具积分边界条件的逐项分数阶耦合系统解的存在性和Ulam型稳定性
Existence and Ulam Type Stability of Solutions for Sequential Fractional Coupled Systems with Integral Boundary Conditions
摘要: 本文研究了一类逐项分数阶耦合系统边值问题解的存在性和Ulam型稳定性,在积分边界条件下通过给出合适的假设条件利用压缩映射原理和Schauder不动点定理得到系统解的存在唯一性,给出了边值问题Ulam型稳定性的结论。最后给出例子说明结论的可行性。
Abstract: In this paper, we study a class of fractional order boundary value problem solution of the coupled system existence and Ulam type stability, in the integral boundary conditions by giving appropriate assumptions using compression mapping principle and Schauder fixed point theorem for the existence and uniqueness of the solution of the system, to give the conclusion of the boundary value problem of Ulam type stability. Example is given to illustrate the conclusion feasibility.
文章引用:陈张丽, 贾梅. 具积分边界条件的逐项分数阶耦合系统解的存在性和Ulam型稳定性[J]. 应用数学进展, 2022, 11(4): 1903-1915. https://doi.org/10.12677/AAM.2022.114207

1. 引言

分数阶微分方程描述了工程和科学领域的许多现象,例如物理学、化学、生物学、经济学、控制理论、信号和图像处理、空气动力学、粘弹性、电磁学和流变学等,因此分数阶微分方程具有广泛的应用,相关文献见 [1] - [9]。具有逐项分数阶导数的微分方程在振动理论中有重大的意义,见参考文献 [10] [11]。在分数阶微分方程研究的过程中,经常用到的导数为Riemann-Liouville导数和Caputo导数,由于适型导数具有整数阶导数的良好性质,因此适型导数从2014年被提出之后便开始被广泛研究,相关文献见 [12] [13]。分数阶微分方程解的存在性和唯一性已经被大量研究,相关结论见参考文献 [14] [15]。Hyers-Ulam稳定性最早出现于函数方程,1940年,Ulam在文献 [16] 中第一次提出了“Ulam Stability problem”。1941年,在文献 [17] 中这一问题被Hyers D H在Banach空间中解决了,自此以后这种稳定性就被称为Hyers-Ulam稳定性,并开始被广泛研究。相关文献见 [17] [18] [19] [20] [21]。目前,在适型导数下,有关逐项分数阶耦合系统的Ulam型稳定性的研究我们还未见到。因此本文致力于研究带积分边界条件的逐项分数阶微分方程耦合系统的Ulam型稳定性以及解的存在唯一性。

本文研究下列耦合系统:

{ D s α 1 + 1 x 1 ( t ) + λ 1 D s α 1 x 1 ( t ) = f 1 ( t , x 1 ( t ) , x 2 ( t ) ) , t ( 0 , 1 ) , D s α 2 + 1 x 2 ( t ) + λ 2 D s α 2 x 2 ( t ) = f 2 ( t , x 1 ( t ) , x 2 ( t ) ) , t ( 0 , 1 ) , x 1 ( 0 ) = 0 , x 1 ( 1 ) = 0 1 g 1 ( s , x 2 ( s ) ) d s , x 2 ( 0 ) = 0 , x 2 ( 1 ) = 0 1 g 2 ( s , x 1 ( s ) ) d s (1)

解的存在唯一性以及Ulam型稳定性,其中, D s α i + 1 D s α i 分别表示阶数为 α i + 1 α i 的适型分数阶导数, 0 < α i 1 λ i + ( i = 1 , 2 ) f 1 , f 2 : [ 0 , 1 ] × 2 是连续函数, g 1 , g 2 C ( [ 0 , 1 ] × , )

2. 前期准备

定义2.1 (见 [12] ) 假设 γ ( n , n + 1 ] u : [ 0 , ) ,对于 t > 0 ,u是n阶可微的。则把u的 γ 阶导数定义为:

D s γ u ( t ) = lim ε 0 u ( n ) ( t + ε t n + 1 γ ) u ( n ) ( t ) ε .

所给定的右侧极限存在。

如果在 ( 0 , a ) 上u是 γ 阶可微的,其中 a > 0 lim t 0 + D s γ u ( t ) 存在,然后定义 D s γ u ( 0 ) = lim t 0 + D s γ u ( t )

如果 λ = 1 ,则 D s 1 u ( t ) = u ( t ) ,见参考文献 [12]。

引理2.1 (见 [13] ) 令 γ ( 0 , 1 ] ,当 t > 0 时, f , g γ 阶可微的,则

D s γ ( f g ) = f D s γ ( g ) + g D s γ ( f ) .

引理2.2 (见 [22] ) 令 t > 0 ,函数 u ( t ) γ 阶可微的当且仅当u是 ( n + 1 ) 阶可微的,进一步说,下面的关系式成立:

D s γ ( u ) = t n + 1 γ u ( n + 1 ) ( t ) .

引理2.3 (见 [13] ) 令 f : [ a , ) ,在 ( a , ) 上是两次可微的, 0 < γ 1 γ 2 1 使得 1 < γ 1 + γ 2 2 ,则有下列关系式成立:

D s γ 1 + γ 2 f ( t ) = ( D s γ 1 D s γ 2 f ) ( t ) ( 1 γ 2 ) ( t a ) γ 2 D s γ 1 f ( t ) .

在本文中 0 < γ 1 1 γ 2 = 1 满足 0 < γ 1 γ 2 1 1 < γ 1 + γ 2 2 ,即 D s γ 1 + 1 f ( t ) = ( D s γ 1 D s 1 f ) ( t )

定义2.2 (见 [12] ) 连续函数 f : ( 0 , ) γ ( n , n + 1 ) 阶分数积分定义为:

I s γ f ( t ) = I s n + 1 ( t γ n 1 f ( t ) ) = 1 n ! 0 t ( t s ) n s γ n 1 f ( s ) d s

其中 I s n + 1 n + 1 重积分算子。

引理2.4 (见 [12] ) 设 γ ( n , n + 1 ) f : ( 0 , ) 连续,则 t > 0 D s γ I s γ f ( t ) = f ( t )

引理2.5 (见 [22] ) 设 γ ( n , n + 1 ) u C ( 0 , + ) 具有 γ 阶导数,则

I s γ D s γ u ( t ) = u ( t ) + c 0 + c 1 t + + c n t n ,

这里 c k k = 0 , 1 , , n

引理2.6 设 h 1 , h 2 C [ 0 , 1 ] a , b ,则逐项分数阶微分方程边值问题:

{ D s α 1 + 1 x 1 ( t ) + λ 1 D s α 1 x 1 ( t ) = h 1 ( t ) , t ( 0 , 1 ) , D s α 2 + 1 x 2 ( t ) + λ 2 D s α 2 x 2 ( t ) = h 2 ( t ) , t ( 0 , 1 ) , x 1 ( 0 ) = 0 , x 1 ( 1 ) = a , x 2 ( 0 ) = 0 , x 2 ( 1 ) = b (2)

存在唯一解

x 1 ( t ) = 0 1 G 1 ( t , s ) h 1 ( s ) d s + a ( 1 e λ 1 t ) 1 e λ 1 ,

x 2 ( t ) = 0 1 G 2 ( t , s ) h 2 ( s ) d s + b ( 1 e λ 2 t ) 1 e λ 2 ,

其中

G i ( t , s ) = s α i 1 λ i ( 1 e λ i ) { ( e λ i e λ i t ) ( 1 e λ i s ) , 0 < s t 1 , ( 1 e λ i t ) ( 1 e λ i ( s 1 ) ) , 0 t < s 1 , i = 1 , 2.

证明:已知 D s α 1 + 1 x 1 ( t ) + λ 1 D s α 1 x 1 ( t ) = h 1 ( t ) ,根据引理2.3得

D s α 1 + 1 x 1 ( t ) + λ 1 D s α 1 x 1 ( t ) = D s α 1 ( D s 1 + λ 1 ) x 1 ( t ) = h 1 ( t ) .

对上式两边作用 I s α 1 ,可得:

( D s 1 + λ 1 ) x 1 ( t ) = I s α 1 h 1 ( t ) c 0 .

D s 1 ( e λ 1 t x 1 ( t ) ) = e λ 1 t I s α 1 h 1 ( t ) c 0 e λ 1 t .

对上式两边从0到t积分可得:

e λ 1 t x 1 ( t ) = x 1 ( 0 ) c 0 0 t e λ 1 s d s + 0 t e λ 1 s ( 0 s τ α 1 1 h 1 ( τ ) d τ ) d s .

从而

x 1 ( t ) = e λ 1 t x 1 ( 0 ) c 0 λ 1 ( 1 e λ 1 t ) + 0 t e λ 1 ( s t ) ( 0 s τ α 1 1 h 1 ( τ ) d τ ) d s .

代入边界条件 x 1 ( 0 ) = 0 可得,

x 1 ( t ) = c 0 λ 1 ( 1 e λ 1 t ) + 0 t e λ 1 ( s t ) ( 0 s τ α 1 1 h 1 ( τ ) d τ ) d s = c 0 λ 1 ( 1 e λ 1 t ) + 1 λ 1 0 t ( 1 e λ 1 ( τ t ) ) τ α 1 1 h 1 ( τ ) d τ .

由边界条件 x 1 ( 1 ) = a 有,

x 1 ( 1 ) = c 0 λ 1 ( 1 e λ 1 ) + 1 λ 1 0 1 ( 1 e λ 1 ( τ 1 ) ) τ α 1 1 h 1 ( τ ) d τ = a .

可得

c 0 = 1 1 e λ 1 ( 0 1 ( 1 e λ 1 ( τ 1 ) ) τ α 1 1 h 1 ( τ ) d τ a λ 1 ) .

从而

x 1 ( t ) = 1 λ 1 0 t ( 1 e λ 1 ( s t ) ) s α 1 1 h 1 ( s ) d s 1 e λ 1 t λ 1 ( 1 e λ 1 ) 0 1 ( 1 e λ 1 ( s 1 ) ) s α 1 1 h 1 ( s ) d s + a ( 1 e λ 1 t ) 1 e λ 1 = 0 1 G 1 ( t , s ) h 1 ( s ) d s + a ( 1 e λ 1 t ) 1 e λ 1 .

类似可得

x 2 ( t ) = 0 1 G 2 ( t , s ) h 2 ( s ) d s + b ( 1 e λ 2 t ) 1 e λ 2 .

证毕。

由引理2.6可得下面的引理成立。

引理2.7 边值问题(1)与积分方程

x 1 ( t ) = 0 1 G 1 ( t , s ) f 1 ( s , x 1 ( s ) , x 2 ( s ) ) d s + 1 e λ 1 t 1 e λ 1 0 1 g 1 ( s , x 2 ( s ) ) d s ,

x 2 ( t ) = 0 1 G 2 ( t , s ) f 2 ( s , x 1 ( s ) , x 2 ( s ) ) d s + 1 e λ 2 t 1 e λ 2 0 1 g 2 ( s , x 1 ( s ) ) d s ,

等价。

通过简单计算易得引理2.8成立。

引理2.8 对于 t [ 0 , 1 ] s C ( 0 , 1 ] 核函数 G 1 ( t , s ) , G 2 ( t , s ) 满足

1) 0 G i ( t , s ) s α i 1 λ i , i = 1 , 2.

2) | G i ( t , s ) t | s α i 1 e λ i , i = 1 , 2.

由引理2.8可知,

0 0 1 G i ( t , s ) d s 1 λ i 0 1 s α i 1 d s = 1 λ i α i N i , i = 1 , 2.

0 1 | G i ( t , s ) t | d s = e λ i α i , i = 1 , 2.

引理2.9 (见 [23] ) (Banach压缩映射原理)设 ( X , ρ ) 是一个完备的度量空间, F X 为闭集,映射 P : F F ,如果存在 0 < K < 1 ,使得对所有的 x , y X 都有

ρ ( P x , P y ) K ρ ( x , y ) ,

则在F中存在唯一点 x * ,满足 P x * = x * ,称 x * 为映射P的不动点,并称P为压缩映射。

引理2.10 (见 [24] ) (Schauder不动点定理)令X是一个实赋范线性空间, Ω ¯ X 为非空有界闭凸子集,又 F : Ω ¯ Ω ¯ 为全连续算子,则F在 Ω ¯ 中有不动点。

3. 解的存在唯一性

对于 x C [ 0 , 1 ] ,定义范数 x = max t [ 0 , 1 ] | x ( t ) | ,则 C [ 0 , 1 ] 为Banach空间,令 X = C [ 0 , 1 ] × C [ 0 , 1 ] ,定义范数 ( x 1 , x 2 ) = x 1 + x 2 ,则X在 ( x 1 , x 2 ) 下为Banach空间。

定义算子 T : X X ,这里 T ( x 1 , x 2 ) ( t ) = ( T 1 ( x 1 , x 2 ) ( t ) T 2 ( x 1 , x 2 ) ( t ) ) ,其中

T 1 ( x 1 , x 2 ) ( t ) = 0 1 G 1 ( t , s ) f 1 ( s , x 1 ( s ) , x 2 ( s ) ) d s + 1 e λ 1 t 1 e λ 1 0 1 g 1 ( s , x 2 ( s ) ) d s ,

T 2 ( x 1 , x 2 ) ( t ) = 0 1 G 2 ( t , s ) f 2 ( s , x 1 ( s ) , x 2 ( s ) ) d s + 1 e λ 2 t 1 e λ 2 0 1 g 2 ( s , x 1 ( s ) ) d s .

引理4.4 T : X X 为全连续算子。

证明:因为 f 1 , f 2 [ 0 , 1 ] × 2 g 1 , g 2 [ 0 , 1 ] × 上均是连续函数,根据引理2.8,易得 T : X X ,且T是连续的。

Ω X 是有界的,存在非负常数 M 1 , M 2 , M 3 , M 4 ,使得对任意的 t [ 0 , 1 ] ( x 1 , x 2 ) Ω

| f 1 ( t , x 1 , x 2 ) | M 1 , | f 2 ( t , x 1 , x 2 ) | M 2 , | g 1 ( t , x 2 ) | M 3 , | g 2 ( t , x 1 ) | M 4 .

| T 1 ( x 1 , x 2 ) ( t ) | = | 0 1 G 1 ( t , s ) f 1 ( s , x 1 ( s ) , x 2 ( s ) ) d s + 1 e λ 1 t 1 e λ 1 0 1 g 1 ( s , x 2 ( s ) ) d s | 0 1 | G 1 ( t , s ) | | f 1 ( s , x 1 ( s ) , x 2 ( s ) ) | d s + 0 1 | g 1 ( s , x 2 ( s ) ) | d s N 1 M 1 + M 3 .

因此,

T 1 ( x 1 , x 2 ) N 1 M 1 + M 3 .

同理,

T 2 ( x 1 , x 2 ) N 2 M 2 + M 4 .

所以, T ( x 1 , x 2 ) N 1 M 1 + N 2 M 2 + M 3 + M 4 ,故 T ( Ω ) 是一致有界的。

因为 e λ i t ( i = 1 , 2 ) [ 0 , 1 ] 上是连续的,所以, e λ i t ( i = 1 , 2 ) [ 0 , 1 ] 上是一致连续的。因此,对于给定的 ε > 0 ,存在 0 < δ < α i ε 2 M 1 e λ i ,使得对 t 1 , t 2 [ 0 , 1 ] | t 1 t 2 | < δ

| e λ i t 1 e λ i t 2 | < ( 1 e λ i ) ε 2 M 3 .

| T 1 ( x 1 , x 2 ) ( t 2 ) T 1 ( x 1 , x 2 ) ( t 1 ) | = | t 1 t 2 ( T 1 ( x 1 , x 2 ) ) ( t ) d t | = | t 1 t 2 ( 0 1 G 1 ( t , s ) t f 1 ( s , x 1 ( s ) , x 2 ( s ) ) d s + λ 1 e λ 1 t 1 e λ 1 0 1 g 1 ( s , x 2 ( s ) ) d s ) d t | M 1 e λ i α 1 | t 1 t 2 | + M 3 e λ 1 t 1 e λ 1 t 2 1 e λ 1 < ε 2 + ε 2 = ε .

同理, | T 2 ( x 1 , x 2 ) ( t 2 ) T 2 ( x 1 , x 2 ) ( t 2 ) | < ε 。所以 T ( Ω ) 是等度连续的,由Arzela-Ascoli定理知, T ( Ω ) 是相对列紧的。

因此,T是全连续算子。

证毕。

下面给出以下假设:

(H1) 存在常数 l j > 0 ( j = 1 , 2 , 3 , 4 ) ,使得对 t [ 0 , 1 ] x i , y i ( i = 1 , 2 )

| f 1 ( t , x 1 , x 2 ) f 1 ( t , y 1 , y 2 ) | l 1 | x 1 y 1 | + l 2 | x 2 y 2 | ,

| f 2 ( t , x 1 , x 2 ) f 2 ( t , y 1 , y 2 ) | l 3 | x 1 y 1 | + l 4 | x 2 y 2 | ,

| g 1 ( t , x 2 ) g 1 ( t , y 2 ) | l 5 | x 2 y 2 | ,

| g 2 ( t , x 1 ) g 2 ( t , y 1 ) | l 6 | x 1 y 1 | .

(H2) 存在非负函数 a i , b i , c j , d j L 1 [ 0 , 1 ] i = 0 , 1 , 2 j = 0 , 1 ,使得

| f 1 ( t , x 1 , x 2 ) | a 0 ( t ) + a 1 ( t ) | x 1 | + a 2 ( t ) | x 2 | , ( t , x 1 , x 2 ) [ 0 , 1 ] × 2 ,

| f 2 ( t , x 1 , x 2 ) | b 0 ( t ) + b 1 ( t ) | x 1 | + b 2 ( t ) | x 2 | , ( t , x 1 , x 2 ) [ 0 , 1 ] × 2 ,

| g 1 ( t , x 2 ) | c 0 ( t ) + c 1 ( t ) | x 2 | , ( t , x 2 ) [ 0 , 1 ] × ,

| g 2 ( t , x 1 ) | d 0 ( t ) + d 1 ( t ) | x 1 | , ( t , x 1 ) [ 0 , 1 ] × .

为了方便,记

a i 1 = 0 1 a i ( t ) d t , b i 1 = 0 1 b i ( t ) d t , i = 0 , 1 , 2 ,

c j 1 = 0 1 c j ( t ) d t , d j 1 = 0 1 d j ( t ) d t , j = 0 , 1.

Λ 1 = a 0 1 N 1 + b 0 1 N 2 + c 0 1 + d 0 1 ,

Λ 2 = a 1 1 N 1 + b 1 1 N 2 + d 1 1 ,

Λ 3 = a 2 1 N 1 + b 2 1 N 2 + c 1 1 .

定理3.2 假设(H1)成立,如果

N = max { l 1 N 1 + l 3 N 2 + l 6 , l 2 N 1 + l 4 N 2 + l 5 } < 1 ,

则边值问题(1)存在唯一解。

证明:对任意的 ( x 1 , x 2 ) ( y 1 , y 2 ) X t [ 0 , 1 ]

| T 1 ( x 1 , x 2 ) ( t ) T 1 ( y 1 , y 2 ) ( t ) | 0 1 G 1 ( t , s ) | f 1 ( s , x 1 ( s ) , x 2 ( s ) ) f 1 ( s , y 1 ( s ) , y 2 ( s ) ) | d s + 0 1 | g 1 ( s , x 2 ( s ) ) g 1 ( s , y 2 ( s ) ) | d s N 1 ( l 1 x 1 y 1 + l 2 x 2 y 2 ) + l 5 x 2 y 2 l 1 N 1 x 1 y 1 + ( l 2 N 1 + l 5 ) x 2 y 2 .

因此,

T 1 ( x 1 , x 2 ) T 1 ( y 1 , y 2 ) l 1 N 1 x 1 y 1 + ( l 2 N 1 + l 5 ) x 2 y 2 .

类似地,

T 2 ( x 1 , x 2 ) T 2 ( y 1 , y 2 ) ( l 3 N 2 + l 6 ) x 1 y 1 + l 4 N 2 x 2 y 2 .

从而

T ( x 1 , x 2 ) T ( y 1 , y 2 ) ( l 1 N 1 + l 3 N 2 + l 6 ) x 1 y 1 + ( l 2 N 1 + l 4 N 2 + l 5 ) x 2 y 2 N ( x 1 , x 2 ) ( y 1 , y 2 ) .

由于 N < 1 ,所以,根据Banach压缩映射原理可知边值问题(4.1)存在唯一解。

证毕。

定理3.4 假设(H2)成立,如果 max { Λ 2 , Λ 3 } < 1 ,则边值问题(1)在X中至少存在一个解。

证明:因为 max { Λ 2 , Λ 3 } < 1 ,所以, 1 max { Λ 2 , Λ 3 } > 0 。令 r Λ 1 1 max { Λ 2 + Λ 3 }

B r = { ( x 1 , x 2 ) | ( x 1 , x 2 ) X , ( x 1 , x 2 ) r } ,因此 B r 在X上是有界闭凸集。

对于 ( x 1 , x 2 ) B r , t [ 0 , 1 ] ,我们有

| T 1 ( x 1 , x 2 ) ( t ) | 0 1 | G 1 ( t , s ) | | f 1 ( s , x 1 ( s ) , x 2 ( s ) ) | d s + 0 1 | g 1 ( s , x 2 ( s ) ) | d s N 1 a 0 1 + c 0 1 + N 1 a 1 1 x 1 + ( N 1 a 2 1 + c 1 1 ) x 2 .

同理

| T 2 ( x 1 , x 2 ) ( t ) | N 2 b 0 1 + d 0 1 + N 2 b 2 1 x 2 + ( N 2 b 1 1 + d 1 1 ) x 1 .

因此,

| T 1 ( x 1 , x 2 ) ( t ) | N 1 a 0 1 + c 0 1 + N 1 a 1 1 x 1 + ( N 1 a 2 1 + c 1 1 ) x 2 + N 2 b 0 1 + d 0 1 + N 2 b 2 1 x 2 + ( N 2 b 1 1 + d 1 1 ) x 1 = Λ 1 + Λ 2 x 1 + Λ 3 x 2 Λ 1 + max { Λ 2 , Λ 3 } ( x 1 , x 2 ) .

所以

T ( x 1 , x 2 ) Λ 1 1 max { Λ 2 , Λ 3 } r .

从而, T ( B r ) B r

通过引理3.1可知,T是全连续的,由Schauder不动点定理可知边值问题(1)在X中至少存在一个解,且其解是有界的。

证毕。

4. 解的Ulam型稳定性

这部分主要研究边值问题(1)的Ulam-Hyers稳定性和Ulam-Hyers-Rassias稳定性。

对任意 ε 1 , ε 2 > 0 φ 1 , φ 2 C ( [ 0 , 1 ] , + ) ,记 ε = max { ε 1 , ε 2 } > 0 φ ( t ) = max { φ 1 ( t ) , φ 2 ( t ) } ,则 φ C ( [ 0 , 1 ] , + ) ,考虑:

{ | D s α 1 + 1 z 1 ( t ) + λ 1 D s α 1 z 1 ( t ) f 1 ( t , z 1 ( t ) , z 2 ( t ) ) | ε 1 , 0 < α 1 1 , t [ 0 , 1 ] , | D s α 2 + 1 z 2 ( t ) + λ 2 D s α 2 z 2 ( t ) f 2 ( t , z 1 ( t ) , z 2 ( t ) ) | ε 2 , 0 < α 2 1 , t [ 0 , 1 ] , z 1 ( 0 ) = 0 , z 1 ( 1 ) = 0 1 g 1 ( s , z 2 ( s ) ) d s , z 2 ( 0 ) = 0 , z 2 ( 1 ) = 0 1 g 2 ( s , z 1 ( s ) ) d s , (3)

{ | D s α 1 + 1 z 1 ( t ) + λ 1 D s α 1 z 1 ( t ) f 1 ( t , z 1 ( t ) , z 2 ( t ) ) | ε 1 φ 1 ( t ) , 0 < α 1 1 , t [ 0 , 1 ] , | D s α 2 + 1 z 2 ( t ) + λ 2 D s α 2 z 2 ( t ) f 2 ( t , z 1 ( t ) , z 2 ( t ) ) | ε 2 φ 2 ( t ) , 0 < α 2 1 , t [ 0 , 1 ] , z 1 ( 0 ) = 0 , z 1 ( 1 ) = 0 1 g 1 ( s , z 2 ( s ) ) d s , z 2 ( 0 ) = 0 , z 2 ( 1 ) = 0 1 g 2 ( s , z 1 ( s ) ) d s . (4)

注4.1 函数 ( z 1 , z 2 ) X 是不等式组(3)的一个解当且仅当存在一个函数 h i C ( [ 0 , 1 ] , ) , i = 1 , 2 ,使得

1) | h i ( t ) | ε i , t [ 0 , 1 ] ;

2) D s α i + 1 z i ( t ) + λ i D s α i z i ( t ) = f i ( t , z 1 ( t ) , z 2 ( t ) ) + h i ( t ) , t [ 0 , 1 ] ;

3) z 1 ( 0 ) = 0 , z 2 ( 0 ) = 0 , z 1 ( 1 ) = 0 1 g 1 ( s , z 2 ( s ) ) d s , z 2 ( 1 ) = 0 1 g 2 ( s , z 1 ( s ) ) d s .

注4.2 函数 ( z 1 , z 2 ) X 是不等式组(4)的一个解当且仅当存在函数 h ˜ i ( t ) C ( [ 0 , 1 ] , ) , i = 1 , 2 ,使得:

1) | h ˜ i ( t ) | ε i φ i ( t ) , t [ 0 , 1 ] ;

2) D s α i + 1 z i ( t ) + λ i D s α i z i ( t ) = f i ( t , z 1 ( t ) , z 2 ( t ) ) + h ˜ i ( t ) , t [ 0 , 1 ] ;

3) z 1 ( 0 ) = 0 , z 2 ( 0 ) = 0 , z 1 ( 1 ) = 0 1 g 1 ( s , z 2 ( s ) ) d s , z 2 ( 1 ) = 0 1 g 2 ( s , z 1 ( s ) ) d s .

定义4.1 若存在常数 K 1 > 0 ,使得对任意的 ε > 0 ( z 1 , z 2 ) X 满足不等式组(3),存在 ( x 1 , x 2 ) X 为边值问题(1)的解,使得

( z 1 , z 2 ) ( x 1 , x 2 ) K 1 ε .

则称边值问题(1)中的方程具有Ulam-Hyers稳定性。

定义4.2 若存在常数 K 2 > 0 ,使得对任意的 ε > 0 φ ( t ) C ( [ 0 , 1 ] , + ) ( z 1 , z 2 ) X 满足不等式组(4),存在 ( x 1 , x 2 ) X 为边值问题(1)的解,使得:

| z i ( t ) x i ( t ) | K 2 ε φ ( t ) , t [ 0 , 1 ] , i = 1 , 2.

则称边值问题(1)关于 φ ( t ) 具有Ulam-Hyers-Rassias稳定性。

为了方便,记 Δ = ( 1 l 1 N 1 ) ( 1 l 4 N 2 ) ( l 3 N 2 + l 6 ) ( l 2 N 1 + l 5 )

定理4.1 假设(H1)成立,且 N < 1 ,则边值问题(1)是Ulam-Hyers稳定的。

证明:令 ( z 1 , z 2 ) X 是不等式组(3)的解,由于(H1)成立,且 N < 1 ,由定理3.2可得,边值问题(1)有唯一解。令 ( x 1 , x 2 ) X 为边值问题(1)的唯一解,从注4.1可知边值问题

{ D s α 1 + 1 z 1 ( t ) + λ 1 D s α 1 z 1 ( t ) = f 1 ( t , z 1 ( t ) , z 2 ( t ) ) + h 1 ( t ) , t ( 0 , 1 ) , D s α 2 + 1 z 2 ( t ) + λ 2 D s α 2 z 2 ( t ) = f 2 ( t , z 1 ( t ) , z 2 ( t ) ) + h 2 ( t ) , t ( 0 , 1 ) , z 1 ( 0 ) = 0 , z 1 ( 1 ) = 0 1 g 1 ( s , z 2 ( s ) ) d s , z 2 ( 0 ) = 0 , z 2 ( 1 ) = 0 1 g 2 ( s , z 1 ( s ) ) d s

的解可以写为:

z 1 ( t ) = 0 1 G 1 ( t , s ) ( f 1 ( s , z 1 ( s ) , z 2 ( s ) ) + h 1 ( s ) ) d s + 1 e λ 1 t 1 e λ 1 0 1 g 1 ( s , z 2 ( s ) ) d s ,

z 2 ( t ) = 0 1 G 2 ( t , s ) ( f 2 ( s , z 1 ( s ) , z 2 ( s ) ) + h 2 ( s ) ) d s + 1 e λ 2 t 1 e λ 2 0 1 g 2 ( s , z 1 ( s ) ) d s .

则对于 t [ 0 , 1 ]

| z 1 ( t ) x 1 ( t ) | | 0 1 G 1 ( t , s ) ( f 1 ( s , z 1 ( s ) , z 2 ( s ) ) + h 1 ( s ) ) d s + 1 e λ 1 t 1 e λ 1 0 1 g 1 ( s , z 2 ( s ) ) d s + 0 1 G 1 ( t , s ) f 1 ( s , z 1 ( s ) , z 2 ( s ) ) d s 1 e λ 1 t 1 e λ 1 0 1 g 1 ( s , x 2 ( s ) ) d s | 0 1 | G 1 ( t , s ) | | h 1 ( s ) | d s + 0 1 | G 1 ( t , s ) | | f 1 ( s , x 1 ( s ) , x 2 ( s ) ) f 1 ( s , z 1 ( s ) , z 2 ( s ) ) | d s + 0 1 | g 1 ( s , z 2 ( s ) ) g 1 ( s , x 2 ( s ) ) | d s ε 1 N 1 + N 1 ( l 1 x 1 z 1 + l 2 x 2 z 2 ) + l 5 x 2 z 2 .

从而

( 1 l 1 N 1 ) x 1 z 1 ( l 2 N 1 + l 5 ) x 2 z 2 ε 1 N 1 . (5)

类似地,

( 1 l 4 N 2 ) x 2 z 2 ( l 3 N 2 + l 6 ) x 1 z 1 ε 2 N 2 . (6)

由于 N < 1 ,所以 1 l 1 N 1 > l 3 N 2 + l 6 > 0 1 l 4 N 2 > l 2 N 1 + l 5 > 0 。故 Δ > 0

由(5)可得,

x 1 z 1 1 1 l 1 N 1 ( ε 1 N 1 + ( l 2 N 1 + l 5 ) x 2 z 2 ) . (7)

由(6)可得,

x 2 z 2 1 1 l 4 N 2 ( ε 2 N 2 + ( l 3 N 2 + l 6 ) x 1 z 1 ) . (8)

根据(7)、(8)可得,

x 1 z 1 1 1 l 1 N 1 ( ε 1 N 1 + ( l 2 N 1 + l 5 ) 1 1 l 4 N 2 ( ε 2 N 2 + ( l 3 N 2 + l 6 ) x 1 z 1 ) ) = N 1 1 l 1 N 1 ε 1 + N 2 ( l 2 N 1 + l 5 ) ( 1 l 1 N 1 ) ( 1 l 4 N 2 ) ε 2 + ( l 2 N 1 + l 5 ) ( l 3 N 2 + l 6 ) ( 1 l 1 N 1 ) ( 1 l 4 N 2 ) x 1 z 1 .

所以,

( 1 ( l 2 N 1 + l 5 ) ( l 3 N 2 + l 6 ) ( 1 l 1 N 1 ) ( 1 l 4 N 2 ) ) x 1 z 1 N 1 1 l 1 N 1 ε 1 + N 2 ( l 2 N 1 + l 5 ) ( 1 l 1 N 1 ) ( 1 l 4 N 2 ) ε 2 .

Δ ( 1 l 1 N 1 ) ( 1 l 4 N 2 ) x 1 z 1 N 1 1 l 1 N 1 ε 1 + N 2 ( l 2 N 1 + l 5 ) ( 1 l 1 N 1 ) ( 1 l 4 N 2 ) ε 2 .

从而,

x 1 z 1 ( 1 l 4 N 2 ) N 1 Δ ε 1 + ( l 2 N 1 + l 5 ) N 2 Δ ε 2 1 Δ ( ( 1 l 4 N 2 ) N 1 + ( l 2 N 1 + l 5 ) N 2 ) ε .

代入(8)可得,

x 2 z 2 1 Δ ( ( l 3 N 2 + l 6 ) N 1 + ( 1 l 1 N 1 ) N 2 ) ε .

故,

( x 1 , x 2 ) ( z 1 , z 2 ) = x 1 z 1 + x 2 z 2 1 Δ ( ( 1 l 4 N 2 + l 3 N 2 + l 6 ) N 1 + ( 1 l 1 N 1 + l 2 N 1 + l 5 ) N 2 ) ε = K 1 ε .

这里 K 1 = 1 Δ ( ( 1 l 4 N 2 + l 3 N 2 + l 6 ) N 1 + ( 1 l 1 N 1 + l 2 N 1 + l 5 ) N 2 ) ,因此,耦合系统(1)是Ulam-Hyers稳定的。

证毕。

与定理4.1证明类似可得定理4.2成立。

定理4.2 假设(H1)成立,如果存在 k i > 0 ,使得 0 1 | φ i ( τ ) | d τ k i φ i ( t ) ,且 N < 1 ,则边值问题(1)关于 φ ( t ) 是Ulam-Hyers-Rassias稳定的。

5. 例子

例1 考虑下列逐项分数阶耦合系统边值问题:

{ D s 3 2 x 1 ( t ) + 100 D s 1 2 x 1 ( t ) = x 1 ( t ) + 1 10 cos x 2 ( t ) + 1 , t ( 0 , 1 ) , D s 4 3 x 2 ( t ) + 200 D s 1 3 x 2 ( t ) = π 2 x 2 ( t ) + 3 sin x 1 ( t ) + 1 2 , t ( 0 , 1 ) , x 1 ( 0 ) = 0 , x 1 ( 1 ) = 0 1 ( 1 25 x 2 ( t ) + 1 ) d s , x 2 ( 0 ) = 0 , x 2 ( 1 ) = 0 1 ( 1 100 x 1 ( t ) + 10 π ) d s . (4.4)

α 1 = 1 2 α 2 = 1 3 λ 1 = 100 λ 2 = 200

f 1 ( t , x 1 , x 2 ) = x 1 + 1 10 cos x 2 + 1 , f 2 ( t , x 1 , x 2 ) = π 2 x 2 + 3 sin x 1 + 1 2 ,

g 1 ( t , x 2 ) = 1 25 x 2 + 1 , g 2 ( t , x 1 ) = 1 100 x 1 + 10 π ,

a 0 ( t ) = 1 10 cos t , a 1 ( t ) = 1 100 cos t , a 2 ( t ) = 1 1000 cos t ,

b 0 ( t ) = 1 10 sin t , b 1 ( t ) = 1 100 sin t , b 2 ( t ) = 1 1000 sin t ,

c 0 ( t ) = t , c 1 ( t ) = 1 2 t , d 0 ( t ) = 1 10 t , d 1 ( t ) = 1 100 t .

可得:

1 λ 1 = 0.01 , 1 λ 2 = 0.005 , a 0 1 = 1 10 sin 1 , a 1 1 = 1 100 sin 1 , a 2 1 = 1 1000 sin 1 ,

b 0 1 = 1 10 ( 1 cos 1 ) , b 1 1 = 1 100 ( 1 cos 1 ) , b 2 1 = 1 1000 ( 1 cos 1 ) , c 0 1 = 1 2 ,

c 1 1 = 1 4 , d 0 1 = 1 20 , d 1 1 = 1 200 .

从而 N 1 = 0.02 N 2 = 0.015

又由于,

| f 1 ( t , x 1 , x 2 ) f 1 ( t , y 1 , y 2 ) | | x 1 y 1 | + 1 10 | cos x 2 cos y 2 | x 1 y 1 + 1 10 x 2 y 2 .

| f 1 ( t , x 1 , x 2 ) f 1 ( t , y 1 , y 2 ) | π 2 | x 2 y 2 | + 3 | sin x 1 sin y 1 | π 2 x 2 y 2 + 3 x 1 y 1 .

| g 1 ( t , x 2 ) g 1 ( t , y 2 ) | 1 25 x 2 y 2 .

| g 2 ( t , x 1 ) g 2 ( t , y 1 ) | 1 100 x 1 y 1 .

所以, l 1 = 1 l 2 = 1 10 l 3 = 3 l 4 = π 2 l 5 = 1 25 l 6 = 1 100

由于,

N = max { l 1 N 1 + l 3 N 2 + l 6 , l 2 N 1 + l 4 N 2 + l 5 } = max { 0.0750 , 0.0656 } = 0.0750 < 1.

根据定理3.2可知,边值问题(10)存在唯一解. 根据定理4.1可知,边值问题(10)是Ulam-Hyers稳定的。

通过计算可得:

Λ 2 = a 1 1 N 1 + b 1 1 N 2 + d 1 1 0.0050.

Λ 3 = a 2 1 N 1 + b 2 1 N 2 + c 1 1 0.2500.

因此,

max { Λ 2 , Λ 3 } = Λ 3 0.2500 < 1.

根据定理3.3可知边值问题(10)至少存在一个解。

k 1 = 2 k 2 = 1 φ 1 ( t ) = sin x + π φ 2 ( t ) = cos x + 5 ,从而有

0 1 | φ 1 ( t ) | d t = 0 1 ( sin x ( t ) + π ) d t = π + 1 cos 1.

2 φ 1 ( t ) = 2 π + 2 sin x > π + 1 cos 1 ,故,

0 1 | φ 1 ( t ) | d t 2 φ 1 ( t ) .

同理,

0 1 | φ 2 ( t ) | d t φ 2 ( t ) .

所以,由定理4.2可知边值问题(10)是Ulam-Hyers-Rassias稳定的。

参考文献

[1] 白占兵. 分数阶微分方程边值问题理论及应用[M]. 北京: 中国科学技术出版社, 2013: 1-62.
[2] Kilbas, A., Srivastava, M. and Trujillo, J. (2006) Theory and Applications of Fractional Differential Equations. Elsevier, Amsterdam, 1-20.
[3] 李燕, 刘锡平, 李晓晨, 等. 具逐项分数阶导数的积分边值问题正解的存在性[J]. 上海理工大学学报, 2016, 38(6): 511-516.
[4] 刘帅, 贾梅, 秦小娜. 带积分边值条件的分数阶微分方程解的存在性和唯一性[J]. 上海理工大学学报, 2014, 36(5): 409-415.
[5] 张潇涵, 刘锡平, 贾梅, 等. 混合分数阶p-Laplace算子方程积分边值问题的多解性[J]. 上海理工大学学报, 2018, 40(3): 205-210.
[6] Jia, M., Li, L. and Liu, X.P. (2019) A Class of Nonlocal Problems of Fractional Differential Equations with Composition of Derivative and Parameters. Advances in Difference Equations, 2019, Article No. 280.
https://doi.org/10.1186/s13662-019-2181-6
[7] Ahmad, B. and Luca, R. (2017) Existence of Solutions for a Sequential Fractional Intefro-Differential System with Coupled Integral Boundary Conditions. Chaos, Solitons and Fractals, 104, 378-388.
https://doi.org/10.1016/j.chaos.2017.08.035
[8] Mahmudov, N. and Al-Khateeb, A. (2019) Existence and Ulam-Hyers Stability of Coupled Sequential Fractional Differential Equations with Integral Boundary Conditions. Journal of Inequalities and Applications, 2019, Article No. 165.
https://doi.org/10.1186/s13660-019-2115-6
[9] Wongcharoen, A., Ntouyas, S. and Tariboon, J. (2020) On Coupled Systems for Hilfer Fractional Differential Equations with Nonlocal Integral Boundary Conditions. Journal of Mathematics, 2020, Article ID: 2875152.
https://doi.org/10.1155/2020/2875152
[10] Ahmad, B. and Nieto, J. (2012) Sequential Fractional Differential Equations with Three-Point Boundary Conditions. Computers and Mathematics with Applications, 64, 3046-3052.
https://doi.org/10.1016/j.camwa.2012.02.036
[11] Ahmad, B. and Luca, R. (2018) Existence of Solutions for Sequential Fractional Integro-Differential Equations and Inclusions with Nonlocal Boundary Conditions. Applied Mathematics and Computation, 339, 516-534.
https://doi.org/10.1016/j.amc.2018.07.025
[12] Khalil, R., Al Horani, M., Yousef, A., et al. (2014) A New Definition of Fractional Derivative. Journal of Computational and Applied Mathematics, 264, 65-70.
https://doi.org/10.1016/j.cam.2014.01.002
[13] Abdeljawad, T. (2015) On Conformable Fractional Calculus. Journal of Computational and Applied Mathematics, 279, 57-66.
https://doi.org/10.1016/j.cam.2014.10.016
[14] Wang, Y. and Liu, L.S. (2017) Uniqueness and Existence of Positive Solutions for the Fractional Integro-Differential Equation. Boundary Value Problems, 2017, Article No. 12.
https://doi.org/10.1186/s13661-016-0741-1
[15] Feng, M., Zhang, X. and Ge, W.G. (2011) New Existence Results for Higher-Order Nonlinear Fractional Differential Equation with Integral Boundary Conditions. Boundary Value Problems, 2011, Article ID: 720702.
https://doi.org/10.1186/1687-2770-2011-720702
[16] Ulam, S. (1964) Problems in Modern Mathematics. John Wiley & Sons, New York.
[17] Hyers, D. (1941) On the Stability of the Linear Functional Equation. Proceedings of the National Academy of Sciences of the United States of America, 27, 222-224.
https://doi.org/10.1073/pnas.27.4.222
[18] Wang, J.R., Lv, L. and Zhou, Y. (2011) Ulam Stability and Data Dependence for Fractional Differential Equations with Caputo Derivative. Electronic Journal of Qualitative Theory of Differential Equations, 64, 3389-3405.
https://doi.org/10.14232/ejqtde.2011.1.63
[19] Li, M.M., Wang, J.R. and O’Regan, D. (2019) Existence and Ulam’s Stability for Conformable Fractional Differential Equations with Constant Coefficients. Bulletin of the Malaysian Sciences Society, 42, 1791-1812.
https://doi.org/10.1007/s40840-017-0576-7
[20] Öğrekçi, S. and Başcı, Y. (2021) Ulam Type Stability for Conformable Fractional Differential Equations. Rendiconti del Circolo Matematico di Palermo, 70, 807-817.
https://doi.org/10.1007/s12215-020-00532-3
[21] Almalahi, M., Abdo, M. and Panchal, S. (2021) Existence and Ulam-Hyers Stability Results of a Coupled System of ψ-Hilfer Sequential Fractional Differential Equations. Results in Applied Mathematics, 2021, Article ID: 100142.
https://doi.org/10.1016/j.rinam.2021.100142
[22] Dong, X.Y., Bai, Z.B. and Zhang, S. (2017) Positive Solutions to Boundary Value Problems of p-Laplacian with Fractional Derivative. Boundary Value Problems, 2017, Article No. 5.
https://doi.org/10.1186/s13661-016-0735-z
[23] 郭大钧, 孙经先, 刘兆理. 非线性常微分方程泛函方法[M]. 济南: 山东科学技术出版社, 1995: 49-63.
[24] Krasnoselskii, M. (1964) Positive Solutions of Operator Equations. P. Noordhoff, Groningen, 1-45.