New Advances in Network Constrained Non-convex Optimization
Network constrained nonconvex optimization (NCNO) represents a broad class of problems, ranging from optimization over a physical network, such as an electric power grid or a communication network, to a general constrained optimization with an incidence graph or a factor graph, to dynamic programming on scenario trees. From a unifying perspective, we present a set of new algorithmic tools and schemes to advance the theoretical understanding as well as practical computation of a broad class of NCNO problems. In particular, we present 1) a first globally convergent distributed algorithm for general NCNOs, 2) a new scheme for reformulating rank constraints using matrix minors and a second-order cone programming based fast and strong convex relaxation algorithm for quadratic constrained NCNOs, and 3) if time allows, an effective cutting-plane algorithm for stochastic mixed-integer DP with huge scenario trees supported by a new strong duality result for mixed integer programming. We will present several major applications of these new algorithms in fundamental problems in energy systems, such as AC OPF, stochastic Unit Commitment, and generation expansion, with promising performance improvements.
Andy Sun is an associate professor and the Anderson-Interface Early Career Professor in the H. Milton Stewart School of Industrial and Systems Engineering, Georgia Institute of Technology. He has a broad research interest in optimization and engineering applications, especially nonconvex optimization with network structures, and mathematical modeling of electric energy systems. Dr. Sun’s research has won several awards, including the NSF CAREER award, the HICSS best paper award in 2018, the INFORMS Junior Faculty Interests Group (JFIG) Paper Competition, the INFORMS ENRE Best Publication in Energy, and the INFORMS Georgia B. Dantzig Doctoral Dissertation Award. Dr. Sun has been working with several leading utility companies such as the California ISO, MISO, ISO New England, and the Southern Company on robust power system operations. Dr. Sun received his doctoral degree in Operations Research from Massachusetts Institute of Technology. He was a postdoctoral researcher at IBM Thomas J. Watson Research Center.
Thursday, September 19 at 2:30pm to 3:30pm
John D. Tickle Engineering Building, 410
851 Neyland Dr, Knoxville, TN 37996