A package for using feedback particle filters in nonlinear stochastic filtering and data assimilation problems.

What are feedback particle filters?

Feedback particle filters (FPFs) are a class of sample-based numerical algorithms to approximate the conditional distribution in a nonlinear filtering problem. In contrast to conventional particle filters, which use importance weights, FPFs use feedback control to let the observations guide the particles to the appropriate position.

Further background reading:

Package features

This package's aim is to provide a versatile and efficient feedback particle filter implementation in Julia, with abstractions to flexibly construct, run, and analyze feedback particle filters for a variety of uni- and multivariate filtering problems with both diffusion and point process observations.

In particular, the following features are planned to be implemented in FeedbackParticleFilters: