Interpolation in a derivative strip (Q1362835): Difference between revisions

Let \(s(x)\) be a smooth function that interpolates a data set \((x_i, y_i)\). Interpolation in a derevative strip means that the interpolant \(s(x)\) should fulfil the restriction \[ L_i\leq s'(x) \leq U_i, \qquad x\in\langle x_{i-1},x_i\rangle \] provided the bounds \(L_i<U_i\) are strictly compatible with slopes \[ \tau_i= (y_i-y_{i-1})/h_i:\;L_i<\tau_{i-1}, \;\tau_i<U_i. \] The aim of this paper is to prove that restricted interpolation, under compatibility conditions, is always possible. Direct algorithms for constructing the desired strip interpolants are described. The main idea is to use a placement procedure for the additional knots. Then the splines resulting from the existence proof can be improved by means of optimization. Results of numerical tests are also given.