We begin with a traditional model predictive control problem using the l1 norm in the objective function, and then allow the model parameters to be stochastic, with discrete distributions and finite support. We apply the ...