Recursive grid methods to compute value sets for transfer functions with parametric uncertainty

The most straightforward way to compute value sets for uncertain transfer functions, where each uncertain parameter belongs to an interval, is the grid method. Each parameter interval is gridded equidistantly, and the transfer function values are computed in the grid points. Here, recursive extensions to the grid method are proposed whereby the gridding is adapted locally such that a prescribed distance is achieved between neighbouring points in the value set. Also presented is the so called pruning algorithm which finds the outer border of a value set defined by a set of points whose nearest neighbour lies within a prescribed distance. The pruning algorithm is part of the recursive grid methods, but can also be used independently with other methods to compute value sets.