Restart of accelerated first-order methods and the implementation of MPC in embedded systems

In this seminar we present results in two different but closely related topics. The first part is centered on restart schemes, which are schemes that are applied to accelerated first-order methods to improve their convergence. We show the theoretical guarantees of the schemes and numerical results highlighting the advantages that can be obtained with them.The second part is centered on the implementation of model predictive control (MPC) in embedded systems. We present efficient solvers for various MPC formulations and introduce an MPC formulation, called harmonic model predictive control, which is particularly suitable for its use with small prediction horizons.

Join at http://imt.lu/seminar