Bayesian post-processor and other enhancements of Subset Simulation for estimating failure probabilities in high dimensions

• Published in: Computers & structures Vol. 92; pp. 283 - 296
• Main Authors: Zuev, Konstantin M., Beck, James L., Au, Siu-Kui, Katafygiotis, Lambros S.
• Format: Journal Article
• Language: English
• Published: Kidlington Elsevier Ltd 01.02.2012; Elsevier
• Subjects: Algorithms; Applied sciences; Bayesian approach; Computer simulation; Estimating; Exact sciences and technology; Failure; Fluctuation phenomena, random processes, noise, and brownian motion; Markov chain Monte Carlo; Mathematical models; Mathematics; Monte Carlo methods; Operational research and scientific management; Operational research. Management science; Optimization; Physics; Polymethyl methacrylates; Probability and statistics; Probability theory and stochastic processes; Reliability engineering; Reliability theory. Replacement problems; Risk theory. Actuarial science; Sciences and techniques of general use; Special processes (renewal theory, markov renewal processes, semi-markov processes, statistical mechanics type models, applications); Statistical physics, thermodynamics, and nonlinear dynamical systems; Stochastic simulation methods; Subset Simulation; Reliability engineering; Stochastic simulation methods; Subset Simulation; Bayesian approach; Markov chain Monte Carlo; Ergodic theory; Stochastic model; Conditional distribution; Conditional probability; Stochastic method; Modeling; Parameter space; Markov chain; System with n degrees of freedom; Optimal strategy; Uncertain system; Elastoplasticity; Sampling; Bayes estimation; Metropolis algorithm; Prior information; Monte Carlo method; Imprecise probability; Rupture; Ergodicity; Risk analysis; System reliability

► The Modified Metropolis algorithm (MMA) for sampling from complex PDFs is analyzed. ► An optimal scaling strategy for MMA is presented. ► A theoretical basis for the optimal value of the conditional probability in Subset Simulation is provided. ► A Bayesian post-processor for the original Subset S...