Lipschitz Based Lower Bound Construction for Surrogate Optimization

Bounds play a vital role in guiding optimization algorithms by enhancing convergence, improving solution quality, and quantifying optimality gaps. While Lipschitz-based lower bounds are well-established, their effectiveness is often constrained by the function’s topological properties. To address these limitations, we propose an approach that integrates nonlinear distance metrics with surrogate approximations, yielding more adaptive and … Read more

Column Generation based Alternating Direction Methods for solving MINLPs

Traditional decomposition based branch-and-bound algorithms, like branch-and-price, can be very efficient if the duality gap is not too large. However, if this is not the case, the branch-and-bound tree may grow rapidly, preventing the method to find a good solution. In this paper, we present a new decompositon algorithm, called ADGO (Alternating Direction Global Optimization … Read more