
We propose a new definition of continuous approximate solution to initial value problem for differential equations involving variable order Caputo fractional derivative based on the classical definition of solution of integer order (or constant fractional order) differential equation. Some examples are presented to illustrate these theoretical results.
Citation: Shuqin Zhang, Jie Wang, Lei Hu. On definition of solution of initial value problem for fractional differential equation of variable order[J]. AIMS Mathematics, 2021, 6(7): 6845-6867. doi: 10.3934/math.2021401
[1] | Zoubida Bouazza, Sabit Souhila, Sina Etemad, Mohammed Said Souid, Ali Akgül, Shahram Rezapour, Manuel De la Sen . On the Caputo-Hadamard fractional IVP with variable order using the upper-lower solutions technique. AIMS Mathematics, 2023, 8(3): 5484-5501. doi: 10.3934/math.2023276 |
[2] | Zahra Pirouzeh, Mohammad Hadi Noori Skandari, Kamele Nassiri Pirbazari, Stanford Shateyi . A pseudo-spectral approach for optimal control problems of variable-order fractional integro-differential equations. AIMS Mathematics, 2024, 9(9): 23692-23710. doi: 10.3934/math.20241151 |
[3] | Kangqun Zhang . Existence and uniqueness of positive solution of a nonlinear differential equation with higher order Erdélyi-Kober operators. AIMS Mathematics, 2024, 9(1): 1358-1372. doi: 10.3934/math.2024067 |
[4] | H. H. G. Hashem, Hessah O. Alrashidi . Qualitative analysis of nonlinear implicit neutral differential equation of fractional order. AIMS Mathematics, 2021, 6(4): 3703-3719. doi: 10.3934/math.2021220 |
[5] | Najat Almutairi, Sayed Saber . Chaos control and numerical solution of time-varying fractional Newton-Leipnik system using fractional Atangana-Baleanu derivatives. AIMS Mathematics, 2023, 8(11): 25863-25887. doi: 10.3934/math.20231319 |
[6] | Umair Ali, Sanaullah Mastoi, Wan Ainun Mior Othman, Mostafa M. A Khater, Muhammad Sohail . Computation of traveling wave solution for nonlinear variable-order fractional model of modified equal width equation. AIMS Mathematics, 2021, 6(9): 10055-10069. doi: 10.3934/math.2021584 |
[7] | Özge Arıbaş, İsmet Gölgeleyen, Mustafa Yıldız . On the solvability of an initial-boundary value problem for a non-linear fractional diffusion equation. AIMS Mathematics, 2023, 8(3): 5432-5444. doi: 10.3934/math.2023273 |
[8] | Naimi Abdellouahab, Keltum Bouhali, Loay Alkhalifa, Khaled Zennir . Existence and stability analysis of a problem of the Caputo fractional derivative with mixed conditions. AIMS Mathematics, 2025, 10(3): 6805-6826. doi: 10.3934/math.2025312 |
[9] | Zhi-Yuan Li, Mei-Chun Wang, Yu-Lan Wang . Solving a class of variable order nonlinear fractional integral differential equations by using reproducing kernel function. AIMS Mathematics, 2022, 7(7): 12935-12951. doi: 10.3934/math.2022716 |
[10] | Pinghua Yang, Caixia Yang . The new general solution for a class of fractional-order impulsive differential equations involving the Riemann-Liouville type Hadamard fractional derivative. AIMS Mathematics, 2023, 8(5): 11837-11850. doi: 10.3934/math.2023599 |
We propose a new definition of continuous approximate solution to initial value problem for differential equations involving variable order Caputo fractional derivative based on the classical definition of solution of integer order (or constant fractional order) differential equation. Some examples are presented to illustrate these theoretical results.
In this paper, we research a continuous approximate solution of the following variable order initial value problem
{CDp(t,x(t))0+x(t)=f(t,x(t)),0<t≤T,x(0)=u0, | (1.1) |
where 0<p(t,x(t))<1, u0∈R, p(t,x(t)) and f(t,x(t)) are given real-valued functions, CDp(t,x(t))0+ denotes variable order Caputo fractional derivative defined by
CDp(t,x(t))0+x(t)=I1−p(t,x(t))0+x′(t), | (1.2) |
and I1−p(t,x(t))0+ is variable order Riemann-Liouville fractional integral defined by
I1−p(t,x(t))0+x(t)=∫t0(t−s)−p(t,x(t))Γ(1−p(t,x(t))x(s)ds,t>0. | (1.3) |
For details, please refer to [1,2].
Fractional calculus has been acknowledged as an extremely powerful tool in describing the natural behavior and complex phenomena of practical problems due to its applications in [3,4,5,6,7]. However, the constant fractional order calculus is not the ultimate tool to model the phenomena in nature. Therefore, variable order fractional calculus is proposed. Moreover, variable order fractional differential equations provide better descriptions for nonlocal phenomena with varying dynamics than constant order differential equations and are extensively researched in [1,2,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30,31]. Among these, there are many works dealing with numerical methods for some class of variable fractional order differential equations, for instance, [1,2,8,9,10,12,13,14,15,16,17,18,19,21,22,23,24,25,26,30,31]. In particular, variable order fractional boundary value problems are considered by numerical method base on reproducing kernel theory in [30,31].
There are several definitions of variable order fractional integrals and derivatives in [1,2]. We notice that when the order p(t) is a constant function p, variable order Riemann-Liouville fractional derivative and integral are exactly constant order fractional derivative and integral. It is well known that the Riemann-Liouville fractional integral has the law of exponents, i.e. Iα0+Iβ0+(⋅)=Iβ0+Iα0+(⋅)=Iα+β0+(⋅),α>0,β>0. Based on the law of exponents, we can obtain some properties which are associated with fractional derivative and integral. For this reason, fractional order differential equations are transformed into equivalent integral equations. Thus some results of nonlinear functional analysis (for instance, some fixed point theorems) have been applied to considering the existence of solution of fractional order differential equations (see, e.g. [3,20,32,33,34] and the references therein). However, the law of exponents doesn't hold for variable order fractional integral. For example, in [21,22,23,24],
Ig(t)0+Ih(t)0+(⋅)≠Ih(t)0+Ig(t)0+(⋅), |
Ig(t)0+Ih(t)0+(⋅)≠Ih(t)+g(t)0+(⋅), |
where h(t) and g(t) are both general nonnegative functions.
Then we will consider the properties which are interrelated with variable order fractional integral and variable order fractional derivative by given some examples.
Example 1.1. Let p(t)=t4+14, q(t)=34−t4, f(t)=t,0≤t≤2. Now, we calculate Ip(t)0+Iq(t)0+f(t)|t=1, Iq(t)0+Ip(t)0+f(t)|t=1 and Ip(t)+q(t)0+f(t)|t=1, where Ip(t)0+ and Iq(t)0+ is defined in (1.3).
For 1≤t≤2, we have
Ip(t)0+Iq(t)0+f(t)=∫t0(t−s)t4+14−1Γ(t4+14)∫s0(s−τ)34−s4−1τΓ(34−s4)dτds=∫t0(t−s)t4−34s74−s4Γ(t4+14)Γ(114−s4)ds, |
so
Ip(t)0+Iq(t)0+f(t)|t=1=∫10(1−s)−12s74−s4Γ(12)Γ(114−s4)ds≈0.4757, |
yet
Iq(t)0+Ip(t)0+f(t)=∫t0(t−s)34−t4−1Γ(34−t4)∫s0(s−τ)s4+14−1τΓ(s4+14)dτds=∫t0(t−s)−14−t4ss4+54Γ(34−t4)Γ(94+s4)ds, |
Iq(t)0+Ip(t)0+f(t)|t=1=∫10(1−s)−12ss4+54Γ(12)Γ(94+s4)ds≈0.5283, |
and
Ip(t)+q(t)0+f(t)|t=1=I10+f(t)|t=1=∫10sds=0.5. |
Therefore,
Ip(t)0+Iq(t)0+f(t)|t=1≠Ip(t)+q(t)0+f(t)|t=1, |
Iq(t)0+Ip(t)0+f(t)|t=1≠Ip(t)+q(t)0+f(t)|t=1, |
Ip(t)0+Iq(t)0+f(t)|t=1≠Iq(t)0+Ip(t)0+f(t)|t=1. |
Example 1.1 illustrates that the law of exponents of the variable order Riemann-Liouville fractional integral doesn't hold when the order is non-constant continuous function.
Example 1.2. Let p(t)={12 0≤t≤1,13, 1<t≤6,, q(t)={12, 0≤t≤1,23, 1<t≤6, and f(t)=t,0≤t≤6. We'll consider Ip(t)0+Iq(t)0+f(t)|t=4, Iq(t)0+Ip(t)0+f(t)|t=4 and Ip(t)+q(t)0+f(t)|t=4, where Ip(t)0+ and Iq(t)0+ are defined in (1.3).
For 1≤t≤4, we have
Ip(t)0+Iq(t)0+f(t)=∫10(t−s)p(t)−1Γ(p(t))∫s0(s−τ)12−1τΓ(12)dτds+∫t1(t−s)p(t)−1Γ(p(t))(∫10(s−τ)12−1τΓ(12)dτ+∫s1(s−τ)23−1τΓ(23)dτ)ds=∫10(t−s)p(t)−1s32Γ(p(t))Γ(52)ds+∫t1(t−s)p(t)−1Γ(p(t))4s323−23(s−1)12(2s+1)π12ds+∫t1(t−s)p(t)−1Γ(p(t))3(s−1)23(3s+2))10Γ(23)ds, |
thus,
Ip(t)0+Iq(t)0+f(t)|t=4=∫10(4−s)−23s32Γ(13)Γ(52)ds+∫41(4−s)−23Γ(13)4s323−23(s−1)12(2s+1)π12ds+∫41(4−s)−23Γ(13)3(s−1)23(3s+2))10Γ(23)ds≈7.8626. |
By the same way, we get
Iq(t)0+Ip(t)0+f(t)=∫10(t−s)q(t)−1Γ(q(t))∫s0(s−τ)12−1τΓ(12)dτds+∫t1(t−s)q(t)−1Γ(q(t))(∫10(s−τ)12−1τΓ(12)dτ+∫s1(s−τ)13−1τΓ(13)dτ)ds=∫10(t−s)q(t)−1s32Γ(q(t))Γ(52)ds+∫t1(t−s)q(t)−1Γ(q(t))4s323−23(s−1)12(2s+1)π12ds+∫t1(t−s)q(t)−1Γ(q(t))3(s−1)13(3s+1)4Γ(13)ds, |
Iq(t)0+Ip(t)0+f(t)|t=4=∫10(4−s)−13s32Γ(23)Γ(52)ds+∫41(4−s)−13Γ(23)4s323−23(s−1)12(2s+1)π12ds+∫41(4−s)−13Γ(23)3(s−1)13(3s+1)4Γ(13)ds≈8.1585, |
and
Ip(t)+q(t)0+f(t)|t=4=∫40(4−s)p(4)+q(4)−1sΓ(p(4)+q(4))ds=∫40sds=8. |
As a result, we deduce
Ip(t)0+Iq(t)0+f(t)|t=4≠Ip(t)+q(t)0+f(t)|t=4, |
Iq(t)0+Ip(t)0+f(t)|t=4≠Ip(t)+q(t)0+f(t)|t=4, |
Ip(t)0+Iq(t)0+f(t)|t=4≠Iq(t)0+Ip(t)0+f(t)|t=4. |
Example 1.2 shows that the law of exponents of the variable order Riemann-Liouville fractional integral doesn't hold when the order is piecewise constant function defined in the same partition.
Example 1.3. Let p(t)=t4+14, f(t)=t,0≤t≤3. Now, we consider Ip(t)0+CDp(t)0+f(t)|t=2 and CDp(t)0+Ip(t)0+f(t)|t=2.
By (1.2) and (1.3), we have
Ip(t)0+CDp(t)0+f(t)=∫t0(t−s)t4+14−1Γ(t4+14)∫s0(s−τ)−s4−14Γ(34−s4)dτds=∫t0(t−s)t4−34s34−s4Γ(t4+14)Γ(74−s4)ds, |
Ip(t)0+CDp(t)0+f(t)|t=2=∫20(2−s)−14s34−s4Γ(34)Γ(74−s4)ds≈1.91596≠f(t)|t=2−f(0)=2, |
which implies that Ip(t)0+CDp(t)0+ is different with the result of constant order fractional derivative and integral, that is,
Iα0+CDα0+g(t)=g(t)−g(0),0<t≤b, | (1.4) |
where 0<α<1, g∈C1[0,b],0<b<+∞.
On the other hand, we have
CDp(t)0+Ip(t)0+f(t)|t=2=I1−p(t)0+ddtIp(t)0+f(t)|t=2=∫t0(t−s)−t4−14Γ(34−t4)dds(s54+s4Γ(94+s4))ds|t=2=∫t0(t−s)−t4−14Γ(34−t4)[ss4+54(s4+54s+log(s)4)Γ(s4+94)−ss4+54Γ′(94+s4)4Γ2(94+s4)]ds|t=2≈1.91365≠f(t)|t=2=2, |
which illustrates that CDp(t)0+Ip(t)0+ is different with the result of constant order fractional derivative and integral, that is,
CDα0+Iα0+h(t)=h(t),0<t≤b, | (1.5) |
where 0<α<1, h∈C1[0,b],0<b<+∞.
Example 1.3 verifies that the properties (1.4) and (1.5) of constant order fractional calculus don't hold for variable order fractional calculus when the order is a continuous function.
Example 1.4. Let p(t)={12, 0≤t≤1,13, 1<t≤6,, f(t)=t,0≤t≤6. Now, we consider Ip(t)0+CDp(t)0+f(t)|t=4 and CDp(t)0+Ip(t)0+f(t)|t=4.
By (1.2) and (1.3), for 2≤t≤6, we have
Ip(t)0+CDp(t)0+f(t)=∫t0(t−s)p(t)−1Γ(p(t))∫s0(s−τ)−p(s)Γ(1−p(s))dτds=∫10(t−s)p(t)−1Γ(p(t))∫s0(s−τ)−12Γ(12)dτds+∫t1(t−s)p(t)−1Γ(p(t))[∫10(s−τ)−12Γ(12)dτ+∫s1(s−τ)−13Γ(23)dτ]ds=∫t0(t−s)p(t)−1s12Γ(p(t))Γ(32)ds−∫t1(t−s)p(t)−1(s−1)12Γ(p(t))Γ(32)ds+∫t1(t−s)p(t)−1(s−1)23Γ(p(t))Γ(53)ds, |
so
Ip(t)0+CDp(t)0+f(t)|t=4=∫40(4−s)−23s12Γ(13)Γ(32)ds−∫41(4−s)−23(s−1)12Γ(13)Γ(32)ds+∫41(4−s)−23(s−1)23Γ(13)Γ(53)ds≈3.7194≠f(t)|t=4−f(0)=4. |
On the other hand, for 2≤t≤6, we get
CDp(t)0+Ip(t)0+f(t)=I1−p(t)0+ddtIp(t)0+f(t)=∫t0(t−s)−p(t)Γ(1−p(t))dds∫s0(s−τ)p(s)−1τΓ(p(s))dτds=∫10(t−s)−p(t)Γ(1−p(t))dds∫s0(s−τ)p(s)−1τΓ(p(s))dτds+∫t1(t−s)−p(t)Γ(1−p(t))dds∫s0(s−τ)p(s)−1τΓ(p(s))dτds=∫10(t−s)−p(t)s12Γ(1−p(t))Γ(32)ds+∫t1(t−s)−p(t)Γ(1−p(t))dds∫10(s−τ)−12τΓ(12)dτds+∫t1(t−s)−p(t)Γ(1−p(t))dds∫s1(s−τ)−23τΓ(13)dτds=∫10(t−s)−p(t)s12Γ(1−p(t))Γ(32)ds+∫t1(t−s)−p(t)Γ(1−p(t))1−2s+2(s−1)12s12π12(s−1)12ds+∫t1(t−s)−p(t)Γ(1−p(t))(3s−2)(s−1)−23Γ(13)ds, |
so
CDp(t)0+Ip(t)0+f(t)|t=4=∫10(4−s)−13s12Γ(23)Γ(32)ds+∫41(4−s)−13Γ(23)1−2s+2(s−1)12s12π12(s−1)12ds+∫41(4−s)−13Γ(23)(3s−2)(s−1)−23Γ(13)ds≈4.0331≠f(t)|t=4=4. |
Example 1.4 verifies the properties (1.4) and (1.5) of constant order fractional calculus is impossible for Ip(t)0+CDp(t)0+f(t) and CDp(t)0+Ip(t)0+f(t) when the order is a piecewise function.
Hence, we can claim that variable order fractional integral defined by (1.3) has no law of exponents. Moreover, for general functions p(t) and f(t), the representations of CDp(t)0+Ip(t)0+f(t) and Ip(t)0+CDp(t)0+f(t) are not clear. These obstacles make it difficult for us to transform variable order fractional differential equation into equivalent integral equation. As a result, it is almost impossible that some nonlinear functional analysis classical methods such as fixed point theorems are applied to prove the existence of solution of the corresponding integral equation. To the best of our knowledge, there are few works ([8,23,24,27]) to deal with the existence of solutions to variable order fractional differential equations.
In [18], authors considered the following numerical solutions for variable order fractional functional boundary value problem
{Dα(x)u(x)+a(x)u′(x)+b(x)u(x)+c(x)u(τ(x))=f(x),0≤x≤1,u(0)=λ0,u(1)=λ1, | (1.6) |
where Dα(x) is the variable order Caputo fractional derivative defined by
Dα(x)u(x)=1Γ(2−α(x))∫x0(x−s)1−α(x)u″(s)ds,1≤α(x)<2; |
a(x),b(x),c(x)∈C2[0,1]; α(x),τ(x),f(x)∈C[0,1]; and λ0,λ1∈R. When α(x)=5+sin(x)4, a(x)=cos(x), b(x)=4, c(x)=5, τ(x)=x2, f(x)=2x2−α(x)Γ(3−α(x))+5x4+4x2+2xcos(x) and λ0=0,λ1=1, the exact solution of problem (1.6) was given by
u(x)=x2. |
In (1.6), if we take f(x)=6x6+5x4+4x2+2xcos(x), it is almost impossible to obtain its exact solution. In fact, we don't even know whether the solution to problem (1.6) exists.
In [8], authors discussed the existence of solution for a generalized fractional differential equation with non-autonomous variable order operators
{cDq(t,x(t))tx(t)=f(t,x(t)),x(c)=x0, | (1.7) |
where x∈Rn is the state vector, f:R×Rn→Rn is a vector field, c∈R is the lower terminal, x0 is the initial value, 0<q(t,x(t))⩽1, and cDq(t,x(t))t is the variable order Riemann-Liouville fractional differential operator defined as follows
cDq(t,x(t))tx(t)=ddt∫tc(t−s)−q(s,x(s))Γ(1−q(s,x(s))x(s)ds. |
In [8], authors claimed that, by [35], the initial value problem (1.7) is equivalent to the integral equation
{x(t)=∫tc(t−s)q(s,x(s))−1Γ(q(s,x(s))f(s,x(s))ds+Ψ(f,−q,a,c,t)t>c,x(c)=x0, | (1.8) |
where Ψ:Rn×R×R×R×R→Rn is the intialization function [35] which is needed in the general form given in (1.8) and the scalar a is used for characterizing the initial period such that for a<t<c, the initial information is given, and for t<a we consider f(t)=0. Given a, one can develop a closed analytical formula for Ψ [35].
However, in our opinion, there is no theoretical basis for this assertion. Because there are not contents related to variable order fractional integral and derivative in [35].
In [27], authors considered the existence results of solution to the initial value problem (1.1), in which f(t,x(t)):[0,T]×R→R and p:[0,T]×R→(0,1) are both continuous functions, and 0<p1≤p(t,x(t))≤p2<1, u0∈R. By means of Arzela-Ascoli theorem, authors obtained that the following sequence
{xn(t)=xn−1(t)+∫t−Tn0(t−s)−p(t,x(t))Γ(1−p(t,x(t)))xn−1(s)ds−f[t,∫t−Tn0xn−1(s)ds+u0],t∈(Tn,T],xn(t)=0,t∈[0,Tn] | (1.9) |
existed a subsequence still denoted by the sequence {xn} which uniformly converged to a continuous function x∗. Set x(t)=∫t0x∗(s)ds+u0, then authors obtained the problem (1.1) existed one solution x(t). But, we find that it has fatal errors in these analysis procedure, that is the result of uniformly bounded of sequence {xn}. This is easy to be overlooked. In our opinion the sequence {xn} is non-uniformly bounded. As a result, the existence result of solution to the initial value problem (1.1) is not obtained by Arzela-Ascoli theorem. On the other hand, it is almost impossible to transform the initial value problem (1.1) into equivalent integral equation.
Base on these facts, how to deal with the existence of solution of variable order fractional differential equations is a principal problem to be solved. In this paper, according to the classical definition of solution of integer order(or constant fractional order) differential equation, we propose a new definition of continuous approximate solution to the problem (1.1) under two kinds of variable order p(t,x(t)) and p(t).
The paper is organized as follows. In section 2, a new definition of approximate solution to the problem (1.1) for variable order p(t,x(t)) is proposed and we provide an example to demonstrate the definition. Section 3 is devoted to introduce another definition of approximate solution for p(t), then an example is given to illustrate the theoretical result.
Throughout this section, we assume that
(A1): p:[0,T]×R→(0,1) is a continuous function;
(A2): f:[0,T]×R→R is a continuous differential function.
We begin with definitions and characters of solution of integer order and constant fractional order.
Remark 2.1. According to the definition of solution x(t) of differential equation, it should be defined in the interval, in which the equation is satisfied. For instance, a function x(t) is called a solution of the following initial value problem
{x′(t)=t2sinx+t3x3,0<t≤T,x(0)=0 |
if x is defined in the interval [0,T], satisfying the equation and the initial value condition x(0)=0.
Remark 2.2. Fractional operators are typical nonlocal operators, which can well describe the memory and global correlation of physical processes. Hence, for initial value problem of fractional order differential equations, its solution in given interval is affected by the state of solution in the preceding intervals.
By Remark 2.1, a function x:[0,T]→R is called a solution of the initial value problem (1.1) if x(t) satisfies the equation (1.1) and x(0)=u0. However, based on analysis above, we are faced with extreme difficulties in considering the existence of solution of initial value problem (1.1) in sense of this definition. Thus, a new continuous approximate solution of problem (1.1) is proposed.
We start off by analyzing the problem (1.1) based on the facts above.
Let
p1=p(0,u0). | (2.1) |
We consider the initial value problem defined in the interval [0,T] as following
{CDp10+x(t)=f(t,x),0<t≤T,x(0)=u0. | (2.2) |
Let x1∈C1(0,T]∩C[0,T] be a solution of the initial value problem (2.2)(by standard way, we know that the initial value problem (2.2) exists continuous solution in [0,T] under some assumptions on f). Since x1(t) is right continuous at point 0, then for arbitrary small ε>0, there exists δ01>0 such that
|x1(t)−x1(0)|<ε, for 0<t≤δ01. | (2.3) |
And because p(t,x1(t)) is right continuous at point (0,x1(0))=(0,u0), then together with (2.3) and (2.1), for the above ε, there exists δ0>0 such that
|p(t,x1(t))−p(0,x1(0))|=|p(t,x1(t))−p(0,u0)|=|p(t,x1(t))−p1|<ε, for 0<t≤δ0. | (2.4) |
If δ0<T, we take δ0≐T1 and continue next procedure. Otherwise, we take T1=T and end this procedure.
We assume that δ0<T, and then let
p2=p(T1,x1(T1)). | (2.5) |
In order to consider the existence of solution to (1.1) in the interval [T1,T2], we let
x′(t)=∫t0x1(s)ds,t∈(0,T1]. | (2.6) |
By (2.6) and integration by parts, we denote
∫T10(t−s)−p2x′(s)Γ(1−p2)ds=∫T10(t−s)1−p2x1(s)Γ(2−p2)ds−(t−T1)1−p2Γ(2−p2)∫T10x1(s)ds≐φx1(t). | (2.7) |
Hence, we may consider the initial value problem defined in the interval [T1,T] as following
{CDp2T1+x(t)=f(t,x)−φx1(t),T1<t≤T,x(T1)=x1(T1), | (2.8) |
where x1(t) is the solution of the initial value problem (2.2) and φx1 is the function defined by (2.7).
Let x2∈C1(T1,T]∩C[T1,T] be a solution of the problem (2.8) (by standard way, we know that the problem (2.8) exists continuous solution in [T1,T] under some assumptions on f). Since x2(t) is right continuous at point T1, then for the above ε, there exists δ11>0 such that
|x2(t)−x2(T1)|<ε, for T1<t≤T1+δ11. | (2.9) |
And because p(t,x2(t)) is right continuous at point (T1,x2(T1))=(T1,x1(T1)), then together with (2.9), for the above ε, there exists δ1>0 such that
|p(t,x2(t))−p(T1,x2(T1))|=|p(t,x2(t))−p(T1,x1(T1))|<ε, for T1<t≤T1+δ1. | (2.10) |
If T1+δ1<T, we take T1+δ1≐T2 and continue next procedure. Otherwise, we take T2=T and end this procedure. Obviously, according to (2.10) and (2.5), we obtain
|p(t,x2(t))−p2|<ε, for T1<t≤T2. | (2.11) |
We assume that T1+δ1<T, and then let
p3=p(T2,x2(T2)). | (2.12) |
We let
x′(t)={∫t0x1(s)ds,t∈(0,T1],∫tT1x2(s)ds,t∈(T1,T2]. | (2.13) |
Thus, by (2.13) and integration by parts, we have
∫TiTi−1(t−s)−p3x′(s)Γ(1−p3)ds=∫TiTi−1(t−s)1−p3xi(s)Γ(2−p3)ds−(t−Ti)1−p3Γ(2−p3)∫TiTi−1xi(s)ds≐ϕxi(t), | (2.14) |
i=1,2,T0=0.
We may consider the initial value problem in the interval [T2,T] as following
{CDp3T2+x(t)=f(t,x)−ϕx1(t)−ϕx2(t),T2<t≤T,x(T2)=x2(T2), | (2.15) |
where xi(t) is the solution of the initial value problems (2.2) and (2.8) respectively, and ϕxi(i=1,2) is the function defined by (2.14).
Let x3∈C1(T2,T]∩C[T2,T] be a solution of the problem (2.15)(by standard way, we know that the problem (2.15) exists continuous solution in [T2,T] under some assumptions on f). Since x3(t) is right continuous at point T2, then for the above ε, there exists δ21>0 such that
|x3(t)−x3(T2)|<ε, for T2<t≤T2+δ21. | (2.16) |
And because p(t,x3(t)) is right continuous at point (T2,x3(T2))=(T2,x2(T2)), then together with (2.16), for the above ε, there exists δ2>0 such that
|p(t,x3(t))−p(T2,x3(T2))|=|p(t,x3(t))−p(T2,x2(T2))|<ε, for T2<t≤T2+δ2. | (2.17) |
If T2+δ2<T, we take T2+δ2≐T3 and continue next procedure. Otherwise, we take T3=T and end this procedure. Obviously, according to (2.17) and (2.12), it holds
|p(t,x3(t))−p3|<ε, for T2<t≤T3. | (2.18) |
We assume T2+δ2<T, and then let
p4=p(T3,x3(T3)). | (2.19) |
We let
x′(t)={∫t0x1(s)ds,t∈(0,T1],∫tT1x2(s)ds,t∈(T1,T2],∫tT2x3(s)ds,t∈(T2,T3]. | (2.20) |
By (2.20) and integration by parts, we get
ωxi(t)≐∫TiTi−1(t−s)−p4x′(s)Γ(1−p4)ds=∫TiTi−1(t−s)1−p4xi(s)Γ(2−p4)ds−(t−Ti)1−p4Γ(2−p4)∫TiTi−1xi(s)ds, | (2.21) |
i=1,2,3,T0=0.
We may consider the initial value problem in the interval [T3,T] as following
{CDp4T3+x(t)=f(t,x)−ωx1(t)−ωx2(t)−ωx3(t),T3<t≤T,x(T3)=x3(T3), | (2.22) |
where xi(t) are solutions of the initial value problems (2.2), (2.8) and (2.15) respectively, and ωxi(i=1,2,3) is the function defined by (2.21).
Since [0,T] is a finite interval, we continue this procedure and could finish it by finite steps. That is, there exists δn∗−2>0, δn∗−1>0 (n∗∈N) such that Tn∗−2+δn∗−2≐Tn∗−1<T, Tn∗−1+δn∗−1≥T≐Tn∗. Then we have intervals [0,T1],[T1,T2], [T2,T3], ⋯, [Tn∗−2,Tn∗−1], [Tn∗−1,T], and solutions xi∈C1(Ti−1,T]∩C[Ti−1,T] of the following initial value problem defined in the interval [Ti−1,T]
{CDpiTi−1+x(t)=f(t,x)−ψx1(t)−ψx2(t)−⋯−ψxi−1(t),Ti−1<t≤T,x(Ti−1)=xi−1(Ti−1), | (2.23) |
where pi=p(Ti−1,xi−1(Ti−1)) satisfying
|p(t,xi(t))−pi|<ε, for Ti−1<t≤Ti, | (2.24) |
and
ψxj(t)≐∫TjTj−1(t−s)−pix′(s)Γ(1−pi)ds=∫TjTj−1(t−s)1−pixj(s)Γ(2−pi)ds−(t−Tj)1−piΓ(2−pi)∫TjTj−1xj(s)ds, | (2.25) |
j=1,2,…,i−1, i=5,6,⋯,n∗, T0=0,Tn∗=T. For details, please refer to the analysis of problems (2.2), (2.8), (2.15) and (2.22).
From the arguments above, we obtain a function x∗∈C[0,T] defined by
x∗(t)={x1(t),0≤t≤T1,x2(t),T1≤t≤T2,⋮xn∗(t),Tn∗−1≤t≤T. | (2.26) |
Now, we propose the new definition of approximate solution to the initial value problem (1.1), which is crucial in our work.
Definition 2.1. If there exists natural number n∗∈N and intervals [0,T1],(T1,T2],⋯, (Tn∗−1,T], and initial value problems (2.2), (2.8), (2.15), (2.22), (2.23) exist solutions x1∈C1(0,T]∩C[0,T], x2∈C1(T1,T]∩C[T1,T], x3∈C1(T2,T]∩C[T2,T], x4∈C1(T3,T]∩C[T3,T], xi∈C1(Ti−1,T]∩C[Ti−1,T] (i=5,⋯,n∗) respectively, then the function x∈C[0,T] defined by
x(t)={x1(t),0≤t≤T1,x2(t),T1≤t≤T2,x3(t),T2≤t≤T3,⋮xn∗(t),Tn∗−1≤t≤T | (2.27) |
is called an approximate solution of the initial value problem (1.1).
Remark 2.3. If x1(t),x2(t),⋯,xn∗(t) are both unique, then we say x(t) defined by (2.27) is unique approximate solution of the initial value problem (1.1).
Remark 2.4. Based on Remark 2.1, Definition 2.1 seems suitable.
Remark 2.5. From Definition 2.1, we notice that approximate solution x(t) of the initial value problem (1.1) in interval [T2,T3] is x3(t) which is a solution of the initial value problem (2.15). Obviously, the state of x3(t) is affected by the state of x1(t) and x2(t). That is, the state of x(t) in interval [T2,T3] is affected by the state of x(t) in interval [0,T2]. Hence, Definition 2.1 is suitable and reasonable according to Remark 2.2.
Remark 2.6. In our previous analysis, we chose functions (2.6), (2.13), etc, so that we obtain the initial value problems (2.8), (2.15), etc. Such a choice must meet the following three reasons at the same time. The first reason is operability, for instance, choosing function (2.13) enable us to calculate functions ϕx1(t),ϕx2(t), and obtain the initial value problem (2.15) defined in [T2,T]. The second reason is for fitting Remark 2.1 and Remark 2.5. If we take x′(t)=x1(T1) for 0≤t≤T1 (here x1(t) is the solution of the initial value problem (2.2)), we may easily calculate function φx1(t)=x1(T1)(t1−p1−(t−T1)1−p1)Γ(2−p1), and then have the initial value problem (2.8) with such φx1(t). However, we see that the state of solution of problem (2.8) is only affected by x1(T1), but not affected by the state of x1(t),0≤t≤T1. The third reason is the rationality of the obtained equation. For instance, according to the definition the Caputo fractional derivative CDp2T1+x(t)=1Γ(1−p2)∫tT1(t−s)−p2x′(s)ds, the solution of initial value problem (2.8) exists only if the term f(t,⋅)−φx1(t) is absolutely continuous in the interval [T1,T], otherwise one can not obtain the existence of solution of the initial value problems (2.8). If we take x(t)=x1(t) for 0≤t≤T1 (here x1(t) is the solution of initial value problem (2.2)), we may easily get function φx1(t)=∫T10(t−s)−p2x′1(s)dsΓ(1−p2), and then obtain initial value problem (2.8) with such φx1(t). However, we can't obtain the existence of solution of initial value problem (2.8) with this φx1(t).
Example 2.1. According to Definition 2.1, we consider the approximate solution of the following initial value problem
{CD12+t1000(1+t2)+x(t)3(1+x2(t))0+x(t)=12000Γ(32)t12,0<t≤1,x(0)=0. | (2.28) |
By the definition of the variable order Caputo fractional derivative, there is no way to obtain explicit expression of its solution, even hardly conventional methods to study the existence of its solution. Next, according to Definition 2.1, we try to seek its approximate solution.
Here p(t,x(t))=12+t1000(1+t2)+x(t)3(1+x2(t)), then we take p1=p(0,0)=12.
First, we consider initial value problem as following
{CD120+x(t)=12000Γ(32)t12,0<t≤1,x(0)=0. | (2.29) |
By simple calculation, the solution of the problem (2.29) is given by
x1(t)=12000Γ(32)Γ(12)∫t0(t−s)−12s12ds=12000t,0≤t≤1. | (2.30) |
Since x1(t) is right continuous at point 0, then for ε=0.002, we take δ01=12 such that
|x1(t)−x1(0)|=12000t≤14000<0.002, for 0<t≤δ01=12. | (2.31) |
Notice that p(t,x1(t)) is right continuous at point (0,x1(0))=(0,0), then together with (2.31), for the above ε=0.002, we take δ0=12, such that
|p(t,x1(t))−p(0,0)|=|p(t,x1(t))−p1|≤t1000+x1(t)3=12000+112000<0.002=ε, for 0<t≤12. |
We take point T1=δ0=12, and let
p2=p(T1,x1(T1))=p(12,x1(12))=12+12500+400048000003. | (2.32) |
We denote
φx1(t)=∫120(t−s)1−p2x1(s)dsΓ(2−p2)−(t−12)1−p2∫120x1(s)dsΓ(2−p2),12<t≤1. | (2.33) |
Next, we consider the following initial value problem
{CDp212+x(t)=12000Γ(32)t12−φx1(t),12<t≤1,x(12)=x1(12)=14000. | (2.34) |
From the facts of constant order fractional calculus, the solution of the initial value problem (2.34) is
x2(t)=14000+12000Γ(32)Γ(p2)∫t12(t−s)p2−1s12ds−1Γ(p2)∫t12(t−s)p2−1φx1(s)ds. | (2.35) |
We notice that
|φx1(t)|=12000Γ(2−p2)|∫120(t−s)1−p2sds−(t−12)1−p2∫120sds|≤12000Γ(2−p2)[∫120sds+∫120sds]<12000Γ(2−p2). |
Since x2(t) is right continuous at point T1=12, then for ε=0.002, we take δ11=12 such that for 12<t≤1, we have
|x2(t)−x2(12)|≤12000Γ(32)Γ(p2)∫t12(t−s)p2−1s12ds+1Γ(p2)∫t12(t−s)p2−1|φx1(s)|ds≤12000Γ(32)Γ(p2)∫t12(t−s)p2−1ds+12000Γ(2−p2)Γ(p2)∫t12(t−s)p2−1ds≤12000Γ(32)Γ(1+p2)(t−12)p2+12000Γ(2−p2)Γ(1+p2)(t−12)p2≤12000Γ(32)Γ(1+p2)+12000Γ(2−p2)Γ(1+p2)≈0.00096<0.002. |
Because p(t,x2(t)) is right continuous at point (T1,x2(T1))=(12,x2(12)), then together with the estimation above, for ε=0.002, we take δ1=12 such that for 12<t≤1,
|p(t,x2(t))−p(T1,x2(T1))|=|t1000(1+t2)−121000(1+(12)2)+x2(t)3(1+x22(t))−x2(12)3(1+x22(12))|≤11000|t−12|+13|x2(t)−x2(12)|≤12000+23000<0.002. |
From the arguments above, the function x∗∈C[0,1] defined by
x∗(t)={12000t,0≤t≤12,14000+∫t12(t−s)p2−1s12ds2000Γ(32)Γ(p2)−∫t12(t−s)p2−1φx1(s)dsΓ(p2),12≤t≤1 | (2.36) |
is the continuous approximate solution of problem (2.28) according to Definition 2.1(see Figure 1).
In this section, we study the initial value problem (1.1) for p(t,x(t))≡p(t). Since p(t) is a function of one variable, we may propose another definition of approximate solution of the initial value problem (1.1) such that we can simplify the analysis process in the section 2.
The following result is crucial for us to propose another definition of approximate solution of the initial value problem (1.1) for the particular order p(t).
Lemma 3.1. Assume that p:[0,T]→(0,1) is a continuous function, then for arbitrary small ε>0, there exists natural number n∗ and intervals [0,T1],(T1,T2],⋯,(Tn∗−1,T] and a piecewise function α:[0,T]→(0,1) defined by
α(t)={p1, t∈[0,T1],p2, t∈(T1,T2],⋮pn∗, t∈(Tn∗−1,T], | (3.1) |
where pk=p(Tk−1)(k=1,2,⋯,n∗,T0=0), such that for arbitrary small ε>0,
|p(t)−α(t)|<ε, 0≤t≤T. | (3.2) |
Proof. Since p(t) is right continuous at point 0, then for arbitrary small ε>0, there exists δ0>0 such that
|p(t)−p(0)|<ε, for 0≤t≤δ0. | (3.3) |
We take point δ0≐T1 (if T1<T, we consider continuity of p(t) at point T1, otherwise, we end this procedure). Since p(t) is right continuous at point T1, then for the above ε, there exists δ1>0 such that
|p(t)−p(T1)|<ε, for T1<t≤T1+δ1. | (3.4) |
We take point T1+δ1≐T2 (if T2<T, we consider continuity of p(t) at point T2, otherwise, we end this procedure). Since p(t) is right continuous at point T2, then for the above ε, there exists δ2>0 such that
|p(t)−p(T2)|<ε, for T2<t≤T2+δ2. | (3.5) |
Since [0,T] is a finite interval, we continue this analysis procedure and could finish it by finite steps. That is, there exists δn∗−2>0 and δn∗−1>0 (n∗∈N) such that Tn∗−2+δn∗−2≐Tn∗−1<T and Tn∗−1+δn∗−1≥T. Then we have intervals [0,T1],[T1,T2], ⋯, [Tn∗−2,Tn∗−1], [Tn∗−1,T], such that for the above ε,
|p(t)−p(Ti−1)|<ε, for Ti−1<t≤Ti, | (3.6) |
i=1,2,…,n∗, T0=0 and Tn∗=T.
We denote
p(0)≐p1,p(T1)≐p2,p(T2)≐p3,p(T3)≐p4,⋯,p(Tn∗−1)≐pn∗. | (3.7) |
Thus, we define a piecewise function α:[0,T]→(0,1) as following
α(t)={p1, t∈[0,T1],p2, t∈(T1,T2],⋮pn∗, t∈(Tn∗−1,T]. | (3.8) |
From (3.3)–(3.8), we obtain that for the arbitrary small ε>0,
|p(t)−α(t)|<ε, 0≤t≤T. |
The proof is completed.
Similar to the analysis in the section 2, we consider approximate solution of the initial value problem (1.1) in the following sense: if p(t) and α(t) satisfy (3.2), then solution x(t) of the following initial value problem
{CDα(t)0+x(t)=f(t,x),0<t≤T,x(0)=u0 | (3.9) |
is called the approximate solution of the initial value problem (1.1).
We start off by analyzing the problem (3.9), and then propose a new definition of continuous approximate solutions to the initial value problem (1.1) with p(t,x(t))=p(t).
For the initial value problem (3.9) in the interval [0,T1], by (3.1), we have the initial value problem
{CDp10+x(t)=f(t,x),0<t≤T1,x(0)=u0. | (3.10) |
By similar analysis in section 2, we may consider the initial value problem defined in the interval [T1,T2] as following
{CDp2T1+x(t)=f(t,x)−φx1(t),T1<t≤T2,x(T1)=x1(T1), | (3.11) |
where x1(t) is the solution of the initial value problem (3.10) and
∫T10(t−s)−p2x′(s)Γ(1−p2)ds=∫T10(t−s)1−p2x1(s)Γ(2−p2)ds−(t−T1)1−p2Γ(2−p2)∫T10x1(s)ds≐φx1(t), |
in which x′(s)=∫s0x1(τ)dτ.
Using the same method, we may consider the initial value problem defined in the interval [T2,T3] as following
{CDp3T2+x(t)=f(t,x)−ϕx1(t)−ϕx2(t),T2<t≤T3,x(T2)=x2(T2), | (3.12) |
where x1(t) is the solution of the initial value problem (3.10), x2(t) is the solution of the initial value problem (3.11) and ϕxi(t) is the function defined by
∫TiTi−1(t−s)−p3x′(s)Γ(1−p3)ds=∫TiTi−1(t−s)1−p3xi(s)Γ(2−p3)ds−(t−Ti)1−p3Γ(2−p3)∫TiTi−1xi(s)ds≐ϕxi(t), |
in which x′(s)=∫sTi−1xi(τ)dτ, i=1,2,T0=0.
Similarly, we may consider the initial value problem defined in the interval [Ti−1,Ti] as following
{CDpiTi−1+x(t)=f(t,x)−ψx1(t)−ψx2(t)−⋯−ψxi−1(t),Ti−1<t≤Ti,x(Ti−1)=xi−1(Ti−1), | (3.13) |
where x1(t) is the solution of the initial value problem (3.10), x2(t) is the solution of the initial value problem (3.11), x3(t) is the solution of the initial value problem (3.12), xi(t) is the solution of the initial value problem (3.13) and ψxj is the function defined by
∫TjTj−1(t−s)−pix′(s)Γ(1−pi)ds=∫TjTj−1(t−s)1−pixj(s)Γ(2−pi)ds−(t−Tj)1−piΓ(2−pi)∫TjTj−1xj(s)ds≐ψxj(t), |
in which x′(s)=∫sTj−1xj(τ)dτ, j=1,2,⋯,i−1, i=4,⋯,n∗, T0=0,Tn∗=T.
Based on the arguments above, we propose the definition of solution to the initial value problem (3.9), which is crucial in our work.
Definition 3.1. If the initial value problems (3.10), (3.11), (3.12) and (3.13) exist solutions x1:[0,T1]→R, x2:[T1,T2]→R, x3:[T2,T3]→R, xi:[Ti−1,Ti]→R(i=4,⋯,n∗, Tn∗=T) respectively, then we call function x:[0,T]→R defined by
x(t)={x1(t),0≤t≤T1,x2(t),T1≤t≤T2,x3(t),T2≤t≤T3,⋮xn∗(t),Tn∗−1≤t≤T | (3.14) |
is a solution of the initial value problem (3.9).
Remark 3.1. If x1(t),x2(t),⋯,xn∗(t) are unique, then we say x(t) defined by (3.14) is unique solution of the initial value problem (3.9).
The following is the definition of the approximate solution of the initial value problem (1.1) with p(t,x(t))≡p(t).
Definition 3.2. We call the solution of initial value problem (3.9) defined by Definition 3.1 is an (unique) approximate solution of initial value problem (1.1) with p(t,x(t))=p(t) if α(t) is defined by Lemma 3.1.
Remark 3.2. For the initial value problem (1.1) with p(t,x(t))≡p(t), its approximate solution defined by Definition 3.2 is consistent with by Definition 2.1.
Example 3.1. We consider the following initial value problem for linear equation
{CD13+t1000(1+t2)0+x(t)=t,0<t≤1,x(0)=0. | (3.15) |
By the definition of the variable order Caputo fractional derivative, there is no way to obtain explicit expression of its solution, even hardly conventional methods to study the existence of its solution. Next, we try to seek its approximate solution in the sense of Definition 3.2.
Here p(t)=13+t1000(1+t2). Obviously, p(t) is continuous on [0,1] and 0<p(t)<1.
By the right continuity of function p(t) at point 0, for given arbitrary small ε=0.00035, taking δ0=13, when 0≤t≤δ0=13, we have
|p(t)−p(0)|=|t1000(1+t2)|≤t1000≤δ01000<13000<ε. |
Then, we get T1=δ0=13. By the right continuity of function p(t) at point T1, for the above ε, taking δ1=13, when T1<t≤T1+δ1(13<t≤23), by differential mean value theorem, we have
|p(t)−p(T1)|=|t1000(1+t2)−T11000(1+T21)|≤|1−ξ21000(1+ξ2)2||t−T1|≤1+ξ21000(1+ξ2)2|t−T1|≤11000|t−T1|≤δ11000<13000<ε, |
where T1<ξ<t<23.
We let T2=T1+δ1=23. By the right continuity of function p(t) at point T2, for ε=0.00035, taking δ2=13, when T2<t≤T2+δ2(23<t≤1), we have
|p(t)−p(T2)|=|t1000(1+t2)−T21000(1+T22)|≤δ21000<13000<ε. |
We see that T2+δ2=1, hence, we obtain three intervals [0,13], (13,23], (23,1] and piecewise constant function α(t) defined by
α(t)={p1=p(0)=13, t∈[0,13],p2=p(13)=1000930000, t∈(13,23],p3=p(23)=650919500, t∈(23,1], |
which satisfies
|α(t)−p(t)|<0.00035=ε. | (3.16) |
Thus, according to Definition 3.2, we first consider the following initial value problem
{CDp10+x(t)=t,0<t≤13,x(0)=0. | (3.17) |
By the facts of constant order fractional calculus, the unique solution of the initial value problem (3.17) is
x1(t)=1Γ(73)t43,0≤t≤13. | (3.18) |
Let
φx1(t)=∫130(t−s)1−p2x1(s)Γ(2−p2)ds−(t−13)1−p2Γ(2−p2)∫130x1(s)ds. | (3.19) |
Next, we consider the following initial value problem
{CDp213+x(t)=t−φx1(t),13<t≤23,x(13)=x1(13), | (3.20) |
where x1 and ψx1 are given by (3.18) and (3.19) respectively. By simple calculation, we get the solution of the initial value problem (3.20) as following
x2(t)=∫t13(t−s)p2−1Γ(p2)(s−φx1(s))ds+x1(13),13≤t≤23. | (3.21) |
Let
{ψx1(t)=∫130(t−s)1−p3x1(s)Γ(2−p3)ds−(t−13)1−p3Γ(2−p3)∫130x1(s)ds,ψx2(t)=∫2313(t−s)1−p3x2(s)Γ(2−p3)ds−(t−23)1−p3Γ(2−p3)∫2313x2(s)ds. | (3.22) |
Finally, we consider the following initial value problem
{CDp323+x(t)=t−ψx1(t)−ψx2(t),23<t≤1,x(23)=x2(23), | (3.23) |
where x1 is given by (3.18), x2 is presented by (3.21) and ψxi(i=1,2) is given by (3.22). Thus, the solution of the initial value problem (3.23) is obtained by
x3(t)=∫t23(t−s)p3−1Γ(p3)(s−ψx1(s)−ψx2(s))ds+x2(23),23≤t≤1. | (3.24) |
According to Definition 3.2, the approximate solution of initial value problem (3.15) is given by
x(t)={x1(t)=1Γ(73)t43,0≤t≤13,x2(t)=∫t13(t−s)p2−1Γ(p2)(s−φx1(s))ds+x1(13),13≤t≤23x3(t)=∫t23(t−s)p3−1Γ(p3)(s−ψx1(s)−ψx2(s))ds+x2(23),23≤t≤1. | (3.25) |
Obviously, x(t) defined by (3.25) is continuous.
Example 3.2. We consider the continuous approximate solution of initial value problem (3.15) in the sense of Definition 2.1.
According Definition 2.1, let
p0=p(0)=13. | (3.26) |
We first consider the following initial value problem
{CDp10+x(t)=t,0<t≤1,x(0)=0. | (3.27) |
By the arguments above, we know that the solution of initial value problem (3.27) is
x1(t)=1Γ(73)t43,0≤t≤1. | (3.28) |
Since p(t) is right continuous at point 0, for ε=0.00035, there exists δ0=13 such that
|p(t)−p(0)|<ε,0<t≤δ0=13. |
We take T1=δ0=13. Let
p2=p(T1)=p(13). | (3.29) |
By Definition 2.1, we next seek the solution of the initial value problem
{CDp213+x(t)=t−φx1(t),13<t≤1,x(13)=x1(13), | (3.30) |
where x1(t) is the function given by (3.28), and φx1(t) is the function defined by
φx1(t)=∫130(t−s)1−p2x1(s)Γ(2−p2)ds−(t−13)1−p2Γ(2−p2)∫130x1(s)ds. | (3.31) |
Thus, the solution of the initial value problem (3.30) is
x2(t)=∫t13(t−s)p2−1Γ(p2)(s−φx1(s))ds+x1(13),13≤t≤1. | (3.32) |
Since p(t) is right continuous at point T1, for ε=0.00035, there exists δ1=13 such that
|p(t)−p(T1)|=|p(t)−p(13)|<ε,13<t≤13+δ1=23. | (3.33) |
We take T2=T1+δ1=23. Let
p3=p(T2)=p(23). | (3.34) |
According to Definition 2.1, then we seek the solution of the initial value problem
{CDp323+x(t)=t−ψx1(t)−ψx2(t),23<t≤1,x(23)=x2(23), | (3.35) |
where x1(t) is given by (3.28), x2(t) is presented by (3.32) and ψxi(t)(i=1,2) is the function as followings
{ψx1(t)=∫130(t−s)1−p3x1(s)Γ(2−p3)ds−(t−13)1−p3Γ(2−p3)∫130x1(s)ds,ψx2(t)=∫2313(t−s)1−p3x2(s)Γ(2−p3)ds−(t−23)1−p3Γ(2−p3)∫2313x2(s)ds. | (3.36) |
Thus, the solution of the initial value problem (3.35) is
x3(t)=∫t23(t−s)p3−1Γ(p3)(s−ψx1(s)−ψx2(s))ds+x2(23),23≤t≤1, | (3.37) |
where ψxi(i=1,2) is given by (3.36).
Hence, the approximate solution of initial value problem (3.15) in the sense of Definition 2.1 is given by
x(t)={x1(t)=1Γ(73)t43,0≤t≤13,x2(t)=∫t13(t−s)p2−1Γ(p2)(s−φx1(s))ds+x1(13),13≤t≤23,x3(t)=∫t23(t−s)p3−1Γ(p3)(s−ψx1(s)−ψx2(s))ds+x2(23),23≤t≤1. | (3.38) |
Obviously, x(t) defined by (3.38) is continuous.
From the analysis above, for the same ε and the same interval [0,13], [13,23], [23,1], continuous approximate solution of initial value problem (3.15) obtained by Definition 3.2 is consistent with which is obtained by Definition 2.1.
This paper propose a new definition of continuous approximate solution to initial value problem for differential equations involving variable order Caputo fractional derivative on finite interval. It provide a new method to consider the existence of solution to more general variable order fractional differential equations. This method is different from common numerical methods and it also can be apply to variable order fractional boundary value problem.The new results generalize some existing results in the literature.
The authors would like to thank the referees for the helpful suggestions. The research are supported by the National Natural Science Foundation of China (No. 12071302) and the Fundamental Research Funds for the Central Universities (No. 2021YJSLX01).
The authors declare no conflict of interest.
[1] | D. Valério, J. Sá da Costa, Variable-order fractional derivative and their numerical approximations Signal Process., 91 (2011), 470–483. |
[2] |
D. Tavares, R. Almeida, D. F. M. Torres, Caputo derivatives of fractional variable order: Numerical approximations, Commun. Nonlinear Sci., 35 (2016), 69–87. doi: 10.1016/j.cnsns.2015.10.027
![]() |
[3] | A. A. Kilbas, H. M. Srivastava, J. J. Trujillo, Theory and applications of fractional differential equations, Elsevier, 2006. |
[4] | K. B. Oldham, J. Spanier, The fractional calculus: Integrations and differentiations of arbitrary order, New York: Academic Press, 1974. |
[5] | I. Podlubny, Fractional differential equations, San Diego: Academic Press, 1999. |
[6] | K. Diethelm, The analysis of fractional differential equations, Springer Science & Business Media, 2010. |
[7] | A. Atangana, Fractional operators with constant and variable order with application to geo-hydrology, New York: Academic Press, 2017. |
[8] |
A. Razminia, A. F. Dizaji, V. J. Majd, Solution existence for non-autonomous variable-order fractional differential equations, Math. Comput. Model., 55 (2012), 1106–1117. doi: 10.1016/j.mcm.2011.09.034
![]() |
[9] | A. A. Alikhanov, Boundary value problems for the equation of the variable order in differential and difference settings, Appl. Math. Comput., 219 (2012), 3938–3946. |
[10] |
A. Babaei, H. Jafari, S. Banihashemi, Numerical solution of variable order fractional nonlinear quadratic integro-differential equations based on the sixth-kind Chebyshev collocation method, J. Comput. Appl. Math., 377 (2020), 112908. doi: 10.1016/j.cam.2020.112908
![]() |
[11] |
C. J. Zúniga-Aguilar, H. M. Romero-Ugalde, J. F. Gómez-Aguilar, R. F. Escobar-Jiménez, Solving fractional differential equations of variable-order involving operator with Mittag-Leffler kernel using artificial neural networks, Chaos Soliton. Fract., 103 (2017), 382–403. doi: 10.1016/j.chaos.2017.06.030
![]() |
[12] |
C. M. Chen, F. Liu, V. Anh, I. Turner, Numerical schemes with high spatial accuracy for a variable-order anomalous subdiffusion equation, SIAM J. Sci. Comput., 32 (2010), 1740–1760. doi: 10.1137/090771715
![]() |
[13] |
D. Sierociuk, W. Malesza, M. Macias, Derivation, interpretation, and analog modelling of fractional variable order derivative definition, Appl. Math. Model., 39 (2015), 3876–3888. doi: 10.1016/j.apm.2014.12.009
![]() |
[14] |
H. Hassani, J. A. Tenreiro Machado, E. Naraghirad, An efficient numerical technique for variable order time fractional nonlinear Klein-Gordon equation, Appl. Numer. Math., 154 (2020), 260–272. doi: 10.1016/j.apnum.2020.04.001
![]() |
[15] |
H. Hassani, Z. Avazzadeh, J. A. Tenreiro Machado, Numerical approach for solving variable order space-time fractional telegraph equation using transcendental Bernstein series, Eng. Comput., 36 (2020), 867–878. doi: 10.1007/s00366-019-00736-x
![]() |
[16] |
J. Vanterler da C. Sousa, E. Capelas de Oliverira, Two new fractional derivatives of variable order with non-singular kernel and fractional differential equation, Comput. Appl. Math., 37 (2018), 5375–5394. doi: 10.1007/s40314-018-0639-x
![]() |
[17] |
J. F. Gómez-Aguilar, Analytical and numerical solutions of a nonlinear alcoholism model via variable-order fractional differential equations, Physica. A, 494 (2018), 52–57. doi: 10.1016/j.physa.2017.12.007
![]() |
[18] |
J. Yang, H. Yao, B. Wu, An efficient numerical method for variable order fractional functional differential equation, Appl. Math. Lett., 76 (2018), 221–226. doi: 10.1016/j.aml.2017.08.020
![]() |
[19] |
M. Hajipour, A. Jajarmi, D. Baleanu, H. Sun, On an accurate discretization of a variable-order fractional reaction-diffusion equation, Commun. Nonlinear Sci., 69 (2019), 119–133. doi: 10.1016/j.cnsns.2018.09.004
![]() |
[20] |
R. M. Ganji, H. Jafari, D. Baleanu, A new approach for solving multi variable orders differential equations with Mittag-Leffler kernel, Chaos Soliton. Fract., 130 (2020), 109405. doi: 10.1016/j.chaos.2019.109405
![]() |
[21] |
S. G. Samko, Fractional integration and differentiation of variable order, Anal. Math., 21 (1995), 213–236. doi: 10.1007/BF01911126
![]() |
[22] |
S. G. Samko, B. Boss, Integration and differentiation to a variable fractional order, Integr. Transf. Spec. F., 1 (1993), 277–300. doi: 10.1080/10652469308819027
![]() |
[23] | S. Zhang, S. Sun, L. Hu, Approximate solutions to initial value problem for differential equation of variable order, JFCA, 9 (2018), 93–112. |
[24] | S. Zhang, The uniqueness result of solutions to initial value problem of differential equations of variable-order, RACSAM Rev. R. Acad. A, 112 (2018), 407–423. |
[25] |
W. Malesza, M. Macias, D. Sierociuk, Analyitical solution of fractional variable order differential equations, J. Comput. Appl. Math., 348 (2019), 214–236. doi: 10.1016/j.cam.2018.08.035
![]() |
[26] |
Y. Kian, E. Soccorsi, M. Yamamoto, On time-fractional diffusion equations with space-dependent variable order, Ann. Henri Poincaré, 19 (2018), 3855–3881. doi: 10.1007/s00023-018-0734-y
![]() |
[27] |
J. Jiang, H. Chen, J. L. G. Guirao, D. Cao, Existence of the solution and stability for a class of variable fractional order differential systems, Chaos Soliton. Fract., 128 (2019), 269–274. doi: 10.1016/j.chaos.2019.07.052
![]() |
[28] | R. Almeida, D. Tavares, D. Torres, The variable-order fractional calculus of variations, Springer International Publishing, 2019. |
[29] |
H. G. Sun, A. Chang, Y. Zhang, W. Chen, A review on variable-order fractional differential equations: mathematical foundations, physical models, numerical methods and applications, Fract. Calc. Appl. Anal., 22 (2019), 27–59. doi: 10.1515/fca-2019-0003
![]() |
[30] |
X. Li, B. Wu, A new reproducing kernel method for variable order fractional boundary value problems for functional differential equations, J. Comput. Appl. Math., 311 (2017), 387–393. doi: 10.1016/j.cam.2016.08.010
![]() |
[31] |
X. Li, B. Wu, A numerical technique for variable fractional functional boundary value problems, Appl. Math. Lett., 43 (2015), 108–113. doi: 10.1016/j.aml.2014.12.012
![]() |
[32] |
J. Deng, Z. Deng, Existence of solutions of initial value problems for nonlinear fractional differential equations, Appl. Math. Lett., 32 (2014), 6–12. doi: 10.1016/j.aml.2014.02.001
![]() |
[33] | X. Dong, Z. Bai, S. Zhang, Positive solutions to boundary value problems of p-Laplacian with fractional derivative, Bound. Value Probl., 5 (2017), 1–15. |
[34] |
Z. Bai, S. Zhang, S. Sun, Y. Chun, Monotone iterative method for a class of fractional differential equations, Electron. J. Differ. Eq., 2016 (2016), 1–8. doi: 10.1186/s13662-015-0739-5
![]() |
[35] | T. T. Hartley, C. F. Lorenzo, Fractional system identification: An approach using continuous order distributions, NASA Glenn Research Center, 1999. |
1. | Ravi P. Agarwal, Snezhana Hristova, Donal O’Regan, Impulsive Control of Variable Fractional-Order Multi-Agent Systems, 2024, 8, 2504-3110, 259, 10.3390/fractalfract8050259 | |
2. | John R. Graef, Kadda Maazouz, Moussa Daif Allah Zaak, A Comprehensive Study of the Langevin Boundary Value Problems with Variable Order Fractional Derivatives, 2024, 13, 2075-1680, 277, 10.3390/axioms13040277 |