Recursive algorithms for inner ellipsoidal approximation of convex polytopes. (Q1413940)

The problem of constructing inner ellipsoidal approximations of a convex polytope (described by linear inequalities) is studied. Two recursive algorithms are proposed. The first one solves the feasibility problem, while the second one deals with the optimization problem of finding the ``largest'' inscribed ellipsoid. The main idea of the proposed algorithms is to process in an iterative way the constraints describing the polytope. This allows the authors to overcome some computational and memory storage problems arising for the case of a large number of constraints.