Application of reproducing kernel algorithm for solving second-order, two-point fuzzy boundary value problems (Q1703687)

The paper is concerned with the analytic and numerical solution of the two-point fuzzy boundary value problem \[ y''(x) + f \left(x, y(x), y'(x) \right)= g \left(x, y(x), y'(x) \right) \text{ on } (a,b) \] subject to boundary conditions \[ y(a)=\alpha, \quad y(b)=\beta, \] where \(a\) and \(b\) are real numbers, \(\alpha\) and \(\beta\) are fuzzy numbers, and \(f,g,y\) are fuzzy-valued functions. First, the concept of fuzzy analysis is briefly reviewed. Then the problem is reformulated using \(r\)-cut representations of functions under the assumption of strongly generalized differentiability and appropriate Hilbert spaces with reproducing kernels are introduced. An orthonormal system of functions is constructed and it is used for the representation of the analytic solution as an infinite sum of functions and an approximate solution as a finite sum of functions. These solutions can be computed directly in linear cases. For nonlinear equations the authors propose an iterative method and prove its convergence. Numerical experiments and a numerical comparison with the results computed by the homotopy analysis method and the Adomian decomposition method are presented.