2018/11/09 by Soumaya Azzi, Azzi, Soumaya, Yuanyuan Huang +5 · 2 citations
Computer Science · Decision Sciences · Mathematics · #Advanced Multi-Objective Optimization Algorithms #FOS: Computer and information sciences #Other Statistics (stat.OT) #Simulation Techniques and Applications #Statistical Distribution Estimation and Applications
paper · pdf · doi:10.48550/arxiv.1811.04079
openalex publication_date 2018/11/09 · openalex created_date 2025/10/10 · openalex updated_date 2026/07/28
Metamodeling of complex numerical systems has recently attracted the interest\nof the mathematical programming community. Despite the progress in high\nperformance computing, simulations remain costly, as a matter of fact, the\nassessment of the exposure to radio frequency electromagnetic fields is\ncomputationally prohibitive since one simulation can require hours. Moreover,\nin many engineering problems, carrying out deterministic numerical operations\nwithout considering uncertainties can lead to unreliable designs. In this paper\nwe focus on the surrogate modeling of a particular type of computational models\ncalled stochastic simulators. In contrast to deterministic simulators which\nyield a unique output for each set of input parameters, stochastic simulators\ninherently contain some sources of randomness and the output at a given point\nis a probability density function. Characterizing the stochastic simulators is\neven more time consuming. This paper represents stochastic simulators as a\nstochastic process and describes a metamodeling approach based on the\nKarhunen-Lo `eve spectral decomposition.\n