A hybrid kernel functions collocation approach for boundary value problems with Caputo fractional derivative (Q6167222)

Using ML RKF (Mittag-Leffler Reproducing Kernel Functions) and Sobolev EKF, a new hybrid collocation approach is presented for fractional Caputo fractional boundary value problems. One of the main features is that the method can improve the stability of the ML kernel function based method, and the accuracy of the Sobolev kernel based technique for Caputo fractional problems. The approach can give highly accurate numerical solutions with a small number of nodes. The present work improves [\textit{J. A. Rosenfeld} and \textit{W. E. Dixon}, SIAM J. Numer. Anal. 55, No. 3, 1201--1217 (2017; Zbl 1375.26018)] in which Mittag-Leffler RKHS (Reproducing Kernel Hilbert Spaces) were introduced and kernelized ABM was presented to solve initial value problems with Caputo fractional derivative but the stability of the method is unsatisfactory.