Trigonometric Interpolation Based Approach for Second Order Fredholm Integro-Differential Equations
Xiaorong \surname Zou · 2025
Read Trigonometric Interpolation Based Approach for Second Order Fredholm Integro-Differential Equations by Xiaorong \surname Zou (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2508-09413.txt.
Introduction
Xiaorong Zou extends a trigonometric interpolation method so it can solve a class of problems that mix derivatives and integrals in two dimensions. Trigonometric interpolation was of late adapted to non-periodic functions and to ordinary differential equations. Integro-differential problems of Fredholm type are harder, with kernels that may even be singular.
The improved algorithm approximates functions on a two-dimensional grid. Its accuracy stays tied to how smooth the target is, as tests with numbers confirm. Put to work on second-order Fredholm problems, it converges fast with just a moderate grid. It handles singular kernels and general boundary conditions. The design extends to Volterra problems and to ones of higher order. Tests across types of boundary condition all did quite well.
Catalog record
| CREATOR | Xiaorong \surname Zou |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 35 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2508-09413
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Trigonometric Interpolation Based Approach for Second Order Fredholm Integro-Differential Equations Xiaorong \surname Zou Abstract A trigonometric interpolation algorithm for non-periodic functions has been recently proposed and applied to study general ordinary differential equation (ODE). This paper enhances the algorithm to approximate functions in 2 2 dim space. Performance of the enhanced algorithm is expected to be similar as in 1 1 -dim case and achieve accuracy aligned with smoothness of the target function, which is confirmed by numerical examples. As an application, the 2 2 -dim trigonometric interpolation method is used to develop an algorithm for the solution of a second order Fredholm integro-differential equation (FIDE). There are several advantages of the algorithm. First of all, it converges quickly and high accuracy can be achieved with moderate size of grid points; Secondly, it can effectively address singularities of kernel functions and work well with general boundary conditions. Finally, it can be enhanced to copy with other IDE such as Volterra IDE or IDE with high order ODE component. The tests conducted in this paper include various boundary conditions with both continuous kernels and integrable ones with singularity. Decent performance is observed across all covered scenarios with moderate size of grid points. Classification: MSC2000 keywords Trigonometric Interpolation, Integro-differential Equation, Fast Fourier Transform {opening} Primary 65T40; Secondary 45B05 1 Introduction A new trigonometric interpolation algorithm was recently introduced in [ 12 ] . It leverages Fast Fourier Transformation (FFT) to achieve optimal computational efficiency and converges at speed aligned with smoothness of underlying function. In addition, it can be used to approximate non-periodic functions defined on bounded intervals. Considering the analytic attractiveness of trigonometric polynomial, especially in handling differential and integral operations, the proposed trigonometric estimation of a general function is expected to be used in a wide spectrum. In this paper, we continue on applications of the trigonometric interpolation algorithm to solve integro-differential equations (IDEs) that contains an integral and derivative of unknown target function. IDEs appear in mathematical physics, applied mathematics and engineering ( [ 1 ] - [ 4 ] ). In general, analytical solutions of IDEs are not feasible and certain approximations are required. The concepts of IDEs have motivated a large size of research work for numerical solutions in recent years. Several methods have been established and among them are the Haar Wavelet method [ 7 ] , the Tau method [ 6 ] , the Taylor polynomial ( [ 5 ] ), the Spline collocation method [ 8 ] and Chebyshev polynomial method ( [ 9 ] , [ 10 ] , [ 11 ] ). Fredholm IDE (FIDE) and Volterra IDE (VIDE) are two most popular types of IDEs. The difference is that integration region of FIDE is fixed while VIDE bears a variable integration region. We shall address VIDE separately in [ 15 ] and aim to solve the following FIDE in this paper. y^{\prime\prime}(x)=p(x)y^{\prime}+q(x)y(x)+r(x)+\mu(x)\int^{e}_{s}k(x,t)y(t)dt,\quad x\in[s,e]. (1) \displaystyle d_{11}y(s)+d_{12}y^{\prime}(s)+d_{13}y(e)+d_{14}y^{\prime}(e) \displaystyle= \displaystyle\alpha, (2) \displaystyle d_{21}y(s)+d_{22}y^{\prime}(s)+d_{23}y(e)+d_{24}y^{\prime}(e) \displaystyle= \displaystyle\beta, (3) where p,q,r,\mu is continuous differential on [s,e] , the kernel function k(x,y) is integrable with respect to y on [s,t] for any x\in[s,e] , the rank of matrix D:=(d_{ij})_{1\leq i\leq 2,1\leq j\leq 4} is 2 , and \alpha,\beta are two real numbers. Recently, trigonometric interpolation has been used to solve second order ODE with mixed boundary conditions [ 13 ] and [ 14 ] . The proposed methods leverages the attractiveness of trigonometric polynomial for the connection among target function y and its derivatives y^{\prime},y^{\prime\prime} . For non-linear ODE, The coefficients of trigonometric polynomial can be identified effectively through optimization process whose gradient vectors can be carried out by fast Fourier transformation [ 13 ] . For linear ODE, ODE can be converted to linear algebraic system and the coefficients can be solved directly. The method is flexible to handle general mixed boundary conditions and can be extended to high-order ODE [ 14 ] . We shall follow the same idea to solve FIDE ( 1 - 3 ). In order to address the integral component in Eq. ( 1 ), we first enhance interpolation algorithm developed in [ 12 ] for 2 -variable functions so a kernel k(x,y) can be approximated by a 2 -dim trigonometrical polynomial. The algorithm developed in [ 14 ] can then be enhanced to find a trigonometric approximation of FIDE ( 1 - 3 ). There are some advantage of proposed trigonometric interpolation method. It produces accurate approximation as shown by numerical examples conducted in Section 4 ; it can effectively address singularity issue in kernel function as described in Subsection 2 ; it works for various boundary conditions. Finally, it can be enhanced to work for other IDE such as Volterra IDE or IDEs with high order ODE component. The rest of paper is organized as follows. In Section 2 , we start with a brief review on trigonometric interpolation algorithm developed in [ 12 ] ; then enhance it for two variable functions so it can be used to approximate kernel function k(x,t) in Eq. ( 1 ); the section is closed by numerical testing results on some kernels to demonstrate performance of the enhanced algorithm. Section 3 is denoted to develop Algorithm 3.1 to convert FIDE ( 1 - 3 ) into a linear algebraic system. It begins with reviewing related results in [ 14 ] to attack the ODE component in FIDE. Integral component is addressed in Subsection 3.1 and 3.2 for continuous and integrable kernels respectively. The proposed algorithm is summarized in Subsection 3.3 . Section 4 demonstrates quite a few numerical examples to assess the performance of Algorithm 3.1 . Tests covers four types of boundary conditions with both continuous kernels and integrable kernels with singularity. The results show decent accuracy for all covered scenarios with moderate grid point size. The summary is made on Section 5 . 2 Trigonometric Interpolation on Non-Periodic Functions 2.1 Trigonometric Interpolation with one-variable Functions In this subsection, we review relevant results of trigonometric interpolation algorithm developed in [ 12 ] starting with following interpolation algorithm on periodic functions. Algorithm 2.1 . Let f(x) be an odd periodic function 1 1 1 Similar results for even periodic function is also available in [ 12 ] . with period 2b and N=2M=2^{q+1} for some integer q\geq 1 and x_{j},y_{j} are defined by \displaystyle x_{j} \displaystyle:= \displaystyle-b+j\lambda,\quad\lambda=\frac{2b}{N},\quad 0\leq j<N, (4) \displaystyle y_{j} \displaystyle:= \displaystyle f(x_{j}). (5) Then there is a unique M-1 degree trigonometric polynomial \displaystyle f_{M}(x) \displaystyle= \displaystyle\sum_{0<j<M}a_{j}\sin\frac{j\pi x}{b}, \displaystyle a_{j} \displaystyle= \displaystyle\frac{2}{N}\sum_{0<k<N}(-1)^{j}y_{k}\sin\frac{2\pi jk}{N},\quad 0<j<M such that it fits to all grid points, i.e. f_{M}(x_{k})=y_{k},\quad 0\leq k<N. One can computer coefficients by Inverse Fast Fourier Transform (ifft): \{a_{j}(-1)^{j}\}_{0}^{N-1}=2\times Imag(ifft(\{y_{k}\}_{k=0}^{N-1})). Algorithm 2.1 has been enhanced so it can be applied to a nonperiodic function f whose K+1 -th derivative f^{(K+1)}(x) exists over a bounded interval [s,e] . To seek for a periodic extension with same smoothness, we assume that f can be extended smoothly such that f^{(K+1)} exists and is bounded over [s-\delta,e+\delta] for certain \delta>0 . Such a periodic extension of f can be achieved by a cut-off smooth function h(x) with following property: \displaystyle h(x)=\left\{\begin{array}[]{cc}1&x\in[s,e],\\ 0&\mbox{$x<s-\delta$ or $x>e+\delta$}.\\ \end{array}\right. A cut-off function with closed-form analytic expression is proposed in [