Fast Rank-Revealing Method for Solving Large Global Optimization Problems and Its Applications

We develop a rapid rank-revealing technique to address a wide range of global optimization problems with a huge search space. This method is versatile, serving effectively for both nonlinear, nonconvex systems and linear, and convex ones. Its performance is shown to surpass prevailing optimization methods like genetic algorithms (GAs), stochastic gradient descent algorithms, and Bayesian optimization (BO) in terms of both execution speed and the ability to identify global optima. Moreover, this method’s usefulness extends beyond theoretical benchmark functions, as it has been successfully applied to tackle real-world optimization challenges, specifically in the domains of electromagnetic design and integrated system design optimization.