论文部分内容阅读
This talk concerns function-space MCMC methods for sampling the posterior distribution arising from Bayesian inverse problems.A general class of operator-weighted proposal distributions are introduced,which are dimension(and covariance)-independent,and may include local gradient information.