 |
Center for Nonlinear Studies |
Dynamic optimization problems include differential equations as constraints and allow users to incorporate dynamic models directly within an optimization framework. However, implementing and solving this class of optimization problem is challenging, particularly for non-experts, due to the reformulations that must be applied in order to solve these problems using general algebraic optimization solvers. Pyomo.dae is a modeling package meant to address this challenge by allowing users to formulate dynamic optimization problems using a natural and intuitive syntax and providing general implementations of reformulations which can be used to automatically generate algebraic approximations of dynamic models that can be solved using off-the-shelf optimization solvers. This talk gives an overview of the dynamic optimization modeling flexibility provided by pyomo.dae and the included reformulations. In addition, we discuss new features being developed for the package and demonstrate pyomo.dae on a variety of examples. Finally, we show how pyomo.dae can be combined with other Pyomo modeling extensions, such as PySP for stochastic programming, in order to quickly and concisely implement and solve sophisticated, cutting-edge optimization problems incorporating dynamics.
Host: Carleton Coffrin