Abstract
In this article, an effective method is given to solve nonlinear two-dimensional Volterra integral equations of the second kind. First, we find the solution of integral equation in terms of reproducing kernel functions in series, then by truncating the series an approximate solution obtained. In addition, the calculation of Fourier coefficients solution of the integral equation in terms of reproducing kernel functions is notable. Numerical examples are presented, and their results are compared with the analytical solution to demonstrate the validity and applicability of the method.
Similar content being viewed by others
Avoid common mistakes on your manuscript.
Introduction
Many problems in engineering and mechanics appear to be two-dimensional integral equations. For example, it is usually necessary to solve Fredholm integral equations in the calculation of plasma physics [1]. Mckee et al. [2] revealed that a class of nonlinear telegraph equations is equivalent to two-dimensional Volterra integral equations. Graham [3] showed that two-dimensional Ferdholm integral equation used in solving the problems of electrical engineering. Some other applications of two-dimensional integral equations can be found in [2, 4].
In this paper, we consider the nonlinear two-dimensional Volterra integral equation of the second kinds as follows:
where
Here, u(x, t) is unknown function and \(u(x,t), f(x,t) \in W_2^{(1,1)}(D), N(.)\) the continues terms in \(W_2^{(1,1)}(D)\), \(W_2^{(1,1)}(D)\) is a reproducing kernel space. Little numerical methods for nonlinear integral equations are written. In this context, the methods used are the block-by-block method [5], block-pulse functions [6], and rationalized Harr function [7].
In this article, we obtain presentation of exact solution nonlinear two-dimensional Volterra integral equation of the second kind in reproducing kernel space, and then, approximate solution is obtained by cutting series. Of course, coefficients of the series on presentation of exact solution obtained with a numerical calculation. The error of the approximate solution is monotone deceasing in the sense of \(\Vert .\Vert _{W_2^{(1,1)}}\).
The fundamental principles of the method
Below are some definitions and theorems that we used in the next sections.
Definition 2.1
[9] The function space \(W_2^1[a,b]\) is defined as follows:
Definition 2.2
[9] The inner product and norm in the function space \(W_2^1[a,b]\) are defined as follows:
For any functions \(f(x), g(x) \in W_2^1 [a,b],\)
and
It is easy to prove that \(W_2^1[a,b]\) is an inner space. At [9] prove that function space \(W_2^1[a,b]\) is a Hilbert space and also it is a reproducing kernel space.
Suppose \(R_y(x)\) is the reproducing kernel of \(W_2^1[a,b]\); in this case, we have for any \(f(x)\in W_2^1[a,b]\):
With calculation, we obtain that
Let set \(D=[a,b]\times [c,d] \subset \mathcal R^2\).
Definition 2.3
[9] The binary function space is defined as
Definition 2.4
[9] The inner product and norm in \(W_2^{(1,1)}(D)\) is defined as follows:
and
It is easy to prove that \(W_2^{(1,1)}(D)\) is an inner space. At [9] prove that function space \(W_2^{(1,1)}(D)\) is a Hilbert space and also it is a reproducing kernel space which has reproduced kernel:
where \(R_\xi (x),Q_\eta (y)\) are the reproducing kernels of \(W_2^1[a,b]\) and \(W_2^1[c,d]\), respectively.
Therefore, for each \(f(x,y) \in W_2^{(1,1)}(D)\)
Thus, if \(K_{(y,z)}(x,t)\) is reproducing kernel of \(W_2^{(1,1)}(D)\), then \(K_{(y,z)}(x,t)=R_y(x)Q_z(t)\), where \(R_y(x)\) and \(Q_z(t)\) are reproducing kernels in \(W_2^1[a,b]\) and \(W_2^1[c,d]\), respectively.
We will display the solution of Eq. (1) in reproducing kernel space \(W_2^{(1,1)}(D)\) for this purpose let \(\phi _i(x,t)=K_{(x_i,t_i)}(x,t)\), where \(\{(x_i,t_i)\}_{i=1}^{\infty }\) is dense in region D. From the property of the reproducing kernel, we have
Theorem 2.1
If \(\{(x_i,t_i)\}_{i=1}^\infty\) is dense in the region D, then \(\{\phi _i(x,t)\}_{i=1}^\infty\) is the complete function system of \(W_2^{(1,1)}(D)\).
Proof
Suppose \(u(x,t) \in W_2^{(1,1)}(D)\) if
Then, we have \(u(x,t)=0\) from the density of \(\{(x_i,t_i)\}_{i=1}^ \infty\) and continuity of u(x, t).\(\diamond\)
By applying Gram–Schmidt orthonormalization process for \(\{\phi _i(x,t)\}_{i=1}^\infty\)
where \(\alpha _{ik}\) are coefficients of Gram–Schmidt orthonormalization and \(\{\bar{\phi }_i(x,t)\}_{i=1}^\infty\) is a normal orthogonal basis of \(W_2^{(1,1)}(D)\) [9].
Now, the following theorem is obtained.
Theorem 2.2
We assume \(\{(x_i,t_i)\}_{i=1}^\infty\) be dense in region D. If Eq. (1) has a unique solution, then it is as follows:
Proof
Suppose u(x, t) is the solution of Eq. (1). From Theorem 2.1 and since \(\{\bar{\phi }_i(x,t)\}_{i=1}^\infty\) is a normal orthogonal basis of \(W_2^{(1,1)}(D)\), therefore, we can write
The proof is complete [9].\(\diamond\)
Implementations of the method
In this section, a method will be presented to calculate the solution (2) of the Eq. (1). To this end rewrite (2) as
where
\(A_i\) is unknown, because u(x, t) is an unknown function. \(A_i\) with a numerical calculation can be approximated by known \(B_i\). For this purpose, put initial function \(u_1(x,t)=f(x,t)\) and n-term approximation for u(x, t) is defined as follows:
where
Lemma 3.1
If
\((n \rightarrow \infty )\)
and
\((x_n,t_n) \rightarrow (y,z)\), then
\(u_n(x_n,t_n) \rightarrow \bar{u}(y,z) (n \rightarrow \infty )\)
Proof
We have
For absolute first on the right side of the above inequality, we have
From the convergence of sequence \(\{u_n(x,t)\}_{n=1}^\infty\), we conclude that there exists a constant N, such that
when \(n \ge N\). At the same time, it can be proved that
when \(n \rightarrow \infty\). Thus, \(\mid u_n(x_n,t_n) - u_n(y,z) \mid \rightarrow 0\) when \((x_n,t_n) \rightarrow (y,z)\). From \(\Vert u\Vert _c \le M \Vert u\Vert _{W_2^{(1,1)}}\) for any \((y,z) \in D\), it holds that
when
Therefore, here, we conclude that
when \((x_n,t_n) \rightarrow (y,z)\) [9].\(\diamond\).
Using the continuation of N(.), it will be obtained that \(N(u_n(x_n,t_n)) \rightarrow N(\bar{u}(y,z))\) when \(n\rightarrow \infty\). In addition, this shows that
From the method listed above, convergence theorem will be obtained.
Theorem 3.1
Suppose that the sequence \(\{\Vert u_n(x,t) \Vert _{W_2^{(1,1)}}\}_{n=1}^\infty\) is bounded in (3), if \(\{ (x_i,t_i) \}_{i=1}^ \infty\) is dense in D, then n-term approximate solution \(u_n(x,t)\) converges to the exact solution u(x, t) of Eq. (1) and exact solution is expressed as
where \(B_i\) is given by (4).
Proof
The theorem is proved in three steps:
(a) At this step, we provide that the sequence \(\{u_n(x,t)\}_{n=1}^\infty\) with formula (3) is converged. For this purpose from (4), we have
From recent equality and using the orthogonality of \(\{ \bar{\phi }_i(x,t) \}_{i=1}^ \infty\), it follows
Thus, sequence \(\{\Vert u_n(x,t)\Vert _{W_2^{(1,1)}}\}_{n=1}^\infty\) is monotone increasing. In other hand, sequence \(\{\Vert u_n(x,t)\Vert _{W_2^{(1,1)}}\}_{n=1}^\infty\) is bounded; therefore, this sequence is convergent. Therefore, there exists a constant c, such that
This illustrate that
Using the orthogonality of \(\{u_{n+1}(x,t)-u_n(x,t)\}_{n=1}^\infty\) if \(m > n\), then
Considering that \(W_2^{(1,1)}(D)\) is complete, so it follows
![](http://media.springernature.com/full/springer-static/image/art%3A10.1007%2Fs40096-017-0219-z/MediaObjects/40096_2017_219_Equ54_HTML.gif)
Therefore
(b) We define the projection operator as follows:
Thus
We claim
To this end, we have
Hence
(c) It is obvious that \(\bar{u}(x,t)\) is the solution of Eq. (1). From (6), it follows
By multiplying both sides of (7) by \(\alpha _{nj}\) and summing for j from 1 to n, we have
From (8) and (2), if \(n=1\), then
If \(n=2\), then
\(\alpha _{21}\bar{u}(x_1,t_1)+\alpha _{22}\bar{u}(x_2,t_2)=B_2\)
\(=\alpha _{21}(f(x_1,t_1)+Fu_2(x_1,t_1))+\alpha _{22}(f(x_2,t_2)+Fu_2(x_2,t_2))\)
\(=\alpha _{21}(f(x_1,t_1)+F\bar{u}_2(x_1,t_1))+\alpha _{22}(f(x_2,t_2)+Fu_2(x_2,t_2))\)
\(=\alpha _{21}\bar{u}(x_1,t_1))+\alpha _{22}(f(x_2,t_2)+Fu_2(x_2,t_2))\). From the above equality, it follows
In the same way, we obtain that
Since \(\{ (x_i,t_i) \}_{i=1}^\infty\) is dense in D, therefore, for any\((y,z)\in D\), there exists a subsequence \(\{ (x_{n_k},t_{n_k}) \}_{k=1}^\infty\) converging to (y, z). From Lemma 3.1 and above equality, we obtain that
Thus, \(\bar{u}(x,t)\) is the solution of Eq. (1) and
\(\square\)
Theorem 3.2
Suppose u(x, t) is the solution of Eq. (1) and \(r_n(x,t)\) is the error in the approximate solution \(u_{n+1}(x,t)\), where \(u_{n+1}(x,t)\) is given by (3). Then, \(r_n(x,t)\) is monotone decreasing in the sense of \(\Vert .\Vert _{W_2^{(1,1)}}\).
Proof
If the u(x, t) and \(u_{n+1}(x,t)\) are, respectively, functions in (1) and (6), in this case, we have
This illustrate that the error \(r_n(x,t)\) is monotone decreasing in the sense of \(\Vert .\Vert _{W_2^{(1,1)}}\) [9]. \(\square\)
Numerical examples
We implement the method presented in this article for some examples. We obtain absolute error of the approximate solution values in the selected points \((x,t)=(\frac{1}{2^i}, \frac{1}{2^i})( i=1,2,\ldots ,6)\). Of course, the choice of points is completely customized. Examples of references listed have been selected, so that we can compare the results in here with results in its references.
Example 4.1
[8] Consider the following nonlinear two-dimensional Volterra integral equation:
where
The exact solution is \(u(x,t)=x\sin {(t)}\). Table 1 illustrates the numerical results for this example. In addition, for points \((x_i,t_i)=(0.001i,0.001i)\quad( i=1,2,\ldots ,1000)\), the maximum error is as follows:
where
Example 4.2
[7] Consider the following nonlinear two-dimensional Volterra integral equation:
where
The exact solution is \(u(x,t)=x+t\). Table 2 illustrates the numerical results for this example. In addition, for points \((x_i,t_i)=(0.001i,0.001i)( i=1,2,\ldots ,1000)\), the maximum error is as follows:
where
Example 4.3
[8] Consider the following nonlinear two-dimensional Volterra integral equation:
where
The exact solution of this problem is \(u(x,t)=x^2+t^2\). Table 3 illustrates the numerical results for this example. In addition, for points \((x_i,t_i)=(0.001i,\,0.001i)\,( i=1,2,\ldots ,1000)\), the maximum error is as follows:
where
Conclusion and comments
This paper deals with a computational method for approximate solution of Volterra integral equations of the second kind based on the expansion of the solution as series of reproducing kernel functions. The advantage of the present method (compared with methods based on basis sets of different kinds) is not limitation on the nonlinear term. Considering that absolute errors of approximate solution in given points are small enough, so it follows that the presentation method in this article is right. The codes were written in Maple. We think that this method can be generalized to the new inner multiply that it provides reproducing kernel space.
References
Farengo, R., Lee, Y.C., Guzdar, P.N.: An electromagnetic integral equation: application to microtearing modes. Phys. Fluids. 26, 3515–3523 (1983)
Mckee, S., Tang, T., Diogo, T.: An Euler-type method for two-dimensional Volterra integral equations of the first kind. IMA J. Numer. Anal. 20, 423–440 (2000)
Graham, I.G.: Collocation method for two-dimensional weakly singular integral equations. J. Aust. Math. Soc. (Ser. B) 22, 456–473 (1981)
Hanson, R.J., Phillips, J.L.: Numerical solution of two-dimensional integral equations using linear elements. SIAM J. Numer. Anal. 15(1), 113–121 (1978)
Mirzaee, F., Rafei, Z.: The block by block method for the numerical solution of the nonlinear two-dimensional Volterra integral equations. J. King Saud Univ. Sci. 23, 191–195 (2011)
Mirzaee, F., Hoseini, A.A.: A computational method based on hybrid of block-pulse functions and Taylor series for solving two-dimensional nonlinear integral equations. Alex. Eng. J. 53, 185–190 (2014)
Babolian, E., Bazm, S., Lima, P.: Numerical solution of nonlinear two-dimensional integral equations using rationalized Harr functions. Commun. Nonlinear Sci. Numer. Simul. 16, 1164–1175 (2011)
Maleknejad, K., Sohrabi, B.B.: Two-dimensional FCBS: application to nonlinear Volterra integral equations. In: Proceedings of the World Congress on Engineering (WCE), vol II, 1–3 July 2009, London
Cui, M., Lin, Y.: Nonlinear Numerical Analysis in the Reproducing Kernel Space. Nova Science Publishers, New York (2008)
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.
About this article
Cite this article
Fazli, A., Allahviranloo, T. & Javadi, S. Numerical solution of nonlinear two-dimensional Volterra integral equation of the second kind in the reproducing kernel space. Math Sci 11, 139–144 (2017). https://doi.org/10.1007/s40096-017-0219-z
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s40096-017-0219-z