New Computer Experiment Designs Using Continuum Random Cluster Point Process

Abstract

In this paper, we propose a new approach for building computer experiment designs using the continuum random cluster point process, also referred to as the connected component Markov point process. Our method involves generating designs through the Markov Chain Monte Carlo method (MCMC) and the Random Walk Metropolis Hastings algorithm (RWMH algorithm), which can be easily scaled to meet various objectives. We have conducted a comprehensive study on the convergence of the Markov chain and compared our approach with existing computer experiment designs. Overall, our approach offers a novel and flexible solution for constructing computer experiment designs.