x- \int^{t}_{0}A Y_{s}\,ds- \int^{t}_{0}B(Y_{s}+z_{s})\,ds- \int^{t}_{0}R(Y_{s}+z_{s})\,ds \\ &{}+ \int^{t}_{0} \int _{U}F(X_{s}+z_{s},z)\lambda(dz,dt). \end{aligned}$$
(3.2)
Theorem 3.1
If hypotheses (\(\mathrm{H}_{1}\))–(\(\mathrm{H}_{3}\)) hold, then for
\(\forall x\in \mathscr {D}(A^{\alpha})\), there exists a unique solution
X
of (3.1) such that, for a.s. \(\omega\in\varOmega\),
The transition semigroup associated with (3.2) is a Feller Markov process.
$$X_{.}(\omega)-z_{.}(\omega)\in C\bigl([0,T],\mathscr {D}\bigl(A^{\alpha}\bigr) \bigr)\cap L^{\frac {4}{1-2\alpha}}\bigl(0,T;\mathscr {D}\bigl(A^{\frac{1}{4}+\frac{\alpha}{2}}\bigr)\bigr) \cap L^{2}\bigl(0,T;\mathscr {D}\bigl(A^{\frac{1}{2}+\alpha}\bigr)\bigr). $$
Proof
Our strategy is to utilize the Galerkin method. Denote \(\mathbf{E}_{n}=\operatorname{span}\{e_{1},e_{2},\ldots,e_{n}\}\), \(P_{n}:\mathbf {H}\rightarrow\mathbf{E}_{n}\) is an orthonormal projection. Consider the following ordinary differential equation:
where \(z_{t}^{n}\) is the Ornstein–Uhlenbeck process satisfying
Multiplying equation (3.3) by \(Y^{n}_{t}\) yields
where \(C_{1}\) is a positive constant, whose value may change from one line to another. By the way, we will abuse notation in this way in the sequel. By Gronwall’s inequality, we get
From this we conclude that \(\{Y^{n}_{t}\}_{n\geq1}\) is bounded in \(L^{\infty}(0,T;\mathbf{H})\cap L^{2}(0,T;\mathbf{V})\), because \(z^{n}_{t}\) has continuous trajectories with values in \(\mathscr {D}(A^{\frac {1}{4}})\subset[L^{4}(\mathscr {O})]^{2}\). Consequently, \(Y^{n}_{t}\) cannot explode in finite time.
$$ \textstyle\begin{cases} \frac{dY_{t}^{n}}{dt}=-A Y_{s}^{n}-P_{n}B(Y_{s}^{n}+z_{s}^{n})-P_{n}R(Y_{s}^{n}+z_{s}^{n})+\int_{U}F(Y_{s}^{n}+z_{s}^{n},u)\lambda (du), \\ Y^{n}_{0}=P_{n}x, \end{cases} $$
(3.3)
$$\textstyle\begin{cases} dz^{n}_{t}+Az^{n}_{t}\,dt=P_{n}Q\, dW_{t}, \\ z^{n}_{0}=0. \end{cases} $$
$$\begin{aligned}& \frac{1}{2}\frac{d}{dt} \bigl\vert Y^{n}_{t} \bigr\vert ^{2}+ \bigl\Vert Y^{n}_{t} \bigr\Vert ^{2} \\& \quad = - \bigl(P_{n}B \bigl(Y_{t}^{n}+z_{t}^{n} \bigr),Y^{n}_{t} \bigr)- \bigl(P_{n}R \bigl(Y_{t}^{n}+z_{t}^{n} \bigr),Y^{n}_{t} \bigr)+ \int _{U} \bigl(F \bigl(Y_{s}^{n}+z_{s}^{n},u \bigr),Y^{n}_{t} \bigr)\lambda(du) \\& \quad \leq \bigl\vert \bigl(B \bigl(Y_{t}^{n},Y_{t}^{n} \bigr),z^{n}_{t} \bigr) \bigr\vert + \bigl\vert \bigl(B \bigl(z_{t}^{n},Y_{t}^{n} \bigr),z^{n}_{t} \bigr) \bigr\vert + \bigl\vert Y_{t}^{n}+z_{t}^{n} \bigr\vert \bigl\vert Y_{t}^{n} \bigr\vert + \int _{U} \bigl(F \bigl(Y_{s}^{n}+z_{s}^{n},u \bigr),Y^{n}_{t} \bigr)\lambda(du) \\& \quad \leq \frac{1}{2} \bigl\Vert Y_{t}^{n} \bigr\Vert ^{2}+C_{1} \bigl\vert z^{n}_{t} \bigr\vert ^{4}_{L^{4}} \bigl\vert Y^{n}_{t} \bigr\vert ^{2}+C_{1} \bigl\vert z^{n}_{t} \bigr\vert ^{4}_{L^{4}}+ \bigl\vert Y_{t}^{n}+z_{t}^{n} \bigr\vert \bigl\vert Y_{t}^{n} \bigr\vert \\& \qquad {} + \int_{U} \bigl\vert Y_{t}^{n} \bigr\vert \bigl\vert F \bigl(Y_{s}^{n}+z_{s}^{n},u \bigr)-F(0,u)+F(0,u) \bigr\vert \lambda(du) \\& \quad \leq \frac{1}{2} \bigl\Vert Y_{t}^{n} \bigr\Vert ^{2}+C_{1} \bigl\vert z^{n}_{t} \bigr\vert ^{4}_{L^{4}} \bigl\vert Y^{n}_{t} \bigr\vert ^{2}+C_{1} \bigl\vert z^{n}_{t} \bigr\vert ^{4}_{L^{4}}+ \bigl\vert Y_{t}^{n}+z_{t}^{n} \bigr\vert \bigl\vert Y_{t}^{n} \bigr\vert +\lambda (U) \bigl\vert Y_{t}^{n} \bigr\vert ^{2} \\& \qquad {} +\frac{2K}{\lambda^{\alpha}_{1}} \bigl[ \bigl\vert Y_{t}^{n} \bigr\vert ^{2}+ \bigl\vert z_{t}^{n} \bigr\vert ^{2} \bigr]+\frac{C}{\lambda ^{\alpha}_{1}} \\& \quad \leq \frac{1}{2} \bigl\Vert Y_{t}^{n} \bigr\Vert ^{2}+ \biggl[2+\lambda(U)+C_{1} \bigl\vert z^{n}_{t} \bigr\vert ^{4}_{L^{4}}+ \frac {2K}{\lambda^{\alpha}_{1}} \biggr] \bigl\vert Y^{n}_{t} \bigr\vert ^{2}+ \biggl(1+\frac{2K}{\lambda^{\alpha}_{1}} \biggr) \bigl\vert z_{t}^{n} \bigr\vert ^{2} \\& \qquad {} +C_{1} \bigl\vert z^{n}_{t} \bigr\vert ^{4}_{L^{4}}+\frac{C}{\lambda^{\alpha}_{1}}, \end{aligned}$$
$$\begin{aligned}& \sup_{0\leq t\leq T} \bigl\vert Y^{n}_{t} \bigr\vert ^{2}+ \int^{T}_{0} \bigl\Vert Y_{t}^{n} \bigr\Vert ^{2}\,dt \\& \quad \leq \vert x \vert ^{2}e^{\int^{T}_{0}[4+2\lambda(U)+2C_{1} \vert z^{n}_{t} \vert ^{4}_{L^{4}}+\frac {4K}{\lambda^{\alpha}_{1}}]\,dt} \\& \qquad {}+ \int^{T}_{0}e^{\int^{t}_{0}[4+2\lambda(U)+2C_{1} \vert z^{n}_{t} \vert ^{4}_{L^{4}}+\frac {4K}{\lambda^{\alpha}_{1}}]\,ds} \biggl[ \biggl(2+ \frac{4K}{\lambda^{\alpha}_{1}} \biggr) \bigl\vert z_{t}^{n} \bigr\vert ^{2}+2C_{1} \bigl\vert z^{n}_{t} \bigr\vert ^{4}_{L^{4}}+\frac{2C}{\lambda^{\alpha}_{1}} \biggr] \,dt. \end{aligned}$$
(3.4)
Multiplying equation (3.3) by \(A^{2\alpha}Y^{n}_{t}\) and applying the interpolation inequality
we have
Gronwall’s inequality implies
Since \(z^{n}_{t}\) has continuous trajectories with values in \(\mathscr {D}(A^{\frac{1}{4}+\frac{\alpha}{2}})\subset \mathscr {D}(A^{\alpha})\), we derive from this and (3.4) that \(\{Y^{n}_{t}\}_{n\geq1}\) is bounded in \(L^{\infty}(0,T;\mathscr {D}(A^{\alpha}))\cap L^{2}(0,T;\mathscr {D}(A^{\frac {1}{2}+\alpha}))\). Noting the interpolation inequality
we get \(Y\in L^{\frac{4}{1-2\alpha}}(0,T;\mathscr {D}(A^{\frac {1}{4}+\frac{\alpha}{2}}))\). Utilizing weakly and weakly ∗-convergent subsequences and noting that \(z^{n}\) is strongly convergent to z, one obtains the existence of a solution \(Y\in L^{\infty}(0,T;\mathscr {D}(A^{\alpha}))\cap L^{2}(0,T;\mathscr {D}(A^{\frac{1}{2}+\alpha}))\cap L^{\frac{4}{1-2\alpha}}(0,T;\mathscr {D}(A^{\frac{1}{4}+\frac{\alpha}{2}}))\). By the classical argument (cf. [17, Chap. 3]), we get \(Y\in C(0,T;\mathscr {D}(A^{\alpha}))\).
$$ \bigl\vert A^{\frac{1}{4}+\frac{\alpha}{2}}Y^{n}_{t} \bigr\vert ^{2}\leq C_{1} \bigl\vert A^{\alpha}Y^{n}_{t} \bigr\vert \bigl\vert A^{\frac{1}{2}}Y^{n}_{t} \bigr\vert , $$
(3.5)
$$\begin{aligned}& \frac{1}{2}\frac{d}{dt} \bigl\vert A^{\alpha}Y^{n}_{t} \bigr\vert ^{2}+ \bigl\vert A^{\frac{1}{2}+\alpha }Y^{n}_{t} \bigr\vert ^{2} \\& \quad = - \bigl(P_{n}B \bigl(Y_{t}^{n}+z_{t}^{n} \bigr),A^{2\alpha}Y^{n}_{t} \bigr)- \bigl(A^{\alpha }P_{n}R \bigl(Y_{t}^{n}+z_{t}^{n} \bigr),A^{\alpha}Y^{n}_{t} \bigr) \\& \qquad {} + \int_{U} \bigl(A^{\alpha}F \bigl(Y_{s}^{n}+z_{s}^{n},u \bigr),A^{\alpha}Y^{n}_{t} \bigr)\lambda(du) \\& \quad \leq C_{1} \bigl\vert A^{\frac{1}{4}+\frac{\alpha}{2}} \bigl(Y^{n}_{t}+z^{n}_{t} \bigr) \bigr\vert ^{2} \bigl\vert A^{\alpha +\frac{1}{2}}Y^{n}_{t} \bigr\vert + \bigl\vert \bigl(A^{\alpha}P_{n}R \bigl(Y_{t}^{n}+z_{t}^{n} \bigr),A^{\alpha}Y^{n}_{t} \bigr) \bigr\vert \\& \qquad {} + \biggl\vert \int_{U} \bigl(A^{\alpha}F \bigl(Y_{s}^{n}+z_{s}^{n},u \bigr),A^{\alpha}Y^{n}_{t} \bigr)\lambda (du) \biggr\vert \\& \quad \leq C_{1} \bigl\vert A^{\frac{1}{4}+\frac{\alpha}{2}} \bigl(Y^{n}_{t}+z^{n}_{t} \bigr) \bigr\vert ^{2} \bigl\vert A^{\alpha +\frac{1}{2}}Y^{n}_{t} \bigr\vert + \bigl\vert A^{\alpha} \bigl(Y_{t}^{n}+z_{t}^{n} \bigr) \bigr\vert \bigl\vert A^{\alpha}Y^{n}_{t} \bigr\vert +\lambda(U) \bigl\vert A^{\alpha}Y^{n}_{t} \bigr\vert ^{2} \\& \qquad {} +2K \bigl[ \bigl\vert A^{\alpha}Y^{n}_{t} \bigr\vert ^{2}+ \bigl\vert A^{\alpha}z^{n}_{t} \bigr\vert ^{2} \bigr]+C \\& \quad \leq 2C_{1} \bigl\vert A^{\alpha}Y^{n}_{t} \bigr\vert \bigl\vert A^{\frac{1}{2}} Y^{n}_{t} \bigr\vert \bigl\vert A^{\alpha+\frac {1}{2}} Y^{n}_{t} \bigr\vert +2C_{1} \bigl\vert A^{\frac{1}{4}+\frac{\alpha}{2}} z^{n}_{t} \bigr\vert ^{2} \bigl\vert A^{\alpha +\frac{1}{2}} Y^{n}_{t} \bigr\vert \\& \qquad {} + \bigl(2+\lambda(U)+2K \bigr) \bigl\vert A^{\alpha}Y^{n}_{t} \bigr\vert ^{2}+(2K+1) \bigl\vert A^{\alpha}z^{n}_{t} \bigr\vert ^{2}+C \\& \quad \leq \frac{1}{2} \bigl\vert A^{\frac{1}{2}+\alpha}Y^{n}_{t} \bigr\vert ^{2}+C_{2} \bigl\vert A^{\frac{1}{2}} Y^{n}_{t} \bigr\vert ^{2} \bigl\vert A^{\alpha}Y^{n}_{t} \bigr\vert ^{2}+C_{2} \bigl\vert A^{\frac{1}{4}+\frac{\alpha}{2}} z^{n}_{t} \bigr\vert ^{4}+ \bigl(2+\lambda(U)+2K \bigr) \bigl\vert A^{\alpha}Y^{n}_{t} \bigr\vert ^{2} \\& \qquad {} +(2K+1) \bigl\vert A^{\alpha}z^{n}_{t} \bigr\vert ^{2}+C \\& \quad \leq \frac{1}{2} \bigl\vert A^{\frac{1}{2}+\alpha}Y^{n}_{t} \bigr\vert ^{2}+ \bigl(2+\lambda (U)+2K+C_{2} \bigl\vert A^{\frac{1}{2}} Y^{n}_{t} \bigr\vert ^{2} \bigr) \bigl\vert A^{\alpha}Y^{n}_{t} \bigr\vert ^{2}+C_{2} \bigl\vert A^{\frac {1}{4}+\frac{\alpha}{2}} z^{n}_{t} \bigr\vert ^{4} \\& \qquad {} +(2K+1) \bigl\vert A^{\alpha}z^{n}_{t} \bigr\vert ^{2}+C. \end{aligned}$$
$$\begin{aligned}& \sup_{0\leq t\leq T} \bigl\vert A^{\alpha}Y^{n}_{t} \bigr\vert ^{2}+ \int^{T}_{0} \bigl\vert A^{\frac {1}{2}+\alpha}Y_{t}^{n} \bigr\vert ^{2}\,dt \\& \quad \leq \bigl\vert A^{\alpha}x \bigr\vert ^{2}e^{\int^{T}_{0}[4+2\lambda(U)+4K+2C_{2} \vert A^{\frac{1}{2}} Y^{n}_{t} \vert ^{2}]\,dt} + \int^{T}_{0}e^{\int^{t}_{0}[4+2\lambda(U)+4K+2C_{2} \vert A^{\frac{1}{2}} Y^{n}_{s} \vert ^{2}]\,ds} \bigl[2C_{2} \bigl\vert A^{\frac{1}{4}+\frac{\alpha}{2}} z^{n}_{t} \bigr\vert ^{4} \\& \qquad {}+2(2K+1) \bigl\vert A^{\alpha}z^{n}_{t} \bigr\vert ^{2}+2C \bigr]\,dt. \end{aligned}$$
(3.6)
$$ \bigl\vert A^{\frac{1}{4}+\frac{\alpha}{2}}Y^{n}_{t} \bigr\vert \leq C_{1} \bigl\vert A^{\alpha}Y^{n}_{t} \bigr\vert ^{\frac {1}{2}+\alpha} \bigl\vert A^{\frac{1}{2}+\alpha}Y^{n}_{t} \bigr\vert ^{\frac{1}{2}-\alpha}, $$
(3.7)
Finally, we examine the uniqueness. Suppose that \(Y^{(1)}\) and \(Y^{(2)}\) are two solutions, then
Therefore, by Lemma 2.3,
Applying Gronwall’s inequality, we arrive at the uniqueness. □
$$\begin{aligned}& \frac {d(Y^{(1)}_{t}-Y^{(2)}_{t})}{dt}+A \bigl(Y^{(1)}_{t}-Y^{(2)}_{t} \bigr)+B \bigl(Y^{(1)}_{t}-Y^{(2)}_{t},Y^{(1)}_{t}+z_{t} \bigr) +B \bigl(Y^{(2)}_{t}+z_{t},Y^{(1)}_{t}-Y^{(2)}_{t} \bigr) \\& \qquad {}+R \bigl(Y^{(1)}_{t}-Y^{(2)}_{t},Y^{(1)}_{t}+z_{t} \bigr) +R \bigl(Y^{(2)}_{t}+z_{t},Y^{(1)}_{t}-Y^{(2)}_{t} \bigr) \\& \quad = \int_{U} \bigl[F \bigl(Y^{(2)}_{t}+z_{t},u \bigr)-F \bigl(Y^{(1)}_{t}+z_{t},u \bigr) \bigr] \lambda(du). \end{aligned}$$
$$\begin{aligned}& \frac{1}{2}\frac{d \vert Y^{(1)}_{t}-Y^{(2)}_{t} \vert ^{2}}{dt}+ \bigl\Vert Y^{(1)}_{t}-Y^{(2)}_{t} \bigr\Vert ^{2} \\& \quad \leq \bigl\vert \bigl(B \bigl(Y^{(1)}_{t}-Y^{(2)}_{t},Y^{(1)}_{t}+z_{t} \bigr),Y^{(1)}_{t}-Y^{(2)}_{t} \bigr) \bigr\vert + \bigl\vert \bigl(B \bigl(Y^{(2)}_{t}+z_{t},Y^{(1)}_{t}-Y^{(2)}_{t} \bigr),Y^{(1)}_{t}-Y^{(2)}_{t} \bigr) \bigr\vert \\& \qquad {} + \bigl\vert \bigl(R \bigl(Y^{(1)}_{t}-Y^{(2)}_{t},Y^{(1)}_{t}+z_{t} \bigr),Y^{(1)}_{t}-Y^{(2)}_{t} \bigr) \bigr\vert + \bigl\vert \bigl(R \bigl(Y^{(2)}_{t}+z_{t},Y^{(1)}_{t}-Y^{(2)}_{t} \bigr),Y^{(1)}_{t}-Y^{(2)}_{t} \bigr) \bigr\vert \\& \qquad {} + \int_{U} \bigl\vert \bigl(F \bigl(Y^{(2)}_{t}+z_{t},u \bigr)-F \bigl(Y^{(1)}_{t}+z_{t},u \bigr),Y^{(1)}_{t}-Y^{(2)}_{t} \bigr) \bigr\vert \lambda(du) \\& \quad \leq C_{1} \bigl\Vert Y^{(1)}_{t}-Y^{(2)}_{t} \bigr\Vert \bigl\vert Y^{(1)}_{t}-Y^{(2)}_{t} \bigr\vert _{L^{4}} \bigl\vert Y^{(1)}_{t}+z_{t} \bigr\vert _{L^{4}} \\& \qquad {}+C_{1} \bigl\Vert Y^{(1)}_{t}-Y^{(2)}_{t} \bigr\Vert \bigl\vert Y^{(1)}_{t}-Y^{(2)}_{t} \bigr\vert _{L^{4}} \bigl\vert Y^{(2)}_{t}+z_{t} \bigr\vert _{L^{4}} \\& \qquad {} + \bigl\vert Y^{(1)}_{t}-Y^{(2)}_{t} \bigr\vert \cdot \bigl\vert Y^{(1)}_{t}+z_{t} \bigr\vert \cdot \bigl\vert Y^{(1)}_{t}-Y^{(2)}_{t} \bigr\vert + \bigl\vert Y^{(1)}_{t}-Y^{(2)}_{t} \bigr\vert \cdot \bigl\vert Y^{(2)}_{t}+z_{t} \bigr\vert \cdot \bigl\vert Y^{(1)}_{t}-Y^{(2)}_{t} \bigr\vert \\& \qquad {} + \int_{U} \bigl\vert F \bigl(Y^{(2)}_{t}+z_{t},u \bigr)-F \bigl(Y^{(1)}_{t}+z_{t},u \bigr) \bigr\vert \cdot \bigl\vert Y^{(1)}_{t}-Y^{(2)}_{t} \bigr\vert \lambda(du) \\& \quad \leq \frac{1}{4} \bigl\Vert Y^{(1)}_{t}-Y^{(2)}_{t} \bigr\Vert ^{2}+C_{2} \bigl\vert Y^{(1)}_{t}-Y^{(2)}_{t} \bigr\vert _{L^{4}}^{2} \bigl\vert Y^{(1)}_{t}+z_{t} \bigr\vert _{L^{4}}^{2}+C_{2} \bigl\vert Y^{(1)}_{t}-Y^{(2)}_{t} \bigr\vert _{L^{4}}^{2} \bigl\vert Y^{(2)}_{t}+z_{t} \bigr\vert _{L^{4}}^{2} \\& \qquad {} + \bigl[ \bigl\vert Y^{(1)}_{t}+z_{t} \bigr\vert + \bigl\vert Y^{(2)}_{t}+z_{t} \bigr\vert \bigr] \bigl\vert Y^{(1)}_{t}-Y^{(2)}_{t} \bigr\vert ^{2}+\lambda (U) \bigl\vert Y^{(1)}_{t}-Y^{(2)}_{t} \bigr\vert ^{2}+\frac{1}{\lambda^{\alpha}} \bigl\vert Y^{(1)}_{t}-Y^{(2)}_{t} \bigr\vert ^{2} \\& \quad \leq \frac{1}{4} \bigl\Vert Y^{(1)}_{t}-Y^{(2)}_{t} \bigr\Vert ^{2}+C_{2} \bigl\Vert Y^{(1)}_{t}-Y^{(2)}_{t} \bigr\Vert \bigl\vert Y^{(1)}_{t}-Y^{(2)}_{t} \bigr\vert \bigl\vert Y^{(1)}_{t}+z_{t} \bigr\vert _{L^{4}}^{2} \\& \qquad {} +C_{2} \bigl\Vert Y^{(1)}_{t}-Y^{(2)}_{t} \bigr\Vert \bigl\vert Y^{(1)}_{t}-Y^{(2)}_{t} \bigr\vert \bigl\vert Y^{(2)}_{t}+z_{t} \bigr\vert _{L^{4}}^{2} \\& \qquad {} + \bigl[ \bigl\vert Y^{(1)}_{t}+z_{t} \bigr\vert + \bigl\vert Y^{(2)}_{t}+z_{t} \bigr\vert \bigr] \bigl\vert Y^{(1)}_{t}-Y^{(2)}_{t} \bigr\vert ^{2}+\lambda (U) \bigl\vert Y^{(1)}_{t}-Y^{(2)}_{t} \bigr\vert ^{2}+\frac{1}{\lambda^{\alpha}} \bigl\vert Y^{(1)}_{t}-Y^{(2)}_{t} \bigr\vert ^{2} \\& \quad \leq \frac{1}{2} \bigl\Vert Y^{(1)}_{t}-Y^{(2)}_{t} \bigr\Vert ^{2}+C_{3} \bigl[ \bigl\vert Y^{(1)}_{t}+z_{t} \bigr\vert _{L^{4}}^{4}+ \bigl\vert Y^{(2)}_{t}+z_{t} \bigr\vert _{L^{4}}^{4} \bigr] \bigl\vert Y^{(1)}_{t}-Y^{(2)}_{t} \bigr\vert ^{2} \\& \qquad {} + \bigl[ \bigl\vert Y^{(1)}_{t}+z_{t} \bigr\vert + \bigl\vert Y^{(2)}_{t}+z_{t} \bigr\vert \bigr] \bigl\vert Y^{(1)}_{t}-Y^{(2)}_{t} \bigr\vert ^{2}+\lambda (U) \bigl\vert Y^{(1)}_{t}-Y^{(2)}_{t} \bigr\vert ^{2}+\frac{1}{\lambda^{\alpha}} \bigl\vert Y^{(1)}_{t}-Y^{(2)}_{t} \bigr\vert ^{2}. \end{aligned}$$
Theorem 3.2
If hypotheses (\(\mathrm{H}_{1}\))–(\(\mathrm{H}_{3}\)) hold, then for
\(\forall x\in \mathscr {D}(A^{\alpha})\), there exists a unique solution
X
of (2.3) such that, for a.s. \(\omega\in\varOmega\),
The transition semigroup associated with (2.3) is a Feller Markov process.
$$X_{.}(\omega)-z_{.}(\omega)\in C\bigl([0,T],\mathscr {D}\bigl(A^{\alpha}\bigr) \bigr)\cap L^{\frac {4}{1-2\alpha}}\bigl(0,T;\mathscr {D}\bigl(A^{\frac{1}{4}+\frac{\alpha}{2}}\bigr)\bigr) \cap L^{2}\bigl(0,T;\mathscr {D}\bigl(A^{\frac{1}{2}+\alpha}\bigr)\bigr). $$
Proof
Theorem 3.1 yields that for \(x\in \mathscr {D}(A^{\alpha})\), equation (3.2) has a unique solution X satisfying \(X_{.}-z_{.}\in C([0,T],\mathscr {D}(A^{\alpha}))\cap L^{\frac{4}{1-2\alpha}}(0,T;\mathscr {D}(A^{\frac{1}{4}+\frac{\alpha}{2}}))\cap L^{2}(0,T;\mathscr {D}(A^{\frac {1}{2}+\alpha}))\). Hence we can define on \([0,\sigma_{1}]\) that
where \(P_{t}\) is a stable Poisson point process on \(\mathbf{R}^{+}\times U\) with intensity measure \(\lambda(du)\,dt\).
$$X^{(1)}_{t}= \textstyle\begin{cases} X(t),&t< \sigma_{1}, \\ X_{\sigma_{1}-}+F(X_{\sigma_{1}-},P_{\sigma_{1}}),&t=\sigma_{1}, \end{cases} $$
Next, on \([\sigma_{1},\sigma_{2})\) define
\(\tilde{P}_{t}\) is still a stable point process with intensity measure \(\lambda(du)\,dt\). Likewise, equation (3.2) has a unique solution X̃ satisfying \(\tilde{X}_{.}-z_{.}\in C([0,T],\mathscr {D}(A^{\alpha}))\cap L^{\frac{4}{1-2\alpha}}(0,T;\mathscr {D}(A^{\frac{1}{4}+\frac{\alpha }{2}}))\cap L^{2}(0,T;\mathscr {D}(A^{\frac{1}{2}+\alpha}))\) corresponding to the initial value \(\tilde{X}_{0}\). It is reasonable to define on \([0,\sigma_{2}]\) that
Fixing \(T>0\), the Poisson point process \(P_{t}\) has only finite jumps on \([0,T]\). Therefore, choosing the first integer N to satisfy \(\sigma _{N+1}\geq T\), the method above can be iterated finite times to get \(X^{(N)}_{t}\). It is straightforward to show that \(X^{(N)}_{t}\) is the unique solution satisfying equation (2.3). □
$$\begin{aligned}& \tilde{X}_{0}=X^{(1)}_{\sigma_{1}}I_{(\sigma_{1}< \infty)}, \qquad \tilde{\sigma }_{2}=(\sigma_{2}-\sigma_{1})I_{\sigma_{1}< \infty}+ \infty I_{(\sigma_{1}=\infty )}, \\& \tilde{\mathscr {F}}_{t}=\mathscr {F}_{\sigma_{1}+t}, \qquad \tilde{P}(t)=(\theta _{\sigma_{1}}P) (t)I_{(\sigma_{1}< \infty)}. \end{aligned}$$
$$X^{(2)}_{t}= \textstyle\begin{cases} X^{(1)}(t), & t\leq\sigma_{1}, \\ \tilde{X}_{t-\sigma_{1}},& \sigma_{1}< t< \sigma_{2}, \\ \tilde{X}_{(\sigma_{2}-\sigma_{1})-}+F(\tilde{X}_{(\sigma_{2}-\sigma _{1})-},P_{\sigma_{2}}), & t=\sigma_{2}. \end{cases} $$
3.2 \(\lambda(\cdot)\) being the σ-finite measure
Theorem 3.3
If hypotheses (\(\mathrm{H}_{1}\))–(\(\mathrm{H}_{3}\)) hold, then for
\(\forall x\in \mathscr {D}(A^{\alpha})\), there exists a unique solution
X
of (2.3) such that, for a.s. \(\omega\in\varOmega\),
The transition semigroup associated with (2.3) is a Feller Markov process.
$$X_{.}(\omega)-z_{.}(\omega)\in C\bigl([0,T],\mathscr {D}\bigl(A^{\alpha}\bigr) \bigr)\cap L^{\frac {4}{1-2\alpha}}\bigl(0,T;\mathscr {D}\bigl(A^{\frac{1}{4}+\frac{\alpha}{2}}\bigr)\bigr) \cap L^{2}\bigl(0,T;\mathscr {D}\bigl(A^{\frac{1}{2}+\alpha}\bigr)\bigr). $$
Proof
For every \(n\geq1\), consider the equation
$$ \textstyle\begin{cases} dY^{n}_{t}+AY^{n}_{t}\,dt+B(Y^{n}_{t}+z_{t})\,dt+R(Y^{n}_{t}+z_{t})\,dt= \int _{U_{n}}F(Y^{n}_{t-}+z_{t-},u)\tilde{N}(dt,du), \\ Y^{n}_{0}=x. \end{cases} $$
(3.8)
Hereafter, we define \(\varXi Z(t):=Z(t)-Z(t-)\). By Itô’s formula in [12], one finds that
Therefore,
where
Let
Lemma 3.1 yields \(\tau_{k_{1}}\nearrow\infty\) as \(k_{1}\rightarrow\infty\). Gronwall’s inequality implies
This yields
Applying Itô’s formula to \(|A^{\alpha}Y^{n}_{t}|^{2}\) implies
Accordingly,
where
Fixing \(k_{1}\) first, let
Equation (3.9) yields \(\tau^{n}_{k_{2}}\nearrow\tau_{k_{1}}\) as \(k_{2}\rightarrow\infty\). By Gronwall’s inequality, we have
This yields
For every \(n,m\geq1\), \(Y^{n}_{t}-Y^{m}_{t}\) satisfies the equation
Recall that \(\varXi Z(t):=Z(t)-Z(t-)\). Applying Itô’s formula to \(|A^{\alpha}(Y^{n}_{t}-Y^{m}_{t})|^{2}\), we have
Therefore
where
Now fixing \(k_{1}\), \(k_{2}\) consecutively, let
Equations (3.5), (3.10), and Lemma 3.1 imply that \(\tau^{m,n}_{k_{3}}\nearrow\tau ^{n}_{k_{2}}\wedge\tau^{m}_{k_{2}}\) as \(k_{3}\rightarrow\infty\). By Gronwall’s inequality, one gets
Noting hypothesis (\(\mathrm{H}_{4}\)), we have
as \(m\rightarrow\infty\).
$$\begin{aligned}& \bigl\vert Y^{n}_{t} \bigr\vert ^{2} \\& \quad = \vert x \vert ^{2}+ \int^{t}_{0} \bigl(2Y_{s-}^{n},dY_{s}^{n} \bigr)+ \sum_{s\leq t} \bigl(\varXi \bigl( \bigl\vert Y_{s}^{n} \bigr\vert ^{2} \bigr)- \bigl(2Y_{s-}^{n},\varXi Y_{s}^{n} \bigr) \bigr) \\& \quad = \vert x \vert ^{2}+ \int ^{t}_{0} \bigl(2Y_{s-}^{n},- \bigl[AY_{s}^{n}+B \bigl(Y_{s}^{n}+z_{s} \bigr)+R \bigl(Y_{s}^{n}+z_{s} \bigr) \bigr]\,ds \bigr) \\& \qquad {} + \int^{t}_{0} \int_{U_{n}} \bigl(2Y_{s-}^{n},F \bigl(Y_{s}^{n}+z_{s},u \bigr) \bigr) \tilde{N}(ds,du) \\& \qquad {} + \int^{t}_{0} \int _{U_{n}} \bigl( \bigl\vert \bigl(Y_{s-}^{n}+F \bigl(Y_{s}^{n}+z_{s},u \bigr) \bigr) \bigr\vert ^{2}- \bigl\vert Y_{s-}^{n} \bigr\vert ^{2} \bigr)N(ds,du) \\& \qquad {} - \int^{t}_{0} \int_{U_{n}} \bigl(2Y_{s-}^{n} ,F \bigl(Y_{s}^{n}+z_{s},u \bigr) \bigr)N(ds,du) \\& \quad = \vert x \vert ^{2}+ \int ^{t}_{0} \bigl(2Y_{s-}^{n},- \bigl[AY_{s}^{n}+B \bigl(Y_{s}^{n}+z_{s} \bigr)+R \bigl(Y_{s}^{n}+z_{s} \bigr) \bigr]\,ds \bigr) \\& \qquad {} + \int^{t}_{0} \int_{U_{n}} \bigl\vert \bigl(2Y_{s-}^{n},F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \bigr) \bigr\vert \tilde{N}(ds,du) \\& \qquad {} + \int^{t}_{0} \int_{U_{n}} \int_{0}^{1} \bigl\vert \bigl(2 \bigl(Y_{s-}^{n}+s'F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \bigr), F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \bigr) \bigr\vert \,ds'N(ds,du) \\& \qquad {} + \int^{t}_{0} \int_{U_{n}} \bigl\vert \bigl(2Y_{s-}^{n} ,F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \bigr) \bigr\vert N(ds,du). \end{aligned}$$
$$\begin{aligned}& \bigl\vert Y^{n}_{t} \bigr\vert ^{2}+2 \int^{t}_{0} \bigl\Vert Y^{n}_{s} \bigr\Vert ^{2}\,ds \\& \quad \leq \vert x \vert ^{2}+2 \int^{t}_{0} \bigl\vert \bigl(B \bigl(Y^{n}_{s}+z_{s} \bigr),Y^{n}_{s} \bigr) \bigr\vert \,ds+2 \int ^{t}_{0} \bigl\vert \bigl(R \bigl(Y^{n}_{s}+z_{s} \bigr), Y^{n}_{s} \bigr) \bigr\vert \,ds \\& \qquad {} +6 \int^{t}_{0} \int_{U_{n}} \bigl\vert \bigl(Y_{s-}^{n}, F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \bigr) \bigr\vert \tilde{N}(ds,du) \\& \qquad {} +2 \int^{t}_{0} \int_{U_{n}} \bigl\vert F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \bigr\vert ^{2}N(ds,du) \\& \qquad {} +2 \int^{t}_{0} \int_{U_{n}} \bigl\vert \bigl(Y_{s-}^{n}, F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \bigr) \bigr\vert \lambda(du)\,ds \\& \quad \leq \vert x \vert ^{2}+ \int_{0}^{t} \bigl\Vert Y^{n}_{s} \bigr\Vert ^{2}\,ds+C_{1} \int ^{t}_{0} \bigl\vert Y_{s}^{n} \bigr\vert ^{2} \vert z_{s} \vert ^{4}_{L^{4}} \,ds+C_{1} \int^{t}_{0} \vert z_{s} \vert ^{4}_{L^{4}}\,ds+M_{t}, \end{aligned}$$
$$\begin{aligned} M_{t} =&6 \int^{t}_{0} \int_{U_{n}} \bigl\vert \bigl(Y_{s-}^{n}, F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \bigr) \bigr\vert \tilde{N}(ds,du) \\ &{} +2 \int^{t}_{0} \int_{U_{n}} \bigl\vert F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \bigr\vert ^{2}N(ds,du) \\ &{} +2 \int^{t}_{0} \int_{U_{n}} \bigl\vert \bigl(Y_{s-}^{n}, F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \bigr) \bigr\vert \lambda(du)\,ds. \end{aligned}$$
$$\tau_{k_{1}}=\inf\biggl\{ t>0; \int^{t}_{0}|z_{s}|^{4}_{L^{4}} \,ds>k_{1}\biggr\} ,\quad k_{1}>1. $$
$$\begin{aligned}& \mathbf{E}\sup_{s\leq\tau_{k_{1}}\wedge t} \bigl\vert Y^{n}_{s} \bigr\vert ^{2}+\mathbf{E} \int ^{\tau_{k_{1}}\wedge t}_{0} \bigl\Vert Y^{n}_{s} \bigr\Vert ^{2}\,ds \\& \quad \leq \biggl[ \vert x \vert ^{2}+C_{1}\mathbf{E} \int^{\tau_{k_{1}}\wedge t}_{0} \vert z_{s} \vert ^{4}_{L^{4}}\,ds+\mathbf{E}\sup_{s\leq\tau_{k_{1}}\wedge t} \vert M_{s} \vert \biggr]e^{C_{1}k_{1}} \\& \quad \leq \biggl[ \vert x \vert ^{2}+C_{1}\mathbf{E} \int^{\tau_{k_{1}}\wedge t}_{0} \vert z_{s} \vert ^{4}_{L^{4}}\,ds \biggr]e^{C_{1}k_{1}} \\& \qquad {} +C_{2}e^{C_{1}k_{1}}\mathbf{E} \int^{\tau_{k_{1}}\wedge t}_{0} \int_{U_{n}} \bigl\vert \bigl(Y_{s-}^{n}, F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \bigr) \bigr\vert \tilde{N}(ds,du) \\& \qquad {} +C_{2}e^{C_{1}k_{1}}\mathbf{E} \int^{\tau_{k_{1}}\wedge t}_{0} \int_{U_{n}} \bigl\vert F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \bigr\vert ^{2}\tilde{N}(ds,du) \\& \qquad {} +C_{2}e^{C_{1}k_{1}}\mathbf{E} \int^{\tau_{k_{1}}\wedge t}_{0} \int_{U_{n}} \bigl\vert \bigl(Y_{s-}^{n}, F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \bigr) \bigr\vert \lambda(du)\,ds \\& \qquad {} +C_{2}e^{C_{1}k_{1}}\mathbf{E} \int^{\tau_{k_{1}}\wedge t}_{0} \int_{U_{n}} \bigl\vert F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \bigr\vert ^{2}\lambda(du)\,ds \\& \quad \leq \biggl[ \vert x \vert ^{2}+C_{1}\mathbf{E} \int^{\tau_{k_{1}}\wedge t}_{0} \vert z_{s} \vert ^{4}_{L^{4}}\,ds \biggr]e^{C_{1}k_{1}} \\& \qquad {} +\varepsilon \mathbf{E}\sup_{s\leq\tau_{k_{1}}\wedge t} \bigl\vert Y^{n}_{s} \bigr\vert ^{2}+C_{3}( \lambda_{1},\varepsilon ,k_{1})\mathbf{E} \int^{\tau _{k_{1}}\wedge t}_{0} \int_{U_{n}} \bigl\vert A^{\alpha}F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \bigr\vert ^{2}\lambda(du)\,ds \\& \quad \leq \biggl[ \vert x \vert ^{2}+C_{1}\mathbf{E} \int^{\tau_{k_{1}}\wedge t}_{0} \vert z_{s} \vert ^{4}_{L^{4}}\,ds \biggr]e^{C_{1}k_{1}} \\& \qquad {} +\varepsilon \mathbf{E}\sup_{s\leq\tau_{k_{1}}\wedge t} \bigl\vert Y^{n}_{s} \bigr\vert ^{2}+2C_{3}( \lambda_{1},\varepsilon ,k_{1})\mathbf{E} \int^{\tau _{k_{1}}\wedge t}_{0} \bigl[K \bigl\vert Y^{n}_{s}+z_{s} \bigr\vert ^{2}+C \bigr]\,ds \\& \quad \leq \biggl[ \vert x \vert ^{2}+C_{1}\mathbf{E} \int^{\tau_{k_{1}}\wedge t}_{0} \vert z_{s} \vert ^{4}_{L^{4}}\,ds \biggr]e^{C_{1}k_{1}} \\& \qquad {} +\varepsilon \mathbf{E}\sup_{s\leq\tau_{k_{1}}\wedge t} \bigl\vert Y^{n}_{s} \bigr\vert ^{2}+4C_{3}K \mathbf{E} \int^{\tau_{k_{1}}\wedge t}_{0} \bigl\vert Y^{n}_{s} \bigr\vert ^{2}\,ds+4C_{3}K \mathbf{E} \int^{\tau_{k_{1}}\wedge t}_{0} \vert z_{s} \vert ^{2}\,ds+2C_{3}Ct. \end{aligned}$$
$$\begin{aligned}& (1-\varepsilon )\mathbf{E}\sup_{s\leq\tau_{k_{1}}\wedge t} \bigl\vert Y^{n}_{s} \bigr\vert ^{2}+\mathbf{E} \int^{\tau_{k_{1}}\wedge t}_{0} \bigl\Vert Y^{n}_{s} \bigr\Vert ^{2}\,ds \\& \quad \leq \biggl[|x|^{2}+C_{1}\mathbf{E} \int^{\tau_{k_{1}}\wedge t}_{0}|z_{s}|^{4}_{L^{4}} \,ds \biggr]e^{C_{1}k_{1}} \\& \qquad {} +4C_{3}K\mathbf{E} \int^{\tau_{k_{1}}\wedge t}_{0} \bigl\vert Y^{n}_{s} \bigr\vert ^{2}\,ds+4C_{3}K \mathbf{E} \int^{\tau_{k_{1}}\wedge t}_{0}|z_{s}|^{2} \,ds+2C_{3}Ct \\& \quad \leq C_{4}(x,\varepsilon ,\lambda_{1},k_{1},K)+C_{5}( \varepsilon ,\lambda_{1},k_{1},K)t. \end{aligned}$$
(3.9)
$$\begin{aligned}& \bigl\vert A^{\alpha}Y^{n}_{t} \bigr\vert ^{2} \\& \quad = \bigl\vert A^{\alpha}x \bigr\vert ^{2}+ \int^{t}_{0} \bigl(2A^{\alpha}Y_{s-}^{n} \cdot A^{\alpha },- \bigl[AY_{s}^{n}+B \bigl(Y_{s}^{n}+z_{s} \bigr)+R \bigl(Y_{s}^{n}+z_{s} \bigr) \bigr]\,ds \bigr) \\& \qquad {} + \int^{t}_{0} \int_{U_{n}} \bigl(2A^{\alpha}Y_{s-}^{n} \cdot A^{\alpha },F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \bigr)\tilde{N}(ds,du) \\& \qquad {} + \int^{t}_{0} \int_{U_{n}} \bigl( \bigl\vert A^{\alpha } \bigl(Y_{s-}^{n}+F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \bigr) \bigr\vert ^{2}- \bigl\vert A^{\alpha}Y_{s-}^{n} \bigr\vert ^{2} \bigr)N(ds,du) \\& \qquad {}- \int^{t}_{0} \int_{U_{n}} \bigl(2A^{\alpha}Y_{s-}^{n} \cdot A^{\alpha},F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \bigr)N(ds,du). \end{aligned}$$
$$\begin{aligned}& \bigl\vert A^{\alpha}Y^{n}_{t} \bigr\vert ^{2}+2 \int^{t}_{0} \bigl\vert A^{\frac{1}{2}+\alpha}Y^{n}_{s} \bigr\vert ^{2}\,ds \\& \quad \leq \bigl\vert A^{\alpha}x \bigr\vert ^{2}+2 \int^{t}_{0} \bigl\vert \bigl(A^{\alpha-\frac {1}{2}}B \bigl(Y^{n}_{s}+z_{s} \bigr),A^{\frac{1}{2}+\alpha}Y^{n}_{s} \bigr) \bigr\vert \,ds+2 \int ^{t}_{0} \bigl\vert \bigl(A^{\alpha}R \bigl(Y^{n}_{s}+z_{s} \bigr),A^{\alpha}Y^{n}_{s} \bigr) \bigr\vert \,ds \\& \qquad {}+6 \int^{t}_{0} \int_{U_{n}} \bigl\vert \bigl(A^{\alpha}Y_{s-}^{n}, A^{\alpha}F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \bigr) \bigr\vert \tilde{N}(ds,du) \\& \qquad {}+2 \int^{t}_{0} \int_{U_{n}} \bigl\vert A^{\alpha}F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \bigr\vert ^{2}N(ds,du) \\& \qquad {}+2 \int^{t}_{0} \int_{U_{n}} \bigl\vert \bigl(A^{\alpha}Y_{s-}^{n}, A^{\alpha}F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \bigr) \bigr\vert \lambda(du)\,ds \\& \quad \leq \bigl\vert A^{\alpha}x \bigr\vert ^{2}+ \int_{0}^{t}C_{1} \bigl\vert A^{\frac{1}{4}+\frac{\alpha }{2}} \bigl(Y^{n}_{s}+z_{s} \bigr) \bigr\vert ^{2} \bigl\vert A^{\alpha+\frac{1}{2}}Y^{n}_{s} \bigr\vert \,ds+ \int^{t}_{0} \bigl\vert A^{\alpha } \bigl(Y_{s}^{n}+z_{s}^{n} \bigr) \bigr\vert \bigl\vert A^{\alpha}Y^{n}_{s} \bigr\vert \,ds+\overline{M}_{t} \\& \quad \leq \bigl\vert A^{\alpha}x \bigr\vert ^{2}+C_{1} \int^{t}_{0} \bigl[ \bigl\vert A^{\alpha}Y^{n}_{s} \bigr\vert \bigl\vert A^{\frac{1}{2}} Y^{n}_{s} \bigr\vert \bigl\vert A^{\alpha+\frac{1}{2}} Y^{n}_{s} \bigr\vert + \bigl\vert A^{\frac{1}{4}+\frac{\alpha}{2}} z_{s} \bigr\vert ^{2} \bigl\vert A^{\alpha+\frac{1}{2}} Y^{n}_{s} \bigr\vert \bigr]\,ds \\& \qquad {}+ \int^{t}_{0} \bigl[ \bigl\vert A^{\alpha}Y^{n}_{s} \bigr\vert ^{2}+ \bigl\vert A^{\alpha}Y^{n}_{s} \bigr\vert \bigl\vert A^{\alpha}z_{s} \bigr\vert \bigr]\,ds+\overline{M}_{t} \\& \quad \leq \bigl\vert A^{\alpha}x \bigr\vert ^{2}+ \int^{t}_{0} \bigl[ \bigl\vert A^{\frac{1}{2}+\alpha }Y^{n}_{s} \bigr\vert ^{2}+C_{2} \bigl\vert A^{\frac{1}{2}} Y^{n}_{s} \bigr\vert ^{2} \bigl\vert A^{\alpha}Y^{n}_{s} \bigr\vert ^{2}+C_{2} \bigl\vert A^{\frac {1}{4}+\frac{\alpha}{2}} z_{s} \bigr\vert ^{4} \bigr] \,ds \\& \qquad {}+ \int^{t}_{0} \bigl[2 \bigl\vert A^{\alpha}Y^{n}_{s} \bigr\vert ^{2}+ \bigl\vert A^{\alpha}z_{s} \bigr\vert ^{2} \bigr]\,ds+ \overline{M}_{t} \\& \quad \leq \bigl\vert A^{\alpha}x \bigr\vert ^{2}+ \int^{t}_{0} \bigl\vert A^{\frac{1}{2}+\alpha}Y^{n}_{s} \bigr\vert ^{2}\,ds+ \int ^{t}_{0} \bigl(2+C_{2} \bigl\vert A^{\frac{1}{2}} Y^{n}_{s} \bigr\vert ^{2} \bigr) \bigl\vert A^{\alpha}Y^{n}_{s} \bigr\vert ^{2}\,ds \\& \qquad {}+ \int ^{t}_{0} \bigl[C_{2} \bigl\vert A^{\frac{1}{4}+\frac{\alpha}{2}} z_{s} \bigr\vert ^{4}+ \bigl\vert A^{\alpha}z_{s} \bigr\vert ^{2} \bigr]\,ds+\overline{M}_{t}, \end{aligned}$$
$$\begin{aligned} \overline{M}_{t} =&6 \int^{t}_{0} \int_{U_{n}} \bigl\vert \bigl(A^{\alpha}Y_{s-}^{n}, A^{\alpha}F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \bigr) \bigr\vert \tilde{N}(ds,du) \\ &{} +2 \int^{t}_{0} \int_{U_{n}} \bigl\vert A^{\alpha}F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \bigr\vert ^{2}N(ds,du) \\ &{} +2 \int^{t}_{0} \int_{U_{n}} \bigl\vert \bigl(A^{\alpha}Y_{s-}^{n}, A^{\alpha}F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \bigr) \bigr\vert \lambda(du)\,ds. \end{aligned}$$
$$\tau^{n}_{k_{2}}=\tau^{n}_{k_{2},k_{1}}=\inf\biggl\{ t>0; \int^{t\wedge\tau_{k_{1}}}_{0} \bigl\Vert Y^{n}_{s} \bigr\Vert ^{2}\,ds>k_{2}\biggr\} \wedge \tau_{k_{1}},\quad k_{2}>1. $$
$$\begin{aligned}& \mathbf{E}\sup_{s\leq\tau^{n}_{k_{2}}\wedge t} \bigl\vert A^{\alpha }Y^{n}_{s} \bigr\vert ^{2}+\mathbf{E} \int^{\tau^{n}_{k_{2}}\wedge t}_{0} \bigl\vert A^{\frac{1}{2}+\alpha }Y^{n}_{s} \bigr\vert ^{2}\,ds \\& \quad \leq \biggl[ \bigl\vert A^{\alpha}x \bigr\vert ^{2}+ \mathbf{E} \int^{\tau^{n}_{k_{2}}\wedge t}_{0} \bigl[C_{2} \bigl\vert A^{\frac{1}{4}+\frac{\alpha}{2}} z_{s} \bigr\vert ^{4}+ \bigl\vert A^{\alpha}z_{s} \bigr\vert ^{2} \bigr]\,ds+ \mathbf{E}\sup_{s\leq\tau^{n}_{k_{2}}\wedge t} \vert \overline {M}_{s} \vert \biggr]e^{2+C_{2}k_{2}} \\& \quad \leq \biggl[ \bigl\vert A^{\alpha}x \bigr\vert ^{2}+ \mathbf{E} \int^{\tau^{n}_{k_{2}}\wedge t}_{0} \bigl[C_{2} \bigl\vert A^{\frac{1}{4}+\frac{\alpha}{2}} z_{s} \bigr\vert ^{4}+ \bigl\vert A^{\alpha}z_{s} \bigr\vert ^{2} \bigr]\,ds \biggr]e^{2+C_{2}k_{2}}+e^{2+C_{2}k_{2}}\mathbf{E}\sup_{s\leq\tau ^{n}_{k_{2}}\wedge t} \vert \overline{M}_{s} \vert \\& \quad \leq \biggl[ \bigl\vert A^{\alpha}x \bigr\vert ^{2}+ \mathbf{E} \int^{\tau^{n}_{k_{2}}\wedge t}_{0} \bigl[C_{2} \bigl\vert A^{\frac{1}{4}+\frac{\alpha}{2}} z_{s} \bigr\vert ^{4}+ \bigl\vert A^{\alpha}z_{s} \bigr\vert ^{2} \bigr]\,ds \biggr]e^{2+C_{2}k_{2}} \\& \qquad {} +C_{3}e^{2+C_{2}k_{2}}\mathbf{E} \int^{\tau^{n}_{k_{2}}\wedge t}_{0} \int _{U_{n}} \bigl\vert \bigl(A^{\alpha}Y_{s-}^{n}, A^{\alpha}F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \bigr) \bigr\vert \tilde{N}(ds,du) \\& \qquad {} +C_{3}e^{2+C_{2}k_{2}}\mathbf{E} \int^{\tau^{n}_{k_{2}}\wedge t}_{0} \int _{U_{n}} \bigl\vert A^{\alpha}F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \bigr\vert ^{2}\tilde{N}(ds,du) \\& \qquad {} +C_{3}e^{2+C_{2}k_{2}}\mathbf{E} \int^{\tau^{n}_{k_{2}}\wedge t}_{0} \int _{U_{n}} \bigl\vert \bigl(A^{\alpha}Y_{s-}^{n}, A^{\alpha}F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \bigr) \bigr\vert \lambda(du)\,ds \\& \qquad {} +C_{3}e^{2+C_{2}k_{2}}\mathbf{E} \int^{\tau^{n}_{k_{2}}\wedge t}_{0} \int _{U_{n}} \bigl\vert A^{\alpha}F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \bigr\vert ^{2}\lambda(du)\,ds \\& \quad \leq \biggl[ \bigl\vert A^{\alpha}x \bigr\vert ^{2}+ \mathbf{E} \int^{\tau^{n}_{k_{2}}\wedge t}_{0} \bigl[C_{2} \bigl\vert A^{\frac{1}{4}+\frac{\alpha}{2}} z_{s} \bigr\vert ^{4}+ \bigl\vert A^{\alpha}z_{s} \bigr\vert ^{2} \bigr]\,ds \biggr]e^{2+C_{2}k_{2}} \\& \qquad {} +\varepsilon \mathbf{E}\sup_{s\leq\tau^{n}_{k_{2}}\wedge t} \bigl\vert A^{\alpha }Y^{n}_{s} \bigr\vert ^{2}+C_{4}( \varepsilon ,k)\mathbf{E} \int^{\tau^{n}_{k_{2}}\wedge t}_{0} \int _{U_{n}} \bigl\vert A^{\alpha}F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \bigr\vert ^{2}\lambda(du)\,ds \\& \quad \leq \biggl[ \bigl\vert A^{\alpha}x \bigr\vert ^{2}+ \mathbf{E} \int^{\tau^{n}_{k_{2}}\wedge t}_{0} \bigl[C_{2} \bigl\vert A^{\frac{1}{4}+\frac{\alpha}{2}} z_{s} \bigr\vert ^{4}+ \bigl\vert A^{\alpha}z_{s} \bigr\vert ^{2} \bigr]\,ds \biggr]e^{2+C_{2}k_{2}} \\& \qquad {} +\varepsilon \mathbf{E}\sup_{s\leq\tau^{n}_{k_{2}}\wedge t} \bigl\vert A^{\alpha }Y^{n}_{s} \bigr\vert ^{2}+2C_{4}( \varepsilon ,k)\mathbf{E} \int^{\tau^{n}_{k_{2}}\wedge t}_{0} \bigl[K \bigl\vert Y^{n}_{s}+z_{s} \bigr\vert ^{2}+C \bigr]\,ds \\& \quad \leq \biggl[ \bigl\vert A^{\alpha}x \bigr\vert ^{2}+ \mathbf{E} \int^{\tau^{n}_{k_{2}}\wedge t}_{0} \bigl[C_{2} \bigl\vert A^{\frac{1}{4}+\frac{\alpha}{2}} z_{s} \bigr\vert ^{4}+ \bigl\vert A^{\alpha}z_{s} \bigr\vert ^{2} \bigr]\,ds \biggr]e^{2+C_{2}k_{2}} \\& \qquad {} +\varepsilon \mathbf{E}\sup_{s\leq\tau^{n}_{k_{2}}\wedge t} \bigl\vert A^{\alpha }Y^{n}_{s} \bigr\vert ^{2}+ \frac{4C_{4}K}{\lambda_{1}}\mathbf{E} \int^{\tau^{n}_{k_{2}}\wedge t}_{0} \bigl\Vert Y^{n}_{s} \bigr\Vert ^{2}\,ds \\& \qquad {}+4C_{4}K \mathbf{E} \int^{\tau^{n}_{k_{2}}\wedge t}_{0} \vert z_{s} \vert ^{2}\,ds+2C_{4}Ct. \end{aligned}$$
$$\begin{aligned}& (1-\varepsilon )\mathbf{E}\sup_{s\leq\tau^{n}_{k_{2}}\wedge t} \bigl\vert A^{\alpha }Y^{n}_{s} \bigr\vert ^{2}+ \mathbf{E} \int^{\tau^{n}_{k_{2}}\wedge t}_{0} \bigl\vert A^{\frac{1}{2}+\alpha }Y^{n}_{s} \bigr\vert ^{2}\,ds \\& \quad \leq \biggl[ \bigl\vert A^{\alpha}x \bigr\vert ^{2}+ \mathbf{E} \int^{\tau^{n}_{k_{2}}\wedge t}_{0} \bigl[C_{2} \bigl\vert A^{\frac{1}{4}+\frac{\alpha}{2}} z_{s} \bigr\vert ^{4}+ \bigl\vert A^{\alpha}z_{s} \bigr\vert ^{2} \bigr]\,ds \biggr]e^{2+C_{2}k_{2}} \\& \qquad {} +\frac{4C_{4}Kk_{2}}{\lambda_{1}}+4C_{4}K \mathbf{E} \int^{\tau^{n}_{k_{2}}\wedge t}_{0} \vert z_{s} \vert ^{2}\,ds+2C_{4}Ct \\& \quad \leq C_{5}(x,\varepsilon ,\alpha,\lambda_{1},k_{2},K)+C_{6}( \varepsilon ,\alpha,\lambda _{1},k_{2},K)t. \end{aligned}$$
(3.10)
$$ \textstyle\begin{cases} d(Y^{n}_{t}-Y^{m}_{t})+A(Y^{n}_{t}-Y^{m}_{t})\,dt+B(Y^{n}_{t}+z_{t})\,dt \\ \qquad {}-B(Y^{m}_{t}+z_{t})\,dt+R(Y^{n}_{t}-Y^{m}_{t})\,dt \\ \quad = \int_{U_{n}\setminus U_{m}}F(Y^{n}_{t-}+z_{t-},u)\tilde{N}(dt,du) \\ \qquad {}+ \int_{U_{m}}[F(Y^{n}_{t-}+z_{t-},u)-F(Y^{m}_{t-}+z_{t-},u)]\tilde {N}(dt,du), \\ Y^{n}_{0}-Y^{m}_{0}=0. \end{cases} $$
(3.11)
$$\begin{aligned}& \bigl\vert A^{\alpha} \bigl(Y^{n}_{t}-Y^{m}_{t} \bigr) \bigr\vert ^{2} \\& \quad = \int^{t}_{0} \bigl(2A^{\alpha} \bigl(Y_{s-}^{n}-Y_{s-}^{m} \bigr)\cdot A^{\alpha },d \bigl(Y_{s}^{n}-Y_{s}^{m} \bigr) \bigr) \\& \qquad {}+ \sum_{s\leq t} \bigl(\varXi \bigl( \bigl\vert A^{\alpha} \bigl(Y_{s}^{n}-Y_{s}^{m} \bigr) \bigr\vert ^{2} \bigr)- \bigl(2A^{\alpha } \bigl(Y_{s-}^{n}-Y_{s-}^{m} \bigr)\cdot A^{\alpha},\varXi \bigl(Y_{s}^{n}-Y_{s}^{m} \bigr) \bigr) \bigr) \\& \quad = \int^{t}_{0} \bigl(2A^{\alpha} \bigl(Y_{s-}^{n}-Y_{s-}^{m} \bigr)\cdot A^{\alpha },- \bigl[A \bigl(Y^{n}_{s}-Y^{m}_{s} \bigr)+B \bigl(Y^{n}_{s}+z_{s} \bigr) \\& \qquad {}-B \bigl(Y^{m}_{s}+z_{s} \bigr)+R \bigl(Y^{n}_{s}-Y^{m}_{s} \bigr) \bigr] \,ds \bigr) \\& \qquad {} + \int^{t}_{0} \int_{U_{n}} \bigl(2A^{\alpha} \bigl(Y_{s-}^{n}-Y_{s-}^{m} \bigr)\cdot A^{\alpha},F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \bigr)\tilde{N}(ds,du) \\& \qquad {} - \int^{t}_{0} \int_{U_{m}} \bigl(2A^{\alpha} \bigl(Y_{s-}^{n}-Y_{s-}^{m} \bigr)\cdot A^{\alpha},F \bigl(Y_{s-}^{m}+z_{s-},u \bigr) \bigr)\tilde{N}(ds,du) \\& \qquad {}+ \int^{t}_{0} \int_{U_{n}\setminus U_{m}} \bigl( \bigl\vert A^{\alpha } \bigl( \bigl(Y_{s-}^{n}-Y_{s-}^{m} \bigr)+F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \bigr) \bigr\vert ^{2}- \bigl\vert A^{\alpha } \bigl(Y_{s-}^{n}-Y_{s-}^{m} \bigr) \bigr\vert ^{2} \bigr)N(ds,du) \\& \qquad {}+ \int^{t}_{0} \int_{U_{m}} \bigl( \bigl\vert A^{\alpha } \bigl( \bigl(Y_{s-}^{n}-Y_{s-}^{m} \bigr)+F \bigl(Y^{n}_{s-}+z_{s-},u \bigr)-F \bigl(Y^{m}_{s-}+z_{s-},u \bigr) \bigr) \bigr\vert ^{2} \\& \qquad {}- \bigl\vert A^{\alpha } \bigl(Y_{s-}^{n}-Y_{s-}^{m} \bigr) \bigr\vert ^{2} \bigr)N(ds,du) \\& \qquad {} - \int^{t}_{0} \int_{U_{n}}(2A^{\alpha} \bigl(Y_{s-}^{n}-Y_{s-}^{m} \bigr)\cdot A^{\alpha},F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \bigr)N(ds,du) \\& \qquad {} + \int^{t}_{0} \int_{U_{m}}(2A^{\alpha} \bigl(Y_{s-}^{n}-Y_{s-}^{m} \bigr)\cdot A^{\alpha},F \bigl(Y_{s-}^{m}+z_{s-},u \bigr) \bigr)N(ds,du). \end{aligned}$$
$$\begin{aligned}& \bigl\vert A^{\alpha} \bigl(Y^{n}_{t}-Y^{m}_{t} \bigr) \bigr\vert ^{2}+2 \int^{t}_{0} \bigl\vert A^{\frac{1}{2}+\alpha } \bigl(Y^{n}_{s}-Y^{m}_{s} \bigr) \bigr\vert ^{2}\,ds \\& \quad \leq 2 \int^{t}_{0} \bigl\vert \bigl(B \bigl(Y^{n}_{s}+z_{s} \bigr),A^{2\alpha} \bigl(Y^{n}_{s}-Y^{m}_{s} \bigr) \bigr) \bigr\vert \,ds+2 \int ^{t}_{0} \bigl\vert \bigl(B \bigl(Y^{m}_{s}+z_{s} \bigr),A^{2\alpha} \bigl(Y^{n}_{s}-Y^{m}_{s} \bigr) \bigr) \bigr\vert \,ds \\& \qquad {}+2 \int^{t}_{0} \bigl\vert \bigl(A^{\alpha}R \bigl(Y^{n}_{s}-Y^{m}_{s} \bigr),A^{\alpha}\bigl(Y^{n}_{s}-Y^{m}_{s} \bigr) \bigr) \bigr\vert \,ds \\& \qquad {} +6 \int^{t}_{0} \int_{U_{n}\setminus U_{m}} \bigl\vert \bigl(A^{\alpha} \bigl(Y_{s-}^{n}-Y_{s-}^{m} \bigr), A^{\alpha}F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \bigr) \bigr\vert \tilde{N}(ds,du) \\& \qquad {} +6 \int^{t}_{0} \int_{U_{m}} \bigl\vert \bigl(A^{\alpha} \bigl(Y_{s-}^{n}-Y_{s-}^{m} \bigr), A^{\alpha} \bigl(F \bigl(Y^{n}_{s-}+z_{s-},u \bigr)-F \bigl(Y^{m}_{s-}+z_{s-},u \bigr) \bigr) \bigr) \bigr\vert \tilde {N}(ds,du) \\& \qquad {} +2 \int^{t}_{0} \int_{U_{n}\setminus U_{m}} \bigl\vert A^{\alpha }F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \bigr\vert ^{2}N(ds,du) \\& \qquad {} +2 \int^{t}_{0} \int_{U_{m}} \bigl\vert A^{\alpha } \bigl(F \bigl(Y^{n}_{s-}+z_{s-},u \bigr)-F \bigl(Y^{m}_{s-}+z_{s-},u \bigr) \bigr) \bigr\vert ^{2}N(ds,du) \\& \qquad {} +2 \int^{t}_{0} \int_{U_{n}\setminus U_{m}} \bigl\vert \bigl(A^{\alpha} \bigl(Y_{s-}^{n}-Y_{s-}^{m} \bigr), A^{\alpha}F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \bigr) \bigr\vert \lambda(du)\,ds \\& \qquad {} +2 \int^{t}_{0} \int_{U_{m}} \bigl\vert \bigl(A^{\alpha} \bigl(Y_{s-}^{n}-Y_{s-}^{m} \bigr), A^{\alpha} \bigl(F \bigl(Y^{n}_{s-}+z_{s-},u \bigr)-F \bigl(Y^{m}_{s-}+z_{s-},u \bigr) \bigr) \bigr) \bigr\vert \lambda (du)\,ds \\& \quad \leq C_{1} \int_{0}^{t} \bigl\vert A^{\alpha+\frac{1}{2}} \bigl(Y^{n}_{s}-Y^{m}_{s} \bigr) \bigr\vert ^{\frac {3}{2}-\alpha} \bigl\vert A^{\alpha} \bigl(Y^{n}_{s}-Y^{m}_{s} \bigr) \bigr\vert ^{\frac{1}{2}+\alpha} \\& \qquad {}\times\bigl[ \bigl\vert A^{\frac {1}{4}+\frac{\alpha}{2}} \bigl(Y^{n}_{s}+z_{s} \bigr) \bigr\vert + \bigl\vert A^{\frac{1}{4}+\frac{\alpha }{2}} \bigl(Y^{m}_{s}+z_{s} \bigr) \bigr\vert \bigr]\,ds +2 \int^{t}_{0} \bigl\vert A^{\alpha} \bigl(Y_{s}^{n}-Y_{s}^{m} \bigr) \bigr\vert ^{2}\,ds+g_{m,n}(t) \\& \quad \leq \int^{t}_{0} \bigl\vert A^{\alpha+\frac{1}{2}} \bigl(Y^{n}_{s}-Y^{m}_{s} \bigr) \bigr\vert ^{2}\,ds \\& \qquad {}+C_{2} \int ^{t}_{0} \bigl\vert A^{\alpha} \bigl(Y^{n}_{s}-Y^{m}_{s} \bigr) \bigr\vert ^{2} \bigl[ \bigl\vert A^{\frac{1}{4}+\frac{\alpha }{2}} \bigl(Y^{n}_{s}+z_{s} \bigr) \bigr\vert + \bigl\vert A^{\frac{1}{4}+\frac{\alpha}{2}} \bigl(Y^{m}_{s}+z_{s} \bigr) \bigr\vert \bigr]^{\frac {1+2\alpha}{4}}\,ds \\& \qquad {} +2 \int^{t}_{0} \bigl\vert A^{\alpha} \bigl(Y_{s}^{n}-Y_{s}^{m} \bigr) \bigr\vert ^{2}\,ds+g_{m,n}(t), \end{aligned}$$
$$\begin{aligned} g_{m,n}(t) =&6 \int^{t}_{0} \int_{U_{n}\setminus U_{m}} \bigl\vert \bigl(A^{\alpha } \bigl(Y_{s-}^{n}-Y_{s-}^{m} \bigr), A^{\alpha}F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \bigr) \bigr\vert \tilde{N}(ds,du) \\ &{}+6 \int^{t}_{0} \int_{U_{m}} \bigl\vert \bigl(A^{\alpha} \bigl(Y_{s-}^{n}-Y_{s-}^{m} \bigr), A^{\alpha} \bigl(F \bigl(Y^{n}_{s-}+z_{s-},u \bigr)-F \bigl(Y^{m}_{s-}+z_{s-},u \bigr) \bigr) \bigr) \bigr\vert \tilde {N}(ds,du) \\ &{} +2 \int^{t}_{0} \int_{U_{n}\setminus U_{m}} \bigl\vert A^{\alpha }F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \bigr\vert ^{2}N(ds,du) \\ &{} +2 \int^{t}_{0} \int_{U_{m}} \bigl\vert A^{\alpha } \bigl(F \bigl(Y^{n}_{s-}+z_{s-},u \bigr)-F \bigl(Y^{m}_{s-}+z_{s-},u \bigr) \bigr) \bigr\vert ^{2}N(ds,du) \\ &{} +2 \int^{t}_{0} \int_{U_{n}\setminus U_{m}} \bigl\vert \bigl(A^{\alpha} \bigl(Y_{s-}^{n}-Y_{s-}^{m} \bigr), A^{\alpha}F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \bigr) \bigr\vert \lambda(du)\,ds \\ &{} +2 \int^{t}_{0} \int_{U_{m}} \bigl\vert \bigl(A^{\alpha} \bigl(Y_{s-}^{n}-Y_{s-}^{m} \bigr), A^{\alpha} \bigl(F \bigl(Y^{n}_{s-}+z_{s-},u \bigr)-F \bigl(Y^{m}_{s-}+z_{s-},u \bigr) \bigr) \bigr) \bigr\vert \lambda(du)\,ds. \end{aligned}$$
$$\begin{aligned} \tau^{m,n}_{k_{3}} =&\tau^{m,n}_{k_{3},k_{2},k_{1}} \\ =&\inf \biggl\{ t>0; \int^{t\wedge{\tau ^{n}_{k_{2}}\wedge\tau^{m}_{k_{2}}}}_{0}\bigl[ \bigl\vert A^{\frac{1}{4}+\frac{\alpha }{2}} \bigl(Y^{n}_{s}+z_{s}\bigr) \bigr\vert + \bigl\vert A^{\frac{1}{4}+\frac{\alpha}{2}}\bigl(Y^{m}_{s}+z_{s} \bigr) \bigr\vert \bigr]^{\frac {1+2\alpha}{4}}\,ds>k_{3}\biggr\} \\ &{}\wedge \tau^{n}_{k_{2}}\wedge\tau^{m}_{k_{2}},\quad k_{3}>1. \end{aligned}$$
$$\begin{aligned}& \mathbf{E}\sup_{s\leq\tau^{m,n}_{k_{3}}\wedge t} \bigl\vert A^{\alpha } \bigl(Y^{n}_{s}-Y^{m}_{s} \bigr) \bigr\vert ^{2}+\mathbf{E} \int^{\tau^{m,n}_{k_{3}}\wedge t}_{0} \bigl\vert A^{\frac {1}{2}+\alpha} \bigl(Y^{n}_{s}-Y^{m}_{s} \bigr) \bigr\vert ^{2}\,ds \\& \quad \leq e^{2+C_{2}k_{3}}\mathbf{E}\sup_{s\leq\tau^{m,n}_{k_{3}}\wedge t} \bigl\vert g_{m,n}(s) \bigr\vert \\& \quad \leq C_{3}e^{2+C_{2}k_{3}}\mathbf{E} \int^{\tau^{m,n}_{k_{3}}\wedge t}_{0} \int _{U_{n}\setminus U_{m}} \bigl\vert \bigl(A^{\alpha} \bigl(Y_{s-}^{n}-Y_{s-}^{m} \bigr), A^{\alpha}F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \bigr) \bigr\vert \tilde{N}(ds,du) \\& \qquad {} +C_{3}e^{2+C_{2}k_{3}}\mathbf{E} \int^{\tau^{m,n}_{k_{3}}\wedge t}_{0} \int _{U_{n}\setminus U_{m}} \bigl\vert A^{\alpha}F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \bigr\vert ^{2}\tilde {N}(ds,du) \\& \qquad {} +C_{3}e^{2+C_{2}k_{3}}\mathbf{E} \int^{\tau^{m,n}_{k_{3}}\wedge t}_{0} \int _{U_{n}\setminus U_{m}} \bigl\vert \bigl(A^{\alpha} \bigl(Y_{s-}^{n}-Y_{s-}^{m} \bigr), A^{\alpha}F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \bigr) \bigr\vert \lambda(du)\,ds \\& \qquad {} +C_{3}e^{2+C_{2}k_{3}}\mathbf{E} \int^{\tau^{m,n}_{k_{3}}\wedge t}_{0} \int _{U_{n}\setminus U_{m}} \bigl\vert A^{\alpha}F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \bigr\vert ^{2}\lambda (du)\,ds \\& \qquad {} +C_{3}e^{2+C_{2}k_{3}}\mathbf{E} \int^{\tau^{m,n}_{k_{3}}\wedge t}_{0} \int _{U_{m}} \bigl\vert \bigl(A^{\alpha} \bigl(Y_{s-}^{n}-Y_{s-}^{m} \bigr), A^{\alpha}F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \\& \qquad {}-A^{\alpha}F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \bigr) \bigr\vert \tilde{N}(ds,du) \\& \qquad {} +C_{3}e^{2+C_{2}k_{3}}\mathbf{E} \int^{\tau^{m,n}_{k_{3}}\wedge t}_{0} \int _{U_{m}} \bigl\vert A^{\alpha}F \bigl(Y_{s-}^{n}+z_{s-},u \bigr)-A^{\alpha }F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \bigr\vert ^{2}\tilde{N}(ds,du) \\& \qquad {} +C_{3}e^{2+C_{2}k_{3}}\mathbf{E} \int^{\tau^{m,n}_{k_{3}}\wedge t}_{0} \int _{U_{m}} \bigl\vert \bigl(A^{\alpha} \bigl(Y_{s-}^{n}-Y_{s-}^{m} \bigr), A^{\alpha}F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \\& \qquad {}-A^{\alpha}F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \bigr) \bigr\vert \lambda(du)\,ds \\& \qquad {} +C_{3}e^{2+C_{2}k_{3}}\mathbf{E} \int^{\tau^{m,n}_{k_{3}}\wedge t}_{0} \int _{U_{m}} \bigl\vert A^{\alpha}F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) -A^{\alpha }F \bigl(Y_{s-}^{n}+z_{s-},u \bigr) \bigr\vert ^{2}\lambda(du)\,ds \\& \quad \leq \varepsilon \mathbf{E}\sup_{s\leq\tau^{m,n}_{k_{3}}\wedge t} \bigl\vert A^{\alpha} \bigl(Y^{n}_{s}-Y^{m}_{s} \bigr) \bigr\vert ^{2} +C_{4}(\varepsilon ,k)\mathbf{E} \int^{\tau ^{m,n}_{k_{3}}\wedge t}_{0} \int_{U_{n}\setminus U_{m}} \bigl\vert A^{\alpha }F \bigl(Y_{s}^{n}+z_{s},u \bigr) \bigr\vert ^{2}\lambda(du)\,ds. \end{aligned}$$
$$\begin{aligned}& (1-\varepsilon )\mathbf{E}\sup_{s\leq\tau^{m,n}_{k_{3}}\wedge t} \bigl\vert A^{\alpha } \bigl(Y^{n}_{s}-Y^{m}_{s} \bigr) \bigr\vert ^{2}+\mathbf{E} \int^{\tau^{m,n}_{k_{3}}\wedge t}_{0} \bigl\vert A^{\frac {1}{2}+\alpha} \bigl(Y^{n}_{s}-Y^{m}_{s} \bigr) \bigr\vert ^{2}\,ds \end{aligned}$$
(3.12)
$$\begin{aligned}& \quad \leq C_{4}t\sup_{x\in\mathbf{H}} \int_{U_{m}^{c}} \bigl\vert A^{\alpha }F(x,u) \bigr\vert ^{2}\lambda(du)\rightarrow0, \end{aligned}$$
(3.13)
Since
and
we conclude that, for any fixed \(t>0\) and \(k_{1}\geq1\),
So \(\{Y^{n}_{\tau_{k_{1}}\wedge s}\}\) is a Cauchy sequence in the space \(\mathscr {D}_{T}\), which is the space of all \(\mathscr {D}(A^{\alpha})\)-valued adapted càdlàg processes \(Y_{t}\) with \(\mathbf{E}(\sup_{t\leq T}|A^{\alpha}Y_{t}|+\int^{T}_{0}|A^{\frac{1}{2}+\alpha}Y_{s}|\,ds)<\infty\) for any positive number T. Consequently, there exists a process \(\overline{Y}^{k_{1}}\in \mathscr {D}_{T}\) such that
It is straightforward to examine that \(\overline{Y}^{k_{1}}\) is a weak solution of (2.3) on \([0,\tau_{k_{1}}]\) (cf. [6]). Set
where \(\tau_{0}=0\). Since \(\tau_{k_{1}}\rightarrow\infty(k\rightarrow\infty )\), Y is a weak solution of (2.3). □
$$\begin{aligned}& \mathbf{E}\sup_{s \leq\tau_{k_{1}}\wedge t} \bigl\vert A^{\alpha}\bigl(Y^{n}_{s}-Y^{m}_{s} \bigr) \bigr\vert \\& \quad \leq \Bigl(\mathbf{E}\sup_{s\leq\tau_{k_{3}}^{m,n}\wedge t} \bigl\vert A^{\alpha}\bigl(Y^{n}_{s}-Y^{m}_{s} \bigr) \bigr\vert ^{2} \Bigr)^{\frac{1}{2}}+ \Bigl(\mathbf{E}\sup _{s\leq\tau _{k_{1}}\wedge t} \bigl\vert A^{\alpha}\bigl(Y^{n}_{s}-Y^{m}_{s} \bigr) \bigr\vert ^{2} \Bigr)^{\frac{1}{2}} \bigl[\mathbf {P} \bigl(t> \tau^{m,n}_{k_{3}} \bigr) \bigr]^{\frac{1}{2}} \end{aligned}$$
$$\begin{aligned}& \mathbf{E} \int_{0}^{\tau_{k_{1}}\wedge t} \bigl\vert A^{\frac{1}{2}+\alpha } \bigl(Y^{n}_{s}-Y^{m}_{s} \bigr) \bigr\vert \,ds \\& \quad \leq \biggl(\mathbf{E} \int_{0}^{\tau_{k_{3}}^{m,n}\wedge t} \bigl\vert A^{\frac {1}{2}+\alpha} \bigl(Y^{n}_{s}-Y^{m}_{s} \bigr) \bigr\vert ^{2}\,ds \biggr)^{\frac{1}{2}} \\& \qquad {}+ \biggl(\mathbf {E} \int_{0}^{\tau_{k_{1}}\wedge t} \bigl\vert A^{\frac{1}{2}+\alpha} \bigl(Y^{n}_{s}-Y^{m}_{s} \bigr) \bigr\vert ^{2}\,ds \biggr)^{\frac{1}{2}} \bigl[\mathbf{P} \bigl(t>\tau ^{m,n}_{k_{3}} \bigr) \bigr]^{\frac{1}{2}}, \end{aligned}$$
$$ \lim_{m\rightarrow\infty, n\geq m} \biggl[\mathbf{E}\sup_{s\leq\tau_{k_{1}}\wedge t} \bigl\vert A^{\alpha}\bigl(Y^{n}_{s}-Y^{m}_{s} \bigr) \bigr\vert +\mathbf{E} \int _{0}^{\tau_{k_{1}}\wedge t} \bigl\vert A^{\frac{1}{2}+\alpha} \bigl(Y^{n}_{s}-Y^{m}_{s} \bigr) \bigr\vert \,ds \biggr]=0. $$
$$\lim_{n\rightarrow\infty} \biggl[\mathbf{E}\sup_{s\leq\tau_{k_{1}}\wedge T} \bigl\vert A^{\alpha}\bigl(Y^{n}_{s}- \overline{Y}^{k_{1}}_{s}\bigr) \bigr\vert +\mathbf{E} \int_{0}^{\tau _{k_{1}}\wedge t} \bigl\vert A^{\frac{1}{2}+\alpha} \bigl(Y^{n}_{s}-\overline {Y}^{k_{1}}_{s} \bigr) \bigr\vert \,ds \biggr]=0. $$
$$Y=\sum^{\infty}_{k_{1}=1}\overline{Y}^{k_{1}}I_{\{[\tau_{k_{1}-1},\tau_{k_{1}})\}}, $$
4 Invariant measure
This section aims to show that, although the fragile solution is in a somewhat weaker sense than the classical weak solution, we can utilize it to prove the existence of an invariant measure following classical routes, say by making use of the Krylov–Bogoliubov averaging procedure with energy (compactness) estimates.
Theorem 4.1
Assume that hypotheses (\(\mathrm{H}_{1}\))–(\(\mathrm{H}_{4}\)) hold, then there exists an invariant measure
μ
for the transition semigroup
\(P_{t}\)
associated with (2.3).
Proof
Let \(X_{t}=X(t,t_{0})\) be the solution of the following equation:
Consider the Ornstein–Uhlenbeck process \(z_{t}^{\gamma}\) satisfying \(dz_{t}^{\gamma}=(-A-\gamma)z_{t}^{\gamma}\,dt+Q\,dW_{t}\). Make Ornstein–Uhlenbeck transformation of (4.1) by letting \(X(t,t_{0})=Z^{\gamma}(t,t_{0})+z_{t}^{\gamma}\), then \(Z_{t}^{\gamma}=Z^{\gamma}(t,t_{0})\) satisfies
$$\begin{aligned} X_{t} =&X_{t_{0}}- \int^{t}_{t_{0}}A X_{s}\,ds- \int^{t}_{t_{0}}B(X_{s})\,ds- \int^{t}_{t_{0}}R(X_{s})\,ds+ \int^{t}_{t_{0}}Q\,dW_{s} \\ &{}+ \int^{t}_{t_{0}} \int_{U}F(X_{s},u)\tilde{N}(ds,du). \end{aligned}$$
(4.1)
$$ \textstyle\begin{cases} dZ_{t}^{\gamma}=-AZ_{t}^{\gamma}\,dt-B(Z_{t}^{\gamma}+z_{t}^{\gamma})\,dt-R(Z_{t}^{\gamma}+z_{t}^{\gamma})\,dt \\ \hphantom{dZ_{t}^{\gamma}={}}{}+ \int_{U}F(Z_{t}^{\gamma}+z_{t}^{\gamma},u)\tilde{N}(dt,du), \\ Z_{t_{0}}^{\gamma}=-z_{t_{0}}^{\gamma}. \end{cases} $$
(4.2)
Recall that \(\varXi Z(t):=Z(t)-Z(t-)\). Applying Itô’s formula to \(|A^{\alpha}Z_{t}^{\gamma}(t)|^{2}\) yields
where
In the following proof, we will utilize the interpolation inequality
the proof of which refers to [8, Lemma 4.1]. Therefore,
Hence
where \(\varepsilon \in(0,1)\) and the last inequality is derived by combining hypotheses (\(\mathrm{H}_{2}\)), (\(\mathrm{H}_{3}\)) with the Young inequality.
$$\begin{aligned} \bigl\vert A^{\alpha}Z_{t}^{\gamma}\bigr\vert ^{2} =& \bigl\vert A^{\alpha}Z_{t_{0}}^{\gamma}\bigr\vert ^{2}+ \int^{t}_{t_{0}}\big\langle 2A^{\alpha}Z_{s-}^{\gamma}\cdot A^{\alpha},dZ_{s}^{\gamma}\big\rangle \\ &{}+ \sum_{s\leq t}\bigl(\varXi \bigl( \bigl\vert A^{\alpha}Z_{s}^{\gamma}\bigr\vert ^{2} \bigr)-\big\langle 2A^{\alpha}Z_{s-}^{\gamma}\cdot A^{\alpha},\varXi Z_{s}^{\gamma}\big\rangle \bigr) \\ =& \bigl\vert A^{\alpha}Z_{t_{0}}^{\gamma}\bigr\vert ^{2}+ \int^{t}_{t_{0}} \bigl(2A^{\alpha}Z_{s-}^{\gamma}\cdot A^{\alpha},- \bigl[AZ_{s}^{\gamma}+B \bigl(Z_{s}^{\gamma}+z_{s}^{\gamma}\bigr)+R \bigl(Z_{s}^{\gamma}+z_{s}^{\gamma}\bigr) \bigr] \,ds \bigr) \\ &{}+ \int^{t}_{t_{0}} \int_{U} \bigl(2A^{\alpha}Z_{s-}^{\gamma}\cdot A^{\alpha },F \bigl(Z_{s-}^{\gamma}+z_{s-}^{\gamma},u \bigr) \bigr)\tilde{N}(ds,du) \\ &{}+ \int^{t}_{t_{0}} \int_{U} \bigl( \bigl\vert A^{\alpha} \bigl(Z_{s-}^{\gamma}+F \bigl(Z_{s-}^{\gamma}+z_{s-}^{\gamma},u \bigr) \bigr) \bigr\vert ^{2}- \bigl\vert A^{\alpha}Z_{s-}^{\gamma}\bigr\vert ^{2} \bigr)N(ds,du) \\ &{}- \int^{t}_{t_{0}} \int_{U} \bigl(2A^{\alpha}Z_{s-}^{\gamma}\cdot A^{\alpha},F \bigl(Z_{s-}^{\gamma}+z_{s-}^{\gamma},u \bigr) \bigr)N(ds,du) \\ =& \bigl\vert A^{\alpha}Z_{t_{0}}^{\gamma}\bigr\vert ^{2}+ \int^{t}_{t_{0}} \bigl(2A^{\alpha}Z_{s-}^{\gamma}\cdot A^{\alpha},- \bigl[AZ_{s}^{\gamma}+B \bigl(Z_{s}^{\gamma}+z_{s}^{\gamma}\bigr)+R \bigl(Z_{s}^{\gamma}+z_{s}^{\gamma}\bigr) \bigr] \,ds \bigr)+\widetilde{M}_{t}, \end{aligned}$$
$$\begin{aligned} \widetilde{M}_{t} =& \int^{t}_{t_{0}} \int_{U} \bigl(2A^{\alpha}Z_{s-}^{\gamma}\cdot A^{\alpha},F \bigl(Z_{s-}^{\gamma}+z_{s-}^{\gamma},u \bigr) \bigr)\tilde{N}(ds,du) \\ &{}+ \int^{t}_{t_{0}} \int_{U} \bigl( \bigl\vert A^{\alpha} \bigl(Z_{s-}^{\gamma}+F \bigl(Z_{s-}^{\gamma}+z_{s-}^{\gamma},u \bigr) \bigr) \bigr\vert ^{2}- \bigl\vert A^{\alpha}Z_{s-}^{\gamma}\bigr\vert ^{2} \bigr)N(ds,du) \\ &{}- \int^{t}_{t_{0}} \int_{U} \bigl(2A^{\alpha}Z_{s-}^{\gamma}\cdot A^{\alpha},F \bigl(Z_{s-}^{\gamma}+z_{s-}^{\gamma},u \bigr) \bigr)N(ds,du). \end{aligned}$$
$$ \bigl\vert \bigl(B(u,v),A^{2\alpha}w \bigr) \bigr\vert \leq C_{0} \bigl\vert A^{\frac{1}{2}+\alpha}w \bigr\vert \bigl\vert A^{\frac{1}{4}+\frac{\alpha}{2}}u \bigr\vert \bigl\vert A^{\frac {1}{4}+\frac{\alpha}{2}}v \bigr\vert , $$
(4.3)
$$\begin{aligned}& \bigl\vert A^{\alpha}Z_{t}^{\gamma}\bigr\vert ^{2}+2 \int^{t}_{t_{0}} \bigl\vert A^{\alpha+\frac {1}{2}}Z_{s}^{\gamma}\bigr\vert ^{2}\,ds \\& \quad \leq \bigl\vert A^{\alpha}Z_{t_{0}}^{\gamma}\bigr\vert ^{2}+C_{1} \int^{t}_{t_{0}} \bigl\vert A^{\alpha+\frac {1}{2}}Z_{s}^{\gamma}\bigr\vert \bigl\vert A^{\frac{\alpha}{2}+\frac{1}{4}} \bigl(Z_{s}^{\gamma}+z_{s}^{\gamma}\bigr) \bigr\vert ^{2}\,ds \\& \qquad {} +C_{1} \int^{t}_{t_{0}} \bigl( \bigl\vert A^{\alpha}Z_{s}^{\gamma}\bigr\vert ^{2}+ \bigl\vert A^{\alpha}Z_{s}^{\gamma}\bigr\vert \bigl\vert A^{\alpha}z_{s}^{\gamma}\bigr\vert \bigr)\,ds+ \widetilde{M}_{t} \\& \quad \leq \bigl\vert A^{\alpha}Z_{t_{0}}^{\gamma}\bigr\vert ^{2}+C_{2} \int^{t}_{t_{0}} \bigl[ \bigl\vert A^{\alpha+\frac {1}{2}}Z_{s}^{\gamma}\bigr\vert \bigl\vert A^{\frac{\alpha}{2}+\frac{1}{4}}Z_{s}^{\gamma}\bigr\vert ^{2} + \bigl\vert A^{\alpha+\frac{1}{2}}Z_{s}^{\gamma}\bigr\vert \bigl\vert A^{\frac{\alpha}{2}+\frac {1}{4}}z_{s}^{\gamma}\bigr\vert ^{2} \\& \qquad {} + \bigl\vert A^{\alpha}Z_{s}^{\gamma}\bigr\vert ^{2}+ \bigl\vert A^{\alpha}z_{s}^{\gamma}\bigr\vert ^{2} \bigr]\,ds+\widetilde {M}_{t} \\& \quad \leq \bigl\vert A^{\alpha}Z_{t_{0}}^{\gamma}\bigr\vert ^{2}+C_{3} \int^{t}_{t_{0}} \bigl[ \bigl\vert A^{\alpha }Z_{s}^{\gamma}\bigr\vert ^{4\alpha} \bigl\vert A^{\alpha+\frac{1}{2}}Z_{s}^{\gamma}\bigr\vert ^{3-4\alpha} + \bigl\vert A^{\alpha+\frac{1}{2}}Z_{s}^{\gamma}\bigr\vert \bigl\vert A^{\frac{\alpha}{2}+\frac {1}{4}}z_{s}^{\gamma}\bigr\vert ^{2} \\& \qquad {} + \bigl\vert A^{\alpha}Z_{s}^{\gamma}\bigr\vert ^{2}+ \bigl\vert A^{\alpha}z_{s}^{\gamma}\bigr\vert ^{2} \bigr]\,ds+\widetilde {M}_{t} \\& \quad \leq \bigl\vert A^{\alpha}h \bigr\vert ^{2}+ \int^{t}_{t_{0}} \bigl\vert A^{\alpha+\frac{1}{2}}Z_{s}^{\gamma}\bigr\vert ^{2}\,ds+C_{4} \int^{t}_{t_{0}} \bigl\vert A^{\alpha}Z_{s}^{\gamma}\bigr\vert ^{2}\,ds \\& \qquad {} +C_{4} \int^{t}_{t_{0}} \bigl( \bigl\vert A^{\alpha}z_{s}^{\gamma}\bigr\vert ^{2}+ \bigl\vert A^{\frac{\alpha }{2}+\frac{1}{4}}z_{s}^{\gamma}\bigr\vert ^{2} \bigr)\,ds+\widetilde{M}_{t}. \end{aligned}$$
$$\begin{aligned}& \mathbf{E}\sup_{s\leq t} \bigl\vert A^{\alpha}Z_{s}^{\gamma}\bigr\vert ^{2}+\mathbf{E} \int ^{t}_{t_{0}} \bigl\vert A^{\alpha+\frac{1}{2}}Z_{s}^{\gamma}\bigr\vert ^{2}\,ds \\& \quad \leq \bigl\vert A^{\alpha}Z_{t_{0}}^{\gamma}\bigr\vert ^{2}+C_{4}\mathbf{E} \int ^{t}_{t_{0}} \bigl\vert A^{\alpha}Z_{s}^{\gamma}\bigr\vert ^{2}\,ds+C_{4}\mathbf{E} \int ^{t}_{t_{0}} \bigl( \bigl\vert A^{\alpha}z_{s}^{\gamma}\bigr\vert ^{2}+ \bigl\vert A^{\frac{\alpha}{2}+\frac {1}{4}}z_{s}^{\gamma}\bigr\vert ^{2} \bigr)\,ds \\& \qquad {} +\mathbf{E} \int^{t}_{t_{0}} \int_{U} \bigl(2A^{\alpha}Z_{s-}^{\gamma}\cdot A^{\alpha},F \bigl(Z_{s-}^{\gamma}+z_{s-}^{\gamma},u \bigr) \bigr)\tilde{N}(ds,du) \\& \qquad {} +\mathbf{E} \int^{t}_{t_{0}} \int_{U} \bigl( \bigl\vert A^{\alpha} \bigl(Z_{s-}^{\gamma}+F \bigl(Z_{s-}^{\gamma}+z_{s-}^{\gamma},u \bigr) \bigr) \bigr\vert ^{2}- \bigl\vert A^{\alpha}Z_{s-}^{\gamma}\bigr\vert ^{2} \bigr)N(ds,du) \\& \qquad {} -\mathbf{E} \int^{t}_{t_{0}} \int_{U} \bigl(2A^{\alpha}Z_{s-}^{\gamma}\cdot A^{\alpha},F \bigl(Z_{s-}^{\gamma}+z_{s-}^{\gamma},u \bigr) \bigr)N(ds,du) \\& \quad \leq \bigl\vert A^{\alpha}Z_{t_{0}}^{\gamma}\bigr\vert ^{2}+C_{4}\mathbf{E} \int ^{t}_{t_{0}} \bigl\vert A^{\alpha}Z_{s}^{\gamma}\bigr\vert ^{2}\,ds+C_{4}\mathbf{E} \int ^{t}_{t_{0}} \bigl( \bigl\vert A^{\alpha}z_{s}^{\gamma}\bigr\vert ^{2}+ \bigl\vert A^{\frac{\alpha}{2}+\frac {1}{4}}z_{s}^{\gamma}\bigr\vert ^{2} \bigr)\,ds \\& \qquad {} +\mathbf{E} \int^{t}_{t_{0}} \int_{U} \bigl\vert \bigl(2A^{\alpha}Z_{s-}^{\gamma}\cdot A^{\alpha},F \bigl(Z_{s-}^{\gamma}+z_{s-}^{\gamma},u \bigr) \bigr) \bigr\vert \tilde{N}(ds,du) \\& \qquad {} +\mathbf{E} \int^{t}_{t_{0}} \int_{U} \biggl\vert \int_{0}^{1} \bigl(2A^{\alpha } \bigl(Z_{-s}+s'F \bigl(Z_{s-}^{\gamma}+z_{s-}^{\gamma},u \bigr) \bigr)\cdot A^{\alpha}, \\& \qquad F \bigl(Z_{s-}^{\gamma}+z_{s-}^{\gamma},u \bigr) \bigr)\,ds' \biggr\vert N(ds,du) \\& \qquad {} +\mathbf{E} \int^{t}_{t_{0}} \int_{U} \bigl\vert \bigl(2A^{\alpha}Z_{s-}^{\gamma}\cdot A^{\alpha},F \bigl(Z_{s-}^{\gamma}+z_{s-}^{\gamma},u \bigr) \bigr) \bigr\vert N(ds,du) \\& \quad \leq \bigl\vert A^{\alpha}Z_{t_{0}}^{\gamma}\bigr\vert ^{2}+C_{4}\mathbf{E} \int ^{t}_{t_{0}} \bigl\vert A^{\alpha}Z_{s}^{\gamma}\bigr\vert ^{2}\,ds+C_{4}\mathbf{E} \int ^{t}_{t_{0}} \bigl( \bigl\vert A^{\alpha}z_{s}^{\gamma}\bigr\vert ^{2}+ \bigl\vert A^{\frac{\alpha}{2}+\frac {1}{4}}z_{s}^{\gamma}\bigr\vert ^{2} \bigr)\,ds \\& \qquad {} +C_{5} \biggl[\mathbf{E} \int^{t}_{t_{0}} \int_{U} \bigl\vert A^{\alpha}Z_{s}^{\gamma}\bigr\vert ^{2}\cdot \bigl\vert A^{\alpha}F \bigl(Z_{s}^{\gamma}+z_{s}^{\gamma},u \bigr) \bigr\vert ^{2}\lambda(du)\,ds \biggr]^{\frac{1}{2}} \\& \qquad {} +C_{5} \biggl[\mathbf{E} \int^{t}_{t_{0}} \int_{U} \bigl\vert A^{\alpha}F \bigl(Z_{s}^{\gamma}+z_{s}^{\gamma},u \bigr) \bigr\vert ^{2}\cdot \bigl\vert A^{\alpha}F \bigl(Z_{s}^{\gamma}+z_{s}^{\gamma},u \bigr) \bigr\vert ^{2}\lambda (du)\,ds \biggr]^{\frac{1}{2}} \\& \quad \leq \bigl\vert A^{\alpha}Z_{t_{0}}^{\gamma}\bigr\vert ^{2}+\varepsilon \mathbf{E}\sup_{s\leq t} \bigl\vert A^{\alpha}Z_{t}^{\gamma}\bigr\vert ^{2}+C_{6}\mathbf{E} \int^{t}_{t_{0}} \bigl\vert A^{\alpha }Z_{s}^{\gamma}\bigr\vert ^{2}\,ds \\& \qquad {} +C_{6}\mathbf{E} \int^{t}_{t_{0}} \bigl( \bigl\vert A^{\alpha}z_{s}^{\gamma}\bigr\vert ^{2}+ \bigl\vert A^{\frac {\alpha}{2}+\frac{1}{4}}z_{s}^{\gamma}\bigr\vert ^{2} \bigr)\,ds, \end{aligned}$$
Henceforth,
If \(t_{0}\leq-1\),
Repeating the argument similarly in Lemma 3.1, (\(\mathrm{H}_{1}\)) implies that \(z_{t}^{\gamma}\) has a continuous version in \(\mathscr {D}(A^{\frac{1}{4}+\frac{\alpha}{2}})\), and therefore, also has a continuous version in \(\mathscr {D}(A^{\alpha})\). Furthermore, \(\mathbf{E}\int^{0}_{-1}(|A^{\alpha}z_{s}^{\gamma}|^{2}+|A^{\frac{\alpha}{2}+\frac{1}{4}}z_{s}^{\gamma}|^{2})\,ds\) is finite. Following almost the same argument as [4, Proposition15.4.3], it can be proved that there exists \(\gamma >0\) such that \(|A^{\frac{1}{2}}Z^{\gamma}(-1,t_{0})|^{2}\) is finite a.e. for \(t_{0}\leq -1\).
$$\begin{aligned}& (1-\varepsilon )\mathbf{E}\sup_{s\leq t} \bigl\vert A^{\alpha}Z_{s}^{\gamma}\bigr\vert ^{2}+ \mathbf {E} \int^{t}_{t_{0}} \bigl\vert A^{\alpha+\frac{1}{2}}Z_{s}^{\gamma}\bigr\vert ^{2}\,ds \\& \quad \leq \bigl\vert A^{\alpha}Z_{t_{0}}^{\gamma}\bigr\vert ^{2}+C_{6}\mathbf{E} \int ^{t}_{t_{0}} \bigl\vert A^{\alpha}Z_{s}^{\gamma}\bigr\vert ^{2}\,ds+C_{6}\mathbf{E} \int ^{t}_{t_{0}} \bigl( \bigl\vert A^{\alpha}z_{s}^{\gamma}\bigr\vert ^{2}+ \bigl\vert A^{\frac{\alpha}{2}+\frac {1}{4}}z_{s}^{\gamma}\bigr\vert ^{2} \bigr)\,ds. \end{aligned}$$
(4.4)
$$\begin{aligned} \mathbf{E} \bigl\vert Z^{\gamma}(0,t_{0}) \bigr\vert ^{2} \leq& e^{2C_{6}} \biggl[ \bigl\vert A^{\alpha}Z^{\gamma}(-1,t_{0}) \bigr\vert ^{2}+C_{6}\mathbf{E} \int ^{0}_{-1} \bigl( \bigl\vert A^{\alpha}z_{s}^{\gamma}\bigr\vert ^{2}+ \bigl\vert A^{\frac{\alpha}{2}+\frac {1}{4}}z_{s}^{\gamma}\bigr\vert ^{2} \bigr)\,ds \biggr] \\ \leq& e^{2C_{7}} \biggl[ \bigl\vert A^{\frac{1}{2}}Z^{\gamma}(-1,t_{0}) \bigr\vert ^{2}+C_{6}\mathbf{E} \int^{0}_{-1} \bigl( \bigl\vert A^{\alpha}z_{s}^{\gamma}\bigr\vert ^{2}+ \bigl\vert A^{\frac{\alpha}{2}+\frac {1}{4}}z_{s}^{\gamma}\bigr\vert ^{2} \bigr)\,ds \biggr]. \end{aligned}$$
Therefore, both \(\sup_{t_{0}\leq -1}\mathbf{E}|A^{\alpha}Z^{\gamma}(0,t_{0})|^{2}\) and \(\sup_{t_{0}\leq -1}\mathbf{E}|A^{\alpha}X(0,t_{0})|^{2}\) are finite. Following the classical arguments (cf. [12]), we can prove the tightness of \(\{\mu_{T}(\cdot)=\frac{1}{T}\int_{-T}^{0}P(s,x,\cdot), T>0\} \), which implies the existence of an invariant measure μ. □
5 Discussion
In this section, we will generalize the result of well-posedness for stochastic Boussinesq equation (2.1) to some of the stochastic hydrodynamical systems, such as 2D-stochastic Navier–Stokes equation, 2D magneto-hydrodynamic equation, 2D Boussinesq model for the Bénard convection, 2D magnetic Bénard problem, and so on. We adopt the notations and assumptions proposed in [2] and refer to [11, 14, 21, 22] for more examples.
We will introduce an abstract framework for the stochastic hydrodynamical systems. Let H be a separable Hilbert space with the norm \(|\cdot|\), and the operator A be an unbounded self-adjoint positive linear operator on H. Denote \(V=\operatorname{Dom}(A^{\frac{1}{2}})\) with the norm \(\|v\|=|A^{\frac {1}{2}}v|\). Let \(V'\) be the dual of V. For any \(u\in V\), \(v\in V'\), the duality product between V and \(V'\) is denoted by \(\langle u, v\rangle \). Suppose that the mapping \(B: V\times V\to V'\) satisfies the following assumptions:
-
(H(i)) \(B(\cdot,\cdot): V\times V\to V'\) is a continuous bilinear mapping.
-
(H(ii)) For any \(u_{i}\in V\), \(i=1,2,3\), there holds$$ \bigl\langle B(u_{1},u_{2}),u_{3} \bigr\rangle =- \bigl\langle B(u_{1},u_{3}),u_{2} \bigr\rangle . $$(5.1)
-
(H(iii)) There exists a Banach space \(\mathscr {H}\) which satisfies(iii-1)\(V\subset \mathscr {H} \subset H\);(iii-2)there exists a positive constant \(\alpha_{0}\) such that$$ \|v\|_{\mathscr {H}}^{2}\leq\alpha_{0} |v| \cdot\|v\|, \quad \mbox{for any }v\in V; $$(5.2)(iii-3)for any \(\eta>0\), there exists a positive constant \(C_{\eta}\) such that$$ \bigl\vert \bigl\langle B(u_{1},u_{2}),u_{3} \bigr\rangle \bigr\vert \leq\eta\|u_{3}\|^{2}+C_{\eta} \|u_{1}\| ^{2}_{\mathscr {H}}\cdot\|u_{2} \|^{2}_{\mathscr {H}},\quad u_{i}\in V, i=1,2,3. $$(5.3)
As we will see, the majority of stochastic hydrodynamical systems with Lévy noise, such as stochastic two-dimensional Navier–Stokes equation, stochastic two-dimensional Boussinesq equations, stochastic two-dimensional magnetic Bénard equations, and stochastic two-dimensional magneto-hydrodynamic equations, can be represented uniformly as the following stochastic evolution equation (see [20]):
where \(R(\cdot)\) and Q are linear bounded operators in H, \(W(\cdot )\) is an H-valued Brownian motion, F is a measurable mapping from some measurable space X to H, Ñ is a compensated Poisson measure on \([0,\infty)\times X\) with intensity measure ν. Additionally, we need
$$ \textstyle\begin{cases} du(t)+[Au(t)+B(u(t),u(t))+R(u(t))]\,dt \\ \quad =Q\,dW(t)+\int_{X} F(u(t-),x)\tilde{N}(dt,dx), \\ u(0)=u_{0}, \end{cases} $$
(5.4)
-
(H(iv)) There exist positive constants C and K such that, for some \(\alpha\in[1/4,1/2)\),
-
(H(iv-1)) ν is a σ-finite measure on \(\mathscr {B}(X)\), \(Q:{H}\rightarrow {H}\) is a linear bounded operator with range \(\mathscr {R}(Q)\) dense in \(\mathscr {D}(A^{\frac{1}{4}+\frac{\alpha}{2}})\) and \(\mathscr {D}(A^{2\alpha})\subset \mathscr {R}(Q)\subset \mathscr {D}(A^{\frac{1}{4}+\frac{\alpha}{2}+\varepsilon })\) for some \(\varepsilon >0\);
-
(H(iv-2)) \(\int_{U}|A^{\alpha}F(0,u)|^{2}\lambda(du)=C\);
-
(H(iv-3))\(\int_{U}|A^{\alpha}(F(x,u)-F(y,u))|^{2}\lambda (du)\leq K|A^{\alpha}(x-y)|^{2}\);
-
(H(iv-4)) \(\sup_{x\in H}\int_{U^{c}_{m}}|A^{\alpha}F(x,u)|^{2}\lambda(du)\rightarrow0\), as \(m\rightarrow\infty\).
-
Repeating similar arguments within Sect. 3 and Sect. 4 with minor modification, we can obtain the following result.
Theorem 5.1
Assume that hypotheses H(i)–H(iv) hold. For every
\(u_{0}\in \mathscr {D}(A^{\alpha})\), there exists a unique fragile solution
u
to equation (5.4). Denote by
\(P_{t}\)
the Markov semigroup on
H
generated by the solution flow. Then
\(P_{t}\)
is a Feller Markov process. Furthermore, there exists at least one invariant measure
μ
for the transition semigroup
\(P_{t}\).
Acknowledgements
The authors would like to thank the editor and the reviewers for giving us constructive suggestions which would help us both in English and in depth to improve the quality of the paper.
Competing interests
The authors declare to have no competing interests.
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
