A new higher order accurate reproducing kernel-based approach for boundary value problems (Q2186777)

Traditionally, many methods, including finite difference techniques, shooting methods and spectral methods have been employed for solving boundary value problems (BVPs), numerically. Recently, on the basis of the theory of reproducing kernal Hilbert spaces (RKHSs), the method of reproducing kernel functions (RKF) has been used for solving systems of BVPs. These methods employ the Gram-Schmidt orthogonalization. The existing RKF-based numerical techniques for BVP only possess globally quadratic convergence and three order convergence in RKHS \(W^3\) and \(W^4\), respectively. In the present work, the authors propose a new higher order accurate RKF-based approach. Its global convergence can reach orders of \(4\) and \(6\), respectively in the RKHS mentioned earlier. A discussion on the error analysis is presented. The effectiveness of the proposed methods is demonstrated by numerical experiments.