Optimal chaos control through reinforcement learning

A general purpose chaos control algorithm based on reinforcement learning i s introduced and applied to the stabilization of unstable periodic orbits i n various chaotic systems and to the targeting problem. The algorithm does not require any information about the dynamical system nor about the locati on of periodic orbits. Numerical tests demonstrate good and fast performanc e under noisy and nonstationary conditions. (C) 1999 American Institute of Physics. [S1054-1500(99)00703-X].