A Fast Rank-Revealing Method for Solving High-Dimensional Global Optimization Problems

We develop a fast rank-revealing method for solving generic high-dimensional global optimization problems, suitable for nonlinear, non-convex systems in addition to linear and convex ones. The method is shown to significantly outperform state-of-the-art optimization techniques such as the Machine Learning (ML)-based Bayesian Optimization (BO) in run time as well as in the capability of finding a global optimum. It has been successfully applied to solve real-world floorplanning and placement problems in heterogeneously integrated system design.