LibBi is used for state-space modelling and Bayesian inference on high-performance computer hardware, including multi-core CPUs, many-core GPUs (graphics processing units) and distributed-memory clusters.
The staple methods of LibBi are based on sequential Monte Carlo (SMC), also known as particle filtering. These methods include particle Markov chain Monte Carlo (PMCMC) and SMC\(^2\). Other methods include the extended Kalman filter and some parameter optimisation routines.
LibBi consists of a C++ template library, as well as a parser and compiler, written in Perl, for its own modelling language.