R code for implementing an Ensemble Kalman Filter

This blog post will demonstrate how to implement an Ensemble Kalman Filter (EnKF) in R.
Similar to filters such as the Extended Kalman Filter (EKF) and the Unscented Kalman Filter (UKF), the EnKF may be used for solving nonlinear filtering problems. However, the EnKF may be applied when solving nonlinear (and linear) filtering problems with a large number of states.

Ensemble Kalman Filter Continue reading R code for implementing an Ensemble Kalman Filter

R code for implementing a Gaussian sum filter

The code below implements a Gaussian sum filter (GSF) in R. The implemented GSF employs a bank of Gauss-Hermite Kalman filters (GHKF).

A GSF may be used when the process noise, measurement noise, or a posteriori state distribution is expected to follow some non-Gaussian distribution. The GSF approximates such non-Gaussian distributions with a mixture (or sum) of Gaussian distributions.

If necessary, the implemented Gaussian sum filter performs pruning as a Gaussian mixture reduction technique.

Gaussian sum filter

Continue reading R code for implementing a Gaussian sum filter

R code for forecasting with the bootstrap filter

In one of my previous blog posts I demonstrated how to implement a particle  (bootstrap) filter in R.
In this post I will demonstrate how to predict future system states and observations with the particle/bootstrap filter.

Bootstrap filter forecasting

Continue reading R code for forecasting with the bootstrap filter