MATLAB code for the numerical case studies in
T. J. Meijer and A. Rantzer, "Dual Control: On Exploration–Exploitation in Linear Systems," Annual Review of Control, Robotics, and Autonomous Systems, vol. 10, 2027.
See the paper for details.
Requires Manopt (http://manopt.org/).
Run Main.m. It simulates the three controllers and plots the results (Figure 4 of the paper). The model, noise type and parameters are set at the top of the file.
Main.m: simulation and plotting (entry point)wRLS_CE_LQR.m: weighted recursive least squares CE-LQRCEC.m: regret rate minimizing certainty-equivalence controllerMinimax.m: minimax optimal dual controllerdependencies/: helper functions
@article{meijer2027dual,
title = {Dual Control: On Exploration--Exploitation in Linear Systems},
author = {T.J.~{Meijer} and A.~{Rantzer}},
journal = {Annual Review of Control, Robotics, and Autonomous Systems},
volume = {10},
pages = {1--26},
year = {2027}
}Tomas J. Meijer (tomas.meijer@control.lth.se) and Anders Rantzer, Department of Automatic Control, Lund University. Released under the MIT License.