Local Discontinuous Galerkin Method for the Time-Fractional KdV Equation with the Caputo-Fabrizio Fractional Derivative ()
1. Introduction
Fractional differential equations have become increasingly important due to their deep scientific and engineering background to correctly model challenging phenomena such as long-range time memory effects, mechanical systems, control systems, etc. [1] [2]. In recent years, variable-order fractional calculus has been found in some physical processes such as algebraic structure and noise reduction. Variable-order fractional calculus is a natural choice to provide an effective mathematical framework for describing complex problems and has many advantages in describing the memory properties of systems [3] - [9].
Fractional partial differential equations can describe abnormal physical phenomena more accurately than integer partial differential equations, which have attracted more and more attention. However, it is difficult to obtain analytical solutions to fractional partial differential equations when the fractional derivatives are known. Therefore, we need to consider efficient numerical methods such as the finite element method [10] [11] [12] [13], discontinuous Galerkin method [14] [15] [16] [17] [18], spectral method [19] [20], and finite difference method [21] [22] [23], finite volume method [24] [25]. Wei [26] studied the exact numerical scheme of a class of variable-order fractional diffusion equations, using the fractional derivatives of Caputo-Fabrizio and the theoretical analysis by the local discontinuous Galerkin method. Du [27] proposed different difference schemes for multi-dimensional variable-order time fractional subdiffusion equations and found a special point approximation for the variable-order time Caputo derivative. It is proved that the resulting difference scheme is uniquely solvable. Li et al. [28] carried out a numerical study on three typical Caputo-type partial differential equations using the finite difference method/local discontinuous Galerkin finite element method.
The KdV equation was first proposed by Boussinesq in 1877, and it is a typical dispersion nonlinear partial differential equation. The nonlinear KdV equation was derived by Korteweg and de Vries in 1895 [29], and it describes the propagation of waves in various nonlinear dispersive media. Since then, the KdV equation has been widely used in various physical phenomena and engineering modeling, such as nonlinear wave interactions [30], interfacial electrohydrodynamics [31], plasma physics, geology, etc. Numerous numerical methods have been proposed to solve this equation, such as finite difference schemes [32] [33], pseudospectral methods [34], thermal equilibrium integration methods [35], and discontinuous Galerkin methods [36] [37]. For sufficiently smooth solutions, the following literature does some numerical work on the fractional time KdV equation. Wei et al. [38] proposed the LDG finite element method of the KdV-Burgers-Kuramoto equation, using variable-order Riemann-Liouville fractional derivatives, and proved the unconditional stability and convergence of the scheme. Zhang [39] constructed an efficient numerical scheme for solving linearized fractional KdV equations on unbounded spaces. The non-local fractional derivatives are obtained by exponentiating the convolution kernel and approximately evaluating the initial boundary value problem.
In this paper, the Korteweg-de Vries equation (KdV) with Caputo-Fabrizio fractional derivatives is constructed
(1.1)
where the fractional derivative orders
,
,
and
are smooth functions.
and
are positive constants. In addition, the solutions in this paper are periodic or compactly supported.
The Caputo-Fabrizio fractional derivative in (1.1) is defined as
(1.2)
There are many definitions of fractional derivatives, of which the most widely used are Riemann-Liouville fractional derivatives and Caputo fractional derivatives. The Caputo-Fabrizio fractional derivative used in this paper was proposed by Caputo and Fabrizio [40] in 2015. Compared with the Caputo fractional derivative model, the Caputo-Fabrizio fractional derivative model can describe different scales and configurations of matter. The Caputo-Fabrizio fractional-order derivatives have been widely used by researchers such as Ann Al Sawoor et al. [41] who studied the asymptotic stability of linear and interval linear fractional-order neutral delay differential systems described by the Caputo-Fabrizio fractional derivatives.
The key to the KdV equation LDG method is to rewrite the equation into a first-order equation system by introducing two auxiliary variables. The LDG method was first introduced by Cockburn and Shu to solve the convection-diffusion equation [42]. One of its advantages is that its solution and spatial derivatives have optimal (k + 1) order convergence on the L2 norm. Yan and Shu [43] developed a numerical method for LDG for general KdV-type equations involving third-order derivatives. Wei and He [44] used the LDG finite element method to solve the time-fractional KdV equation problem, discretized using finite differences in time and local discontinuous Galerkin methods in space. In [45], the authors established the L2 conservative LDG numerical scheme and compared it with the dissipative LDG scheme of the KdV type equation to show the dissipative induced phase error. In [46], Baccouch investigated the nonlinear KdV partial differential equation LDG numerical scheme. The results show that the LDG solution is superconvergent to a special Gauss-Radau projection of the exact solution.
The structure of this paper is as follows. In Section 2, some basic notation and mathematical foundations are introduced. Section 3 mainly introduces discrete methods and constructs the LDG scheme. Section 4 presents the stability and convergence results of the scheme. In Section 5, we give numerical experiments to illustrate the accuracy of our proposed format. Finally, we summarize and discuss our results in Section 6.
2. Preliminaries
2.1. Notations and Projection
Divide the interval
as
. For
, define
, and
,
.
We divide the interval
evenly into time steps
,
are mesh points.
The left and right limits of u at
are denoted by
and
, respectively. Where
is in the right cell
, and
is in the left cell
. Define
The associated discontinuous Galerkin space
is defined as follows
In proving the error estimate, we will use two projections on the one-dimensional interval
.
Denoted as
,
(2.1)
and
,
(2.2)
and
(2.3)
For the above projection
, it can be obtained from the standard approximation theory [47] [48] [49] [50],
(2.4)
where
or
. We want to denote all element boundary points in one-dimensional space by
. Furthermore, we have the following definition [51]
In this paper, C is a positive constant, which may take different values in different positions.
represents the scalar inner product over
,
represents the correlation norm. When
, we drop it.
2.2. Numerical Flux
In this paper, we will use the flux
, which is related to the discontinuous Galerkin spatial discretization.
is a monotonic numerical flux that depends on the left and right limits of the function
at point
, satisfying the following conditions:
1) It is local Lipschitz continuous, so
is bounded when the function
is in a bounded region;
2) It is consistent with the flux
, i.e.,
;
3) It is a function with monotonic properties, the first parameter is a non-decreasing function, and the second parameter is a non-increasing function.
3. The LDG Schemes
This section introduces the LDG method for the time-fractional KdV Equation (1.1).
First, we discretize the fractional derivative in the time direction
(3.1)
where
is the truncation error in the time direction,
and
.
By further calculation we can get
(3.2)
Lemma 3.1. [52] [53] When
, the truncation error
satisfies the following estimation
(3.3)
has the following properties
(3.4)
and
(3.5)
Rewrite the Equation (1.1) as a first-order system of equations,
(3.6)
represent approximate solutions of
, respectively.
. Find
such that for the test function
, we have
(3.7)
where
.
The hat function in the element boundary term resulting from the integral by parts in (3.7) is the numerical flux. To ensure stability, we can take the following alternating numerical fluxes
(3.8)
The choice of flux (3.8) is not unique, only
and
can take the opposite sides [54].
The fluxes
are monotonic fluxes as described in Section 2.2. Examples of monotonic fluxes suitable for local discontinuous Galerkin methods can be found [55] [56]. For example, we can use the Lax-Friedrich flux, which consists of
In the next section, we discuss the stability and convergence of the numerical Scheme (3.7).
4. Stability and Convergence
To simplify the notation, we consider the case of
in the numerical analysis.
Theorem 4.1. Under periodic or tightly supported boundary conditions, the fully discrete LDG scheme (3.7) is unconditionally stable, and the numerical solution
satisfies
(4.1)
Proof. Add the three equations in the scheme (3.7),
(4.2)
Substitute the test function
into the scheme (4.2), using flux (3.8) and the Cauchy-Schwarz inequality, we get
which is
(4.3)
here
The above scheme can be calculated by
(4.4)
For nonlinear terms, let
. Using the mean value theorem and the monotonicity of liquidity yields
, where
is a value between
and
.
Substituting (4.4) into (4.3), we get
(4.5)
Prove Theorem 4.1 by mathematical induction. Let
in the scheme (4.5), we have
since
we can get the following result
which means
Suppose the following inequalities hold
Next prove
.
From (4.5) we get
(4.6)
Therefore, we have
In summary, Theorem 4.1 is proved.
Next, we will state the error estimates of the equation
in the linear case, and use (3.8) as the flux choice. We have the following theorem.
Theorem 4.2. Let
is the exact solution of the problem (1.1),
is
is smooth enough. Let
be the numerical solution of the fully discrete LDG scheme (3.7), then there are the following error estimates
(4.7)
C is a positive constant that depends on
.
Proof. Denote
(4.8)
The above
can be estimated by the inequality (2.4). Next, we mainly discuss
.
We can easily verify that the exact solution of the partial differential Equation (1.1) satisfies the following
(4.9)
Select the flux (3.8), and subtract the Equations (3.7) and (4.9) to get the error equation
(4.10)
Substitute (4.2) into (4.10) to get
(4.11)
Using the projection property (2.1) - (2.3) and the test functions
,
, and
in (4.11), the following equality holds
(4.12)
Note that
, the following equation can be obtained
(4.13)
where
Using the Cauchy-Schwarz inequality, we have
(4.14)
Use
,
which is
(4.15)
Multiply both sides of the formula by
,
According to
, we can get
(4.16)
From Lemma 3.1, it can be known that
, and
are defined for simplicity.
Assume that the following estimates hold
(4.17)
We also prove it by mathematical induction. When
, the following inequality holds,
therefore
(4.18)
Then assume that
(4.19)
According to (4.16) and (4.18), we can get
so
(4.20)
Since
,
(4.21)
Combining the triangle inequality and the projection property (2.4), it can be seen that the Theorem 4.2 holds.
5. Numerical Experiment
In this section, discussing the effectiveness of the above scheme for solving KdV equations, we consider the following numerical example with initial values and periodic boundary conditions
(5.1)
where
,
Now we can check that the exact solution is
In the following numerical calculations, we will provide the results of the above examples under different
conditions using piecewise
polynomials to validate our method. The detailed results for the time and space directions are listed below, with
,
for time step and space step, respectively.
In order to reflect the spatial accuracy of the scheme, Figure 1 and Figure 2 adopt a fixed small time step
and the variable space step
. Selecting different
, the accuracy of
norm and
norm of piecewise
polynomial can reach the optimal order. Table 1 examines the convergence rate in the time direction of the LDG method, we
choose a sufficiently small space step
and a variable time step
. It can be seen from Table 1 that it has first-order convergence in time, which is also consistent with the theoretical results.
Table 1. For different order
when
,
, use the piecewise P2 polynomial to test the time accuracy.
Figure 1. L2 errors and L∞ errors VS h, order for
,
, piecewise P1 and P2 polynomial.
Figure 2. L2 errors and L∞ errors VS h, order for
,
, piecewise P1 and P2 polynomial.
6. Conclusion
This paper discusses the solution of a class of time-fractional KdV equations by the LDG method under the Caputo-Fabrizio fractional derivative. We derive the stability and error estimates of the proposed scheme. Numerical results demonstrate the effectiveness and good numerical performance of the method. In the future, we will consider generalizing this scheme to two-dimensional or high-dimensional cases.