ANTIGONE (Algorithms for coNTinuous / Integer Global Optimisation of Nonlinear Equations), is a computational framework for the deterministic global optimisation of mixed-integer nonlinear programs (nonconvex MINLP). The approach: reformulates user input; detects special mathematical structure; globally optimises the transformed problem. ANTIGONE is an evolution of the Global Mixed-Integer Quadratic Optimizer (GloMIQO) from quadratic to general nonconvex terms. This presentation highlights the ANTIGONE cutting plane strategies, which expand the GloMIQO branch-and-cut framework from aggregated quadratic terms to aggregated general nonlinear terms. We discuss the hierarchy of cutting planes integrated into ANTIGONE including: convex/concave outer approximation; high-dimensional edge-concave/edge-convex cuts; alphaBB relaxations; the reformulation-linearization technique. Data is presented for globally optimising a range of MINLP test cases using ANTIGONE; these examples include problems from standard libraries and more recent examples from the open literature. |