Optimized Population Monte Carlo

Adaptive importance sampling (AIS) methods are increasingly used for the approximation of distributions and related intractable integrals in context Bayesian inference. Population Monte Carlo (PMC) algorithms a subclass AIS methods, widely due to their ease adaptation. In this paper, we propose novel algorithm that exploits benefits PMC framework includes more efficient adaptive mechanisms, exploiting geometric information target distribution. particular, adapts location scale parameters set densities (proposals). At each iteration, adapted by combining versatile resampling strategy (i.e., using previous weighted samples) with an advanced optimization-based scheme. Local second-order distribution is incorporated through preconditioning matrix acting as scaling metric onto gradient direction. A damped Newton approach adopted ensure robustness The resulting also update proposals. We discuss several key theoretical foundations proposed approach. Finally, show successful performance method three numerical examples, involving challenging distributions.