An online learning algorithm that adaptively designs a decentralized linear quadratic regulator when the system model is unknown, achieving a regret that scales sublinearly with the time horizon.
