viella4585/kalman.jl

Generic interface for Kalman filters in Julia

Kalman

A generic interface for Kalman filters in Julia.

Note that the TimeModels.jl also has an implementation of Kalman filters for time series analysis. This implementation (Kalman.jl) focuses on applications of Kalman filters to online data assimilation problems, and intends to develop a single API for both linear and nonlinear Kalman filters. Kalman.jl is an unregistered package still in active development, so bugs may be common and changes rapid. For a more stable and tested API, check out TimeModels.jl.

Installation

	Pkg.clone("git://github.com/wkearn/Kalman.jl")

Until this package is registered, simply Pkg.clone from this repository

Usage

using Kalman

x0 = State([0.0],[1.0]')
f = LinearModel([1]',[1]',[1e-10]')
z = LinearObservationModel([1]',[0.01]')

kf0 = BasicKalmanFilter(x0,f,z)
kf = kf0

y = map(i->Observation([i]),-0.37727+sqrt(0.01)*randn(50))

Note that in Julia v0.4 and greater, you can replace the Observation{T} with a Nullable{Vector{T}} which allows you to represent missing measurements.

x = zeros(Float64,50)
p = ones(Float64,50)

for i = 1:49
	kf = predictupdate(kf,y[i])
	x[i+1] = kf.x.x[1]
	p[i+1] = kf.x.p[1]
end

)

Roadmap

Filters which have been implemented list the Kalman.jl type which implements them.

Kalman Filter

• Linear, time-invariant filter with regular updating: BasicKalmanFilter
  • Missing measurements supported with Nullable{Vector{T}} observations

Extended Kalman Filter

• Nonlinear, time-invariant filter with regular updating: BasicExtendedKalmanFilter

Unscented Kalman Filter

Ensemble Kalman Filter