Hamiltonian Cycles and Nonconvex Optimisation
(V Ejov and J Filar with J Gondzio)
Hamiltonian Cycles are characterised as global optima of certain (highly structured) nonconvex optimisation problems.
These characterisations derive from an embedding of the Hamiltonian cycle problem in a Markov decision process. Heuristics based on interior point and branch & bound method are investigated.
This project is funded by an ARC Discovery Grant.