Numerical approximation of Volterra integral equations with highly oscillatory kernels

Abstract

The Volterra integral equations (VIEs) with oscillatory kernels arise in several applied problems and need to be treated with a computational method have multiple characteristics. In the literature (Zaheer-ud-Din et al., 2022; Li et al., 2012), the Levin method combined with multiquadric radial basis functions (MQ-RBFs) and Chebyshev polynomials are well-known techniques for treating oscillatory integrals and integral equations with oscillatory kernels. The numerical experiments show that the Levin method with MQ-RBFs and Chebyshev polynomials produces dense and ill-conditioned matrices, specifically in the case of large data and high frequency. Therefore, the main task in this study is to combine the Levin method with compactly supported radial basis functions (CS-RBFs), which produce sparse and well-conditioned matrices, and subsequently obtain a stable, efficient, and accurate algorithm to treat VIEs. The theoretical error bounds of the method are derived and verified numerically. Although the error bounds obtained are not improved significantly, alternatively, a stable and efficient algorithm is obtained. Several numerical experiments are performed to validate the capabilities of the proposed method and compare it with counterpart methods (Zaheer-ud-Din et al., 2022; Li et al., 2012).