Towards solving MINLPs within a Constraint Integer Programming Framework: Algorithms and Applications
by Ambros Gleixner and Stefan Vigerske
Abstract: We present recent extensions to the constraint integer programming framework SCIP for solving mixed-integer nonlinear programmes.
The talk focuses on linearisation and convexification techniques, reformulations, and primal heuristics for handling quadratic and univariate nonlinear constraints in an LP-based branch-and-cut algorithm.
The algorithms are motivated and illustrated by real-world applications from open pit mine scheduling, sheet metal design, and the operation of water supply networks.
Bio: Ambros is currently the main developer of SoPlex (a linear programming solver developed at the Zuse Institute Berlin) and has worked extensively with Lagrangian relaxations for mining applications. Stefan is an expert on MINLP; he developed LaGO (a COIN-OR MINLP solver) and is the chief developer of the nonlinear components in SCIP. Stefan is also involved in GAMS and is project manager of several COIN-OR projects
For More Information: contact: Christina Burt. email: firstname.lastname@example.org