Adaptive optics based on analog parallel stochastic optimization technique

Summary form only given. In this paper we consider both theoretical and experimental aspects of adaptive wavefront phase distortion compensation using direct system performance metric optimization. A brief review of model free optimization techniques in adaptive optics, with an emphasis on gradient descent methods based on stochastic approximation of the true gradient is presented. The method is extended to include different requirements for wavefront control, as well as information from different wavefront sensor types. It is shown that both can be incorporated into a more general gradient descent optimization paradigm, giving rise to a variety of new control system architectures.