A new sequential approach for solving the integro-differential equation via Haar wavelet bases (Q2359073)
From MaRDI portal
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | A new sequential approach for solving the integro-differential equation via Haar wavelet bases |
scientific article |
Statements
A new sequential approach for solving the integro-differential equation via Haar wavelet bases (English)
0 references
27 June 2017
0 references
The authors present a numerical method to approximate the solution of the nonlinear mixed Volterra-Fredholm integro-differential equations of the second kind: \[ \lambda u'(x) = f(x) + \alpha \int_0^1 K_1(x,t,u(t))dt + \beta\int_0^x K_2(x,t,u(t))dt, \] where \(x,t\in[0,1]\), \(u\in C([0,1],\mathbb{R})\), \(|\alpha|+|\beta|\neq0\), \(u(0)=u_0\), \(\lambda\in \mathbb{R}\backslash\{0\}\), \(f:[0,1]\rightarrow\mathbb{R}\), and \(K_1,K_2:[0,1]^2\times \mathbb{R}\rightarrow \mathbb{R}\). By integration on both sides of the above equation, one can define an operator \(T\) as follows: \[ [Tu](x):=\frac{1}{\lambda}\left[u_0+\int_0^x f(t) dt+\alpha \int_0^x\int_0^1 K_1(s,t,u(t))dt ds + \beta\int_0^x\int_0^s K_2(s,t,u(t))dtds\right]. \] By imposition assumptions on \(T\) and the Banach fixed point theorem, the solution of the integro-differential equation is given by the fixed point \(u\) of the operator \(T\), that is, \(u=\lim_{n\rightarrow\infty} T^n(u_0)\). More precisely, one can define a sequence \(u_i\) (\(i\geq 1\)) iteratively by: \[ \begin{multlined} u_i = Tu_{i-1} = \frac{1}{\lambda}\left[u_0+\int_0^x f(t) dt+\alpha \int_0^x\int_0^1 K_1(s,t,u_{i-1}(t))dt ds\right.\\\left. + \beta\int_0^x\int_0^s K_2(s,t,u_{i-1}(t))dtds\right], \end{multlined} \] and then \(u=\lim_{i\rightarrow \infty}u_i\). By using the rational Haar wavelets on the interval \([0,1]\), the paper then projects the function \(K_1,K_2\) onto the orthogonal Haar wavelet system \(\{h_0(t),\dots,h_{m-1}(t)\}\otimes\{h_0(s),\dots,h_{m-1}(s)\}\) at level \(i\) by \[ [Q_m(K_1)](t,s) :=\sum_{k=0}^{m-1}\sum_{\ell=0}^{m-1} K_{k\ell} h_k(t) h_\ell(s) \] and \[ [Q_m(K_2)](t,s) :=\sum_{k=0}^{m-1}\sum_{\ell=0}^{m-1} L_{k\ell} h_k(t) h_\ell(s). \] In such a way, the \(i\)th iteration solution \(u_i\) is for \(i\geq 1\) approximated by \[ u_i\approx \frac{1}{\lambda}\left[u_0+\int_0^x f(t) dt+\alpha \int_0^x\int_0^1 [Q_m(K_1)](t,s)dt ds + \beta\int_0^x\int_0^s [Q_m(K_2)](t,s)dtds\right]. \] Since \([Q_m(K_1)](t,s)\) and \([Q_m(K_2)](t,s)\) are then piece-wise constants, the integration can be computed efficiently. The paper then continues to show the error analysis between the exact solution \(u\) and the \(i\)th iteration approximation solution \(u_i\), and show that the \(i\)th iteration approximation solution converges to the exact solution under certain conditions. The authors finally demonstrate some numerical examples and comparisons to illustrate the effectiveness of their method.
0 references
rationalized Haar wavelet
0 references
operational matrix
0 references
fixed point theorem
0 references
error analysis
0 references
nonlinear mixed Volterra-Fredholm integro-differential equations of the second kind, numerical examples
0 references
0 references
0 references
0 references
0 references