Bivariate $C^2$ cubic spline quasi-interpolants on uniform Powell-Sabin triangulations of a rectangular domain

In this paper we construct discrete quasi-interpolants based on C^2 cubic multi-box splines on uniform Powell–Sabin triangulations of a rectangular domain. The main problem consists in finding the coefficient functionals associated with boundary multi-box splines (i.e. multi-box splines whose supports overlap with the domain) involving data points inside or on the boundary of the domain and giving the optimal approximation order. They are obtained either by minimizing an upper bound for the infinity norm of the operator w.r.t. a finite number of free parameters, or by inducing the superconvergence of the gradient of the quasi-interpolant at some specific points of the domain. Finally, we give norm and error estimates and we provide some numerical examples illustrating the approximation properties of the proposed operators.