Solving system of second-order BVPs using a new algorithm based on reproducing kernel Hilbert space (Q2301382)

The authors study the following system of second-order boundary value problems (BVP): \[ \left\{ \begin{array}{ll} \mathcal{L}_1(u_1(t))=\mathcal{N}_1(t,U(t))+\mathcal{F}_1(t), \\[4pt] \qquad \quad \qquad\vdots \\[4pt] \mathcal{L}_k(u_k(t))=\mathcal{N}_k(t,U(t))+\mathcal{F}_k(t). \end{array}\right. \] subject to the following boundary conditions: \[ u_i(a)=u_i(b)=\gamma, \] where \(t \in \Omega = [a,b]\), \(\mathcal{L}_d\) and \(\mathcal{N}_d\) for \(d=1,\dots,k\) are linear and nonlinear differential operators respectively, \(\mathcal{F}_d(t)\) are given functions and \(U(t) = (u_1(t), u_2(t), \dots , u_k(t))^T\) is an unknown vector functions to be determined. Here, a new technique based on the Reproducing Kernel Method (RKM) has developed to solve the above system. The authors define the reproducing kernel spaces and their kernels. Also, some preliminaries and notations on the RKM without the Gram-Schmidt orthogonalization algorithm are provided. Since the RKM with Gram-Schmidt orthogonalization process is a very time-consuming method, here the orthogonalization process has been removed. The RKM without the Gram-Schmidt orthogonalization process significantly reduces the CPU time and increases the accuracy of the approximate solutions. The convergence theorem and error analysis are also given. An algorithm is presented to indicate the steps of implementing the proposed scheme. Some numerical examples are included to show the high accuracy of the present method.