Journal of Function Spaces

Volume 2017, Article ID 3679526, 8 pages

https://doi.org/10.1155/2017/3679526

## A Compact Difference Scheme for Solving Fractional Neutral Parabolic Differential Equation with Proportional Delay

^{1}School of Statistics & Mathematics, Zhongnan University of Economics and Law, Wuhan 430073, China^{2}School of Finance, Zhongnan University of Economics and Law, Wuhan 430073, China

Correspondence should be addressed to Yanli Zhou; moc.liamg@7058uohzly and Xiangyu Ge; moc.361@eg_uygnaix

Received 5 July 2017; Accepted 17 September 2017; Published 18 October 2017

Academic Editor: Xinguang Zhang

Copyright © 2017 Wei Gu et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

#### Abstract

A linearized compact finite difference scheme is constructed for solving the fractional neutral parabolic differential equation with proportional delay. By the energy method, the unconditional stability of the scheme is proved, and the convergence order of the scheme is proved to be . A numerical test is also conducted to validate the accuracy and efficiency of the numerical algorithm.

#### 1. Introduction

In the past few years, more and more scholars have been attracted to the research of delay partial differential equations (DPDEs) [1–3]. However, most DPDEs have no exact solutions. Constructing efficient numerical methods for DPDEs is of great importance [4–7]. For details on numerically solving neutral delay parabolic differential equations (NDPDEs), the reader is referred to [5, 8]. Recently, fractional delay partial differential equations have been of great interest due to their application in automatic control, population dynamics, economics, and so forth [9, 10]. For details on numerical solutions to fractional delay partial differential equations, we refer the reader to [11, 12]. The work in [11] considers the numerical method without theoretical analysis, and the work in [12] considers the numerical method for a type of semilinear fractional partial differential equation with time delay.

In this paper, we consider the following fractional neutral parabolic differential equation with proportional delay: where is a constant, .

Let for , where . Then, satisfies the following equation:where .

For simplicity, we consider the following fractional neutral parabolic differential equation with delay instead of (1):where is a constant delay term. Time fractional partial derivative is defined in the Caputo sense by the following:where is the Gamma function.

In this paper, a linearized compact finite difference scheme is constructed for solving (3)–(5). By the energy method, the unconditional stability of the scheme is then proved, and the convergence order of the scheme is proved to be . A numerical test is also conducted to validate the accuracy and efficiency of the numerical algorithm.

The rest of the paper is organized as follows. In Section 2, a compact difference scheme is constructed to solve (3)–(5). Section 3 considers the solvability, convergence, and stability of the provided difference scheme. In Section 4, a numerical test is presented to illustrate the validity of the theoretical results. Section 5 gives a brief conclusion of this paper.

#### 2. The Construction of the Compact Difference Scheme

Throughout this paper, assume . Function is sufficiently smooth and satisfies where are arbitrary real numbers and and are positive constants.

First, let and be two positive integers; then, we take , ( is a positive integer), , . Define , where , , . Denote throughout this paper. Letbe the grid function space defined on . The following notations are used: where

For the time fractional derivative, we have the following lemma.

Lemma 1 (see [13]). *Suppose , ; it holds that and satisfies the following lemma.*

Lemma 2 (see [14]). *Assume ; then, it holds that*(1)* decreases monotonically as increases, and ;*(2)*, .*

*Lemma 3 (see [5]). Suppose ; then, one haswhere .*

*Considering (3) at the point , we have *

*From Lemma 1, we obtain where *

*From Taylor expansion, we have where , in between and .*

*Substituting (13) and (15) into (12) and applying the operator on both sides of (12), we obtain where From Lemma 3 and Taylor expansion, we have Substituting (18) into (16), we have where *

*Noticing and (7), we can easily obtain *

*Discretizing the initial and boundary conditions of (4) and (5), we obtain Replacing by in (19), (22), and (23) and omitting , we can obtain the following compact difference scheme: *

*3. The Solvability, Convergence, and Stability of the Difference Scheme*

*Define the following grid function space on : If , we introduce the following inner products and corresponding norms: *

*It is easy to obtain the following lemma.*

*Lemma 4 (see [12]). , one has .*

*Lemma 5 (see [15]). , one has *

*The following lemma will be used in the proof of the stability and convergence analysis.*

*Lemma 6 (see [15]). Assume that is a nonnegative sequence and satisfies then, where and are nonnegative constants.*

*Theorem 7. The difference scheme (24)–(26) has a unique solution.*

*Proof. *Denote ; the difference scheme (24)–(26) is a linear tridiagonal system , where only depends on , , and and is independent of . where . We can see that is a strictly diagonally dominant coefficient matrix. Thus, scheme (24)–(26) has a unique solution.

*To discuss the stability of the difference scheme (24)–(26), we consider the following problem: where is the perturbation caused by . The following difference scheme can be obtained for solving (33):*

*Denote *

*Definition 8. *Assume that satisfies (24)–(26) and satisfies (34)–(36); then, a numerical scheme for (3)–(5) is stable if one has where is a bounded constant independent of and .

*Theorem 9. Assume is the solution of (3)–(5); the difference scheme (24)–(26) is stable with respect to the initial perturbation of ; that is, , where is a positive constant independent of and .*

*Proof. *Subtracting (34)–(36) from (24)–(26), respectively, we can obtain the following equations: where

Multiplying on both sides of (39) and summing up for from to , we obtain Then, each term of (42) will be estimated. From the discrete Green formula and inequality , we have From the discrete Green formula and inequality , we have From the Cauchy-Schwarz inequality, Lemma 4, and (7), we have Substituting (43)–(45) into (42) and taking in (45), we obtain where Lemma 5 has been used.

Multiplying (46) by , we have Denote Noticing for , we have . Then, from (47), we obtain From Lemmas 2 and 6, we have From Lemma 5, we have The proof is completed.

*Denote ; by subtracting (24)–(26) from (19), (22), and (23), respectively, the following error equations can be obtained: where *

*Similar to the proof of Theorem 9, the following convergence result can be obtained.*

*Theorem 10. Assume is the solution of (3)–(5) and is the solution of (24)–(26). Then, one has where is a positive constant independent of and .*

*4. Numerical Test*

*In this section, a numerical test is used to validate the performance of scheme (24)–(26). Denote the maximum error at all grid points as the convergence order in time and space is defined, respectively, as For , we require to be fixed and small enough, while for , should be fixed and small enough.*

*Example 1. *Consider the following problem: the exact solution of (56) is , and From Table 1, we can see the maximum errors between the numerical solution and the exact solution in the temporal directions for , respectively, where the spatial step is fixed to be . The results show that the temporal convergence order matches well the theoretical convergence order of .

Table 2 shows the maximum errors in the spatial directions for when the temporal step is fixed at . From the results, we can see that the spatial convergence order is 4, which coincides with the theoretical result.

Figure 1 gives the error plane for , respectively. From this figure, we can see that the error becomes larger when a larger is taken.