Package KFAS provides funchtions for Kalman filtering, state,
disturbance and simulation smoothing, forecasting and simulation
of state space models. All functions can use exact diffuse
initialisation when distributions of some or all elements of initial
state vector are unknown. Filtering, state smoothing and simulation
functions use sequential processing algorithm, which is faster than
standard approach, and it also allows singularity of prediction
error variance matrix. KFAS also contains function for computing
the likelihood of exponential family state space models and function
for state smoothing of exponential family state space models.