1. Introduction
Partial differential equations (PDEs) serve as a valuable tool for modeling a wide range of phenomena encountered in science and engineering. Their significance is evident in diverse fields including fluid dynamics, quantum theory and plasma physics [
1].
We consider the following PDE [
2]:
with boundary and initial conditions
In the above equation, u denotes wave displacement at particular position x with time t. Constants are known parameters. The force term characterizes external influences on the system.
Numerous types of equations may be constructed by altering the values of involved constants and force term in Equation (
1). As an example, if we set
, then (
1) becomes a PHI-Four equation. This equation holds significant importance in the field of mathematical physics and Cosmology. Triki and Wazwaz investigated the exact solution for bright and dark solitons and they also used the technique “sine–cosine ansatz” to find the solutions of more than one type of PHI-Four equations [
1]. The authors in [
3] explored a novel set of solutions for a generalized Fitzhugh Nagumo model using ansatz and tanh methods.
Furthermore, by taking the values of parameters
, Equation (
1) transforms to an Allen–Cahn equation. The Allen–Cahn equation serves as a versatile model with applications across diverse fields, including bio-mathematics, plasma-physics, quantum mechanics and image processing [
4,
5,
6]. The authors in [
7] studied the Jacobi Gauss Lobatto technique to solve the PHI-Four model. Ehsani et al. [
8] applied the homotopy perturbation method to solve a PHI-Four equation. The authors employed the solitary wave ansatz method to derive soliton solutions for a PHI-Four model with non-topological and topological nature in [
9]. Alofi introduced a generalized tanh method to obtain solutions of both the Drinfeld–Sokolov system and the PHI-Four model in [
10]. The authors in [
11] formulated the latest solutions for Boussinesq, RLW and PHI-Four models. Sassaman and Biswas employed the soliton perturbation theory to explore solutions for both the Klein–Gordon and PHI-Four equations. Najafi investigated the soliton solution of the aforementioned equation utilizing He’s variational method [
12].
The PHI-Four model is used in Cosmology, helping in understanding the evolution of the universe and the formation of cosmic structures. This model has also been used in statistical physics to study critical phenomena such as the behavior of a system near its critical point. The authors in [
4,
5,
6] studied a wavelet-based method for solving Newell–Whitehead and Allen–Cahn equations. Considering non-periodic boundary conditions, the Allen–Cahn equation was solved by Ishtiaq et al. [
13]. In condensed matter physics, the Allen–Cahn equation finds application in depicting phase transitions within materials. It has been applied in image processing for denoising and image segmentation in computer science. The Allen–Cahn equation has also been applied to model pattern formation in biological systems, such as formation of spatial patterns in the animal coat. Both equations are applied to study the dynamics of defects in materials.
The inception of spline approximation, in its current manifestation, can be traced back to the pioneering work of Schoenberg in 1946 [
14]. Up to 1960, there was some research that mentioned spline functions. A few of the major figures in the development of spline are Ahlberg and Nilson [
15], Birkhoff and Garabedian [
16], Loscalzo and Talbot [
17], Malaren [
18], Rubin and Khosla [
19] and Schoenberg [
20]. Sokolnikoff (1956) offered a succinct but highly readable history of the evolution of beam theory. Some of the pioneers in the utilization of spline functions for achieving smooth numerical solutions to ordinary differential equations (ODEs) and partial differential equations (PDEs) are presented in [
21,
22,
23,
24,
25,
26,
27,
28]. The non-polynomial spline approach has been used more often recently to solve partial and ordinary differential equations. The work in [
29,
30,
31,
32,
33,
34] explains the numerical solutions of a system of second-order differential equations using non-polynomial splines, fourth-order problems using B splines and singular boundary value problems. Research on the fifth order can be found in Islam et al. [
35], Siddiqi and Akram [
36] and Siddiqi et al. [
37]. The work of Akram and Siddiqi [
38] yields sixth-order BVPs. According to Ramadan et al. [
39], Rashidinia and Mahmoodi [
40], non-polynomial splines can be used to solve parabolic equations numerically. For the solution of fifth-order boundary value problems, Kasi and Ballem [
41] employed the finite element approach incorporating the Galerkin method with the quartic B spline as the basis function. Alam et al. [
42] studied the RBF approximation method for the time-fractional FitzHugh–Nagumo equation. Radmanesh and Ebadi [
43] used the local RBF method for solving fractional integral equations. The authors in [
44] solved the fractional fascioliasis disease model by Fibonacci polynomials.
The spline method stands out due to its numerous advantages in comparison to other numerical approaches. It provides a complete description of both function and its rate of change throughout the entire interval. In contrast, the finite difference method only provides functional values at specified knots, while the finite element method necessitates the computation of quadrature, which is not a requirement in the spline method. However, undesirable oscillations of these functions may be expected between the data points. It is important to note that splines of lower orders may result in reduced accuracy, while higher-order splines can lead to an increase in computational cost. The addition of tension to the polynomial splines overcomes this problem. Non-polynomial or tension splines were first introduced by Daniel G. Schweikert [
45]. These splines were designed to get rid of extraneous inflection points in curve fitting [
46]. The desirable scheme could be derived by choosing various tension parameter values over the domain. Finally, by applying non-polynomial splines, we can increase accuracy while using the same computing effort [
47]. The paper’s outline is as follows. In
Section 2, derivation and discretization of the proposed method is provided. In
Section 3, stability of the method is explained using the von Neumann technique, and convergence analysis is discussed in detail. Accuracy and efficiency of the suggested methodology is shown by studying several mathematical problems in
Section 4. Lastly,
Section 5 includes the concluding remarks of the study.
2. The Proposed Methodology
In this portion, the proposed method is discussed, utilized to study the approximate solution of the hyperbolic telegraph equation. For this purpose, first, we transform (
1) into the system of equations given below:
with conditions
Using finite difference for the temporal part and a
-weighted scheme for the spatial part of Systems (
4) and (
5), the following equations can be derived:
where
,
,
and
k is the time step. Substituting
, we can obtain
Equivalently, the above equation can be written as
In order to find the approximate solution of the system under consideration, the technique of non-polynomial splines is implemented.
For this goal, first, we define a partition in given by , where and h is the step size.
For the solution of Equation (
1), the unknown
is approximated by
using the non-polynomial splines as follows:
In the above equation,
where
are unknowns to be determined. According to the definition of splines,
From Equations (
12) and (
13), the following relations can be obtained:
where
. Now, from Equation (
14), we can obtain
The use of condition
produces
Putting values from Equation (
15) in Equation (
16), the following recurrence relation can be obtained:
where
Combining Equation (
17) at time levels
n and
, a newly obtained equation is given by
Now, substituting values from (
10) in (
18), we can obtain
The coefficients involved in (
19) are given as follows:
For finding the solution of the system, Equations (
8) and (
19) along with conditions given in Equations (
6) and (
7) are used.