Edward P. Gatzke

Software

Matlab - CPLEX Interfaces - Allows for direct linking of CPLEX callable libraries for use in Matlab. Currently available interfaces include: Linear Programming (LP), Mixed-Integer Linear Programming (MILP), Quadratic Programming (QP), and Mixed-Integer Quadratic Programming (MIQP), as well as a home-grown MIQP solver which uses the CPLEX QP solver.

GLOBO - Global optimization for NLP and MINLP problems using determinstic branch-and-bound methods.  Uses linear or piecewise-linear outer approximation for relaxation of the nonconvex constraints.  Can use a variety of local NLP methods, LP methods, and MILP methods for solving subproblems.

Mixed-Integer Model Predictive Control with State Estimation - Given a linear state space model of a system, this MPC algorithm used propositional logic and formulates a MILP or MIQP problem to be solved online. This allows for prioritized discrete constraints on process inputs, outputs, or states. Requires Matlab and a MILP or MIQP solver.

Mixed-Integer Formulation for Hybrid Model Identification of Dynamic Systems - Given SISO dynamic data and a switching criteria, this allows for determination of the optimal switching point and best model parameters for an affine dynamic system. Currently supports diagonal Volterra or linear moving average models.