Convexity of nonlinear image of a small ball with applications to optimization (Q5945282)

The basic result of the paper states that under specified conditions an image created by a nonlinear map of a small ball in the Hilbert space is a convex set. The idea is simple: the ball is strongly convex, thus its image under a linear map is strongly convex as well. But it cannot loose convexity for a nonlinear map which is close enough to its linearization. As the author notes, the result also holds if the ball is replaced by any other strongly convex set (e.g. by a nondegenerate ellipsoid). This result has numerous applications in optimization and control. The author proves the duality theory for nonconvex mathematical programming problems with extra constraint \(\|x-a\|\leq\varepsilon\). A special algorithm for such problems is described. Various applications to control problems based on the convexity of the reachable set for nonlinear systems with ``small power control'' are outlined in the last section of the paper.