On Similiarities Between Two Global Optimization Algorithms Based on Different (Bayesian and Lipschitzian) Approaches
摘要
We consider two global optimization algorithms based on different, frequently considered as opposite, approaches. One of them is the actually earliest proposed global optimization algorithm based on the Baesian approach. It is the so-called P-algorithm where a Wiener process is used as a model of the objective functions. The second considered algorithm is the Pijavskij–Shubert algorithm, which is one of the earliest in Lipschitz optimization. We show that both algorithms can be interpreted in terms of each other with respect to the apropriately defined models of the objective functions.