RcppSMC - support for modern Monte Carlo methods

Sequential Monte Carlo (SMC) methods are a general class of Monte Carlo procedures for sampling from sequences of probability distributions. Basic examples of these algorithms, termed particle filters, are frequently used in a variety of fields including signal processing, economics, and systems biology.

Recent methodological developments, one example being Particle MCMC and Particle Gibbs methods, have gained considerable popularity. However, they require algorithmic extensions and are particularly computationally expensive. Hence, software support for these procedures is currently in high demand.

This project aims at providing straightforward implementation of modern SMC/Particle MCMC methods by extending the C++ template class library of RcppSMC which is currently not supporting these methods.

Benefits for the R community are expected, firstly, as SMC practitioners who already use R can reduce execution time as well as time necessary for implementation in their everyday work. SMC developers, currently not using R as their first programming language of choice, hopefully see the benefits of this package and thus might consider using R more often in this domain in the future.