Next Article in Journal
Integrating Multifractal Features into Machine Learning for Improved Prediction
Previous Article in Journal
New Perspectives of Hermite–Hadamard–Mercer-Type Inequalities Associated with ψk-Raina’s Fractional Integrals for Differentiable Convex Functions
 
 
Font Type:
Arial Georgia Verdana
Font Size:
Aa Aa Aa
Line Spacing:
Column Width:
Background:
Article

A Fast Finite Difference Method for 2D Time Fractional Mobile/Immobile Equation with Weakly Singular Solution

1
School of Mathematical Sciences, Liaocheng University, Liaocheng 252059, China
2
School of Mathematics, Shandong University, Jinan 250100, China
*
Author to whom correspondence should be addressed.
Fractal Fract. 2025, 9(4), 204; https://doi.org/10.3390/fractalfract9040204
Submission received: 18 February 2025 / Revised: 13 March 2025 / Accepted: 25 March 2025 / Published: 26 March 2025

Abstract

This paper presents a fast Crank–Nicolson L1 finite difference scheme for the two-dimensional time fractional mobile/immobile diffusion equation with weakly singular solution at the initial moment. First, the time fractional derivative is discretized using the Crank–Nicolson formula on uniform meshes, and a local truncation error estimate is provided. The spatial derivative is discretized using the central difference quotient on uniform meshes. Then, energy analysis methods are utilized to provide an optimal error estimates. On the other hand, the numerical scheme is optimized based on the sum-of-exponentials approximation, effectively reducing computation and memory requirements. Finally, numerical examples are simulated to verify the effectiveness of the algorithm.

1. Introduction

In recent years, fractional calculus has been extensively applied in various fields, including engineering [1,2], biology [3], physics [4,5], thermodynamics [6], and finance [7]. The inherent nonlocality of fractional calculus allows fractional equations to more accurately describe real-world problems and effectively capture the characteristics of physical phenomena, such as memory, heterogeneity, and genetic properties of materials [8,9], compared to classical integer-order differential equations. These phenomena can be described using corresponding fractional equations, including the fractional cable equation, the fractional diffusion equation, and the fractional Allen–Cahn equation.
Consider the following 2D time fractional mobile/immobile problem:
u t ( x , y , t ) + D t α 0 C u ( x , y , t ) = 2 u x 2 ( x , y , t ) + 2 u y 2 ( x , y , t ) + f ( x , y , t ) , ( x , y ) Ω , t ( 0 , T ] , u ( x , y , t ) = ϕ ( x , y , t ) , ( x , y ) Ω , t [ 0 , T ] , u ( x , y , 0 ) = g ( x , y ) , ( x , y ) Ω .
Here, 0 < α < 1 , Ω = ( 0 , L ) × ( 0 , L ) R 2 is a continuous bounded domain with boundary Ω , and D t α 0 C u denotes the Caputo fractional derivative with respect to t, defined as follows:
D t α 0 C u ( t ) = 1 Γ ( 1 α ) 0 t t s α u ( s ) s d s .
The fractional mobile/immobile transport model describes a range of complex phenomena, including thermal diffusion and underwater sound wave propagation. In physical or mathematical systems with temporal variations, the dynamic behavior of these phenomena is fundamentally similar to the process of thermal diffusion within solids [10]. Schumer et al. [11] first developed the fractional mobile/immobile model, which describes motion time by adding a time derivative term and helps distinguish particle motion and stationary states [12,13]. The equation can also be viewed as a limiting case for controlling continuous time random walks with heavy-tailed random waiting times [11]. The fractional mobile/immobile model has important applications across many fields, especially in groundwater solute modeling [14,15].
Typically, the analytical solution of fractional differential equations is extremely challenging, which has led many scholars to focus on the numerical solution of fractional mobile/immobile models. Liu et al. [16] developed an unconditionally stable high-order compact finite difference scheme for a Caputo fractional derivative-based mobile/immobile transport model and provided a fast solution method using fast Fourier transform. Fardi and Ghasemi [17] solved a two-dimensional fractional mobile/immobile diffusion reaction equation with Caputo fractional derivative using finite difference/spectral methods. Ghasemi et al. [18] utilized the shifted Grünwald–Letnikov formula to discretize the time fractional derivative and the local RBF-FD method to approximate the spatial derivative, calculating fractional mobile/immobile transport models on both regular and irregular domains. Zhao et al. [19] constructed a finite volume element scheme for two-dimensional nonlinear fractional mobile/immobile transport equation on triangular meshes based on the second-order weighted shifted Grünwald difference formula to discretize the time fractional derivative. Jiang et al. [20] proposed an alternating direction implicit (ADI) compact difference scheme for a two-dimensional semilinear time fractional mobile/immobile equation, where the first and fractional time derivatives are treated with the second-order backward differential formula and L1 scheme, respectively, and the spatial derivatives are discretized using the compact scheme. Qiu et al. [21] developed a time two-grid finite difference scheme for a two-dimensional nonlinear time fractional mobile/immobile transportation model with the first and fractional time derivatives processed via the second-order backward differential formula and L1 scheme, respectively, and the spatial derivative is discretized using the compact scheme. Qiao and Cheng [22] established a fast finite difference method for a two-dimensional time variable fractional mobile/immobile equation, where the time fractional derivative is discretized using the Crank–Nicolson scheme and the spatial derivative is discretized using the central difference quotient scheme.
The above research was conducted under the assumption that the solution is sufficiently smooth; however, the analytical solution of time fractional equations may have singularity at t 0 + [23,24]. Some scholars have also studied the solution of Equation (1) under the assumption of weak regularity. Liu et al. [25] established a fully discrete ADI scheme for model (1), where the time first-order derivative is approximated using the backward Euler scheme and the time fractional derivative is approximated using the L1 scheme, and stability and convergence were analyzed. Liu et al. [26] established a characteristic finite element scheme for the time fractional mobile/mobile advection diffusion model, where the optimal L2 error estimation has first-order accuracy in time and second-order accuracy in space. Nong and Chen [27] established two finite element schemes for the two-dimensional fractional mobile/immobile transport equation based on convolutional orthogonal time discretization using the backward Euler (BE) and second-order backward difference (SBD) methods and provided corresponding error estimates for the two schemes. Nong et al. [28] developed an effective Crank–Nicolson compact difference scheme for the two-dimensional time fractional mobile/immobile transport equation based on the improved L1 method and fast discrete sine transform (DST) technology, increasing the convergence order by adding correction terms. Zhang et al. [29] established a weighted shifted Grünwald–Letnikov difference (WSGD) Legendre spectral scheme for the two-dimensional nonlinear fractional mobile/immobile convection diffusion equation. Yin et al. [30] used generalized B D F 2 θ to discretize time derivatives, used the Galerkin finite element method to discretize spatial derivatives, and added correction terms to develop time semidiscrete and fully discrete schemes for the fractional mobile/immobile transfer equation. Zheng and Wang [31] assumed that the problem (1) has a unique solution u C ( Ω ¯ × [ 0 , T ] ) , which has weak singularity at the initial moment; that is, there exists a constant C, such that
( a ) | x 4 u ( x , t ) | C , ( x , t ) [ 0 , L ] × [ 0 , T ] , ( b ) | t l u ( x , t ) | C ( 1 + t α + 1 l ) , ( x , t ) [ 0 , L ] × [ 0 , T ] , l = 0 , 1 , 2 , 3 , ( c ) | t l x 2 u ( x , t ) | C ( 1 + t α + 1 l ) , ( x , t ) [ 0 , L ] × [ 0 , T ] , l = 0 , 1 , 2 ,
An average L1 compact difference scheme is established for Equation (1) on time uniform meshes, and it is proved that the scheme unconditionally converges with a convergence order of O ( τ 2 | ln τ | + h 4 ) for all α ( 0 , 1 ) , where τ and h are the sizes of the temporal and spatial steps, respectively.
In this article, we assume that the solution of Equation (1) satisfies the assumption (3) and has weak singularity. We use the Crank–Nicolson scheme to approximate the first-order and fractional temporal derivatives on uniform meshes and the center difference quotient to approximate the spatial derivatives on uniform meshes, thus constructing a fully discrete scheme for Equation (1) and estimating the local truncation error. Then, we obtain the optimal error estimates by energy analysis method. In addition, we employ an exponential summation approximation of the weak singular kernel in the time fractional derivative term to reduce the computational and storage requirements from O ( M N 2 ) and O ( M N ) to O ( M N N ^ ) and O ( M N ^ ) , respectively, where N and M represent the total number of grid points in time and space, respectively, N ^ = O ( log 2 N ) for T 1 , and N ^ = O ( log N ) for T > > 1 .
For problems exhibiting weak singularity solutions at the initial moment, most of them are solved discretely on time graded meshes. However, this research chose to discretize on uniform meshes, which simplifies both the theoretical analysis and the numerical computation process. From a theoretical perspective, the order of temporal convergence can reach 1, while the results of numerical simulations demonstrate a performance that surpasses the theoretical predictions.
The rest of this paper is as follows. In Section 2, we construct the fully discrete scheme of Equation (1) and provide local truncation error estimates. In Section 3, we employ the energy analysis method to propose optimal error estimates. In Section 4, we optimize the algorithm using the sum-of-exponentials approximation technique. In Section 5, we confirm the effectiveness of the algorithm through numerical experiments. In the conclusion, we summarize the numerical algorithm and results and present an outlook on future research directions.

2. The Discrete Problem

Choose positive integers M, N, representing the number of spatial and temporal partitions, respectively. Define h = L M as the spatial step size, which yields the spatial discrete node Ω h = { ( x l , y j ) | x l = l h , y j = j h , 0 l , j M } . Define τ = T N as the time step size, obtaining the time discrete node Ω τ = { t n | t n = n τ , n = 0 , 1 , 2 , , N } . Further, define t n + 1 2 = t n + t n + 1 2 . U l , j n , u l , j n denote the numerical solution and the true solution at the grid node ( x l , y j , t n ) , respectively.
The problem (1) is discretized at the point ( x l , y j , t k + 1 2 ) , 1 l , j M 1 , k = 0 , 1 , 2 , 3 , , N 1 . By employing the central difference scheme to approximate the spatial derivative, we obtain
2 u x 2 ( x l , y j , t k + 1 2 ) = 1 2 δ x 2 u l , j k + δ x 2 u l , j k + 1 + R 1 l , j k + 1 2 , 2 u y 2 ( x l , y j , t k + 1 2 ) = 1 2 δ y 2 u l , j k + δ y 2 u l , j k + 1 + R 1 l , j k + 1 2 ,
where
δ x 2 u l , j k : = u l + 1 , j k 2 u l , j k + u l 1 , j k h 2 , δ y 2 u l , j k : = u l , j + 1 k 2 u l , j k + u l , j 1 k h 2 ,
and
R 1 l , j k + 1 2 = O ( h 2 + τ 2 ) .
The derivative u t ( x l , y j , t k + 1 2 ) is approximated by the central difference quotient, which is
u t ( x l , y j , t k + 1 2 ) = δ t u l , j k + R 2 l , j k + 1 2 ,
where
δ t u l , j k : = u l , j k + 1 u l , j k τ ,
and
R 2 l , j k + 1 2 = O ( τ 2 ) .
The time fractional derivative D t α 0 C u ( x l , y j , t k + 1 2 ) is approximated by the Crank–Nicolson L1 formula [32]; we obtain
D t α 0 C u ( x l , y j , t 1 2 ) = σ u l , j 1 u l , j 0 2 1 α + R 3 l , j 1 2 : = D N α u l , j 1 2 + R 3 l , j 1 2 , k = 0 , D t α 0 C u ( x l , y j , t k + 1 2 ) = w 1 u l , j k + m = 1 k 1 w k m + 1 w k m u l , j m w k u l , j 0 + σ u l , j k + 1 u l , j k 2 1 α + R 3 l , j k + 1 2 : = D N α u l , j k + 1 2 + R 3 l , j k + 1 2 , k = 1 , 2 , , N 1 .
where σ = τ α Γ ( 2 α ) , and
w m = σ m + 1 2 1 α m 1 2 1 α , m = 1 , 2 , , N 1 .
If the solution of the problem (1) is sufficiently smooth, we have R 3 l , j k + 1 2 = O ( τ 2 α ) , k = 0 , 1 , 2 , , N 1 . However, in this paper, we consider that the solution has weak singularity at t 0 + , which satisfies the assumption (3). Next, we prove the local truncation error R 3 l , j k + 1 2 = O ( τ ) .
The u is approximated by linear interpolation on interval [ t m 1 , t m ] ; we have
L 1 , m u = t m s τ u m 1 + s t m 1 τ u m , s [ t m 1 , t m ] ,
then
u L 1 , m u = 1 2 u ( η m ) ( s t m 1 ) ( s t m ) , η m [ t m 1 , t m ] .
| D N α u l , j k + 1 2 D t α 0 C u ( x l , y j , t k + 1 2 ) | = m = 0 k | T k , m | , k = 0 , 1 , , N 1 .
For k = 0 , we have
| T 0 , 0 | = | 1 Γ ( 1 α ) t 0 t 1 2 u 1 u 0 τ u s t 1 2 s α d s | | T 0 , 0 1 | + | T 0 , 0 2 | .
According to the hypothesis (3) and the partial integration, we obtain
T 0 , 0 1 = 1 Γ ( 1 α ) t 0 t 1 2 u 1 u 0 τ t 1 2 s α d s = 1 Γ ( 1 α ) t 0 t 1 2 1 τ t 0 t 1 u t d t t 1 2 s α d s C τ α t 0 t 1 2 t 1 2 s α d s C τ
and
T 0 , 0 2 = 1 Γ ( 1 α ) t 0 t 1 2 u s t 1 2 s α d s C τ α t 0 t 1 2 t 1 2 s α d s C τ
For k = 1 , 2 , , N 1 , we have
| T k , 0 | = | 1 Γ ( 1 α ) t 0 t 1 u 1 u 0 τ u s t k + 1 2 s α d s | | T k , 0 1 | + | T k , 0 2 | .
According to the hypothesis (3) and the partial integration, we obtain
T k , 0 1 = 1 Γ ( 1 α ) t 0 t 1 u 1 u 0 τ t k + 1 2 s α d s = 1 Γ ( 1 α ) t 0 t 1 1 τ t 0 t 1 u t d t t k + 1 2 s α d s C τ α t 0 t 1 t k + 1 2 s α d s C τ α τ t k + 1 2 t 1 α C τ ,
and
T k , 0 2 = 1 Γ ( 1 α ) t 0 t 1 u s t k + 1 2 s α d s C τ α t 0 t 1 t k + 1 2 s α d s C τ 1 + α t k + 1 2 t 1 α C τ .
In addition, by the assumption (3), the Equation (11) and partial integration, we obtain
m = 1 k 2 1 | T k , m | = m = 1 k 2 1 | 1 Γ ( 1 α ) t m t m + 1 1 2 u ( η m ) ( s t m ) ( s t m + 1 ) t k + 1 2 s α d s | m = 1 k 2 1 | 1 Γ ( 1 α ) t m t m + 1 u ( η m ) ( s t m + 1 2 ) t k + 1 2 s α d s | C m = 1 k 2 1 t m α 1 τ t m t m + 1 t k + 1 2 s α d s C m = 1 k 2 1 τ 2 t m α 1 t k + 1 2 α = C m = 1 k 2 1 τ 2 t m t k + 1 2 α t m 1 C m = 1 k 2 1 τ m 1 C τ ,
and
m = k 2 k 1 | T k , m | = m = k 2 k 1 | 1 Γ ( 1 α ) t m t m + 1 1 2 u ( η m ) ( s t m ) ( s t m + 1 ) t k + 1 2 s α d s | m = k 2 k 1 | 1 Γ ( 1 α ) t m t m + 1 u ( η m ) ( s t m + 1 2 ) t k + 1 2 s α d s | C t k α 1 τ t k 2 t k 1 t k + 1 2 s α d s C τ t k α 1 τ α τ C τ .
Finally, let us consider T k , k , for k = 1 , 2 , , N 1 , still from the assumption (3), the Equation (11) and partial integration, we obtain
| T k , k | = | 1 Γ ( 1 α ) t k t k + 1 2 1 2 u ( η k ) ( s t k ) ( s t k + 1 ) t k + 1 2 s α d s | | 1 2 Γ ( 1 α ) u ( η m ) t k t k + 1 2 t k + 1 2 s α d ( s t k ) ( s t k + 1 ) | C t k α 1 τ 2 t k t k + 1 2 t k + 1 2 s α 1 d s C t k α 1 τ 2 α C τ .
Thus
| D N α u l , j k + 1 2 D t α 0 C u ( x l , y j , t k + 1 2 ) | C τ .
To sum up, the discretization of the problem (1) at point ( x l , y j , t k + 1 2 ) , l , j = 1 , 2 , , M , k = 0 , 1 , 2 , 3 , , N 1 is obtained
u l , j 1 u l , j 0 τ + σ u l , j 1 u l , j 0 2 1 α f l , j 1 2 = 1 2 δ x 2 u l , j 0 + δ x 2 u l , j 1 + δ y 2 u l , j 0 + δ y 2 u l , j 1 + R l , j 1 2 , k = 0 , u l , j k + 1 u l , j k τ + w 1 u l , j k + m = 1 k 1 ( w k m + 1 w k m ) u l , j m w k u l , j 0 + σ u l , j k + 1 u l , j k 2 1 α f l , j k + 1 2 = 1 2 δ x 2 u l , j k + δ x 2 u l , j k + 1 + δ y 2 u l , j k + δ y 2 u l , j k + 1 + R l , j k + 1 2 , k = 1 , 2 , , N 1 .
where R l , j k + 1 2 = R 1 l , j k + 1 2 + R 2 l , j k + 1 2 + R 3 l , j k + 1 2 = O ( τ + h 2 ) , k = 0 , 1 , 2 , , N 1 .
Omitting truncation errors and replacing the exact solution u l , j k with the numerical solution U l , j k , we obtain the discrete scheme for Equation (1) as
U l , j 1 U l , j 0 τ + σ U l , j 1 U l , j 0 2 1 α f l , j 1 2 = 1 2 δ x 2 U l , j 0 + δ x 2 U l , j 1 + δ y 2 U l , j 0 + δ y 2 U l , j 1 , k = 0 , U l , j k + 1 U l , j k τ + w 1 U l , j k + m = 1 k 1 w k m + 1 w k m U l , j m w k U l , j 0 + σ U l , j k + 1 U l , j k 2 1 α f l , j k + 1 2 = 1 2 δ x 2 U l , j k + δ x 2 U l , j k + 1 + δ y 2 U l , j k + δ y 2 U l , j k + 1 , k = 1 , 2 , , N 1 .

3. Error Estimates

For convenience in error analysis, the schemes (12) and (13) are organized as follows.
δ t u l , j k + 1 + L k + 1 u l , j k + 1 = δ x 2 u l , j k + 1 2 + δ y 2 u l , j k + 1 2 + f l , j k + 1 2 + R l , j k + 1 2 , 1 l , j M 1 , k = 0 , 1 , 2 , , N 1 ,
and
δ t U l , j k + 1 + L k + 1 U l , j k + 1 = δ x 2 U l , j k + 1 2 + δ y 2 U l , j k + 1 2 + f l , j k + 1 2 , 1 l , j M 1 , k = 0 , 1 , 2 , , N 1 ,
where
L k + 1 v l , j k + 1 = m = 1 k + 1 ω k , m δ t v l , j m , 0 k N 1 , 1 m k + 1 ,
with
ω k , m = t m 1 min { t m , t k + 1 2 } t k + 1 2 s α Γ ( 1 α ) d s , 0 k N 1 , 1 m k + 1 .
Lemma 1.
For any grid function v k , 0 k N , when 0 < α < 1 log 3 2 , set τ < 2 Γ ( 2 α ) 3 1 α 2 1 1 α , when 1 log 3 2 α < 1 , for any τ > 0 , we have
1 2 δ t ( v k ) 2 + L k ( v k ) 2 v k δ t v k + L k v k , 1 k N .
Proof. 
Let ω ^ k , k + 1 = 1 + ω k , k + 1 , and ω ^ k , m = ω k , m , 1 m k . Through calculation, we have
ω ^ k , k + 1 ω ^ k , k = 1 + τ 1 α 2 1 α Γ ( 1 α ) 2 3 1 α ,
Furthermore, due to
τ Γ ( 1 α ) t k + 1 2 t m 1 α < ω ^ k , m = t m 1 t m ( t k + 1 2 s ) α Γ ( 1 α ) d s < τ Γ ( 1 α ) t k + 1 2 t m α ,
It is easy to prove that when 0 < α < 1 log 3 2 , setting τ < 2 Γ ( 2 α ) 3 1 α 2 1 1 α , and when 1 log 3 2 α < 1 , for any τ > 0 , we have
ω ^ k , k + 1 > ω ^ k , k > > ω ^ k , 1 > 0 , 0 k N 1 .
Then, we prove Lemma 1 using the same method as in the proof of Lemma 3.1 in [31], namely
1 2 δ t ( v k ) 2 + L k ( v k ) 2 v k δ t v k + L k v k , 1 k N .
Let Ω ¯ h = { ( x l , y j ) | 0 l , j M } , Ω h = Ω ¯ h Ω , Ω h = Ω ¯ h Ω , I = { ( l , j ) | ( x l , y j ) Ω h } , I = { ( l , j ) | ( x l , y j ) Ω h } , I ¯ = I I . Let S h = { v | v = v ( x l , y j ) , ( l , j ) I ¯ which is the mesh function on Ω ¯ h , and S ˚ h = { v | v S h , v l , j = 0 , ( l , j ) I } . For any grid function v S h , we introduce the following notation:
δ x v l 1 2 , j = 1 h v l , j v l 1 , j , δ x 2 v l , j = 1 h δ x v l + 1 2 , j δ x v l 1 2 , j , δ y v l , j 1 2 = 1 h v l , j v l , j 1 , δ y 2 v l , j = 1 h δ y v l , j + 1 2 δ y v l , j 1 2 .
For the grid function v , u S ˚ h , we define the following inner products and norms:
( v , u ) = h 2 l = 1 M 1 j = 1 M 1 v l , j u l , j , | | v | | = ( v , v ) 1 2 , ( v , u ) 1 , x = h 2 l = 1 M 1 j = 1 M 1 δ x v l 1 2 , j δ x u l 1 2 , j , | | δ x v | | = ( u , u ) 1 , x 1 2 , ( v , u ) 1 , y = h 2 l = 1 M 1 j = 1 M 1 δ y v l , j 1 2 δ y u l , j 1 2 , | | δ y v | | = ( u , u ) 1 , y 1 2 , ( v , u ) 2 , x = h 2 l = 1 M 1 j = 1 M 1 δ x 2 v l , j δ x 2 u l , j , | | δ x 2 v | | = ( v , v ) 2 , x 1 2 , ( v , u ) 2 , y = h 2 l = 1 M 1 j = 1 M 1 δ y 2 v l , j δ y 2 u l , j , | | δ y 2 v | | = ( v , v ) 2 , y 1 2 , | | h v | | = | | δ x v | | 2 + | | δ y v | | 2 , | | Δ h v | | = | | δ x 2 v | | 2 + | | δ y 2 v | | 2 .
It is easy to verify that for arbitrary grid function v , u S ˚ h , we have
( δ x 2 v , u ) : = h 2 l = 1 M 1 j = 1 M 1 δ x 2 v l , j u l , j = ( u , v ) 1 , x , ( δ y 2 v , u ) : = h 2 l = 1 M 1 j = 1 M 1 δ y 2 v l , j u l , j = ( u , v ) 1 , y .
Theorem 1.
Assume that the solution of the problem (1) satisfies the assumption (3), u l , j k and U l , j k are the solutions of (14) and (15), respectively. τ < 2 Γ ( 2 α ) 3 1 α 2 1 1 α , for 0 < α < 1 log 3 2 , and τ > 0 , for 1 log 3 2 α < 1 , then
| | u k U k | | C T ( τ + h 2 ) , 1 k N .
Proof. 
Let e l , j k = u l , j k U l , j k , then e l , j k = 0 , 0 k N 1 , ( l , j ) I , and e l , j 0 = 0 , ( l , j ) I ¯ . Subtracting Equation (15) from Equation (14) yields the error equation
δ t e l , j k + 1 + L k + 1 e l , j k + 1 = δ x 2 e l , j k + 1 2 + δ y 2 e l , j k + 1 2 + R l , j k + 1 2 , 1 l , j M 1 , k = 0 , 1 , 2 , , N 1 .
Multiply both sides of Equation (18) by h 2 e l , j k + 1 , and sum l and j from 1 to M 1 ; we obtain
δ t | | e k + 1 | | 2 + L k + 1 | | e k + 1 | | 2 δ x 2 ( e k + 1 + e k ) , e k + 1 + δ y 2 ( e k + 1 + e k ) , e k + 1 + 2 R k + 1 2 , e k + 1 .
Due to ω k , m + 1 = ω k 1 , m , ( 1 m k , 1 k N 1 ) , we obtain
L k + 1 | | e k + 1 | | 2 = 1 τ m = 1 k + 1 ω k , m | | e m | | 2 | | e m 1 | | 2 = 1 τ m = 1 k + 1 ω k , m | | e m | | 2 m = 1 k ω k 1 , m | | e m | | 2 .
In addition, based on the inverse estimate h | | Δ h v | | 2 | | h v | | and Young’s inequality, we obtain
δ x 2 ( e k + 1 + e k ) , e k + 1 + δ y 2 ( e k + 1 + e k ) , e k + 1 + 2 R k + 1 2 , e k + 1 | | δ x e k + 1 | | 2 | | δ y e k + 1 | | 2 + | | δ x 2 e k | | + | | δ y 2 e k | | | | e k + 1 | | + 2 | | R k + 1 2 | | | | e k + 1 | | | | h e k + 1 | | 2 + h 2 8 | | δ x 2 e k | | + | | δ y 2 e k | | 2 + 2 h 2 | | e k + 1 | | 2 + 2 | | R k + 1 2 | | | | e k + 1 | | | | h e k + 1 | | 2 + h 2 4 | | Δ h e k | | 2 + 2 h 2 | | e k + 1 | | 2 + 2 | | R k + 1 2 | | | | e k + 1 | | | | h e k + 1 | | 2 + | | h e k | | 2 + 2 h 2 | | e k + 1 | | 2 + 2 | | R k + 1 2 | | | | e k + 1 | | ,
Let E 0 = 0 , and
E k + 1 = 1 τ | | e k + 1 | | 2 + 1 τ m = 1 k + 1 ω k , m | | e m | | 2 + | | h e k + 1 | | 2 .
Combining the Formulas (19)–(22), we obtain
E k + 1 E k + 2 h 2 | | e k + 1 | | + 2 | | R k + 1 2 | | | | e k + 1 | | m = 0 k 2 h 2 | | e m + 1 | | + 2 | | R m + 1 2 | | max 0 m k + 1 | | e m | | .
Based on the definition of E k + 1 and Formula (23), we obtain
| | e k + 1 | | τ m = 0 k 2 h 2 | | e m + 1 | | + 2 | | R m + 1 2 | | ,
Thus, according to the estimate of local truncation error R m + 1 2 and the discrete G r o ¨ n w a l l s inequality, we obtain
| | e k + 1 | | C k τ τ + h 2 C T τ + h 2 .

4. Fast Algorithms

In this section, we utilize the sum-of-exponentials approximation technique to optimize numerical scheme (13). We first present a lemma.
Lemma 2 
([33]). For a given β ( 0 , 2 ) , absolute tolerance error ϵ 1 , cut-off time step size Δ t > 0 and final time T, there exists a positive number N ^ , positive orthogonal nodes s i , and corresponding positive weights w i , such that
| 1 t β i = 1 N ^ w i e s i t | ϵ , t [ Δ t , T ] ,
where the number of orthogonal nodes N ^ satisfies
N ^ = O log 1 ϵ l o g l o g 1 ϵ + l o g T Δ t + log 1 Δ t l o g l o g 1 ϵ + l o g 1 Δ t .
More precisely, the calculation of the Caputo derivative at time [ 0 , t k + 1 2 ] is divided into a historical part (the integral on [ 0 , t k ] ) and a local part (the integral on [ t k , t k + 1 2 ] ). The local part will be approximated directly through linear interpolation, while the historical part can be handled using sum-of-exponentials approximation technology, i.e.,
D t α 0 FC u ( t k + 1 2 ) = 1 Γ ( 1 α ) 0 t k + 1 2 u ( s ) t k + 1 2 s α d s = 1 Γ ( 1 α ) t k t k + 1 2 u ( s ) t k + 1 2 s α d s + 1 Γ ( 1 α ) 0 t k u ( s ) t k + 1 2 s α d s 1 Γ ( 2 α ) τ α 2 1 α u k + 1 u k + 1 Γ ( 1 α ) τ 2 α u k t k + 1 2 α u 0 α i = 1 N ^ w i H h i s t , i ( t k + 1 2 )
By employing linear interpolation and recursive formulas, we have
H h i s t , i ( t k + 1 2 ) = e s i τ H h i s t , i ( t k 1 2 ) + t k 1 t k e s i ( t k + 1 2 s ) u ( s ) d s , k = 1 , 2 , , N 1 ,
and
t k 1 t k e s i ( t k + 1 2 s ) u ( s ) d s e s i ( t k + 1 2 t k 1 ) s i 2 τ ( e s i τ 1 s i τ ) u k 1 + ( 1 e s i τ + s i τ e s i τ ) u k ,
with H h i s t , i ( t 1 2 ) = 0 .
Combining Formulas (4), (7) and (26) and omitting errors, we obtain a fast numerical scheme for Equation (1)
U i , j 1 U i , j 0 τ + σ U i , j 1 U i , j 0 2 1 α f i , j 1 2 + U i , j 0 ( τ 2 ) α Γ ( 1 α ) = 1 2 δ x 2 U i , j 0 + δ x 2 U i , j 1 + δ y 2 U i , j 0 + δ y 2 U i , j 1 , k = 0 , U i , j k + 1 U i , j k τ + σ U i , j k + 1 U i , j k 2 1 α + 1 Γ ( 1 α ) τ 2 α U i , j k t k + 1 2 α U i , j 0 α i = 1 N ^ w i H h i s t , i ( t k + 1 2 ) f i , j k + 1 2 + U i , j 0 t k + 1 2 α Γ ( 1 α ) = 1 2 δ x 2 U i , j k + δ x 2 U i , j k + 1 + δ y 2 U i , j k + δ y 2 U i , j k + 1 , k = 1 , 2 , , N 1 .

5. Numerical Results

In this section, we validate the effectiveness of the Crank–Nicolson L1 scheme (CN-L1) (13) and the fast Crank–Nicolson L1 scheme (CN-L1-SOE) (29) through numerical examples. The numerical results include the errors, the convergence orders, and required CPU time (in seconds) for computation. We simulated the examples using Matlab R2020b on a computer with processor of 2.30 GHz and memory 8 GB.
Define maximum error norm, L 2 error norm
L ( M , N ) = max 0 i , j M , 0 k N | u i , j k U i , j k | , L 2 ( M , N ) = max 0 k N 0 i , j M h 2 ( u i , j k U i , j k ) 2
and temporal, spatial convergence orders
r h = log L ( M 1 , N 1 ) / L ( M 2 , N 1 ) log ( M 2 / M 1 ) , r τ = log L ( M 1 , N 1 ) / L ( M 2 , N 2 ) log ( N 2 / N 1 ) .
Example 1.
Consider
u t ( x , y , t ) + D t α 0 C u ( x , y , t ) = 2 u x 2 ( x , y , t ) + 2 u y 2 ( x , y , t ) + f ( x , y , t ) , ( x , y ) ( 0 , 1 ) 2 , t ( 0 , 1 ] , u ( x , y , t ) = ϕ ( x , y , t ) , ( x , y ) Ω , t [ 0 , 1 ] , u ( x , y , 0 ) = g ( x , y ) , ( x , y ) Ω .
Assume that the exact solution is u ( x , y , t ) = 1 + t 1 + α Γ ( 2 + α ) x 2 ( 1 x ) 2 y 2 ( 1 y ) 2 , which satisfies the assumption (3). The corresponding f ( x , y , t ) , g ( x , y ) and ϕ ( x , y , t ) can be derived, where
f ( x , y , t ) = t α Γ ( 1 + α ) + t Γ ( 2 ) x 2 ( 1 x ) 2 y 2 ( 1 y ) 2 1 + t 1 + α Γ ( 2 + α ) 12 x 2 12 x + 2 y 2 ( 1 y ) 2 1 + t 1 + α Γ ( 2 + α ) 12 y 2 12 y + 2 x 2 ( 1 x ) 2 .
Setting α = 0.6 , by selecting a larger time subdivisions N = 5000 , ensures that the time step size τ = 1 / 5000 is sufficiently small, which is dominated by the spatial error in the numerical simulation. Different numbers of spatial subdivisions M = 2 3 , 2 4 , 2 5 , 2 6 , 2 7 , 2 8 are chosen, and Example 1 was solved using both the Crank–Nicolson L1 scheme (13) and the fast Crank–Nicolson L1 scheme (29). Table 1 lists the L ( M , N ) and L 2 ( M , N ) errors, the spatial convergence orders, and CPU times (in seconds) required by both methods. The data indicate that the spatial convergence orders of both numerical schemes are approximately 2, and that the fast Crank–Nicolson L1 scheme (29) takes significantly less time than the Crank–Nicolson L1 scheme (13). However, due to the data being retained only to four decimal places, the specific differences between the errors and convergence orders calculated by the two algorithms are not displayed. Figure 1 shows that the errors L ( M , N ) and L 2 ( M , N ) calculated by the two algorithms are extremely close, with their magnitudes being only 10 13 ; the differences in convergence orders are also small, with magnitudes of 10 6 .
Setting M = N , then h = τ , the error O ( τ + h 2 ) = O ( τ ) . Choosing different α = 0.2 , 0.4 , 0.6 , 0.8 , we use the Crank–Nicolson L1 scheme (13) and the fast Crank–Nicolson L1 scheme (29) to simulate Example 1. The L ( M , N ) , L 2 ( M , N ) errors, the time convergence orders, and CPU time are shown in Table 2, Table 3, Table 4 and Table 5. The data indicate that for different values of α , both schemes exhibit the second-order convergence rate, which is higher than the theoretical analysis, and that the time required for the fast Crank–Nicolson L1 scheme (29) is smaller than the Crank–Nicolson L1 scheme (13) computation, but the time difference is not significant due to the small number of time divisions selected. Furthermore, Figure 2 shows that the absolute error between the errors (convergence rates) calculated by the two schemes is very small.
Next, we fix the number of spatial partitions M = 10 and α = 0.6 and select larger temporal partitions N. We then use the Crank–Nicolson L1 scheme (13) and the fast Crank–Nicolson L1 scheme (29) to simulate Example 1. The changes in CPU time with respect to the number of temporal partitions are listed in Table 6, and an image of CPU time versus temporal partitions is plotted as shown in Figure 3. The data in Table 6 show that when N = 24,000, the Crank–Nicolson L1 scheme (13) takes about 287 s, while the fast Crank–Nicolson L1 scheme (29) only takes over 4 s. In addition, Figure 3 showsthat the CPU time calculated by the Crank–Nicolson L1 scheme is quadratic in the number of partitions N, while the CPU time required by the fast Crank–Nicolson L1 scheme numerical simulation increases linearly with respect to N.
Example 2.
Consider
u t ( x , y , t ) + D t α 0 C u ( x , y , t ) = 2 u x 2 ( x , y , t ) + 2 u y 2 ( x , y , t ) + f ( x , y , t ) , ( x , y ) ( 0 , 1 ) 2 , t ( 0 , 1 ] , u ( x , y , t ) = ϕ ( x , y , t ) , ( x , y ) Ω , t [ 0 , 1 ] , u ( x , y , 0 ) = g ( x , y ) , ( x , y ) Ω .
Assume that the exact solution is u ( x , y , t ) = 1 + t 1 + α Γ ( 2 + α ) sin π x sin π y , which satisfies the assumption (3). The corresponding f ( x , y , t ) , g ( x , y ) and ϕ ( x , y , t ) can be derived, where
f ( x , y , t ) = t α Γ ( 1 + α ) + t Γ ( 2 ) sin π x sin π y + 2 π 2 sin π x sin π y 1 + t 1 + α Γ ( 2 + α ) .
Setting M = N = 30 , α = 0.6 , the true solution, the numerical solutions computed using the schemes (13) and (29), the absolute errors between the numerical solution and the true solution, and the absolute errors between the two numerical solutions at t = 1 are plotted in Figure 4. The figure indicates that both numerical solutions approximate the true solution well, and the differences between the two numerical solutions are very small, with magnitudes of 10 10 .
Setting N = 5000 , choosing different α = 0.2 , 0.4 , 0.6 , 0.8 and M = 2 4 , 2 5 , 2 6 , 2 7 , 2 8 , we solve Example 2 by the fast Crank–Nicolson L1 scheme (29). The computed L ( M , N ) and L 2 ( M , N ) errors as a function of M are depicted in Figure 5. The figure indicates that for different α , as the spatial discretization is refined, the error gradually decreases, with a spatial convergence order of approximately 2.
Setting M = N , choosing α = 0.2 , 0.4 , 0.6 , 0.8 , we employ the fast Crank–Nicolson L1 scheme (29) to simulate Example 2, and the computed L 2 ( M , N ) errors and the convergence orders are shown in Table 7. The data show that for α = 0.6 , 0.8 , second-order time convergence is achieved, which exceeds the theoretical convergence order; when α is smaller, as the mesh is refined, the temporal convergence order tends to be first order, consistent with the theoretical analysis. In addition, we used the fully discrete alternating direction implicit (ADI) scheme established in [25] to calculate Example 2. The obtained results are presented in Table 8. By comparing with the data in Table 7, it can be observed that the algorithm constructed in this paper yields more accurate computation results.
Next, the algorithm established in this paper calculates Example 2 on time-graded meshes. The time node set is given by Ω τ = { t n | t n = T ( n N ) r , n = 0 , 1 , 2 , , N } , where r is the parameter of the graded meshes. We set M = N and α = 0.6 , selecting different values for r. The results are presented in Table 9. Comparing with the results in Table 7 for α = 0.6 , we find that due to the weak singularity of the solution at the initial moment, selecting an appropriate parameter, such as r = 2 or 4, leads to time nodes that are suitably close to t = 0 , resulting in more accurate calculations compared to those obtained on uniform meshes. However, when the value of r is too large, the distribution of time nodes becomes severely uneven, leading to increased errors.

6. Conclusions

In this paper, we establish a Crank–Nicolson L1 scheme on uniform meshes for the two-dimensional time fractional mobile/immobile equation with weak singularity at the initial moment. The error is estimated using the energy analysis method. The obtained error estimate has a convergence order of O ( h 2 + τ ) , but the numerical simulation calculations yield a spatial convergence order of 2, and the temporal convergence order is superior to the first order as theoretically analyzed.
Furthermore, considering the historical dependence of the time fractional derivative, the numerical simulation of the Crank–Nicolson L1 scheme requires a large amount of memory and expensive computation; we adopted the sum-of-exponentials approximation technology to discretize the fractional derivative and established a fast Crank–Nicolson L1 scheme. Through numerical simulation, it can be found that when the number of time steps N is large, the efficiency of the fast algorithm is significantly higher than the Crank–Nicolson L1 scheme. In addition, the difference between the calculated results of the fast Crank–Nicolson L1 scheme (29) and the Crank–Nicolson L1 scheme (13) is particularly small, but there is no strict error estimate for the fast Crank–Nicolson L1 scheme (29), which we will study in the future. The algorithm proposed in this paper is also applicable to other fractional models and provides a technical foundation for solving the practical problems described by model (1).
The time fractional mobile/immobile models have wide applications in various fields, such as modeling groundwater solute transport, describing thermal diffusion, and ocean sound propagation. Conducting effective numerical algorithm research on these models can promote the solution of practical problems. This paper has systematically investigated the two-dimensional time fractional mobile/immobile model. However, considering that the three-dimensional model is closer to the actual situation, we will further consider extending algorithms to three-dimensional models based on this research.

Author Contributions

Methodology, H.Q.; Validation, H.Q.; Formal analysis, H.Q.; Writing—original draft, H.Q.; Writing—review & editing, A.C.; Supervision, H.Q. and A.C. All authors have read and agreed to the published version of the manuscript.

Funding

This work was supported in part by the Natural Science Foundation of Shandong Province, Grant ZR2022QA038, and the Doctoral Research Foundation of Liaocheng University, Grant 318052155.

Data Availability Statement

The original contributions presented in this study are included in the article. Further inquiries can be directed to the corresponding author.

Conflicts of Interest

The authors declare that they have no competing interests.

References

  1. Baleanu, D.; Güvenç, Z.B.; Machado, J.T. New Trends in Nanotechnology and Fractional Calculus Applications; Springer: Berlin/Heidelberg, Germany, 2010; Volume 10. [Google Scholar]
  2. Baleanu, D.; Machado, J.A.T.; Luo, A.C. Fractional Dynamics and Control; Springer Science & Business Media: Berlin/Heidelberg, Germany, 2011. [Google Scholar]
  3. Babaei, A.; Jafari, H.; Ahmadi, M. A fractional order HIV/AIDS model based on the effect of screening of unaware infectives. Math. Methods Appl. Sci. 2019, 42, 2334–2343. [Google Scholar] [CrossRef]
  4. Cao, J.; Xu, W. Adaptive-coefficient finite difference frequency domain method for time fractional diffusive-viscous wave equation arising in geophysics. Appl. Math. Lett. 2025, 160, 109337. [Google Scholar] [CrossRef]
  5. Cao, J.; Li, S. Two implicit-explicit difference schemes for a time-fractional viscous wave equation arising in geophysics. Fractals 2025. [Google Scholar] [CrossRef]
  6. Kumar, D.; Singh, J.; Tanwar, K.; Baleanu, D. A new fractional exothermic reactions model having constant heat source in porous media with power, exponential and Mittag-Leffler laws. Int. J. Heat Mass Transf. 2019, 138, 1222–1227. [Google Scholar] [CrossRef]
  7. Golbabai, A.; Nikan, O.; Nikazad, T. Numerical analysis of time fractional Black–Scholes European option pricing model arising in financial market. Comput. Appl. Math. 2019, 38, 1–24. [Google Scholar] [CrossRef]
  8. Podlubny, I. An introduction to fractional derivatives, fractional differential equations, to methods of their solution and some of their applications. Math. Sci. Eng. 1999, 198, 340. [Google Scholar]
  9. Uchaikin, V.V. Fractional Derivatives for Physicists and Engineers; Springer: Berlin/Heidelberg, Germany, 2013; Volume 2. [Google Scholar]
  10. Atangana, A.; Baleanu, D. Numerical solution of a kind of fractional parabolic equations via two difference schemes. Abstr. Appl. Anal. 2013, 2013, 828764. [Google Scholar] [CrossRef]
  11. Schumer, R.; Benson, D.A.; Meerschaert, M.M.; Baeumer, B. Fractal mobile/immobile solute transport. Water Resour. Res. 2003, 39. [Google Scholar] [CrossRef]
  12. Zhang, Y.; Benson, D.A.; Reeves, D.M. Time and space nonlocalities underlying fractional-derivative models: Distinction and literature review of field applications. Adv. Water Resour. 2009, 32, 561–581. [Google Scholar] [CrossRef]
  13. Benson, D.A.; Meerschaert, M.M. A simple and efficient random walk solution of multi-rate mobile/immobile mass transport equations. Adv. Water Resour. 2009, 32, 532–539. [Google Scholar] [CrossRef]
  14. Hansen, S.K.; Berkowitz, B. Modeling non-Fickian solute transport due to mass transfer and physical heterogeneity on arbitrary groundwater velocity fields. Water Resour. Res. 2020, 56, e2019WR026868. [Google Scholar]
  15. Zhang, Y.; Zhou, D.; Yin, M.; Sun, H.; Wei, W.; Li, S.; Zheng, C. Nonlocal transport models for capturing solute transport in one-dimensional sand columns: Model review, applicability, limitations and improvement. Hydrol. Process. 2020, 34, 5104–5122. [Google Scholar] [CrossRef]
  16. Liu, Z.; Li, X.; Zhang, X. A fast high-order compact difference method for the fractal mobile/immobile transport equation. Int. J. Comput. Math. 2020, 97, 1860–1883. [Google Scholar]
  17. Fardi, M.; Ghasemi, M. A numerical solution strategy based on error analysis for time-fractional mobile/immobile transport model. Soft Comput. 2021, 25, 11307–11331. [Google Scholar]
  18. Nikan, O.; Machado, J.T.; Golbabai, A.; Nikazad, T. Numerical approach for modeling fractal mobile/immobile transport model in porous and fractured media. Int. Commun. Heat Mass Transf. 2020, 111, 104443. [Google Scholar]
  19. Zhao, J.; Fang, Z.; Li, H.; Liu, Y. Finite volume element method with the WSGD formula for nonlinear fractional mobile/immobile transport equations. Adv. Differ. Equ. 2020, 2020, 360. [Google Scholar]
  20. Jiang, H.; Xu, D.; Qiu, W.; Zhou, J. An ADI compact difference scheme for the two-dimensional semilinear time-fractional mobile–immobile equation. Comput. Appl. Math. 2020, 39, 1–17. [Google Scholar]
  21. Qiu, W.; Xu, D.; Guo, J.; Zhou, J. A time two-grid algorithm based on finite difference method for the two-dimensional nonlinear time-fractional mobile/immobile transport model. Numer. Algorithms 2020, 85, 39–58. [Google Scholar] [CrossRef]
  22. Qiao, H.; Cheng, A. A fast finite difference method for 2D time variable fractional mobile/immobile equation. J. Appl. Math. Comput. 2024, 70, 551–577. [Google Scholar]
  23. McLean, W.; Mustapha, K. A second-order accurate numerical method for a fractional wave equation. Numer. Math. 2007, 105, 481–510. [Google Scholar]
  24. Stynes, M.; O’Riordan, E.; Gracia, J.L. Error analysis of a finite difference method on graded meshes for a time-fractional diffusion equation. SIAM J. Numer. Anal. 2017, 55, 1057–1079. [Google Scholar]
  25. Liu, W.; Chen, H.; Zaky, M. Error analysis of an ADI scheme for the two-dimensional fractal mobile/immobile transport equation with weakly singular solutions. Appl. Numer. Math. 2025, 210, 113–122. [Google Scholar] [CrossRef]
  26. Liu, H.; Zheng, X.; Chen, C.; Wang, H. A characteristic finite element method for the time-fractional mobile/immobile advection diffusion model. Adv. Comput. Math. 2021, 47, 41. [Google Scholar] [CrossRef]
  27. Nong, L.; Chen, A. Numerical schemes for the time-fractional mobile/immobile transport equation based on convolution quadrature. J. Appl. Math. Comput. 2022, 68, 199–215. [Google Scholar] [CrossRef]
  28. Nong, L.; Chen, A.; Yi, Q.; Li, C. Fast Crank-Nicolson compact difference scheme for the two-dimensional time-fractional mobile/immobile transport equation. AIMS Math. 2021, 6, 6242–6254. [Google Scholar] [CrossRef]
  29. Zhang, H.; Jiang, X.; Liu, F. Error analysis of nonlinear time fractional mobile/immobile advection-diffusion equation with weakly singular solutions. Fract. Calc. Appl. Anal. 2021, 24, 202–224. [Google Scholar]
  30. Yin, B.; Liu, Y.; Li, H. A class of shifted high-order numerical methods for the fractional mobile/immobile transport equations. Appl. Math. Comput. 2020, 368, 124799. [Google Scholar]
  31. Zheng, Z.; Wang, Y. An averaged L1-type compact difference method for time-fractional mobile/immobile diffusion equations with weakly singular solutions. Appl. Math. Lett. 2022, 131, 108076. [Google Scholar] [CrossRef]
  32. Karatay, I.; Kale, N.; Bayramoglu, S. A new difference scheme for time fractional heat equations based on the Crank-Nicholson method. Fract. Calc. Appl. Anal. 2013, 16, 892–910. [Google Scholar]
  33. Jiang, S.; Zhang, J.; Zhang, Q.; Zhang, Z. Fast evaluation of the Caputo fractional derivative and its applications to fractional diffusion equations. Commun. Comput. Phys. 2017, 21, 650–678. [Google Scholar]
Figure 1. The absolute error between the errors (the spatial convergence orders) calculated by the two algorithms.
Figure 1. The absolute error between the errors (the spatial convergence orders) calculated by the two algorithms.
Fractalfract 09 00204 g001
Figure 2. The absolute error between the errors (the spatial convergence orders) calculated by the two algorithms.
Figure 2. The absolute error between the errors (the spatial convergence orders) calculated by the two algorithms.
Fractalfract 09 00204 g002
Figure 3. The log-log plot of the CPU time (in seconds) versus the total number of time steps N with M = 10 , α = 0.6 .
Figure 3. The log-log plot of the CPU time (in seconds) versus the total number of time steps N with M = 10 , α = 0.6 .
Fractalfract 09 00204 g003
Figure 4. The exact solution, numerical solution, absolute error between the numerical and exact solutions, and absolute error between the two numerical solutions are presented at t = 1 .
Figure 4. The exact solution, numerical solution, absolute error between the numerical and exact solutions, and absolute error between the two numerical solutions are presented at t = 1 .
Fractalfract 09 00204 g004
Figure 5. The variation in error with M obtained by fast Crank–Nicolson L1 scheme (29) for Example 2.
Figure 5. The variation in error with M obtained by fast Crank–Nicolson L1 scheme (29) for Example 2.
Fractalfract 09 00204 g005
Table 1. The L ( M , N ) , L 2 ( M , N ) errors, convergence rates, and CPU times with N = 5000 , α = 0.6 for Example 1.
Table 1. The L ( M , N ) , L 2 ( M , N ) errors, convergence rates, and CPU times with N = 5000 , α = 0.6 for Example 1.
M CN L 1 CN L 1 SOE
L 2 r h L r h L 2 r h L r h
2 3 1.7696 × 10 4 -3.4297 × 10 4 -1.7696 × 10 4 -3.4297 × 10 4 -
2 4 4.4260 × 10 5 1.99938.5576 × 10 5 2.00284.4260 × 10 5 1.99938.5576 × 10 5 2.0028
2 5 1.1067 × 10 5 1.99982.1385 × 10 5 2.00061.1067 × 10 5 1.99982.1385 × 10 5 2.0006
2 6 2.7668 × 10 6 1.99995.3457 × 10 6 2.00022.7668 × 10 6 1.99995.3457 × 10 6 2.0002
2 7 6.9169 × 10 7 2.00001.3363 × 10 6 2.00016.9169 × 10 7 2.00001.3363 × 10 6 2.0001
2 8 1.7290 × 10 7 2.00023.3404 × 10 7 2.00021.7290 × 10 7 2.00023.3404 × 10 7 2.0002
CPU(s)3379.70311009.9375
Table 2. The L ( M , N ) , L 2 ( M , N ) errors and convergence rates with M = N , α = 0.2 for Example 1.
Table 2. The L ( M , N ) , L 2 ( M , N ) errors and convergence rates with M = N , α = 0.2 for Example 1.
M = N CN L 1 CN L 1 SOE
L 2 r τ L r τ L 2 r τ L r τ
2 3 1.9758 × 10 4 -3.8283 × 10 4 -1.9758 × 10 4 -3.8283 × 10 4 -
2 4 4.9428 × 10 5 1.99909.5467 × 10 5 2.00364.9428 × 10 5 1.99909.5467 × 10 5 2.0036
2 5 1.2360 × 10 5 1.99972.3857 × 10 5 2.00061.2360 × 10 5 1.99972.3857 × 10 5 2.0006
2 6 3.0902 × 10 6 1.99995.9639 × 10 6 2.00013.0902 × 10 6 1.99995.9639 × 10 6 2.0001
2 7 7.7257 × 10 7 1.99991.4910 × 10 6 2.00007.7257 × 10 7 1.99991.4910 × 10 6 2.0000
2 8 1.9316 × 10 7 1.99993.7277 × 10 7 1.99991.9316 × 10 7 1.99993.7277 × 10 7 1.9999
2 9 4.8296 × 10 8 1.99989.3206 × 10 8 1.99984.8296 × 10 8 1.99989.3206 × 10 8 1.9998
C P U ( s ) 1040.8125936.0000
Table 3. The L ( M , N ) , L 2 ( M , N ) errors and convergence rates with M = N , α = 0.4 for Example 1.
Table 3. The L ( M , N ) , L 2 ( M , N ) errors and convergence rates with M = N , α = 0.4 for Example 1.
M = N CN L 1 CN L 1 SOE
L 2 r τ L r τ L 2 r τ L r τ
2 3 1.8598 × 10 4 -3.5984 × 10 4 -1.8598 × 10 4 -3.5984 × 10 4 -
2 4 4.6533 × 10 5 1.99888.9758 × 10 5 2.00334.6533 × 10 5 1.99888.9758 × 10 5 2.0033
2 5 1.1635 × 10 5 1.99982.2428 × 10 5 2.00081.1635 × 10 5 1.99982.2428 × 10 5 2.0008
2 6 2.9082 × 10 6 2.00025.6051 × 10 6 2.00052.9082 × 10 6 2.00025.6051 × 10 6 2.0005
2 7 7.2684 × 10 7 2.00041.4008 × 10 6 2.00057.2684 × 10 7 2.00041.4008 × 10 6 2.0005
2 8 1.8164 × 10 7 2.00063.5004 × 10 7 2.00061.8164 × 10 7 2.00063.5004 × 10 7 2.0006
2 9 4.5387 × 10 8 2.00078.7462 × 10 8 2.00084.5387 × 10 8 2.00078.7462 × 10 8 2.0008
C P U ( s ) 1214.98441002.2813
Table 4. The L ( M , N ) , L 2 ( M , N ) errors and convergence rates with M = N , α = 0.6 for Example 1.
Table 4. The L ( M , N ) , L 2 ( M , N ) errors and convergence rates with M = N , α = 0.6 for Example 1.
M = N CN L 1 CN L 1 SOE
L 2 r τ L r τ L 2 r τ L r τ
2 3 1.7455 × 10 4 -3.3713 × 10 4 -1.7455 × 10 4 -3.3713 × 10 4 -
2 4 4.3664 × 10 5 1.99918.4106 × 10 5 2.00304.3664 × 10 5 1.99918.4106 × 10 5 2.0030
2 5 1.0910 × 10 5 2.00082.1001 × 10 5 2.00181.0910 × 10 5 2.00082.1001 × 10 5 2.0018
2 6 2.7242 × 10 6 2.00185.2422 × 10 6 2.00222.7242 × 10 6 2.00185.2422 × 10 6 2.0022
2 7 6.7969 × 10 7 2.00291.3076 × 10 6 2.00326.7969 × 10 7 2.00291.3076 × 10 6 2.0032
2 8 1.6940 × 10 7 2.00443.2580 × 10 7 2.00491.6940 × 10 7 2.00443.2580 × 10 7 2.0049
2 9 4.2153 × 10 8 2.00678.1033 × 10 8 2.00744.2153 × 10 8 2.00678.1033 × 10 8 2.0074
C P U ( s ) 1164.53131039.2031
Table 5. The L ( M , N ) , L 2 ( M , N ) errors and convergence rates with M = N , α = 0.8 for Example 1.
Table 5. The L ( M , N ) , L 2 ( M , N ) errors and convergence rates with M = N , α = 0.8 for Example 1.
M = N CN L 1 CN L 1 SOE
L 2 r τ L r τ L 2 r τ L r τ
2 3 1.6397 × 10 4 -3.1619 × 10 4 -1.6397 × 10 4 -3.1619 × 10 4 -
2 4 4.0996 × 10 5 1.99997.8864 × 10 5 2.00334.0996 × 10 5 1.99997.8864 × 10 5 2.0033
2 5 1.0231 × 10 5 2.00261.9665 × 10 5 2.00381.0231 × 10 5 2.00261.9665 × 10 5 2.0038
2 6 2.5483 × 10 6 2.00534.8955 × 10 6 2.00612.5483 × 10 6 2.00534.8955 × 10 6 2.0061
2 7 6.3286 × 10 7 2.00961.2149 × 10 6 2.01066.3286 × 10 7 2.00961.2149 × 10 6 2.0106
2 8 1.5638 × 10 7 2.01682.9984 × 10 7 2.01861.5638 × 10 7 2.01682.9984 × 10 7 2.0186
2 9 3.8295 × 10 8 2.02987.3265 × 10 8 2.03303.8295 × 10 8 2.02987.3265 × 10 8 2.0330
C P U ( s ) 1239.85941149.6875
Table 6. CPU times (in seconds) of Crank–Nicolson L1 scheme (13) and fast Crank–Nicolson L1 scheme (29) with M = 10 , α = 0.6 .
Table 6. CPU times (in seconds) of Crank–Nicolson L1 scheme (13) and fast Crank–Nicolson L1 scheme (29) with M = 10 , α = 0.6 .
N3000600012,00024,000
CN-L13.812515.343871.1406286.8281
CN-L1-SOE0.37500.79691.64064.1094
Table 7. The L 2 ( M , N ) errors and convergence rates of the scheme (29) with M = N for Example 2.
Table 7. The L 2 ( M , N ) errors and convergence rates of the scheme (29) with M = N for Example 2.
M = N α = 0.2 α = 0.4 α = 0.6 α = 0.8
L 2 r τ L 2 r τ L 2 r τ L 2 r τ
2 3 1.1205 × 10 2 -1.0330 × 10 2 -9.4627 × 10 3 -8.6820 × 10 3 -
2 4 2.7871 × 10 3 2.00732.5677 × 10 3 2.00832.3463 × 10 3 2.01192.1438 × 10 3 2.0178
2 5 6.9588 × 10 4 2.00186.4047 × 10 4 2.00335.8280 × 10 4 2.00935.2815 × 10 4 2.0212
2 6 1.7392 × 10 4 2.00041.5985 × 10 4 2.00241.4449 × 10 4 2.01201.2887 × 10 4 2.0351
2 7 4.4635 × 10 5 1.96223.9889 × 10 5 2.00273.5676 × 10 5 2.01793.0851 × 10 5 2.0625
2 8 2.2208 × 10 5 1.00719.9491 × 10 6 2.00338.7505 × 10 6 2.02757.1187 × 10 6 2.1156
2 9 1.0790 × 10 5 1.04143.5674 × 10 6 1.47972.1237 × 10 6 2.04281.5211 × 10 6 2.2265
Table 8. The L 2 ( M , N ) errors and convergence rates of the L1-ADI with M = N for Example 2.
Table 8. The L 2 ( M , N ) errors and convergence rates of the L1-ADI with M = N for Example 2.
M = N α = 0.2 α = 0.4 α = 0.6 α = 0.8
L 2 r τ L 2 r τ L 2 r τ L 2 r τ
2 3 3.5346 × 10 2 -2.7729 × 10 2 -1.8927 × 10 2 -1.0410 × 10 2 -
2 4 1.1495 × 10 2 1.62059.3443 × 10 3 1.56926.3885 × 10 3 1.56693.2066 × 10 3 1.6989
2 5 3.2516 × 10 3 1.82192.7202 × 10 3 1.78041.8558 × 10 3 1.78347.9451 × 10 4 2.0129
2 6 8.5323 × 10 4 1.93017.1979 × 10 4 1.91814.7806 × 10 4 1.95683.1457 × 10 4 1.3367
2 7 2.9564 × 10 4 1.52912.5030 × 10 4 1.52391.7607 × 10 4 1.44101.4818 × 10 4 1.0860
2 8 1.7540 × 10 4 0.75321.3547 × 10 4 0.88578.8698 × 10 5 0.98927.1268 × 10 5 1.0561
2 9 9.8947 × 10 5 0.82597.1485 × 10 5 0.92224.4494 × 10 5 0.99533.4639 × 10 5 1.0408
Table 9. The L 2 ( M , N ) errors and convergence rates of the scheme (29) on graded meshes with M = N , α = 0.6 for Example 2.
Table 9. The L 2 ( M , N ) errors and convergence rates of the scheme (29) on graded meshes with M = N , α = 0.6 for Example 2.
M = N r = 2 r = 4 r = 6 r = 8
L 2 r τ L 2 r τ L 2 r τ L 2 r τ
2 3 7.5493 × 10 3 -4.9609 × 10 3 -9.8903 × 10 3 -2.2025 × 10 2 -
2 4 1.8584 × 10 3 2.02231.1735 × 10 3 2.07982.9136 × 10 3 1.76326.7808 × 10 3 1.6996
2 5 4.5629 × 10 4 2.02602.8529 × 10 4 2.04037.8943 × 10 4 1.88391.8362 × 10 3 1.8847
2 6 1.1111 × 10 4 2.03806.9415 × 10 5 2.03912.1274 × 10 4 1.89174.8571 × 10 4 1.9186
2 7 2.6672 × 10 5 2.05861.6752 × 10 5 2.05095.8385 × 10 5 1.86541.2949 × 10 4 1.9072
2 8 6.2511 × 10 6 2.09323.9849 × 10 6 2.07171.6501 × 10 5 1.82313.5246 × 10 5 1.8774
2 9 1.4051 × 10 6 2.15341.0640 × 10 6 1.90514.8379 × 10 6 1.77019.8781 × 10 6 1.8351
Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Share and Cite

MDPI and ACS Style

Qiao, H.; Cheng, A. A Fast Finite Difference Method for 2D Time Fractional Mobile/Immobile Equation with Weakly Singular Solution. Fractal Fract. 2025, 9, 204. https://doi.org/10.3390/fractalfract9040204

AMA Style

Qiao H, Cheng A. A Fast Finite Difference Method for 2D Time Fractional Mobile/Immobile Equation with Weakly Singular Solution. Fractal and Fractional. 2025; 9(4):204. https://doi.org/10.3390/fractalfract9040204

Chicago/Turabian Style

Qiao, Haili, and Aijie Cheng. 2025. "A Fast Finite Difference Method for 2D Time Fractional Mobile/Immobile Equation with Weakly Singular Solution" Fractal and Fractional 9, no. 4: 204. https://doi.org/10.3390/fractalfract9040204

APA Style

Qiao, H., & Cheng, A. (2025). A Fast Finite Difference Method for 2D Time Fractional Mobile/Immobile Equation with Weakly Singular Solution. Fractal and Fractional, 9(4), 204. https://doi.org/10.3390/fractalfract9040204

Article Metrics

Back to TopTop