ABSTRACT
In this paper, we present an effective method under Taylor wavelets and collocation technique to find an approximate solution of linear and non-linear second-order singular value differential equations. Application of this method on various problems confirms the efficiency, easy applicability, and rapid computation. The obtained solution is compared with some other existing numerical solutions. For many problems, this method gives a solution same as the exact solution of the problem which confirms the effectiveness and accuracy of this method. Additionally, we include graphs and figures to demonstrate that the Taylor wavelet method offers better accuracy for a number of problems. MATLAB software is used to process relative data and execute calculations.
1. Introduction
An oscillation that resembles a wave and has an amplitude that starts at zero, rises or falls, and then returns to zero one or more times is called a wavelet. The primary characteristics of the wavelets are multiresolution analysis, density, orthogonality, and compact support. Wavelet-based numerical approaches have gained popularity in recent years, particularly in computer and applied mathematics due to their simplicity and ease of use to find the solution of several classes of differential equations. Wavelet analysis has been used in a variety of disciplines including signal analysis, harmonic analysis, temporal frequency analysis, and others. The study of wavelets is a relatively new and developing field in mathematics.
Differential equations are the formulation of scientific theory for many real-world physical problems. In recent decades, nonlinear equations have drawn a lot of attention since they are utilized in several technical and scientific applications, like gas dynamics, atomic structure, chemical processes, and nuclear physics, for instance [Citation1]. Sometimes, it is impossible to obtain the analytical solution to a given singular differential equation using existing (analytical) methods then we use numerical approaches like the Tau method, operational matrix of integration method, etc. Researchers in several branches of science and engineering have given a lot of attention to wavelets, recently. One of the most frequently used numerical method for solving differential equations is the operational matrix of integration with the collocation method. This approach is based on the operational matrix of integration and collocation methodology, which converts the given problem into a set of algebraic equations. The Newton Rapshon technique can be used to find the unknown coefficients in this system of algebraic equations.
A mathematical model of the thermal behavior of a spherical cloud of gas and nuclear physics similarly faces these problems. It is always crucial to obtain the solution of differential equations because several human disease models are represented in terms of differential equations. The second-order ordinary differential equations (ODEs) have been studied by several researchers, like Tunc and Tunc [Citation2–4], and Hasan and Zhu [Citation5]. Additionally, numerous mathematicians have solved singular value differential equations using different wavelet methods like the Legendre wavelet method [Citation6,Citation7], Haar wavelet collocation method [Citation8], Laguerre wavelets method [Citation9], Chebyshev wavelet operational matrix of integration [Citation10], and Hermite wavelet method [Citation11]. Recently, vivek utilized the Fibonacci wavelet method to solve the system of nonlinear differential equations representing the HIV-infection model [Citation12] and Tunc et al. [Citation13] obtained the solution of Intego-differential equations. Some of the existing methods are described in [Citation14–17] for solving number of singular differential equations. We conclude from an in-depth review of the literature and the conclusions of several research publications that researchers have not used Taylor wavelets to solve the following class of differential equations. This motivated us to implement Taylor wavelets to solve singular differential equations.
Consider a class of the general second-order singular differential equation [Citation18,Citation19]: (1) (1) These types of equations are our focus throughout this research, under the following four conditions: (2) (2) (3) (3) (4) (4) in addition to the most general combined boundary conditions as: (5) (5) where , i=1,2, and , i = 1, 2, 3, 4 be given finite constants with . may be discontinuous at , whereas and R satisfies the following conditions [Citation20]:
(D) and .
(D) and . (for boundary condition (Type-I)).
(D) and . (for boundary conditions (Type-II)).
(D) and on Ω, where .
Existence and uniqueness of the singular boundary value problem defined in (Equation2(2) (2) ) along with boundary conditions (Equation3(3) (3) ) and (Equation4(4) (4) ) is discussed in [Citation18,Citation21–23].
Now consider the singular value problem (Equation1(1) (1) ) along with the condition (Equation4(4) (4) ). If and R satisfies the following conditions [Citation24]:
(D)ν is measurable on ,
(D on ,
(D),
(D)R satisfies the Carathéodory conditions and is bounded. Specifically, there exist a, b with and K>0 such that,
(d)For each is continuous on ;
(d)For each is measurable on ; and
(d).
Then the solution of the initial value problem (Equation2(2) (2) ) along the conditions (Equation5(5) (5) ) exists (see Theorem (1) of [Citation25]). In addition to the above conditions, (Equation2(2) (2) ) has a unique solution, if the function R satisfies the Lipschitz condtion in Λ on [0,1] (see Theorem (2) of [Citation25]) [Citation26].
The condition which guaranteed the uniqueness and existence of Strum-Liouville singular BVP [Citation27] defined by problem (Equation2(2) (2) ) along with the condition (Equation5(5) (5) ) is also discussed by the author. They used the perturbation technique, maximal principle, Schauder's fixed-point theorem, and mixed monotone iterative techniques to prove it. For this purpose, they present two sufficient conditions. Liu and Yu [Citation28] additionally filled the gap which arises to investigate the uniqueness of a positive solution of those BVP in which the Lipschitz condition is not satisfied in the presence of singularity. Chawla and Shivakumar [Citation23] and R. K. Pandey [Citation29–31] published several results on the existence and uniqueness of the singular and initial value problem having the form (Equation1(1) (1) ) with the conditions (Equation3(3) (3) ).
Following is the overview of the article's structure: We discuss the Taylor wavelet characteristics and function approximation in Section 3. In Section 4, we present the convergence theorem and error bound of the Taylor wavelet expansion of the functions, while, Section 5 introduces the operational integration matrix evaluation process. The application of the operational matrix of integration is described in Section 6. Section 7 presents evaluation-based numerical problems to test the effectiveness of the suggested method, and Section 8 draws an overall conclusion.
2. Wavelets
Wavelets are generated from a single function ζ known as ‘mother wavelet’ by dilating and translating it. When both the translation and dilation variables are continuous then the family of continuous wavelets is defined as follows: (6) (6) In particular, if the variables ν and η are restricted to discrete quantities i.e. , and , where ω and θ are natural numbers. Then, (7) (7) defines the discrete family of wavelets and form a basis(wavelet basis) of . In particular, when and , then forms an orthonormal basis.
2.1. Taylor wavelets
Four arguments exist for Taylor wavelets : , where is the order of the Taylor polynomial [Citation32]. Consider the family of the functions defined below on the interval : (8) (8) with Then defines the family of Taylor wavelets for the Taylor polynomial of order r and . The coefficient is for normality, the dilation parameter is and the translation parameter is .
2.2. Function approximation
We can express any arbitrary function using the Taylor wavelet basis as follows [Citation8]: (9) (9) where are the Taylor wavelets and are the Taylor wavelet coefficients. Consider the truncated series of approximation for , (10) (10) where T denotes transposition and are two matrices of order and defined as (11) (11)
3. Convergence analysis
In this section, we present two theorems to discuss the convergence and error analysis of the proposed method.
Theorem 3.1
[Citation33]
Let be a continuous function on the interval such that it is bounded by δ i.e. , for every . Then, the Taylor wavelet coefficients of in Equation (Equation10(10) (10) ) are bounded as: (12) (12) where δ is a constant and λ is given by
Proof.
Consider the following series expansion of using Taylor wavelets as defined in (Equation10(10) (10) ), (13) (13) where E and are given in Equation (Equation11(11) (11) ) and the coefficients can be determined as, (14) (14) Using the definition of Taylor wavelets , we have (15) (15) Let . Let , then Equation (Equation14(14) (14) ) becomes Therefore, (16) (16) Seeing the properties of Taylor polynomials, we can say that (17) (17) Using the assumption in Equations (Equation17(17) (17) ) and (Equation16(16) (16) ), we have (18) (18) Thus the proof of the Theorem 3.1 has been completed. Also, boundedness of the function implies absolutely convergent of the series . Hence the Taylor wavelet approximation of the function is absolutely convergent.
Theorem 3.2
[Citation33]
Let be a continuous function on the interval and for every . Let be the Taylor wavelet series expansions where , be the Taylor wavelet coefficients and Taylor wavelet basis respectively. Then, the bound of the truncated error is given as: where,
Proof.
Any function can be expanded in terms of Taylor wavelets as: If is the expansion truncated by using Taylor wavelets, then the error obtained by truncating the above function can be computed as: (19) (19) From Equation (Equation19(19) (19) ), we can write (20) (20) Now using the result of Theorem 3.1, which is to reduce Equation (Equation20(20) (20) ) as, (21) (21) Let us define (22) (22) Then from Equations (Equation21(21) (21) ) and (Equation22(22) (22) ), we get (23) (23) Therefore, (24) (24) By the definition of Taylor wavelets, we have (25) (25) Integrating Equation (Equation25(25) (25) ) with respect to ρ, we get (26) (26) Let , Equation (Equation26(26) (26) ) becomes (27) (27) But the standard definition of Taylor polynomial implies that, (28) (28) Substituting Equation (Equation28(28) (28) ) in Equation (Equation27(27) (27) ), we get Thus the Theorem 3.2 has been proved. This theorem also implies consistency and stability of the approximation.
4. Operational matrix of integration
Let be the vector consisting of the Taylor wavelets [Citation32] described in Equation (Equation11(11) (11) ), then defines the relation between the integration operator I and the operational matrix of integration S of order where . Here we will discuss some integration characteristics of Taylor wavelets and the Integral operator for and . So we have (29) (29) where is the standard characteristic function defined as and . Therefore, in Equation (Equation29(29) (29) ), for i = 1, we have (30) (30) It is known that (31) (31) Now, substituting Equation (Equation31(31) (31) ) in Equation (Equation29(29) (29) ) for , we have (32) (32) Now, approximating by using Taylor wavelets, we have (33) (33) where Using Equation (Equation33(33) (33) ) in Equations (Equation30(30) (30) ) and (Equation32(32) (32) ), we get and where and Therefore, we get For example, we define some structure of Taylor wavelets and operational matrix of integration for and : Let .
Each row of the operational matrix of integration can be obtained as follows;
Integrate each basis w.r.t. ρ limit from ‘0’ to ‘ρ’.
Express the obtained integration value as a linear combination of the given basis.
5. Application of the operational matrix of integration
Here, we will discuss the application of the operational matrix of integration [Citation32] and the collocation method for the given four types of conditions of the singular value problem defined in Equation (Equation1(1) (1) ).
Consider the boundary conditions of Type I defined in Equation (Equation2(2) (2) ). Assume (34) (34) Integrate Equation (Equation34(34) (34) ) w.r.t. ρ taking limit from 0 to ρ, we get (35) (35) Again apply integration from 0 to ρ on both sides of Equation (Equation35(35) (35) ) with respect to the variable ρ yields, (36) (36) Substitute in Equation (Equation36(36) (36) ) gives (37) (37) Now, apply the conditions given in Equation (Equation2(2) (2) ) and substitute the obtained equation in Equation (Equation36(36) (36) ), so that (38) (38) Further, on substituting the values of the function Λ and all its derivatives obtained by the above procedure, into (Equation1(1) (1) ) we get an algebraic equation of the following form, (39) (39) Collocate the above-obtained equation at n number of collocation points defined by , where to get a system of n number of equations. We can solve these equations for the unknown Taylor wavelets coefficient using Newton's method or fsolve command of MATLAB.
Consider Type II boundary conditions defined in Equation (Equation3(3) (3) ). Assume (40) (40) Integrate Equation (Equation40(40) (40) ) w.r.t. ρ taking limit from 0 to ρ, we get (41) (41) (42) (42) Substitute in Equation (Equation42(42) (42) ), we get (43) (43) Now, apply the conditions given in Equation (Equation3(3) (3) ) and substitute Equation (Equation43(43) (43) ) in Equation (Equation42(42) (42) ), we have (44) (44) Similarly proceed as we have done for case (I) to get our algebraic equation for this case.
Consider Type III boundary conditions defined in Equation (Equation4(4) (4) ). Assume (45) (45) Integrate Equation (Equation45(45) (45) ) w.r.t. ρ taking limit from 0 to ρ, we get (46) (46) Integrate Equation (Equation46(46) (46) ) w.r.t. ρ taking limit from 0 to ρ, we get (47) (47) Consider Type IV boundary conditions defined in Equation (Equation5(5) (5) ). Assume (48) (48) Integrate Equation (Equation48(48) (48) ) w.r.t. ρ taking limit from 0 to ρ, we get (49) (49) (50) (50) Substitute in Equations (Equation49(49) (49) ) and (Equation50(50) (50) ), we have (51) (51) and (52) (52) Using these values of and in Equation (Equation5(5) (5) ) which will turn out into the following equations: (53) (53) (54) (54) Observe that the values of and can be obtained by solving Equations (Equation53(53) (53) ) and (Equation54(54) (54) ), so that we have a truncated expression of and in terms of Taylor wavelet basis and Taylor wavelet coefficients with the provided boundary conditions (Equation5(5) (5) ) as and Substituting the expression of the function Λ and its derivatives obtained by the above procedure in Equation (Equation1(1) (1) ), we get an associated algebraic equation of the following form: (55) (55) Collocate this equation at n number of grid points , where to get a set of n algebraic equations. We can solve this system of n equations for the value of vector E (Taylor wavelet coefficients) by a suitable method. The required numerical solution will next be presented by substituting these coefficients in Equation (Equation50(50) (50) ).
6. Numerical experiments
Problem 1
Take into account the following problem (56) (56) with where The exact solution of this problem is . We obtain a Taylor wavelet solution (TWS) for this problem for different values of n = 6, 7, 9 and 11 with . Table demonstrates that only a few Taylor wavelet basis functions are required to obtain an approximation that is the same as an exact solution with a full agreement up to 7, 8 Digits, while Figure shows the exact and approximate solution. In Figure , error variation can be observed (Table provides the absolute error that occurred in TWS for different values of n).
Problem 2
Take into account the following problem (57) (57) with where The exact solution of (Equation57(57) (57) ) is . We apply the proposed Taylor wavelet method for the parameters n = 6, 7, 9, 10 with . Table , represents a comparison between the values obtained by TWM, while Table , represents error variation for different values of n. In Figure , we have shown the TWS and the exact solution, while in Figure error variation can be observed.
Problem 3
Take into account the following problem (Ex. 5.6 [Citation34]) (58) (58) with Where The exact solution of this problem is . We solve this problem using Taylor wavelet for n = 6, 7, 8, 9 and . Taylor wavelet solution (TWS) is presented in Table in comparison with the exact solution. In Table , TWS is compared with some other existing methods, while error variation for the same values of n and θ is shown in Table . Taylor wavelet solution for and exact solution is graphically shown in Figure . , shows absolute error variation for multiple values of n.
Problem 4
Take into account the following problem with the exact solution . We solve this problem using the Taylor wavelet for n = 6, 7, 9, 11 and . In Table , numerical values are compared to the exact solution (Figure ). In Table error variation for the same values of n and θ is shown. Taylor wavelet solution for n = 11, , and the exact solution is graphically shown in Figure , while in Figure error variation can be observed graphically.
Problem 5
Take into account the following problem (59) (59) with and the exact solution of this problem is , where . We obtain the Taylor wavelet solution (TWS) for this problem for different values of n = 6, 7, 9 and 10 with . Table demonstrates that only a few Taylor wavelet basis functions are required to obtain an approximation that is the same as an exact solution with a full agreement up to 5, 6 Digits, while Figure shows the exact and approximate solution. In Figure error variation of TWS can be observed. (Table , provides the absolute error for multiple values of n).
Problem 6
Take into account the following problem, (60) (60) with Which has the exact solution as . We solve this problem using Taylor wavelets for n = 6, 7, 9, 11 and . Table shows that a very less number of basis functions are required to obtain our truncated solution presented by the Taylor wavelets series of the minimum error. Absolute error variation can be seen for different values inTable (Taylor wavelet approximate solution and absoluter error occured in the proposed method can be graphically observed in Figures and respectively).
7. Conclusion
Using the Taylor wavelets, we tried to implement an operational integration matrix to find the solution to singular value linear and non-linear differential equations under various given conditions. This approach is crucial for the advancement of fresh studies in the discipline of numerical analysis and is advantageous for beginning researchers. When compared to the several numerical solutions, the proposed method performs very satisfactorily on applying several problems and performs well in comparison with other numerical methods. The results of this analysis are summarized as follows:
Comparing the accuracy of the current method to other numerical techniques found in the literature, it is more accurate.
The method's steps are rapid and simple to implement in computer programming, and they may be increased to higher orders with very minor methodological modifications.
The operational matrix of integration is generalized to make this method simple.
The Taylor wavelet technique is a new one that was recently used to construct this operational integration matrix.
In future, researchers can also try this approach to solve any class of differential equations of the scope.
CRediT authorship contribution statement
Vivek developed the fundamental concept of the article, wrote the text, and completed all phases of the research's proofs.
Availability of the Data
All the data associated with the manuscript is included in it.
Acknowledgments
All authors are grateful to the reviewers for their outstanding comments, which encouraged us at each step of the well-read proofs of the article to develop a distinct and excellent script.
Disclosure statement
No potential conflict of interest was reported by the author(s).
References
- Kumar M, Singh N. A collection of computational techniques for solving singular boundary-value problems. Adv Eng Softw. 2009;40(4):288–297. doi: 10.1016/j.advengsoft.2008.04.010
- Tunç C, Tunç O. A note on certain qualitative properties of a second order linear differential system. Appl Math Inf Sci. 2015;9(2):953. doi: 10.12785/amis/090245
- Tunç C, Tunç O. On the boundedness and integration of non-oscillatory solutions of certain linear differential equations of second order. J Adv Res. 2016;7(1):165–168. doi: 10.1016/j.jare.2015.04.005
- Tunç C, Tunç O. A note on the stability and boundedness of solutions to non-linear differential systems of second order. Arab J Basic Appl Sci. 2017;24:169–175. doi: 10.1016/j.jaubas.2016.12.004
- Hasan YQ, Zhu LM. Solving singular initial value problems in the second-order ordinary differential equations. J Appl Sci. 2007;7(17):2505–2508. doi: 10.3923/jas.2007.2505.2508
- Mohammadi F, Hosseini MM. A new Legendre wavelet operational matrix of derivative and its applications in solving the singular ordinary differential equations. J Frank Inst. 2011;348(8):1787–1796. doi: 10.1016/j.jfranklin.2011.04.017
- Iqbal J, Abass R, Kumar P. Solution of linear and nonlinear singular boundary value problems using Legendre wavelet method. Ital J Pure Appl Math. 2018;40:311–328. Available from: https://ijpam.uniud.it/online_issue/201840/30-Iqbal-Abass-Kumar.pdf
- Shiralashetti SC, Deshi AB, Desai PM. Haar wavelet collocation method for the numerical solution of singular initial value problems. Ain Shams Eng J. 2016;7(2):663–670. doi: 10.1016/j.asej.2015.06.006
- Zhou F, Xu X. Numerical solutions for the linear and nonlinear singular boundary value problems using Laguerre wavelets. Adv Differ Equ. 2016;2016(1):1–5. doi: 10.1186/s13662-018-1939-6
- Babolian E, Fattahzadeh F. Numerical solution of differential equations by using Chebyshev wavelet operational matrix of integration. Appl Math Comput. 2007;188(1):417–426. doi: 10.1016/j.amc.2006.10.008
- Shiralashetti SC, Kumbinarasaiah S. Hermite wavelets operational matrix of integration for the numerical solution of nonlinear singular initial value problems. Alex Eng J. 2018;57(4):2591–2600. doi: 10.1016/j.aej.2017.07.014
- Vivek, Kumar M, Mishra SN. A fast Fibonacci wavelet-based numerical algorithm for the solution of HIV-infected CD4+ T cells model. Eur Phys J Plus. 2023;138(5):458. doi: 10.1140/epjp/s13360-023-04062-6
- Tunç C, Tunç O. On the fundamental analyses of solutions to nonlinear integro-differential equations of the second order. Mathematics. 2022;10(22):4235. doi: 10.3390/math10224235
- Yüzbaşı Ş. A numerical approach for solving a class of the nonlinear Lane–Emden type equations arising in astrophysics. Math Methods Appl Sci. 2011;34(18):2218–2230. doi: 10.1002/mma.1519
- Yüzbaşı Ş. A numerical approach for solving the high-order linear singular differential–difference equations. Comput Math with Appl. 2011;62(5):2289–2303. doi: 10.1016/j.camwa.2011.07.016
- Yüzbaşı Ş. A numerical scheme for solutions of a class of nonlinear differential equations. J Taibah Univ Sci. 2017;11(6):1165–1181. doi: 10.1016/j.jtusci.2017.03.001
- Doha EH, Abd-Elhameed WM, Youssri YH. Second kind Chebyshev operational matrix algorithm for solving differential equations of Lane–Emden type. New Astron. 2013;23:113–117. doi: 10.1016/j.newast.2013.03.002
- Bobisud LE. Existence of solutions for nonlinear singular boundary value problems. Appl Anal. 1990;35(1–4):43–57. doi: 10.1080/00036819008839903
- Abukhaled M, Khuri SA, Sayfy A. A numerical approach for solving a class of singular boundary value problems arising in physiology. Int J Numer Anal Model. 2011;8(2):353–363. doi: 10.1186/s13662-015-0572-x
- Singh R, Guleria V, Singh M. Haar wavelet quasilinearization method for numerical solution of Emden–Fowler type equations. Math Comput Simul. 2020;174:123–133. doi: 10.1016/j.matcom.2020.02.004
- Dunninger DR, Kurtz JC. Existence of solutions for some nonlinear singular boundary value problems. J Math Anal. 1986;115(2):396–405. doi: 10.1016/0022-247X(86)90003-X
- Pandey RK, Verma AK. A note on existence-uniqueness results for a class of doubly singular boundary value problems. Nonlinear Anal Theory Methods Appl. 2009;71(7-8):3477–3487. doi: 10.1016/j.na.2009.02.012
- Chawla MM, Shivakumar PN. On the existence of solutions of a class of singular nonlinear two-point boundary value problems. J Comput Appl. 1987;19(3):379–388. doi: 10.1016/0377-0427(87)90206-8
- Shiralashetti SC, Deshi AB. The numerical solution of singular initial value problems using Chebyshev wavelet collocation method. Ain Shams Eng J. 2018;9(4):1451–1456. doi: 10.1016/j.asej.2016.08.015
- Biles DC, Robinson MP, Spraker JS. A generalization of the Lane–Emden equation. J Math Anal. 2002;273(2):654–666. doi: 10.1016/S0022-247X(02)00296-2
- Aslanov A. A singular initial-value problem for second-order differential equations. Abstr Appl Anal. 2014;2014:526549. doi: 10.1155/2014/526549
- Sun J, Zhang G. Nontrivial solutions of singular sublinear Sturm–Liouville problems. J Math Anal. 2007;326(1):242–251. doi: 10.1016/j.jmaa.2006.03.003
- Liu Y, Yu H. Existence and uniqueness of positive solution for singular boundary value problem. Comput Math Appl. 2005;50(1-2):133–143. doi: 10.1016/j.camwa.2005.01.022
- Pandey RK. On a class of regular singular two point boundary value problems. J Math Anal. 1997;208(2):388–403. doi: 10.1006/jmaa.1997.5320
- Pandey RK. On a class of weakly regular singular two-point boundary value problems–II. Differ Equ. 1996;127(1):110–123. doi: 10.1006/jdeq.1996.0064
- Pandey RK. On a class of weakly regular singular two point boundary value problems–I. Nonlinear Anal Theory Methods Appl. 1996;27(1):1–2. doi: 10.1016/0362-546X(95)00006-H
- Keshavarz E, Ordokhani Y, Razzaghi M. The Taylor wavelets method for solving the initial and boundary value problems of Bratu-type equations. Appl Numer Math. 2018;128:205–216. doi: 10.1016/j.apnum.2018.02.001
- Shiralashetti SC, Lamani L. Taylor wavelets operational matrix method for the numerical solution of stochastic Volterra-Fredholm integral equations. Stoch Anal Appl. 2020;24(2):121–240. Available from: https://www.mukpublications.com/resources/3.%20Lata%20Lamani
- Nasab AK, Kılıçman A, Babolian E, et al. Wavelet analysis method for solving linear and nonlinear singular boundary value problems. Appl Math Model. 2013;37(8):5876–5886. doi: 10.1016/j.apm.2012.12.001