数学杂志  2016, Vol. 36 Issue (5): 1083-1090   PDF    
扩展功能
加入收藏夹
复制引文信息
加入引用管理器
Email Alert
RSS
本文作者相关文章
杨洪福
张启敏
与年龄相关的随机分数阶种群系统温和解的存在性、唯一性
杨洪福, 张启敏    
北方民族大学数学与信息科学学院, 宁夏 银川 750021
摘要:本文研究了一类与年龄相关的随机分数阶种群动态系统.利用不动点定理、随机分析和算子半群理论, 讨论了与年龄相关的随机分数阶种群系统温和解的存在性、唯一性.本文是随机整数阶种群系统的推广.
关键词存在性    唯一性    随机分数阶种群系统    温和解    
EXISTENCE, UNIQUENESS FOR STOCHASTIC FRACTIONAL-ORDER AGE-DEPENDENT POPULATION
YANG Hong-fu, ZHANG Qi-min    
School of Mathematics and Information Science, Beifang University for Nationalities, Yinchuan 750021, China
Abstract: In this paper, we study a class of stochastic fractional-order age-dependent population dynamic system. By using fixed point theorem, stochastic analysis and semigroup of operators theory, the main conclusion of the existence and uniqueness of mild solution to stochastic fractional-order age-dependent population equations are obtained. This paper is a generalization of the stochastic integer order population system.
Key words: existence     uniqueness     stochastic fractional-order population system     mild solution    
1 引言

近年来, 随机分数阶微分方程引起了国内外学者的关注, 并被广泛的应用在物理、生物、工程、金融等领域[1-5].然而, 分数阶非线性系统还可能存在与整数阶非线性系统相似的非线性现象, 如自振、多平衡状态、混沌以及更复杂的过渡过程等[6, 7].

另一方面, 分数阶微分方程与分形密切相关, 而分形富含于生物系统中.研究表明, 在传统的整数阶微分方程不能建模的现象中, 分数阶微分方程为此提供了可能性.这里, 特别强调的是, 分数阶与整数阶模型最大区别在于分数阶模型拥有记忆, 而死亡率有长程相似性, 它的主要特征恰恰包含了记忆.因此, 本文首次尝试建立与年龄相关的随机分数阶动力学模型.另外, 已经证实由分数阶微分方程建立的某些生物学模型比整数阶更有优势[8, 9].

现在将分数阶引入与年龄相关的随机种群模型[10].新系统可描述为如下分数阶非线性系统:

$ \begin{eqnarray}\label{E:1.1} \left\{ \begin{array}{ll} D_t^{\alpha}P_t=-\frac{\partial P_t}{\partial a}-\mu(t, a)P_t +f(t, P_t)+g(t, P_t)\frac{d\omega_t}{dt}, & \text{在} \ \ Q=[0, A]\times J\ \ \text{内}, \\ P(0, a)=P_0, & \text{在} \ \ [0, A]\ \ \text{内}, \\ P(t, 0)=\displaystyle\int_0^{A}\beta(t, a)P(t, a) da, & \text{在} \ \ J=[0, T]\ \ \text{内}, \end{array} \right. \end{eqnarray} $ (1.1)

其中 $0 < \alpha < 1$, $a\in [0, A]$表示年龄, $t\in [0, T]$表示时间, $0 < T < \infty$, $P_t:=P(t, a)$ $\beta(t, a)$ $\mu(t, a)$分别表示时刻 $t$年龄为 $a$的种群密度、生育率和死亡率. $f(t, P)$为外部环境对种群系统的影响, 如迁移、地震等突发性灾害对种群引起的影响, $g(t, P)\frac{d \omega_t}{dt}$为随机外界环境对系统的扰动.

$\alpha=1$的情况下, 方程(1.1) 成为经典的与年龄相关的随机种群模型, 该模型通过随机分析理论和数值方法已被广泛研究.例如张启敏等[10]研究了与年龄相关的随机种群方程解的存在性, 唯一性和指数稳定性.李荣华等[11]考虑了带Markovian跳的与年龄相关的随机种群方程数值解的收敛性.马维军和张启敏等[12, 13]讨论了与年龄相关的随机种群方程带分数布朗运动的数值解和Markovian跳的渐近稳定性.杨洪福等在文献[14]中介绍了与年龄相关的随机两种群方程在POD基下的数值解.

本文首次将分数阶引入与年龄相关的随机种群动力学模型, 讨论在满足一定的假设条件下证明了温和解的存在性、唯一性, 得到的结论是文献[10]的推广.

2 预备知识

$ V=H^1([0, A])\equiv \{\varphi | \varphi\in L^2([0, A]), \ \frac{\partial \varphi}{\partial a}\in L^2([0, A]), \text{其中}\frac{\partial \varphi}{\partial a} \text{是广义偏导数}\}, $

$V$是一个Sobolev空间, $\{\omega_t\}_{t\in J}$是定义在完备的概率空间 $(\Omega, \mathcal{A}, \mathbb{P})$取值于可分的Hilbert空间 $K$上的Brown运动,

$ L^p_V=L^p([0, T]; V), L_H^p=L^p([0, T]; H). $

下面, 首先给出两个相关的基本定义.

定义2.1[4]  函数 $f$ $\alpha$分数阶积分定义为

$ I^{\alpha} f(t)=\frac{1}{\Gamma(\alpha)}\int_{t_0}^{t} \frac{f(s)}{(t-s)^{1-\alpha}}ds, \ \ t>0, \ \beta >0, $

其中 $t\geq t_0$ $\alpha>0$, $\Gamma(\cdot)$为Gamma函数.

定义2.2[15, 16]  函数 $f$ $\alpha$阶Caputo分数阶微分定义为

$ D^{\alpha} f(t)=\frac{1}{\Gamma(n-\alpha)}\frac{d^n}{dt^n}\int_{t_0}^{t} \frac{f(s)}{(t-s)^{\alpha+1-n}}ds, \ \ t>0, $

其中 $t\geq t_0$, $n$是一个正整数满足 $n-1 < \alpha < n$.特殊地, 当 $0 < \alpha < 1$时,

$ D^{\alpha} f(t)=\frac{1}{\Gamma(1-\alpha)}\int_{t_0}^{t} \frac{f'(s)}{(t-s)^{\alpha}}ds, \ \ t>0. $

把与年龄相关的随机分数阶种群方程(1.1) 写成如下形式的积分方程[16]

$ \begin{eqnarray} \left\{ \begin{array}{ll} P_t=\mathscr{T}(t)P_0+\displaystyle\int_0^t(t-s)^{\alpha-1}\mathscr{S}(t-s)[-\mu(s, a)P_s+f(s, P_s)]ds\\ \ \ \ \ \ \ \ \ \ \ +\displaystyle\int_0^t(t-s)^{\alpha-1}\mathscr{S}(t-s)g(s, P_s)d\omega_s, \\ P(t, 0)=\displaystyle\int_0^{A}\beta(t, a)P_s da, \end{array} \right. \forall t\in J, \end{eqnarray} $ (2.1)

其中

$ \mathscr{T}(t)=\displaystyle\int_{0}^{\infty}\xi_{\alpha}(\theta)S(t^{\alpha}\theta)d \theta,\ \ \ \ \ \mathscr{S}(t)=\alpha\displaystyle\int_{0}^{\infty}\theta\xi_{\alpha}(\theta)S(t^{\alpha}\theta)d \theta, $

$S(t)$是由一个 $V$中的有界线性算子 $B=\frac{\partial}{\partial a}$生成的 $C_0$ -半群, $\xi_{\alpha}$是定义在 $(0, \infty)$的概率密度函数, 即 $\xi_{\alpha}(\theta)\geq 0$, $\theta\in(0, \infty)$并且

$ \displaystyle\int_{0}^{\infty}\xi_{\alpha}(\theta)d \theta=1. $

定义2.3[17]  如果一个 $\cal F$ $_t$ -循序可测的随机变量 $\{P(t, a)\}_{t\in J}$是方程(1.1) 的一个温和解, 则 $\{P(t, a)\}_{t\in J}$满足相应的积分方程(2.1).

引理2.4[18]  如果算子 $\{\mathscr{T}(t)\}_{t\in J}$ $\{\mathscr{S}(t)\}_{t\in J}$是强连续的, 则对每个 $P_t \in H$, 当 $t_1\rightarrow t_2$, 有

$ \parallel\mathscr{T}(t_2)P_t-\mathscr{T}(t_1)P_t\parallel\rightarrow 0, \parallel\mathscr{S}(t_2)P_t-\mathscr{S}(t_1)P_t\parallel\rightarrow 0. $

为了证明本文的主要结论, 给出以下假设条件:

(ⅰ) $\mu(t, a)$非负可测, 并且

$ \begin{array}{ll} 0\le \mu(t,a)\le \mu_0 < \infty, &\text{在} \ \ Q\ \ \text{内}. \end{array} $

(ⅱ) $\forall t\geq 0$, $\{\mathscr{T}(t)\}_{t\in J}$ $\{\mathscr{S}(t)\}_{t\in J}$是有界线性算子, 即

$ P_t \in I^p(0, T;V)\bigcap L^2(\Omega;C(0, T;H)), $

并且满足 $E\displaystyle\int_{0}^{T}\parallel P_t\parallel^2dt < \infty$, 则存在一个常数 $\gamma>0$, 使得

$ \begin{eqnarray*} &&\parallel \mathscr{T}(t) P_t\parallel\leq M e^{-\gamma t}\parallel P_t\parallel, \\ &&\parallel t^{\alpha-1}\mathscr{S}(t) P_t\parallel\leq \frac{M\alpha}{\Gamma(\alpha+1)} e^{-\gamma t}\parallel P_t\parallel, \ \ \ \ \forall t\geq 0. \end{eqnarray*} $

(ⅲ)设 $f(t, \cdot ) : L^2_H\rightarrow H$是对几乎所有 $t$有定义的非线性算子, 满足

(a.1) $f(t, 0)=0$;

(a.2) $\exists N>0$, 使得

$ \parallel f(t, x)-f(t, y)\parallel^2\leq N\parallel x-y\parallel^2. $

$g(t, \cdot ) : L^2_H\rightarrow $ $\cal{L}$ $ (K, H)$是对几乎所有 $t$有定义的非线性算子, $g(t, x)\in \cal{L}$ $ (K, H)$且满足

(b.1) $g(t, 0)=0$;

(b.2) $\exists L>0$, 使得

$ \parallel g(t, x)-g(t, y)\parallel_2^2\leq L\parallel x-y\parallel^2. $
3 温和解的存在性、唯一性

在本节中, 利用第二部分给出的定义和引理, 证明与年龄相关的随机分数阶种群方程(1.1) 的温和解的存在性、唯一性.定义算子 $\Psi: H\rightarrow H$, 满足如下方程

$ \begin{eqnarray}\label{E:3.1} \begin{array}{ll} \Psi (P_t)=&\mathscr{T}(t)P_0+\displaystyle\int_0^t(t-s)^{\alpha-1}\mathscr{S}(t-s)(-\mu(s, a)P_s)ds\\ &+\displaystyle\int_0^t(t-s)^{\alpha-1}\mathscr{S}(t-s)f(s, P_s)ds +\displaystyle\int_0^t(t-s)^{\alpha-1}\mathscr{S}(t-s)g(s, P_s)d\omega_s. \end{array} \end{eqnarray} $ (3.1)

为了证明文章的主要结论, 先证明如下引理.

引理3.1  对任意的 $P_t\in H$, $\Psi P_t$ $[0, T]$上是 $L^2$ -连续的.

  令 $0 < t_1 < t_2 < T$.对于方程(3.1)的任意固定点 $P_t\in H$, 有

$ \begin{eqnarray} \begin{array}{ll} E\parallel \Psi P_{t_2}-\Psi P_{t_1}\parallel^2\leq 4[E\parallel(\mathscr{T}(t_2)-\mathscr{T}(t_1))P_0\parallel^2+\sum\limits_{i=1}^{3}E\parallel I_{i}^{P}(t_2)-I_{i}^{P}(t_1)\parallel^2]. \end{array} \end{eqnarray} $ (3.2)

$\mathscr{T}(t)$是强连续的, 则当 $t_2-t_1\rightarrow 0$时, 方程(3.2) 右侧第一项也趋于零.接下来, 应用Holder不等式和假设条件(ⅰ)-(ⅲ), 有

$ \begin{array}{ll} &E\parallel I_{1}^{P}(t_2)-I_{1}^{P}(t_1)\parallel^2\\ \leq& 3E\parallel\displaystyle\int_{t_1}^{t_2}-\mu(s, a)(t_2-s)^{\alpha-1}\mathscr{S}(t_2-s)P_sds\parallel^2\\ &+3E\parallel\displaystyle\int_0^{t_{1}}-\mu(s, a)((t_2-s)^{\alpha-1}-(t_1-s)^{\alpha-1})\mathscr{S}(t_2-s)P_sds\parallel^2\\ &+3E\parallel\displaystyle\int_0^{t_{1}}-\mu(s, a)(t_1-s)^{\alpha-1}(\mathscr{S}(t_2-s)-\mathscr{S}(t_1-s))P_sds\parallel^2\\ \leq& 3(t_2-t_1)\mu^2_0(\frac{M\alpha}{\Gamma(1+\alpha)})^2\displaystyle\int_{t_1}^{t_2}e^{-2\gamma(t_2-s)}E\parallel P_s\parallel^2 ds\\ &+3\mu^2_0(\displaystyle\int_0^{t_{1}}\parallel((t_2-s)^{\alpha-1}-(t_1-s)^{\alpha-1})\mathscr{S}(t_2-s)\parallel^2 ds)(\displaystyle\int_{0}^{t_1}E\parallel P_s\parallel^2 ds)\\ &+3\mu_0^2(\displaystyle\int_0^{t_{1}}\parallel(t_1-s)^{\alpha-1}(\mathscr{S}(t_2-s)-\mathscr{S}(t_1-s))\parallel^2ds)(\displaystyle\int_{0}^{t_1}E\parallel P_s\parallel^2 ds). \end{array} $

进一步, 有

$ \begin{array}{ll} &E\parallel I_{2}^{P}(t_2)-I_{2}^{P}(t_1)\parallel^2\\ \leq& 3E\parallel\displaystyle\int_{t_1}^{t_2}(t_2-s)^{\alpha-1}\mathscr{S}(t_2-s)f(s, P_s)ds\parallel^2\\ &+3E\parallel\displaystyle\int_0^{t_{1}}((t_2-s)^{\alpha-1}-(t_1-s)^{\alpha-1})\mathscr{S}(t_2-s)f(s, P_s)ds\parallel^2\\ &+3E\parallel\displaystyle\int_0^{t_{1}}(t_1-s)^{\alpha-1}(\mathscr{S}(t_2-s)-\mathscr{S}(t_1-s))f(s, P_s)ds\parallel^2\\ \leq& 3(t_2-t_1)(\frac{M\alpha}{\Gamma(1+\alpha)})^2\displaystyle\int_{t_1}^{t_2}e^{-2\gamma(t_2-s)}E\parallel f(s, P_s)\parallel^2 ds\\ &+3(\displaystyle\int_0^{t_{1}}\parallel((t_2-s)^{\alpha-1}-(t_1-s)^{\alpha-1})\mathscr{S}(t_2-s)\parallel^2 ds)(\displaystyle\int_{0}^{t_1}E\parallel f(s, P_s)\parallel^2 ds)\\ &+3(\displaystyle\int_0^{t_{1}}\parallel(t_1-s)^{\alpha-1}(\mathscr{S}(t_2-s)-\mathscr{S}(t_1-s))\parallel^2ds)(\displaystyle\int_{0}^{t_1}E\parallel f(s, P_s)\parallel^2 ds). \end{array} $

类似的, 应用Holder不等式、Itô积分和假设条件(ⅰ)-(ⅲ), 可得

$ \begin{array}{ll} &E\parallel I_{3}^{P}(t_2)-I_{3}^{P}(t_1)\parallel^2\\ \leq& 3E\parallel\displaystyle\int_{t_1}^{t_2}(t_2-s)^{\alpha-1}\mathscr{S}(t_2-s)g(s, P_s)d\omega_s\parallel^2\\ &+3E\parallel\displaystyle\int_0^{t_{1}}((t_2-s)^{\alpha-1}-(t_1-s)^{\alpha-1})\mathscr{S}(t_2-s)g(s, P_s)d\omega_s\parallel^2\\ &+3E\parallel\displaystyle\int_0^{t_{1}}(t_1-s)^{\alpha-1}(\mathscr{S}(t_2-s)-\mathscr{S}(t_1-s))g(s, P_s)d\omega_s\parallel^2\\ \leq& 3(\frac{M\alpha}{\Gamma(1+\alpha)})^2\displaystyle\int_{t_1}^{t_2}e^{-2\gamma(t_2-s)}E\parallel g(s, P_s)\parallel_2^2 ds\\ &+3\displaystyle\int_0^{t_{1}}(\parallel((t_2-s)^{\alpha-1}-(t_1-s)^{\alpha-1})\mathscr{S}(t_2-s)\parallel^2) (E\parallel g(s, P_s)\parallel_2^2 )ds\\ &+3\displaystyle\int_0^{t_{1}}(\parallel(t_1-s)^{\alpha-1}(\mathscr{S}(t_2-s)-\mathscr{S}(t_1-s))\parallel^2)(E\parallel g(s, P_s)\parallel_2^2) ds. \end{array} $

因此由 $\mathscr{S}(t)$是强连续的和Lebesgue控制收敛定理, 得出这样的结论:当 $t_2-t_1 \rightarrow 0$时, 方程(3.2)右侧不等式趋于零.因此得出在 $[0, T)$ $\Psi P_t$是右连续.

类似的可以证明在 $[0, T)$ $\Psi P_t$是左连续.因此引理得证.

定理3.2  如果条件(ⅰ)-(ⅲ)是成立的, 并且满足下面的不等式, 则方程(1.1) 存在唯一的温和解 $ P_t\in H$.

$ M_0=3T(\mu_0^2+N+L)(\frac{M\alpha}{\Gamma(\alpha+1)})^2 < 1. $

  应用压缩映射原理证明算子 $\Psi$存在一个不动点.首先, 要证 $\Psi(H)\subset H$.令 $P_t\in H$, 由(3.1)式, 可得

$ \begin{eqnarray} \begin{array}{ll} E\parallel \Psi P_{t}\parallel^2\leq 4[E\parallel\mathscr{T}(t)P_0\parallel^2+\sum\limits_{i=1}^{3}E\parallel I_{i}^{P}(t)\parallel^2]. \end{array} \end{eqnarray} $ (3.3)

现在估计方程(3.3) 右边的项.第一项, 由条件(ⅱ)可得

$ \begin{eqnarray} E\parallel\mathscr{T}(t)P_0\parallel^2\leq M^2e^{-2\gamma t}E\parallel P_0\parallel^2. \end{eqnarray} $ (3.4)

第二项应用Holder不等式和假设条件(ⅰ)-(ⅲ)有

$ \begin{eqnarray}\label{E:3.5} \begin{array}{ll} E\parallel I_{1}^{P}(t)\parallel^2&=E\parallel\displaystyle\int_{0}^{t}-\mu(s, a)(t-s)^{\alpha-1}\mathscr{S}(t-s)P_sds\parallel^2\\ &\leq \mu_0^2(\frac{M\alpha}{\Gamma(1+\alpha)})^2(\displaystyle\int_{0}^{t}e^{-2\gamma (t-s)} ds)(\displaystyle\int_{0}^{t}E\parallel P_s\parallel^2 ds)\\ &\leq \mu_0^2(\frac{M\alpha}{\Gamma(1+\alpha)})^2\displaystyle\int_{0}^{t}E\parallel P_s\parallel^2ds. \end{array} \end{eqnarray} $ (3.5)

进一步又可以得到

$ \begin{eqnarray}\label{E:3.6} \begin{array}{ll} E\parallel I_{2}^{P}(t)\parallel^2&=E\parallel\displaystyle\int_{0}^{t}(t-s)^{\alpha-1}\mathscr{S}(t-s)f(s, P_s)ds\parallel^2\\ &\leq N(\frac{M\alpha}{\Gamma(1+\alpha)})^2\displaystyle\int_{0}^{t}E\parallel P_s\parallel^2ds. \end{array} \end{eqnarray} $ (3.6)

类似的, 应用Itô积分有

$ \begin{eqnarray} \begin{array}{ll} E\parallel I_{3}^{P}(t)\parallel^2&=E\parallel\displaystyle\int_{0}^{t}(t-s)^{\alpha-1}\mathscr{S}(t-s)g(s,P_s)d\omega_s\parallel^2\\ &\leq \displaystyle\int_{0}^{t} E \parallel(t-s)^{\alpha-1}\mathscr{S}(t-s) g(s,P_s)\parallel^2 ds\\ &\leq L(\frac{M\alpha}{\Gamma(1+\alpha)})^2\displaystyle\int_{0}^{t}E\parallel P_s\parallel^2ds. \end{array} \end{eqnarray} $ (3.7)

所以从(3.4) 式到(3.7)式可以得到 $E\parallel \Psi P_{t}\parallel^2 < \infty$.由引理3.1知 $\Psi P_{t}\in H$.因此算子 $\Psi$是从 $H$ $H$的自映射.接下来, 将证明 $\Psi $ $H$上的连续映射.事实上, 对任意的 $P_{1t}, P_{2t} \in H$, 由方程(3.1) 和Itô等距公式, 可得

$ \begin{eqnarray} \begin{array}{ll} E\parallel \Psi P_{1t}-\Psi P_{2t}\parallel^2\leq &3E\parallel\displaystyle\int_{0}^{t}-\mu(s, a)(t-s)^{\alpha-1}\mathscr{S}(t-s)(P_{1s}-P_{2s})ds\parallel^2\\ &+3E\parallel\displaystyle\int_{0}^{t}(t-s)^{\alpha-1}\mathscr{S}(t-s)(f(s, P_{1s})-f(s, P_{2s}))ds\parallel^2\\ &+3E\parallel\displaystyle\int_{0}^{t}(t-s)^{\alpha-1}\mathscr{S}(t-s)(g(s, P_{1s})-g(s, P_{2s}))d\omega_s\parallel^2. \end{array} \end{eqnarray} $ (3.8)

现在估计方程(3.8)右边的项.第一项由Holder不等式和假设条件(ⅰ)-(ⅲ), 有

$ \begin{eqnarray} \begin{array}{ll} &3E\parallel\displaystyle\int_{0}^{t}-\mu(s, a)(t-s)^{\alpha-1}\mathscr{S}(t-s)(P_{1s}-P_{2s})ds\parallel^2\\ \leq &3\mu_0^2(\frac{M\alpha}{\Gamma(1+\alpha)})^2(\displaystyle\int_{0}^{t}e^{-2\gamma (t-s)} ds)(\displaystyle\int_{0}^{t}E\parallel P_{1s}-P_{2s}\parallel^2 ds)\\ \leq &3\mu_0^2(\frac{M\alpha}{\Gamma(1+\alpha)})^2\displaystyle\int_{0}^{t}E\parallel P_{1s}-P_{2s}\parallel^2ds. \end{array} \end{eqnarray} $ (3.9)

类似的, 第二项为

$ \begin{eqnarray} \begin{array}{ll} &3E\parallel\displaystyle\int_{0}^{t}(t-s)^{\alpha-1}\mathscr{S}(t-s)(f(s, P_{1s})-f(s, P_{2s}))ds\parallel^2\\ \leq& 3N(\frac{M\alpha}{\Gamma(1+\alpha)})^2\displaystyle\int_{0}^{t}E\parallel P_{1s}-P_{2s}\parallel^2ds. \end{array} \end{eqnarray} $ (3.10)

接下来, 对方程(3.7) 的第三项应用Itô积分可得

$ \begin{eqnarray}\label{E:3.11} \begin{array}{ll} &3E\parallel\displaystyle\int_{0}^{t}(t-s)^{\alpha-1}\mathscr{S}(t-s)(g(s, P_{1s})-g(s, P_{2s}))d\omega_s\parallel^2\\ \leq &3\displaystyle\int_{0}^{t}E\parallel(t-s)^{\alpha-1}\mathscr{S}(t-s) (g(s, P_{1s})-g(s, P_{2s}))\parallel_2^2 ds\\ \leq &3L(\frac{M\alpha}{\Gamma(1+\alpha)})^2\displaystyle\int_{0}^{t}E\parallel P_{1s}-P_{2s}\parallel^2ds. \end{array} \end{eqnarray} $ (3.11)

因此 $\forall t \in J$由方程(3.9)-(3.11)可以得到

$ \begin{array}{ll} E\parallel \Psi P_{1t}-\Psi P_{2t}\parallel^2\leq 3(\mu_0^2+N+L)(\frac{M\alpha}{\Gamma(\alpha+1)})^2\displaystyle\int_{0}^{t}E\parallel P_{1s}-P_{2s}\parallel^2 ds. \end{array} $

$ \begin{eqnarray} \parallel \Psi P_{1t}-\Psi P_{2t}\parallel^2\leq M_0\parallel P_{1t}-P_{2t}\parallel^2. \end{eqnarray} $ (3.12)

从方程(3.12) 可以得出 $\Psi$是连续映射.由Banach压缩映射原理, 在 $H$上存在唯一不动点 $P_{t}$, 使得 $\Psi P_{t}=P_{t}$.因此可以得到

$ \begin{eqnarray*}P_t&=&\mathscr{T}(t)P_0+\displaystyle\int_0^t(t-s)^{\alpha-1}\mathscr{S}(t-s)[-\mu(s, a)P_s+f(s, P_s)]ds\\ &&+\displaystyle\int_0^t(t-s)^{\alpha-1}\mathscr{S}(t-s)g(s, P_s)d\omega_s.\end{eqnarray*} $

$P_{t}\in H$是方程(1. 1)的温和解.定理得证.

4 结论

本文用分数阶导数代替古典的随机种群系统中关于时间的整数阶导数, 得到了一类随机分数阶种群模型, 该模型较传统的随机种群系统能更好地刻画现实中的种群问题, 尤其对具有遗传性的种群现象.在引入分数阶概念的基础上, 利用不动点定理、随机分析和算子半群理论, 讨论了随机分数阶种群系统温和解的存在性、唯一性.所得到的结论为种群未来的研究提供了一定的理论依据.

参考文献
[1] He J H. Nonlinear oscillation with fractional derivative and its applications[J]. Intern. Confer. Vi brating Engin., 1998: 288–291.
[2] Cui Z J, Yang Z D. Application of homotopy perturbation method to nonlinear fractional population dynamics models[J]. Intern. J. Appl. Math. Comput., 2012, 4(4): 403–412.
[3] Xu H. Analytical approximations for a population growth model with fractional order[J]. Commun. Nonl. Sci. Numer. Simulat., 2009, 14: 1978–1983. DOI:10.1016/j.cnsns.2008.07.006
[4] Shu X B, Lai Y, Chen Y. The existence of mild solutions for impulsive fractional partial difierential equations[J]. Nonl. Anal. TMA, 2011, 74: 2003–2011. DOI:10.1016/j.na.2010.11.007
[5] Liu Kewei, Jiang Wei. Finite time stability of fractional order neutral difierential equations[J]. J. Math., 2014, 1: 43–50.
[6] Wang Y W, Guan Z H, Xiao J W. Impulsive control for synchronization of a class of continuous systems[J]. Chaos, 2004, 14(1): 199–203. DOI:10.1063/1.1644351
[7] 胡建兵, 韩焱, 赵灵冬. 分数阶系统的一种稳定性判定定理及在分数阶统一混沌系统同步中的应用[J]. 物理学报, 2009, 58(7): 4402–4406. DOI:10.7498/aps.58.4402
[8] Cole K S. Electric conductance of biological systems[J]. Proc. Cold Spring Harbor Symp. Quant. Biol., New York: Cold Spring Harbor, 1993: 107-116.
[9] Anastasio T J. The fractional-order dynamics of bainstem vestibulo-oculomotor neurons[J]. Biol. Cyber., 1994, 72: 69–79. DOI:10.1007/BF00206239
[10] Zhang Q M, Liu W A, Nie Z K. Existence, uniqueness and exponential stability for stochastic age-dependent population[J]. Appl. Math. Comput., 2004, 154: 183–201.
[11] Li R L, Leung P, Pang W. Convergence of numerical solutions to stochastic age-dependent popula tion equations with Markovian switching[J]. J. Comput. Appl. Math., 2009, 233: 1046–1055. DOI:10.1016/j.cam.2009.08.113
[12] Ma W J, Zhang Q M, Wang Z P. Asymptotic stability of stochastic age-dependent population equations with Markovian switching[J]. Appl. Math. Comput., 2014, 227: 309–319.
[13] Ma W J, Zhang Q M, Han C Z. Numerical analysis for stochastic age-dependent population equations with fractional Brownian motion[J]. Commun. Nonl. Sci. Numer. Simul., 2012, 17: 1884–1893. DOI:10.1016/j.cnsns.2011.08.025
[14] Yang H F, Zhang Q M, Feng J T. Numerical simulations based on POD for stochastic age-dependent system of two species[J]. Difier. Equ. Dyn. Syst., 2015, 23(4): 433–451. DOI:10.1007/s12591-014-0229-3
[15] Jiang H P, Jiang W. The existence of a positive solution for nonlinear fractional functional difierential equations[J]. J. Math., 2011, 3(31): 440–446.
[16] Chen L P, Chai Y, Wu R C, Ma T D, Zhai H Z. Dynamic analysis of a class of fractional-order neural networks with delay[J]. Neurocomputing, 2013, 11: 190–194.
[17] Sakthivel R, Revathi P, Anthoni S M. Existence of pseudo almost automorphic mild solutions to stochastic fractional difierential equations[J]. Nonl. Anal., 2012, 75: 3339–3347. DOI:10.1016/j.na.2011.12.028
[18] Sakthivel R, Suganya S, Anthoni S M. Approximate controllability of fractional stochastic evolution equations[J]. Comp. Math. Appl., 2012, 63: 660–668. DOI:10.1016/j.camwa.2011.11.024